autocomplete: extend additional name fields used in multimatch queries

package.json

@@ -55,7 +55,7 @@
     "pelias-microservice-wrapper": "^1.10.0",
     "pelias-model": "^9.0.0",
     "pelias-parser": "2.2.0",
-    "pelias-query": "^11.0.0",
+    "pelias-query": "github:pelias/query#additional-name-fields",
     "pelias-sorting": "^1.2.0",
     "predicates": "^2.0.0",
     "regenerate": "^1.4.0",

query/search_pelias_parser.js

@@ -5,7 +5,10 @@ const textParser = require('./text_parser_pelias');
 const config = require('pelias-config').generate().api;
 
 var placeTypes = require('../helper/placeTypes');
-var views = { custom_boosts: require('./view/boost_sources_and_layers') };
+var views = {
+  custom_boosts: require('./view/boost_sources_and_layers'),
+  query_multi_match: require('./view/query_multi_match')
+};
 
 // region_a is also an admin field which can be identified by
 // the pelias_parser. this functionality was inherited from the
@@ -18,10 +21,10 @@ var adminFields = placeTypes.concat(['region_a']);
 var query = new peliasQuery.layout.FilteredBooleanQuery();
 
 // mandatory matches
-query.score( peliasQuery.view.leaf.match('main'), 'must' );
+query.score( views.query_multi_match('match', 'main'), 'must' );
 
 // scoring boost
-const phrase_view = peliasQuery.view.leaf.match_phrase('main');
+const phrase_view = views.query_multi_match('match_phrase', 'main', 'phrase');
 
 query.score( phrase_view );
 query.score( peliasQuery.view.focus( peliasQuery.view.leaf.match_all ) );

query/view/helper.js

@@ -1,5 +1,12 @@
-function toMultiFields(baseField, suffix) {
-  return [baseField, toSingleField(baseField, suffix)];
+const _ = require('lodash');
+
+function toMultiFields(baseField, ...suffix) {
+  return [
+    baseField,
+    ...suffix
+      .filter(e => _.isString(e) && !_.isEmpty(e))
+      .map(suffix => toSingleField(baseField, suffix))
+  ];
 }
 
 function toSingleField(baseField, suffix) {
@@ -9,4 +16,12 @@ function toSingleField(baseField, suffix) {
   return parts.join('.');
 }
 
-module.exports = { toMultiFields, toSingleField };
+function toMultiFieldsWithWildcards(baseField, ...suffix) {
+  const result = [];
+  toMultiFields(baseField, ...suffix).forEach(field => {
+    result.push(field, `${field}_*`);
+  });
+  return result;
+}
+
+module.exports = { toMultiFields, toSingleField, toMultiFieldsWithWildcards };

query/view/ngrams_strict.js

@@ -1,5 +1,5 @@
 const peliasQuery = require('pelias-query');
-const toMultiFields = require('./helper').toMultiFields;
+const toMultiFieldsWithWildcards = require('./helper').toMultiFieldsWithWildcards;
 
 /**
   Ngrams view with the additional properties to enable:
@@ -16,7 +16,10 @@ module.exports = function( vs ){
   }
 
   vs.var('multi_match:ngrams_strict:input', vs.var('input:name').get());
-  vs.var('multi_match:ngrams_strict:fields', toMultiFields(vs.var('ngram:field').get(), vs.var('lang').get()));
+  vs.var('multi_match:ngrams_strict:fields', toMultiFieldsWithWildcards(
+    vs.var('ngram:field').get(),
+    vs.var('lang').get(),
+  ));
 
   vs.var('multi_match:ngrams_strict:analyzer', vs.var('ngram:analyzer').get());
   vs.var('multi_match:ngrams_strict:slop', vs.var('phrase:slop').get());

query/view/phrase_first_tokens_only.js

@@ -1,5 +1,5 @@
 const peliasQuery = require('pelias-query');
-const toMultiFields = require('./helper').toMultiFields;
+const toMultiFieldsWithWildcards = require('./helper').toMultiFieldsWithWildcards;
 
 /**
   Phrase view which trims the 'input:name' and uses ALL BUT the last token.
@@ -18,7 +18,10 @@ module.exports = function( vs ){
 
