address parser initial pass - breaks 68 tests! (ignoring tests for now)

package.json

@@ -46,7 +46,10 @@
     "morgan": "1.5.2",
     "pelias-config": "^0.1.4",
     "microtime": "1.4.0",
-    "pelias-suggester-pipeline": "2.0.2"
+    "pelias-suggester-pipeline": "2.0.2",
+    "extend": "2.0.1",
+    "parse-address": "0.0.4",
+    "addressit": "1.2.1"
   },
   "devDependencies": {
     "ciao": "^0.3.4",

query/search.js

@@ -13,36 +13,94 @@ function generate( params ){
   } 
   
   var query = queries.distance( centroid, { size: params.size } );
+  var input = params.input;
 
   if (params.bbox) {
     query = queries.bbox ( centroid, { size: params.size, bbox: params.bbox } );
   }
 
-  // add search condition to distance query
   query.query.filtered.query = {
     'bool': {
-      'must': [{ 
-          'match': {
-            'name.default': params.input
-          }
-        }
-      ]   
+      'must': [],
+      'should': []
     }
   };
 
-  if (params.input_admin) {
-    var admin_fields = ['admin0', 'admin1', 'admin1_abbr', 'admin2', 'alpha3'];
+  if (params.parsed_input) {
+
     query.query.filtered.query.bool.should = [];
 
-    admin_fields.forEach(function(admin_field) {
-      var match = {};
-      match[admin_field] = params.input_admin;
-      query.query.filtered.query.bool.should.push({
-         'match': match
+    var admin_fields = [];
+    var qb = function(admin_fields, value) {
+      admin_fields.forEach(function(admin_field) {
+        var match = {};
+        match[admin_field] = value;
+        query.query.filtered.query.bool.should.push({
+           'match': match
+        });
       });  
-    });
+    };
+
+    // update input
+    if (params.parsed_input.number && params.parsed_input.street) {
+      input = params.parsed_input.number + ' ' + params.parsed_input.street;
+    } else if (params.parsed_input.admin_parts) {
+      input = params.parsed_input.name;
+    }
+
+    // address
+    // number, street, zip
+    if (params.parsed_input.number) {
+      qb(['address.number'], params.parsed_input.number);
+    } 
+    if (params.parsed_input.street) {
+      qb(['address.street'], params.parsed_input.street);
+    } 
+    if (params.parsed_input.zip) {
+      qb(['address.zip'], params.parsed_input.zip);
+    } 
+
+    // city
+    // admin2, locality, local_admin, neighborhood
+    if (params.parsed_input.admin2) {
+      qb(['admin2'], params.parsed_input.admin2);
+    } else {
+      admin_fields.push('admin2');
+    }
+
+    // state
+    // admin1, admin1_abbr
+    if (params.parsed_input.admin1) {
+      qb(['admin1', 'admin1_abbr'], params.parsed_input.admin1);
+    } else {
+      admin_fields.push('admin1', 'admin1_abbr');
+    }
+
+    // country
+    // admin0, alpha3
+    if (params.parsed_input.admin0) {
+      qb(['admin0', 'alpha3'], params.parsed_input.admin0);
+    } else {
+      admin_fields.push('admin0', 'alpha3');
+    }
+
+    var input_regions = params.parsed_input.regions.join(' ');
+    if (admin_fields.length === 5 &&  input_regions !== params.input) {
+      if (params.parsed_input.admin_parts) {
+        qb(admin_fields, params.parsed_input.admin_parts);
+      } else {
+        qb(admin_fields, input_regions);
+      }
+    }
   }
 
+  // add search condition to distance query
+  query.query.filtered.query.bool.must.push({ 
+    'match': {
+      'name.default': input
+    }
+  });
+
   query.sort = query.sort.concat( sort( params ) );
 
   return query;

sanitiser/_input.js

@@ -1,4 +1,7 @@
 var isObject = require('is-object');
+var parser1  = require('parse-address'); // works well with US addresses
+var parser2  = require('addressit'); // freeform address parser (backup)
+var extend   = require('extend');
 
 // validate inputs, convert types and apply defaults
 function sanitize( req ){
@@ -22,14 +25,54 @@ function sanitize( req ){
   
   req.clean.input = params.input;
 
+  // naive approach
   // for admin matching during query time
   // split 'flatiron, new york, ny' into 'flatiron' and 'new york, ny'
-  var delim_index = params.input.indexOf(delim);
-  if ( delim_index !== -1 ) {
-    req.clean.input = params.input.substring(0, delim_index);
-    req.clean.input_admin = params.input.substring(delim_index + 1).trim();
+  var delimIndex = params.input.indexOf(delim);
+  var parsedAddress0 = {};
+  if ( delimIndex !== -1 ) {
+    parsedAddress0.name = params.input.substring(0, delimIndex);
+    parsedAddress0.admin_parts = params.input.substring(delimIndex + 1).trim();
   }
 
+  // address parsing
+  var parsedAddress1 = parser1.parseAddress(params.input);
+  var parsedAddress2 = parser2(params.input);
+
+  var parsedAddress  = extend(parsedAddress0, parsedAddress1, parsedAddress2);
+
+  var address_parts  =  [ 'name',
+                          'number',
+                          'street',
+                          'city',
+                          'state',
+                          'country',
+                          'zip',
+                          'regions',
+                          'admin_parts'
+                        ];
+  
+  req.clean.parsed_input = {};
+
+  address_parts.forEach(function(part){ 
+    if (parsedAddress[part]) {
+      req.clean.parsed_input[part] = parsedAddress[part];
+    }
+  });
+  
+  // req.clean.parsed_input = {
+  //   name   : parsedAddress.name,
+  //   number : parsedAddress.number,
+  //   street : parsedAddress.street,
+  //   admin2 : parsedAddress.city,
+  //   admin1 : parsedAddress.state,
+  //   admin0 : parsedAddress.country,
+  //   zip    : parsedAddress.zip,
+  //   regions: parsedAddress.regions,
+  //   admin_parts: parsedAddress.admin_parts
+  // }
+
+
   return { 'error': false };
 
 }