compile and run!
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

1392 lines
29 KiB

/* $Id$
*
* isync - IMAP4 to maildir mailbox synchronizer
* Copyright (C) 2000-2002 Michael R. Elkins <me@mutt.org>
* Copyright (C) 2002-2003 Oswald Buddenhagen <ossi@users.sf.net>
24 years ago
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*
* As a special exception, isync may be linked with the OpenSSL library,
* despite that library's more restrictive license.
24 years ago
*/
#include "isync.h"
24 years ago
#include <assert.h>
#include <unistd.h>
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
#include <sys/mman.h>
#include <sys/time.h>
24 years ago
#include <stdlib.h>
#include <stdio.h>
#include <stdarg.h>
24 years ago
#include <errno.h>
#include <string.h>
#include <ctype.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <arpa/inet.h>
#include <netdb.h>
#if HAVE_LIBSSL
# include <openssl/err.h>
#endif
24 years ago
const char *Flags[] = {
"\\Seen",
"\\Answered",
"\\Deleted",
"\\Flagged",
"\\Recent",
"\\Draft"
};
void
free_message (message_t * msg)
{
message_t *tmp;
while (msg)
{
tmp = msg;
msg = msg->next;
if (tmp->file)
free (tmp->file);
free (tmp);
}
}
#if HAVE_LIBSSL
#define MAX_DEPTH 1
SSL_CTX *SSLContext = 0;
/* this gets called when a certificate is to be verified */
static int
verify_cert (SSL * ssl)
{
X509 *cert;
int err;
char buf[256];
int ret = -1;
BIO *bio;
cert = SSL_get_peer_certificate (ssl);
if (!cert)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "Error, no server certificate\n");
return -1;
}
err = SSL_get_verify_result (ssl);
if (err == X509_V_OK)
return 0;
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "Error, can't verify certificate: %s (%d)\n",
X509_verify_cert_error_string (err), err);
X509_NAME_oneline (X509_get_subject_name (cert), buf, sizeof (buf));
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info ("\nSubject: %s\n", buf);
X509_NAME_oneline (X509_get_issuer_name (cert), buf, sizeof (buf));
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info ("Issuer: %s\n", buf);
bio = BIO_new (BIO_s_mem ());
ASN1_TIME_print (bio, X509_get_notBefore (cert));
memset (buf, 0, sizeof (buf));
BIO_read (bio, buf, sizeof (buf) - 1);
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info ("Valid from: %s\n", buf);
ASN1_TIME_print (bio, X509_get_notAfter (cert));
memset (buf, 0, sizeof (buf));
BIO_read (bio, buf, sizeof (buf) - 1);
BIO_free (bio);
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info (" to: %s\n", buf);
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr,
"\n*** WARNING *** There is no way to verify this certificate. It is\n"
" possible that a hostile attacker has replaced the\n"
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
" server certificate. Continue at your own risk!\n"
"\nAccept this certificate anyway? [no]: ");
if (fgets (buf, sizeof (buf), stdin) && (buf[0] == 'y' || buf[0] == 'Y'))
{
ret = 0;
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "\n*** Fine, but don't say I didn't warn you!\n\n");
}
return ret;
}
static int
init_ssl (config_t * conf)
{
patch from Daniel Resare <noa@metamatrix.se>: 1 giving a path to a nonexistant rc-file with the -c argument dumps core The patch adds a check to ensure that the given rc-file is accessible 2 the error messages given from failed openssl calls are bogus The handles the error from SSL_connect () correctly. The bug is understndable since the error handling in openssl is quite obfuscated. Good news is that the documentation manapges has been greatly updated in the latest version (0.9.6). See in particular err(3), ERR_get_error(3) and SSL_get_error(3). Please note that possible SSL_ERROR_SSL type errors from SSL_read() and SSL_write() is not handled. This should also be fixed. 3 connecting using the STARTTLS command with an imap server that is configured only to accept the TLSv1 protocol gives an error because isync sends an SSLv2 Hello message for backwards compability. (This is the case with the uw-imap 2000 that ships with redhat-7.0) I've read RFC2595 several times to see if it says something about compability SSL2/SSL3 hello messages but can't find anything. IMHO the correct thing to do is change the default to not use SSL2/3 compability hello when using the STARTTLS command but use it if the imaps port is used. The patch implements this change 4 repeated calls to SSL_CTX_set_options overwrites the old settings (the values needs to be ORed together) fixed in the patch patch from me@mutt.org: \Recent messages were put in the cur/ directory instead of new/ give error message when the LOGIN command fails
24 years ago
SSL_METHOD *method;
int options = 0;
if (!conf->cert_file)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "Error, CertificateFile not defined\n");
return -1;
}
SSL_library_init ();
SSL_load_error_strings ();
patch from Daniel Resare <noa@metamatrix.se>: 1 giving a path to a nonexistant rc-file with the -c argument dumps core The patch adds a check to ensure that the given rc-file is accessible 2 the error messages given from failed openssl calls are bogus The handles the error from SSL_connect () correctly. The bug is understndable since the error handling in openssl is quite obfuscated. Good news is that the documentation manapges has been greatly updated in the latest version (0.9.6). See in particular err(3), ERR_get_error(3) and SSL_get_error(3). Please note that possible SSL_ERROR_SSL type errors from SSL_read() and SSL_write() is not handled. This should also be fixed. 3 connecting using the STARTTLS command with an imap server that is configured only to accept the TLSv1 protocol gives an error because isync sends an SSLv2 Hello message for backwards compability. (This is the case with the uw-imap 2000 that ships with redhat-7.0) I've read RFC2595 several times to see if it says something about compability SSL2/SSL3 hello messages but can't find anything. IMHO the correct thing to do is change the default to not use SSL2/3 compability hello when using the STARTTLS command but use it if the imaps port is used. The patch implements this change 4 repeated calls to SSL_CTX_set_options overwrites the old settings (the values needs to be ORed together) fixed in the patch patch from me@mutt.org: \Recent messages were put in the cur/ directory instead of new/ give error message when the LOGIN command fails
24 years ago
if (conf->use_tlsv1 && !conf->use_sslv2 && !conf->use_sslv3)
method = TLSv1_client_method ();
else
method = SSLv23_client_method ();
SSLContext = SSL_CTX_new (method);
if (access (conf->cert_file, F_OK))
{
if (errno != ENOENT)
{
perror ("access");
return -1;
}
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr,
"*** Warning, CertificateFile doesn't exist, can't verify server certificates\n");
}
else
if (!SSL_CTX_load_verify_locations
(SSLContext, conf->cert_file, NULL))
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "Error, SSL_CTX_load_verify_locations: %s\n",
ERR_error_string (ERR_get_error (), 0));
return -1;
}
if (!conf->use_sslv2)
