From c08f8adde5fb5d4d819b0ce33904d587e3097e3c Mon Sep 17 00:00:00 2001
From: Tyler Pedelose <pedelose.1@gmail.com>
Date: Tue, 5 Dec 2017 14:31:00 -0500
Subject: [PATCH] Add source for geonames.mil data

---
 helper/type_mapping.js             | 26 +++++++++++++++++---------
 middleware/dedupe.js               |  1 +
 middleware/normalizeParentIds.js   |  6 ++++--
 sanitizer/_geonames_deprecation.js |  8 ++++++++
 sanitizer/_geonames_warnings.js    | 13 +++++++++++++
 5 files changed, 43 insertions(+), 11 deletions(-)

diff --git a/helper/type_mapping.js b/helper/type_mapping.js
index 41303583..3f23cc7f 100644
--- a/helper/type_mapping.js
+++ b/helper/type_mapping.js
@@ -16,7 +16,12 @@ function addStandardTargetsToAliases(standard, aliases) {
  */
 
 // a list of all sources
-var SOURCES = ['openstreetmap', 'openaddresses', 'geonames', 'whosonfirst'];
+var SOURCES = [
+  'openstreetmap',
+  'openaddresses',
+  'geonames',
+  'geonames-mil',
+  'whosonfirst'];
 
 /*
  * A list of alternate names for sources, mostly used to save typing
@@ -25,6 +30,7 @@ var SOURCE_ALIASES = {
   'osm': ['openstreetmap'],
   'oa': ['openaddresses'],
   'gn': ['geonames'],
+  'gnm': ['geonames-mil'],
   'wof': ['whosonfirst']
 };
 
@@ -44,14 +50,16 @@ var SOURCE_MAPPING = addStandardTargetsToAliases(SOURCES, SOURCE_ALIASES);
  * not going to match any records and will return no results.
  */
 var LAYERS_BY_SOURCE = {
- openstreetmap: [ 'address', 'venue', 'street' ],
- openaddresses: [ 'address' ],
- geonames: [ 'country','macroregion', 'region', 'county','localadmin',
-  'locality','borough', 'neighbourhood', 'venue' ],
- whosonfirst: [ 'continent', 'empire', 'country', 'dependency', 'macroregion', 'region',
-   'locality', 'localadmin', 'macrocounty', 'county', 'macrohood', 'borough',
-   'neighbourhood', 'microhood', 'disputed', 'venue', 'postalcode',
-   'continent', 'ocean', 'marinearea']
+  openstreetmap: [ 'address', 'venue', 'street' ],
+  openaddresses: [ 'address' ],
+  geonames: [ 'country', 'macroregion', 'region', 'county','localadmin',
+    'locality','borough', 'neighbourhood', 'venue' ],
+  geonames-mil: [ 'country', 'macroregion', 'region', 'county','localadmin',
+    'locality','borough', 'neighbourhood', 'venue' ],
+  whosonfirst: [ 'continent', 'empire', 'country', 'dependency', 'macroregion', 'region',
+    'locality', 'localadmin', 'macrocounty', 'county', 'macrohood', 'borough',
+    'neighbourhood', 'microhood', 'disputed', 'venue', 'postalcode',
+    'continent', 'ocean', 'marinearea']
 };
 
 /*
diff --git a/middleware/dedupe.js b/middleware/dedupe.js
index 129d157e..41d442a8 100644
--- a/middleware/dedupe.js
+++ b/middleware/dedupe.js
@@ -77,6 +77,7 @@ function isPreferred(existing, candidateReplacement) {
   var trumpsFunc = trumps.bind(null, existing, candidateReplacement);
 
   return trumpsFunc('geonames', 'whosonfirst') || // WOF has bbox and is generally preferred
+         trumpsFunc('geonames-mil', 'whosonfirst') || // WOF has bbox and is generally preferred
          trumpsFunc('openstreetmap', 'openaddresses') || // addresses are better in OA
          trumpsFunc('whosonfirst', 'openstreetmap'); // venues are better in OSM, at this time
 }
diff --git a/middleware/normalizeParentIds.js b/middleware/normalizeParentIds.js
index 67a4874a..ce87da10 100644
--- a/middleware/normalizeParentIds.js
+++ b/middleware/normalizeParentIds.js
@@ -39,12 +39,14 @@ function normalizeParentIds(place) {
         const placetype_ids = _.get(place, `${placeType}_gid`, [null]);
 
         // looking forward to the day we can remove all geonames specific hacks, but until then...
-        // geonames sometimes has its own ids in the parent hierarchy, so it's dangerous to assume that 
+        // geonames sometimes has its own ids in the parent hierarchy, so it's dangerous to assume that
         // it's always WOF ids and hardcode to that
         if (place.source === 'geonames' && place.source_id === placetype_ids[0]) {
           source = place.source;
+        } elseif (place.source === 'geonames-mil' && place.source_id === placetype_ids[0]) {
+          source = place.source; {
         }
-        
+
         place[`${placeType}_gid`] = [ makeNewId(source, placeType, placetype_ids[0]) ];
       }
     });
diff --git a/sanitizer/_geonames_deprecation.js b/sanitizer/_geonames_deprecation.js
index 2cdef1d1..31c50b8a 100644
--- a/sanitizer/_geonames_deprecation.js
+++ b/sanitizer/_geonames_deprecation.js
@@ -29,6 +29,14 @@ function _sanitize( raw, clean, opts ) {
     messages.warnings.push(coarse_reverse_message);
   }
 
+  if (_.isEqual(clean.sources, ['geonamesmil']) || _.isEqual(clean.sources, ['gnm'])) {
+    messages.errors.push(coarse_reverse_message);
+
+  } else if (_.includes(clean.sources, 'geonamesmil') || _.includes(clean.sources, 'gnm')) {
+    clean.sources = _.without(clean.sources, 'geonamesmil', 'gnm');
+    messages.warnings.push(coarse_reverse_message);
+  }
+
   return messages;
 
 }
diff --git a/sanitizer/_geonames_warnings.js b/sanitizer/_geonames_warnings.js
index 3422c4f6..7a006348 100644
--- a/sanitizer/_geonames_warnings.js
+++ b/sanitizer/_geonames_warnings.js
@@ -31,6 +31,19 @@ function _sanitize( raw, clean ){
 
   }
 
+  // the analysis is admin-only, so add errors or warnings if geonames.mil was requested
+  if (_.isEqual(clean.sources, ['geonamesmil'])) {
+    // if requested sources is only geonames, return an error
+    messages.errors.push('input contains only administrative area data, ' +
+      'no results will be returned when sources=geonamesmil');
+
+  } else if (_.includes(clean.sources, 'geonamesmil')) {
+    // if there are other sources besides geonames, return an warning
+    messages.warnings.push('input contains only administrative area data, ' +
+      'geonamesmil results will not be returned');
+
+  }
+
   return messages;
 }