   // set the 'input' variable to all but the last token
   vs.var(`multi_match:${view_name}:input`).set( tokens.join(' ') );
-  vs.var(`multi_match:${view_name}:fields`).set(toMultiFields(vs.var('phrase:field').get(), vs.var('lang').get()));
+  vs.var(`multi_match:${view_name}:fields`, toMultiFieldsWithWildcards(
+    vs.var('phrase:field').get(),
+    vs.var('lang').get(),
+  ));
 
   vs.var(`multi_match:${view_name}:analyzer`).set(vs.var('phrase:analyzer').get());
   vs.var(`multi_match:${view_name}:boost`).set(vs.var('phrase:boost').get());

query/view/query_multi_match.js

@@ -0,0 +1,52 @@
+const _ = require('lodash');
+const toMultiFieldsWithWildcards = require('./helper').toMultiFieldsWithWildcards;
+
+const optional_params = [
+  'boost',
+  'slop',
+  'operator',
+  'analyzer',
+  'cutoff_frequency',
+  'fuzziness',
+  'max_expansions',
+  'prefix_length',
+  'fuzzy_transpositions',
+  'minimum_should_match',
+  'zero_terms_query'
+];
+
+module.exports = (namespace, prefix, type) => {
+  return (vs) => {
+    const input_variable = `${namespace}:${prefix}:input`;
+    const field_variable = `${namespace}:${prefix}:field`;
+
+    if (!vs.isset(input_variable) || !vs.isset(field_variable)) {
+      return null;
+    }
+
+    const query = {
+      multi_match: {
+        query: vs.var(input_variable),
+        fields: toMultiFieldsWithWildcards(
+          vs.var(field_variable).get(),
+          vs.var('lang').get(),
+        )
+      }
+    };
+
+    // optional 'type'
+    if (_.isString(type)) {
+      query.multi_match.type = type;
+    }
+
+    // other optional params
+    optional_params.forEach((param) => {
+      const variable_name = `${namespace}:${prefix}:${param}`;
+      if (vs.isset(variable_name)) {
+        query.multi_match[param] = vs.var(variable_name);
+      }
+    });
+
+    return query;
+  };
+};

test/unit/fixture/autocomplete_boundary_country.js

@@ -5,7 +5,12 @@ module.exports = {
         'constant_score': {
           'filter': {
             'multi_match': {
-              'fields': ['name.default', 'name.en'],
+              'fields': [
+                'name.default',
+                'name.default_*',
+                'name.en',
+                'name.en_*'
+              ],
               'analyzer': 'peliasQuery',
               'query': 'test',
               'boost': 100,

test/unit/fixture/autocomplete_boundary_gid.js

@@ -5,7 +5,12 @@ module.exports = {
         'constant_score': {
           'filter': {
             'multi_match': {
-              'fields': ['name.default', 'name.en'],
+              'fields': [
+                'name.default',
+                'name.default_*',
+                'name.en',
+                'name.en_*'
+              ],
               'analyzer': 'peliasQuery',
               'query': 'test',
               'boost': 100,

test/unit/fixture/autocomplete_custom_boosts.json

@@ -6,7 +6,12 @@
         "must": [
           {
             "multi_match": {
-              "fields": ["phrase.default", "phrase.en"],
+              "fields": [
+                "phrase.default",
+                "phrase.default_*",
+                "phrase.en",
+                "phrase.en_*"
+              ],
               "analyzer": "peliasQuery",
               "query": "foo",
               "boost": 1,

test/unit/fixture/autocomplete_linguistic_bbox_san_francisco.js

@@ -5,7 +5,12 @@ module.exports = {
         'constant_score': {
           'filter': {
             'multi_match': {
-              'fields': ['name.default', 'name.en'],
+              'fields': [
+                'name.default',
+                'name.default_*',
+                'name.en',
+                'name.en_*'
+              ],
               'analyzer': 'peliasQuery',
               'query': 'test',
               'boost': 100,

test/unit/fixture/autocomplete_linguistic_circle_san_francisco.js

@@ -6,7 +6,12 @@ module.exports = {
           'constant_score': {
             'filter': {
               'multi_match': {
-                'fields': ['name.default', 'name.en'],
+                'fields': [
+                  'name.default',
+                  'name.default_*',
+                  'name.en',
+                  'name.en_*'
+                ],
                 'analyzer': 'peliasQuery',
                 'query': 'test',
                 'boost': 100,

test/unit/fixture/autocomplete_linguistic_final_token.js

@@ -3,7 +3,12 @@ module.exports = {
     'bool': {
       'must': [{
         'multi_match': {
-          'fields': ['phrase.default', 'phrase.en'],
+          'fields': [
+            'phrase.default',
+            'phrase.default_*',
+            'phrase.en',
+            'phrase.en_*'
+          ],
           'analyzer': 'peliasQuery',
           'query': 'one',
           'boost': 1,

test/unit/fixture/autocomplete_linguistic_focus.js

@@ -5,7 +5,12 @@ module.exports = {
         'constant_score': {
           'filter': {
             'multi_match': {
-              'fields': ['name.default', 'name.en'],
+              'fields': [
+                'name.default',
+                'name.default_*',
+                'name.en',
+                'name.en_*'
+              ],
               'analyzer': 'peliasQuery',
               'query': 'test',
               'boost': 100,