patch from Daniel Resare <noa@metamatrix.se>: 1 giving a path to a nonexistant rc-file with the -c argument dumps core The patch adds a check to ensure that the given rc-file is accessible 2 the error messages given from failed openssl calls are bogus The handles the error from SSL_connect () correctly. The bug is understndable since the error handling in openssl is quite obfuscated. Good news is that the documentation manapges has been greatly updated in the latest version (0.9.6). See in particular err(3), ERR_get_error(3) and SSL_get_error(3). Please note that possible SSL_ERROR_SSL type errors from SSL_read() and SSL_write() is not handled. This should also be fixed. 3 connecting using the STARTTLS command with an imap server that is configured only to accept the TLSv1 protocol gives an error because isync sends an SSLv2 Hello message for backwards compability. (This is the case with the uw-imap 2000 that ships with redhat-7.0) I've read RFC2595 several times to see if it says something about compability SSL2/SSL3 hello messages but can't find anything. IMHO the correct thing to do is change the default to not use SSL2/3 compability hello when using the STARTTLS command but use it if the imaps port is used. The patch implements this change 4 repeated calls to SSL_CTX_set_options overwrites the old settings (the values needs to be ORed together) fixed in the patch patch from me@mutt.org: \Recent messages were put in the cur/ directory instead of new/ give error message when the LOGIN command fails
24 years ago
options |= SSL_OP_NO_SSLv2;
if (!conf->use_sslv3)
patch from Daniel Resare <noa@metamatrix.se>: 1 giving a path to a nonexistant rc-file with the -c argument dumps core The patch adds a check to ensure that the given rc-file is accessible 2 the error messages given from failed openssl calls are bogus The handles the error from SSL_connect () correctly. The bug is understndable since the error handling in openssl is quite obfuscated. Good news is that the documentation manapges has been greatly updated in the latest version (0.9.6). See in particular err(3), ERR_get_error(3) and SSL_get_error(3). Please note that possible SSL_ERROR_SSL type errors from SSL_read() and SSL_write() is not handled. This should also be fixed. 3 connecting using the STARTTLS command with an imap server that is configured only to accept the TLSv1 protocol gives an error because isync sends an SSLv2 Hello message for backwards compability. (This is the case with the uw-imap 2000 that ships with redhat-7.0) I've read RFC2595 several times to see if it says something about compability SSL2/SSL3 hello messages but can't find anything. IMHO the correct thing to do is change the default to not use SSL2/3 compability hello when using the STARTTLS command but use it if the imaps port is used. The patch implements this change 4 repeated calls to SSL_CTX_set_options overwrites the old settings (the values needs to be ORed together) fixed in the patch patch from me@mutt.org: \Recent messages were put in the cur/ directory instead of new/ give error message when the LOGIN command fails
24 years ago
options |= SSL_OP_NO_SSLv3;
if (!conf->use_tlsv1)
patch from Daniel Resare <noa@metamatrix.se>: 1 giving a path to a nonexistant rc-file with the -c argument dumps core The patch adds a check to ensure that the given rc-file is accessible 2 the error messages given from failed openssl calls are bogus The handles the error from SSL_connect () correctly. The bug is understndable since the error handling in openssl is quite obfuscated. Good news is that the documentation manapges has been greatly updated in the latest version (0.9.6). See in particular err(3), ERR_get_error(3) and SSL_get_error(3). Please note that possible SSL_ERROR_SSL type errors from SSL_read() and SSL_write() is not handled. This should also be fixed. 3 connecting using the STARTTLS command with an imap server that is configured only to accept the TLSv1 protocol gives an error because isync sends an SSLv2 Hello message for backwards compability. (This is the case with the uw-imap 2000 that ships with redhat-7.0) I've read RFC2595 several times to see if it says something about compability SSL2/SSL3 hello messages but can't find anything. IMHO the correct thing to do is change the default to not use SSL2/3 compability hello when using the STARTTLS command but use it if the imaps port is used. The patch implements this change 4 repeated calls to SSL_CTX_set_options overwrites the old settings (the values needs to be ORed together) fixed in the patch patch from me@mutt.org: \Recent messages were put in the cur/ directory instead of new/ give error message when the LOGIN command fails
24 years ago
options |= SSL_OP_NO_TLSv1;
SSL_CTX_set_options (SSLContext, options);
/* we check the result of the verification after SSL_connect() */
SSL_CTX_set_verify (SSLContext, SSL_VERIFY_NONE, 0);
return 0;
}
#endif /* HAVE_LIBSSL */
static int
socket_read (Socket_t * sock, char *buf, size_t len)
{
#if HAVE_LIBSSL
if (sock->use_ssl)
return SSL_read (sock->ssl, buf, len);
#endif
return read (sock->fd, buf, len);
}
static int
socket_write (Socket_t * sock, char *buf, size_t len)
{
#if HAVE_LIBSSL
if (sock->use_ssl)
return SSL_write (sock->ssl, buf, len);
#endif
return write (sock->fd, buf, len);
}
patch from Daniel Resare <noa@metamatrix.se>: 1 giving a path to a nonexistant rc-file with the -c argument dumps core The patch adds a check to ensure that the given rc-file is accessible 2 the error messages given from failed openssl calls are bogus The handles the error from SSL_connect () correctly. The bug is understndable since the error handling in openssl is quite obfuscated. Good news is that the documentation manapges has been greatly updated in the latest version (0.9.6). See in particular err(3), ERR_get_error(3) and SSL_get_error(3). Please note that possible SSL_ERROR_SSL type errors from SSL_read() and SSL_write() is not handled. This should also be fixed. 3 connecting using the STARTTLS command with an imap server that is configured only to accept the TLSv1 protocol gives an error because isync sends an SSLv2 Hello message for backwards compability. (This is the case with the uw-imap 2000 that ships with redhat-7.0) I've read RFC2595 several times to see if it says something about compability SSL2/SSL3 hello messages but can't find anything. IMHO the correct thing to do is change the default to not use SSL2/3 compability hello when using the STARTTLS command but use it if the imaps port is used. The patch implements this change 4 repeated calls to SSL_CTX_set_options overwrites the old settings (the values needs to be ORed together) fixed in the patch patch from me@mutt.org: \Recent messages were put in the cur/ directory instead of new/ give error message when the LOGIN command fails
24 years ago
static void
socket_perror (const char *func, Socket_t *sock, int ret)
{
#if HAVE_LIBSSL
int err;
if (sock->use_ssl)
{
switch ((err = SSL_get_error (sock->ssl, ret)))
{
case SSL_ERROR_SYSCALL:
case SSL_ERROR_SSL:
if ((err = ERR_get_error ()) == 0)
{
if (ret == 0)
fprintf (stderr, "SSL_%s:got EOF\n", func);
else
fprintf (stderr, "SSL_%s:%d:%s\n", func,
errno, strerror (errno));
}
else
fprintf (stderr, "SSL_%s:%d:%s\n", func, err,
ERR_error_string (err, 0));
return;
default:
fprintf (stderr, "SSL_%s:%d:unhandled SSL error\n", func, err);
break;
}
return;
}
#else
(void) sock;
patch from Daniel Resare <noa@metamatrix.se>: 1 giving a path to a nonexistant rc-file with the -c argument dumps core The patch adds a check to ensure that the given rc-file is accessible 2 the error messages given from failed openssl calls are bogus The handles the error from SSL_connect () correctly. The bug is understndable since the error handling in openssl is quite obfuscated. Good news is that the documentation manapges has been greatly updated in the latest version (0.9.6). See in particular err(3), ERR_get_error(3) and SSL_get_error(3). Please note that possible SSL_ERROR_SSL type errors from SSL_read() and SSL_write() is not handled. This should also be fixed. 3 connecting using the STARTTLS command with an imap server that is configured only to accept the TLSv1 protocol gives an error because isync sends an SSLv2 Hello message for backwards compability. (This is the case with the uw-imap 2000 that ships with redhat-7.0) I've read RFC2595 several times to see if it says something about compability SSL2/SSL3 hello messages but can't find anything. IMHO the correct thing to do is change the default to not use SSL2/3 compability hello when using the STARTTLS command but use it if the imaps port is used. The patch implements this change 4 repeated calls to SSL_CTX_set_options overwrites the old settings (the values needs to be ORed together) fixed in the patch patch from me@mutt.org: \Recent messages were put in the cur/ directory instead of new/ give error message when the LOGIN command fails
24 years ago
#endif
if (ret)
perror (func);
else
fprintf (stderr, "%s: unexpected EOF\n", func);
patch from Daniel Resare <noa@metamatrix.se>: 1 giving a path to a nonexistant rc-file with the -c argument dumps core The patch adds a check to ensure that the given rc-file is accessible 2 the error messages given from failed openssl calls are bogus The handles the error from SSL_connect () correctly. The bug is understndable since the error handling in openssl is quite obfuscated. Good news is that the documentation manapges has been greatly updated in the latest version (0.9.6). See in particular err(3), ERR_get_error(3) and SSL_get_error(3). Please note that possible SSL_ERROR_SSL type errors from SSL_read() and SSL_write() is not handled. This should also be fixed. 3 connecting using the STARTTLS command with an imap server that is configured only to accept the TLSv1 protocol gives an error because isync sends an SSLv2 Hello message for backwards compability. (This is the case with the uw-imap 2000 that ships with redhat-7.0) I've read RFC2595 several times to see if it says something about compability SSL2/SSL3 hello messages but can't find anything. IMHO the correct thing to do is change the default to not use SSL2/3 compability hello when using the STARTTLS command but use it if the imaps port is used. The patch implements this change 4 repeated calls to SSL_CTX_set_options overwrites the old settings (the values needs to be ORed together) fixed in the patch patch from me@mutt.org: \Recent messages were put in the cur/ directory instead of new/ give error message when the LOGIN command fails
24 years ago
}
24 years ago
/* simple line buffering */
static int
buffer_gets (buffer_t * b, char **s)
{
int n;
int start = b->offset;
*s = b->buf + start;
for (;;)
{
/* make sure we have enough data to read the \r\n sequence */
if (b->offset + 1 >= b->bytes)
24 years ago
{
if (start != 0)
{
/* shift down used bytes */
*s = b->buf;
24 years ago
assert (start <= b->bytes);
n = b->bytes - start;
24 years ago
if (n)
memmove (b->buf, b->buf + start, n);
b->offset -= start;
b->bytes = n;
start = 0;
}
24 years ago
n =
socket_read (b->sock, b->buf + b->bytes,
sizeof (b->buf) - b->bytes);
24 years ago
if (n <= 0)
{
patch from Daniel Resare <noa@metamatrix.se>: 1 giving a path to a nonexistant rc-file with the -c argument dumps core The patch adds a check to ensure that the given rc-file is accessible 2 the error messages given from failed openssl calls are bogus The handles the error from SSL_connect () correctly. The bug is understndable since the error handling in openssl is quite obfuscated. Good news is that the documentation manapges has been greatly updated in the latest version (0.9.6). See in particular err(3), ERR_get_error(3) and SSL_get_error(3). Please note that possible SSL_ERROR_SSL type errors from SSL_read() and SSL_write() is not handled. This should also be fixed. 3 connecting using the STARTTLS command with an imap server that is configured only to accept the TLSv1 protocol gives an error because isync sends an SSLv2 Hello message for backwards compability. (This is the case with the uw-imap 2000 that ships with redhat-7.0) I've read RFC2595 several times to see if it says something about compability SSL2/SSL3 hello messages but can't find anything. IMHO the correct thing to do is change the default to not use SSL2/3 compability hello when using the STARTTLS command but use it if the imaps port is used. The patch implements this change 4 repeated calls to SSL_CTX_set_options overwrites the old settings (the values needs to be ORed together) fixed in the patch patch from me@mutt.org: \Recent messages were put in the cur/ directory instead of new/ give error message when the LOGIN command fails
24 years ago
socket_perror ("read", b->sock, n);
24 years ago
return -1;
}
b->bytes += n;
24 years ago
}
if (b->buf[b->offset] == '\r')
{
assert (b->offset + 1 < b->bytes);
24 years ago
if (b->buf[b->offset + 1] == '\n')
{
b->buf[b->offset] = 0; /* terminate the string */
b->offset += 2; /* next line */
if (Verbose) {
puts (*s);
fflush (stdout);
}
24 years ago
return 0;
}
}
b->offset++;
}
/* not reached */
}
static int
parse_fetch (imap_t * imap, list_t * list)
{
list_t *tmp;
unsigned int uid = 0;
unsigned int mask = 0;
unsigned int size = 0;
message_t *cur;
if (!is_list (list))
return -1;
for (tmp = list->child; tmp; tmp = tmp->next)
{
if (is_atom (tmp))
{
if (!strcmp ("UID", tmp->val))
{
tmp = tmp->next;
if (is_atom (tmp))
{
uid = atoi (tmp->val);
if (uid < imap->minuid)
{
/* already saw this message */
return 0;
}
else if (uid > imap->maxuid)
imap->maxuid = uid;
}
else
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: unable to parse UID\n");
}
else if (!strcmp ("FLAGS", tmp->val))
{
tmp = tmp->next;
if (is_list (tmp))
{
list_t *flags = tmp->child;
for (; flags; flags = flags->next)
{
if (is_atom (flags))
{
if (!strcmp ("\\Seen", flags->val))
mask |= D_SEEN;
else if (!strcmp ("\\Flagged", flags->val))
mask |= D_FLAGGED;
else if (!strcmp ("\\Deleted", flags->val))
mask |= D_DELETED;
else if (!strcmp ("\\Answered", flags->val))
mask |= D_ANSWERED;
else if (!strcmp ("\\Draft", flags->val))
mask |= D_DRAFT;
else if (!strcmp ("\\Recent", flags->val))
mask |= D_RECENT;
else
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: unknown flag %s\n",
flags->val);
}
else
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: unable to parse FLAGS list\n");
}
}
else
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: unable to parse FLAGS\n");
}
else if (!strcmp ("RFC822.SIZE", tmp->val))
{
tmp = tmp->next;
if (is_atom (tmp))
size = atol (tmp->val);
}
}
}
cur = calloc (1, sizeof (message_t));
cur->next = imap->msgs;
imap->msgs = cur;
if (mask & D_DELETED)
imap->deleted++;
cur->uid = uid;
cur->flags = mask;
cur->size = size;
return 0;
}
static void
parse_response_code (imap_t * imap, char *s)
{
char *arg;
if (*s != '[')
return; /* no response code */
s++;
arg = next_arg (&s);
if (!strcmp ("UIDVALIDITY", arg))
{
arg = next_arg (&s);
imap->uidvalidity = atol (arg);
}
else if (!strcmp ("ALERT", arg))
{
/* RFC2060 says that these messages MUST be displayed
* to the user
*/
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "*** IMAP ALERT *** %s\n", s);
}
}
24 years ago
static int
imap_exec (imap_t * imap, const char *fmt, ...)