test/unit/fixture/autocomplete_linguistic_focus_null_island.js

@@ -5,7 +5,12 @@ module.exports = {
         'constant_score': {
           'filter': {
             'multi_match': {
-              'fields': ['name.default', 'name.en'],
+              'fields': [
+                'name.default',
+                'name.default_*',
+                'name.en',
+                'name.en_*'
+              ],
               'analyzer': 'peliasQuery',
               'query': 'test',
               'boost': 100,

test/unit/fixture/autocomplete_linguistic_multiple_tokens.js

@@ -3,7 +3,12 @@ module.exports = {
     'bool': {
       'must': [{
         'multi_match': {
-          'fields': ['phrase.default', 'phrase.en'],
+          'fields': [
+            'phrase.default',
+            'phrase.default_*',
+            'phrase.en',
+            'phrase.en_*'
+          ],
           'analyzer': 'peliasQuery',
           'query': 'one two',
           'boost': 1,

test/unit/fixture/autocomplete_linguistic_multiple_tokens_complete_numeric.js

@@ -3,7 +3,12 @@ module.exports = {
     'bool': {
       'must': [{
         'multi_match': {
-          'fields': ['phrase.default', 'phrase.en'],
+          'fields': [
+            'phrase.default',
+            'phrase.default_*',
+            'phrase.en',
+            'phrase.en_*'
+          ],
           'analyzer': 'peliasQuery',
           'type': 'phrase',
           'slop': 3,
@@ -15,7 +20,12 @@ module.exports = {
         'constant_score': {
           'filter': {
             'multi_match': {
-              'fields': ['name.default', 'name.en'],
+              'fields': [
+                'name.default',
+                'name.default_*',
+                'name.en',
+                'name.en_*'
+              ],
               'analyzer': 'peliasQuery',
               'query': 'three',
               'boost': 100,

test/unit/fixture/autocomplete_linguistic_one_char_token.js

@@ -5,7 +5,12 @@ module.exports = {
         'constant_score': {
           'filter': {
             'multi_match': {
-              'fields': ['name.default', 'name.en'],
+              'fields': [
+                'name.default',
+                'name.default_*',
+                'name.en',
+                'name.en_*'
+              ],
               'analyzer': 'peliasQuery',
               'query': 't',
               'boost': 100,

test/unit/fixture/autocomplete_linguistic_only.js

@@ -5,7 +5,12 @@ module.exports = {
         'constant_score': {
           'filter': {
             'multi_match': {
-              'fields': ['name.default', 'name.en'],
+              'fields': [
+                'name.default',
+                'name.default_*',
+                'name.en',
+                'name.en_*'
+              ],
               'analyzer': 'peliasQuery',
               'query': 'test',
               'boost': 100,

test/unit/fixture/autocomplete_linguistic_three_char_token.js

@@ -5,7 +5,12 @@ module.exports = {
         'constant_score': {
           'filter': {
             'multi_match': {
-              'fields': ['name.default', 'name.en'],
+              'fields': [
+                'name.default',
+                'name.default_*',
+                'name.en',
+                'name.en_*'
+              ],
               'analyzer': 'peliasQuery',
               'query': 'tes',
               'boost': 100,

test/unit/fixture/autocomplete_linguistic_two_char_token.js

@@ -5,7 +5,12 @@ module.exports = {
         'constant_score': {
           'filter': {
             'multi_match': {
-              'fields': ['name.default', 'name.en'],
+              'fields': [
+                'name.default',
+                'name.default_*',
+                'name.en',
+                'name.en_*'
+              ],
               'analyzer': 'peliasQuery',
               'query': 'te',
               'boost': 100,

test/unit/fixture/autocomplete_linguistic_with_admin.js

@@ -4,7 +4,12 @@ module.exports = {
       'must': [
         {
           'multi_match': {
-            'fields': ['phrase.default', 'phrase.en'],
+            'fields': [
+              'phrase.default',
+              'phrase.default_*',
+              'phrase.en',
+              'phrase.en_*'
+            ],
             'analyzer': 'peliasQuery',
             'type': 'phrase',
             'slop': 3,

test/unit/fixture/autocomplete_single_character_street.js

@@ -3,7 +3,12 @@ module.exports = {
     'bool': {
       'must': [{
         'multi_match': {
-          'fields': ['phrase.default', 'phrase.en'],
+          'fields': [
+            'phrase.default',
+            'phrase.default_*',
+            'phrase.en',
+            'phrase.en_*'
+          ],
           'analyzer': 'peliasQuery',
           'query': 'k road',
           'boost': 1,