{
va_list ap;
char tmp[256];
char buf[256];
char *cmd;
char *arg;
char *arg1;
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
config_t *box;
patch from Daniel Resare <noa@metamatrix.se>: 1 giving a path to a nonexistant rc-file with the -c argument dumps core The patch adds a check to ensure that the given rc-file is accessible 2 the error messages given from failed openssl calls are bogus The handles the error from SSL_connect () correctly. The bug is understndable since the error handling in openssl is quite obfuscated. Good news is that the documentation manapges has been greatly updated in the latest version (0.9.6). See in particular err(3), ERR_get_error(3) and SSL_get_error(3). Please note that possible SSL_ERROR_SSL type errors from SSL_read() and SSL_write() is not handled. This should also be fixed. 3 connecting using the STARTTLS command with an imap server that is configured only to accept the TLSv1 protocol gives an error because isync sends an SSLv2 Hello message for backwards compability. (This is the case with the uw-imap 2000 that ships with redhat-7.0) I've read RFC2595 several times to see if it says something about compability SSL2/SSL3 hello messages but can't find anything. IMHO the correct thing to do is change the default to not use SSL2/3 compability hello when using the STARTTLS command but use it if the imaps port is used. The patch implements this change 4 repeated calls to SSL_CTX_set_options overwrites the old settings (the values needs to be ORed together) fixed in the patch patch from me@mutt.org: \Recent messages were put in the cur/ directory instead of new/ give error message when the LOGIN command fails
24 years ago
int n;
24 years ago
va_start (ap, fmt);
vsnprintf (tmp, sizeof (tmp), fmt, ap);
va_end (ap);
snprintf (buf, sizeof (buf), "%d %s\r\n", ++Tag, tmp);
if (Verbose) {
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
printf (">>> %s", buf);
fflush (stdout);
}
patch from Daniel Resare <noa@metamatrix.se>: 1 giving a path to a nonexistant rc-file with the -c argument dumps core The patch adds a check to ensure that the given rc-file is accessible 2 the error messages given from failed openssl calls are bogus The handles the error from SSL_connect () correctly. The bug is understndable since the error handling in openssl is quite obfuscated. Good news is that the documentation manapges has been greatly updated in the latest version (0.9.6). See in particular err(3), ERR_get_error(3) and SSL_get_error(3). Please note that possible SSL_ERROR_SSL type errors from SSL_read() and SSL_write() is not handled. This should also be fixed. 3 connecting using the STARTTLS command with an imap server that is configured only to accept the TLSv1 protocol gives an error because isync sends an SSLv2 Hello message for backwards compability. (This is the case with the uw-imap 2000 that ships with redhat-7.0) I've read RFC2595 several times to see if it says something about compability SSL2/SSL3 hello messages but can't find anything. IMHO the correct thing to do is change the default to not use SSL2/3 compability hello when using the STARTTLS command but use it if the imaps port is used. The patch implements this change 4 repeated calls to SSL_CTX_set_options overwrites the old settings (the values needs to be ORed together) fixed in the patch patch from me@mutt.org: \Recent messages were put in the cur/ directory instead of new/ give error message when the LOGIN command fails
24 years ago
n = socket_write (imap->sock, buf, strlen (buf));
if (n <= 0)
{
socket_perror ("write", imap->sock, n);
return -1;
}
24 years ago
for (;;)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
next:
24 years ago
if (buffer_gets (imap->buf, &cmd))
return -1;
arg = next_arg (&cmd);
if (*arg == '*')
{
arg = next_arg (&cmd);
if (!arg)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: unable to parse untagged response\n");
return -1;
}
24 years ago
if (!strcmp ("NAMESPACE", arg))
{
imap->ns_personal = parse_list (cmd, &cmd);
imap->ns_other = parse_list (cmd, &cmd);
imap->ns_shared = parse_list (cmd, 0);
}
else if (!strcmp ("OK", arg) || !strcmp ("BAD", arg) ||
!strcmp ("NO", arg) || !strcmp ("BYE", arg) ||
!strcmp ("PREAUTH", arg))
{
parse_response_code (imap, cmd);
}
else if (!strcmp ("CAPABILITY", arg))
{
while ((arg = next_arg (&cmd)))
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
if (!strcmp ("UIDPLUS", arg))
imap->have_uidplus = 1;
22 years ago
else if (!strcmp ("NAMESPACE", arg))
imap->have_namespace = 1;
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
#if HAVE_LIBSSL
else if (!strcmp ("STARTTLS", arg))
imap->have_starttls = 1;
else if (!strcmp ("AUTH=CRAM-MD5", arg))
imap->have_cram = 1;
#endif
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
}
}
else if (!strcmp ("LIST", arg))
{
list_t *list, *lp;
int l;
list = parse_list (cmd, &cmd);
if (list->val == LIST)
for (lp = list->child; lp; lp = lp->next)
if (is_atom (lp) &&
!strcasecmp (lp->val, "\\NoSelect"))
{
free_list (list);
goto next;
}
free_list (list);
(void) next_arg (&cmd); /* skip delimiter */
arg = next_arg (&cmd);
l = strlen (global.folder);
if (memcmp (arg, global.folder, l))
goto next;
arg += l;
if (!memcmp (arg + strlen (arg) - 5, ".lock", 5))
goto next;
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
for (box = boxes; box; box = box->next)
if (!strcmp (box->box, arg))
goto next;
box = malloc (sizeof (config_t));
memcpy (box, &global, sizeof (config_t));
box->path = strdup (arg);
box->box = box->path;
box->next = boxes;
boxes = box;
}
else if ((arg1 = next_arg (&cmd)))
24 years ago
{
if (!strcmp ("EXISTS", arg1))
imap->count = atoi (arg);
else if (!strcmp ("RECENT", arg1))
imap->recent = atoi (arg);
else if (!strcmp ("FETCH", arg1))
24 years ago
{
list_t *list;
24 years ago
list = parse_list (cmd, 0);
24 years ago
if (parse_fetch (imap, list))
{
free_list (list);
return -1;
}
24 years ago
free_list (list);
}
}
else
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: unable to parse untagged response\n");
return -1;
24 years ago
}
}
#if HAVE_LIBSSL
else if (*arg == '+')
{
char *resp;
if (!imap->cram)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error, not doing CRAM-MD5 authentication\n");
return -1;
}
resp = cram (cmd, imap->box->user, imap->box->pass);
if (Verbose) {
printf (">+> %s\n", resp);
fflush (stdout);
}
patch from Daniel Resare <noa@metamatrix.se>: 1 giving a path to a nonexistant rc-file with the -c argument dumps core The patch adds a check to ensure that the given rc-file is accessible 2 the error messages given from failed openssl calls are bogus The handles the error from SSL_connect () correctly. The bug is understndable since the error handling in openssl is quite obfuscated. Good news is that the documentation manapges has been greatly updated in the latest version (0.9.6). See in particular err(3), ERR_get_error(3) and SSL_get_error(3). Please note that possible SSL_ERROR_SSL type errors from SSL_read() and SSL_write() is not handled. This should also be fixed. 3 connecting using the STARTTLS command with an imap server that is configured only to accept the TLSv1 protocol gives an error because isync sends an SSLv2 Hello message for backwards compability. (This is the case with the uw-imap 2000 that ships with redhat-7.0) I've read RFC2595 several times to see if it says something about compability SSL2/SSL3 hello messages but can't find anything. IMHO the correct thing to do is change the default to not use SSL2/3 compability hello when using the STARTTLS command but use it if the imaps port is used. The patch implements this change 4 repeated calls to SSL_CTX_set_options overwrites the old settings (the values needs to be ORed together) fixed in the patch patch from me@mutt.org: \Recent messages were put in the cur/ directory instead of new/ give error message when the LOGIN command fails
24 years ago
n = socket_write (imap->sock, resp, strlen (resp));
free (resp);
patch from Daniel Resare <noa@metamatrix.se>: 1 giving a path to a nonexistant rc-file with the -c argument dumps core The patch adds a check to ensure that the given rc-file is accessible 2 the error messages given from failed openssl calls are bogus The handles the error from SSL_connect () correctly. The bug is understndable since the error handling in openssl is quite obfuscated. Good news is that the documentation manapges has been greatly updated in the latest version (0.9.6). See in particular err(3), ERR_get_error(3) and SSL_get_error(3). Please note that possible SSL_ERROR_SSL type errors from SSL_read() and SSL_write() is not handled. This should also be fixed. 3 connecting using the STARTTLS command with an imap server that is configured only to accept the TLSv1 protocol gives an error because isync sends an SSLv2 Hello message for backwards compability. (This is the case with the uw-imap 2000 that ships with redhat-7.0) I've read RFC2595 several times to see if it says something about compability SSL2/SSL3 hello messages but can't find anything. IMHO the correct thing to do is change the default to not use SSL2/3 compability hello when using the STARTTLS command but use it if the imaps port is used. The patch implements this change 4 repeated calls to SSL_CTX_set_options overwrites the old settings (the values needs to be ORed together) fixed in the patch patch from me@mutt.org: \Recent messages were put in the cur/ directory instead of new/ give error message when the LOGIN command fails
24 years ago
if (n <= 0)
{
socket_perror ("write", imap->sock, n);
return -1;
}
n = socket_write (imap->sock, "\r\n", 2);
if (n <= 0)
{
socket_perror ("write", imap->sock, n);
return -1;
}
imap->cram = 0;
}
#endif
24 years ago
else if ((size_t) atol (arg) != Tag)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: wrong tag\n");
24 years ago
return -1;
}
else
{
arg = next_arg (&cmd);
parse_response_code (imap, cmd);
24 years ago
if (!strcmp ("OK", arg))
return 0;
return -1;
}
}
/* not reached */
}
#ifdef HAVE_LIBSSL
22 years ago
static int
start_tls (imap_t *imap, config_t * cfg)
{
int ret;
/* initialize SSL */
if (init_ssl (cfg))
return 1;
imap->sock->ssl = SSL_new (SSLContext);
SSL_set_fd (imap->sock->ssl, imap->sock->fd);
if ((ret = SSL_connect (imap->sock->ssl)) <= 0)
{
socket_perror ("connect", imap->sock, ret);
return 1;
}
/* verify the server certificate */
if (verify_cert (imap->sock->ssl))
return 1;
imap->sock->use_ssl = 1;
puts ("SSL support enabled");
return 0;
}
#endif
22 years ago
24 years ago
imap_t *
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
imap_connect (config_t * cfg)
24 years ago
{
22 years ago
int s;
struct sockaddr_in addr;
struct hostent *he;
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
imap_t *imap;
char *arg, *rsp;
22 years ago
int preauth;
#if HAVE_LIBSSL
22 years ago
int use_ssl;
#endif
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
int a[2];
imap = calloc (1, sizeof (imap_t));
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
imap->box = cfg;
imap->sock = calloc (1, sizeof (Socket_t));
imap->buf = calloc (1, sizeof (buffer_t));
imap->buf->sock = imap->sock;
imap->sock->fd = -1;
/* open connection to IMAP server */
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
if (cfg->tunnel)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info ("Starting tunnel '%s'...", cfg->tunnel);
fflush (stdout);
if (socketpair (PF_UNIX, SOCK_STREAM, 0, a))
{
perror ("socketpair");
exit (1);
}
if (fork () == 0)
{
if (dup2 (a[0], 0) == -1 || dup2 (a[0], 1) == -1)
{
_exit (127);
}
close (a[0]);
close (a[1]);
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
execl ("/bin/sh", "sh", "-c", cfg->tunnel, 0);
_exit (127);
}
close (a[0]);
imap->sock->fd = a[1];
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info ("ok\n");
}
else
{
memset (&addr, 0, sizeof (addr));
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
addr.sin_port = htons (cfg->port);
addr.sin_family = AF_INET;
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info ("Resolving %s... ", cfg->host);
fflush (stdout);
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
he = gethostbyname (cfg->host);
if (!he)
{
perror ("gethostbyname");
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
goto bail;
}
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info ("ok\n");
addr.sin_addr.s_addr = *((int *) he->h_addr_list[0]);
s = socket (PF_INET, SOCK_STREAM, 0);
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info ("Connecting to %s:%hu... ", inet_ntoa (addr.sin_addr),
ntohs (addr.sin_port));
fflush (stdout);
if (connect (s, (struct sockaddr *) &addr, sizeof (addr)))
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
close (s);
perror ("connect");
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
goto bail;
}
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info ("ok\n");
imap->sock->fd = s;
}
22 years ago
#if HAVE_LIBSSL
use_ssl = 0;
if (cfg->use_imaps) {
if (start_tls (imap, cfg))
goto bail;
use_ssl = 1;
}
#endif
/* read the greeting string */
if (buffer_gets (imap->buf, &rsp))
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: no greeting response\n");
goto bail;
}
arg = next_arg (&rsp);
if (!arg || *arg != '*' || (arg = next_arg (&rsp)) == NULL)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: invalid greeting response\n");
goto bail;
}
22 years ago
preauth = 0;
if (!strcmp ("PREAUTH", arg))
preauth = 1;
else if (strcmp ("OK", arg) != 0)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: unknown greeting response\n");
goto bail;
}
22 years ago
/* let's see what this puppy can do... */
if (imap_exec (imap, "CAPABILITY"))
goto bail;
#if HAVE_LIBSSL
22 years ago
if (!cfg->use_imaps)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
if (cfg->use_sslv2 || cfg->use_sslv3 || cfg->use_tlsv1)
{
/* always try to select SSL support if available */
if (imap->have_starttls)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
if (imap_exec (imap, "STARTTLS"))
goto bail;
22 years ago
if (start_tls (imap, cfg))
goto bail;
use_ssl = 1;
22 years ago
/* to conform to RFC2595 we need to forget all information
* retrieved from CAPABILITY invocations before STARTTLS.
*/
imap->have_uidplus = 0;
imap->have_namespace = 0;
imap->have_cram = 0;
/* imap->have_starttls = 0; */
if (imap_exec (imap, "CAPABILITY"))
goto bail;
}
else
{
if (cfg->require_ssl)
{
fprintf (stderr, "IMAP error: SSL support not available\n");
goto bail;
}
else
fprintf (stderr, "IMAP warning: SSL support not available\n");
}
}
}
#endif
if (!preauth)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info ("Logging in...\n");
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
if (!cfg->pass)
{
/*
* if we don't have a global password set, prompt the user for
* it now.
*/
if (!global.pass)
{
char prompt[80];
sprintf(prompt, "Password (mailbox %s@%s/%s): ",
cfg->user, cfg->host, cfg->box);
global.pass = getpass (prompt);
if (!global.pass)
{
perror ("getpass");
exit (1);
}
if (!*global.pass)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "Skipping %s, no password\n", cfg->path);
global.pass = NULL; /* force retry */
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
goto bail;
}
/*
* getpass() returns a pointer to a static buffer. make a copy
* for long term storage.
*/
global.pass = strdup (global.pass);
}
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
cfg->pass = strdup (global.pass);
}
#if HAVE_LIBSSL
if (imap->have_cram)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info ("Authenticating with CRAM-MD5\n");
imap->cram = 1;
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
if (imap_exec (imap, "AUTHENTICATE CRAM-MD5"))
goto bail;
}
else if (imap->box->require_cram)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: CRAM-MD5 authentication is not supported by server\n");
goto bail;
}
else
#endif
{
#if HAVE_LIBSSL
if (!use_ssl)
#endif
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "*** IMAP Warning *** Password is being sent in the clear\n");
if (imap_exec (imap, "LOGIN \"%s\" \"%s\"", cfg->user, cfg->pass))
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: LOGIN failed\n");
goto bail;
}
24 years ago
}
}
/* get NAMESPACE info */
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
if (cfg->use_namespace && imap->have_namespace)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
if (imap_exec (imap, "NAMESPACE"))
goto bail;
}
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
return imap;
bail:
imap_close (imap);
return 0;
}
static int
mstrcmp (const char *s1, const char *s2)
{
if (s1 == s2)
return 0;
if (!s1 || !s2)
return 1;
return strcmp (s1, s2);
}
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
/* `box' is the config info for the maildrop to sync. `minuid' is the
* minimum UID to consider. in normal mode this will be 1, but in --fast
* mode we only fetch messages newer than the last one seen in the local
* mailbox.
*/
imap_t *
imap_open (config_t * box, unsigned int minuid, imap_t * imap, int imap_create)
{
if (imap)
{
/* determine whether or not we can reuse the existing session */
if (mstrcmp (box->tunnel, imap->box->tunnel) ||
mstrcmp (box->host, imap->box->host) ||
mstrcmp (box->user, imap->box->user) ||
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
box->port != imap->box->port
#if HAVE_LIBSSL
/* ensure that security requirements are met */
|| (box->require_ssl ^ imap->box->require_ssl)
|| (box->require_cram ^ imap->box->require_cram)
#endif
)
{
/* can't reuse */
imap_close (imap);
}
else
{
/* reset mailbox-specific state info */
imap->box = box;
imap->recent = 0;
imap->deleted = 0;
imap->count = 0;
imap->maxuid = 0;
free_message (imap->msgs);
imap->msgs = 0;
goto gotimap;
}
}
if (!(imap = imap_connect (box)))
return 0;
gotimap:
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
if (global.folder)
imap->prefix = !strcmp (box->box, "INBOX") ? "" : global.folder;
else
{
imap->prefix = "";
/* XXX for now assume personal namespace */
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
if (imap->box->use_namespace &&
is_list (imap->ns_personal) &&
is_list (imap->ns_personal->child) &&
is_atom (imap->ns_personal->child->child))
imap->prefix = imap->ns_personal->child->child->val;
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
}
24 years ago
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info ("Selecting IMAP mailbox... ");
fflush (stdout);
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
if (imap_exec (imap, "SELECT \"%s%s\"", imap->prefix, box->box)) {
if (imap_create) {
if (imap_exec (imap, "CREATE \"%s%s\"", imap->prefix, box->box))
goto bail;
if (imap_exec (imap, "SELECT \"%s%s\"", imap->prefix, box->box))
goto bail;
} else
goto bail;
}
info ("%d messages, %d recent\n", imap->count, imap->recent);
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info ("Reading IMAP mailbox index\n");
imap->minuid = minuid;
if (imap->count > 0)
24 years ago
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
if (imap_exec (imap, "UID FETCH %d:* (FLAGS RFC822.SIZE)", minuid))
goto bail;
24 years ago
}
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
return imap;
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
bail:
imap_close (imap);
return 0;
24 years ago
}
void
imap_close (imap_t * imap)
{
if (imap)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
if (imap->sock->fd != -1)
{
imap_exec (imap, "LOGOUT");
close (imap->sock->fd);
}
free (imap->sock);
free (imap->buf);
free_message (imap->msgs);
memset (imap, 0xff, sizeof (imap_t));
free (imap);
}
24 years ago
}
/* write a buffer stripping all \r bytes */
static int
write_strip (int fd, char *buf, size_t len)
{
size_t start = 0;
size_t end = 0;
ssize_t n;
while (start < len)
{
while (end < len && buf[end] != '\r')
end++;
n = write (fd, buf + start, end - start);
if (n == -1)
24 years ago
{
perror ("write");
return -1;
24 years ago
}
else if ((size_t) n != end - start)
{
/* short write, try again */
start += n;
}
else
{
/* write complete */
end++;
start = end;
}
}
return 0;
24 years ago
}
patch from Daniel Resare <noa@metamatrix.se>: 1 giving a path to a nonexistant rc-file with the -c argument dumps core The patch adds a check to ensure that the given rc-file is accessible 2 the error messages given from failed openssl calls are bogus The handles the error from SSL_connect () correctly. The bug is understndable since the error handling in openssl is quite obfuscated. Good news is that the documentation manapges has been greatly updated in the latest version (0.9.6). See in particular err(3), ERR_get_error(3) and SSL_get_error(3). Please note that possible SSL_ERROR_SSL type errors from SSL_read() and SSL_write() is not handled. This should also be fixed. 3 connecting using the STARTTLS command with an imap server that is configured only to accept the TLSv1 protocol gives an error because isync sends an SSLv2 Hello message for backwards compability. (This is the case with the uw-imap 2000 that ships with redhat-7.0) I've read RFC2595 several times to see if it says something about compability SSL2/SSL3 hello messages but can't find anything. IMHO the correct thing to do is change the default to not use SSL2/3 compability hello when using the STARTTLS command but use it if the imaps port is used. The patch implements this change 4 repeated calls to SSL_CTX_set_options overwrites the old settings (the values needs to be ORed together) fixed in the patch patch from me@mutt.org: \Recent messages were put in the cur/ directory instead of new/ give error message when the LOGIN command fails
24 years ago
static int
send_server (Socket_t * sock, const char *fmt, ...)
24 years ago
{
char buf[128];
char cmd[128];
va_list ap;
int n;
va_start (ap, fmt);
vsnprintf (buf, sizeof (buf), fmt, ap);
va_end (ap);
snprintf (cmd, sizeof (cmd), "%d %s\r\n", ++Tag, buf);
if (Verbose) {
printf (">>> %s", cmd);
fflush (stdout);
}
n = socket_write (sock, cmd, strlen (cmd));
if (n <= 0)
{
socket_perror ("write", sock, n);
return -1;
}
return 0;
24 years ago
}
int
imap_fetch_message (imap_t * imap, unsigned int uid, int fd)
{
char *cmd;
char *arg;
size_t bytes;
size_t n;
char buf[1024];
24 years ago
send_server (imap->sock, "UID FETCH %d BODY.PEEK[]", uid);
24 years ago
for (;;)
{
if (buffer_gets (imap->buf, &cmd))
return -1;
if (*cmd == '*')
{
/* need to figure out how long the message is
* * <msgno> FETCH (RFC822 {<size>}
*/
next_arg (&cmd); /* * */
next_arg (&cmd); /* <msgno> */
arg = next_arg (&cmd); /* FETCH */
if (strcasecmp ("FETCH", arg) != 0)
{
/* this is likely an untagged response, such as when new
* mail arrives in the middle of the session. just skip
* it for now.
*
* eg.,
* "* 4000 EXISTS"
* "* 2 RECENT"
*
*/
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
info ("IMAP info: skipping untagged response: %s\n", arg);
continue;
}
while ((arg = next_arg (&cmd)) && *arg != '{')
;
if (!arg)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: parse error getting size\n");
return -1;
}
bytes = strtol (arg + 1, 0, 10);
/* dump whats left over in the input buffer */
n = imap->buf->bytes - imap->buf->offset;
if (n > bytes)
{
/* the entire message fit in the buffer */
n = bytes;
}
/* ick. we have to strip out the \r\n line endings, so
* i can't just dump the raw bytes to disk.
*/
if (write_strip (fd, imap->buf->buf + imap->buf->offset, n))
{
/* write failed, message is not delivered */
return -1;
}
24 years ago
bytes -= n;
24 years ago
/* mark that we used part of the buffer */
imap->buf->offset += n;
24 years ago
/* now read the rest of the message */
while (bytes > 0)
{
n = bytes;
if (n > sizeof (buf))
n = sizeof (buf);
n = socket_read (imap->sock, buf, n);
if (n > 0)
{
if (write_strip (fd, buf, n))
{
/* write failed */
return -1;
}
bytes -= n;
24 years ago
}
else
{
socket_perror ("read", imap->sock, n);
return -1;
24 years ago
}
}
buffer_gets (imap->buf, &cmd);
24 years ago
}
else
{
arg = next_arg (&cmd);
if (!arg || (size_t) atoi (arg) != Tag)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: wrong tag\n");
return -1;
}
arg = next_arg (&cmd);
if (!strcmp ("OK", arg))
return 0;
return -1;
}
}
/* not reached */
24 years ago
}
/* add flags to existing flags */
int
imap_set_flags (imap_t * imap, unsigned int uid, unsigned int flags)
{
char buf[256];
int i;
buf[0] = 0;
for (i = 0; i < D_MAX; i++)
{
if (flags & (1 << i))
snprintf (buf + strlen (buf),
sizeof (buf) - strlen (buf), "%s%s",
(buf[0] != 0) ? " " : "", Flags[i]);
}
return imap_exec (imap, "UID STORE %d +FLAGS.SILENT (%s)", uid, buf);
24 years ago
}
int
imap_expunge (imap_t * imap)
{
return imap_exec (imap, "EXPUNGE");
24 years ago
}
int
imap_copy_message (imap_t * imap, unsigned int uid, const char *mailbox)
{
return imap_exec (imap, "UID COPY %u \"%s%s\"", uid, imap->prefix,
mailbox);
}
int
imap_append_message (imap_t * imap, int fd, message_t * msg)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
char *fmap;
int extra, uid, tuidl = 0;
char flagstr[128], tuid[128];
char *s;
size_t i;
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
size_t start;
size_t len, sbreak = 0, ebreak = 0;
char *arg;
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
struct timeval tv;
pid_t pid = getpid();
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
len = msg->size;
/* ugh, we need to count the number of newlines */
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fmap = (char *)mmap (0, len, PROT_READ, MAP_PRIVATE, fd, 0);
if (!fmap)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
perror ("mmap");
return -1;
}
extra = 0, i = 0;
if (!imap->have_uidplus)
{
nloop:
start = i;
while (i < len)
if (fmap[i++] == '\n')
{
extra++;
if (i - 1 == start)
{
sbreak = ebreak = i - 1;
goto mktid;
}
if (!memcmp (fmap + start, "X-TUID: ", 8))
{
extra -= (ebreak = i) - (sbreak = start) + 1;
goto mktid;
}
goto nloop;
}
/* invalid mesasge */
goto bail;
mktid:
gettimeofday (&tv, 0);
tuidl = sprintf (tuid, "X-TUID: %08lx%05lx%04x\r\n",
tv.tv_sec, tv.tv_usec, pid);
extra += tuidl;
}
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
for (; i < len; i++)
if (fmap[i] == '\n')
extra++;
flagstr[0] = 0;
if (msg->flags)
{
if (msg->flags & D_DELETED)
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
strcat (flagstr," \\Deleted");
if (msg->flags & D_ANSWERED)
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
strcat (flagstr," \\Answered");
if (msg->flags & D_SEEN)
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
strcat (flagstr," \\Seen");
if (msg->flags & D_FLAGGED)
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
strcat (flagstr," \\Flagged");
if (msg->flags & D_DRAFT)
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
strcat (flagstr," \\Draft");
flagstr[0] = '(';
strcat (flagstr,") ");
}
send_server (imap->sock, "APPEND %s%s %s{%d}",
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
imap->prefix, imap->box->box, flagstr, len + extra);
if (buffer_gets (imap->buf, &s))
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
goto bail;
if (*s != '+')
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: expected `+' from server (aborting)\n");
goto bail;
}
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
i = 0;
if (!imap->have_uidplus)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
n1loop:
start = i;
while (i < sbreak)
if (fmap[i++] == '\n')
{
socket_write (imap->sock, fmap + start, i - 1 - start);
socket_write (imap->sock, "\r\n", 2);
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
goto n1loop;
}
socket_write (imap->sock, tuid, tuidl);
i = ebreak;
}
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
n2loop:
start = i;
while (i < len)
if (fmap[i++] == '\n')
{
socket_write (imap->sock, fmap + start, i - 1 - start);
socket_write (imap->sock, "\r\n", 2);
goto n2loop;
}
socket_write (imap->sock, fmap + start, len - start);
socket_write (imap->sock, "\r\n", 2);
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
munmap (fmap, len);
for (;;)
{
if (buffer_gets (imap->buf, &s))
return -1;
arg = next_arg (&s);
if (*arg == '*')
{
/* XXX just ignore it for now */
}
else if (atoi (arg) != (int) Tag)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: wrong tag\n");
return -1;
}
else
{
arg = next_arg (&s);
if (strcmp (arg, "OK"))
return -1;
arg = next_arg (&s);
if (*arg != '[')
break;
arg++;
if (strcasecmp ("APPENDUID", arg))
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: expected APPENDUID\n");
break;
}
arg = next_arg (&s);
if (!arg)
break;
if (atoi (arg) != (int) imap->uidvalidity)
{
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
fprintf (stderr, "IMAP error: UIDVALIDITY doesn't match APPENDUID\n");
return -1;
}
arg = next_arg (&s);
if (!arg)
break;
uid = strtol (arg, &s, 10);
if (*s != ']')
{
/* parse error */
break;
}
return uid;
}
}
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
/* didn't receive an APPENDUID */
send_server (imap->sock,
"UID SEARCH HEADER X-TUID %08lx%05lx%04x",
tv.tv_sec, tv.tv_usec, pid);
uid = 0;
for (;;)
{
if (buffer_gets (imap->buf, &s))
return -1;
arg = next_arg (&s);
if (*arg == '*')
{
arg = next_arg (&s);
if (!strcmp (arg, "SEARCH"))
{
arg = next_arg (&s);
if (!arg)
fprintf (stderr, "IMAP error: incomplete SEARCH response\n");
else
uid = atoi (arg);
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
}
}
else if (atoi (arg) != (int) Tag)
{
fprintf (stderr, "IMAP error: wrong tag\n");
return -1;
}
else
{
arg = next_arg (&s);
if (strcmp (arg, "OK"))
return -1;
return uid;
}
}
return 0;
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
bail:
munmap (fmap, len);
return -1;
}
Bunch 'o patches from Oswald Buddenhagen: i implemented some cool stuff (tm). first, the long missing "create server-side missing mailboxes". -C now creates both local and remote boxes; -L and -R create only local/remote. second, i implemented a 1:1 remote:local folder mapping (-1) with an optional INBOX exception (inbox/-I). the remote folder is specified with the folder keyword (or -F switch) and takes precedence over the namespace setting. the local directory with the mailboxes can now be specified on the command line, too (-M). another patch: - made the -1 switch settable permanently (OneToOne). after all, you usually define your mailbox layout once forever. removed -A, as it is semantically -a modified by -1. - cleaned up message output a bit. still, the quiet variable should be used throughout the program. at best, create some generic output function, which obeys a global verbosity level variable. - optimized + cleaned up configuration parser slightly - minor cleanups add an (almost) unique id to every uploaded message and search for it right after. i thought about using the message-id, but a) it is not guaranteed to be unique in a mailbox (imagine you edit a mail and store the dupe in the same box) and b) some mails (e.g., postponed) don't even have one. a downside of the current implementation is, that this id-header remains in the mailbox, but given that it wastes only 27 bytes per mail and removing it would mean several roundtrips more, this seems acceptable. i changed the line-counting loop to use a mmapped file instead of reading it in chunks, as it makes things simpler and is probably even faster for big mails. the amount of goto statements in my code may be scary, but c is simply lacking a multi-level break statement. :) this is the "shut up" patch. :) it makes the -q option consequent, so to say. additionally it adds an -l option which gathers all defined/found mailboxes and just outputs the list. don't ask what i need it for. ;)
23 years ago
int
imap_list (imap_t * imap)
{
return imap_exec (imap, "LIST \"\" \"%s*\"", global.folder);
}