Add "Railway Station alias" to improve searches

config/venue_normalization.js

@@ -0,0 +1,69 @@
+
+/**
+  expansion and normalization of venue names to allow for consistent searching and more effective synonym use
+**/
+
+var mapping = [
+    {
+        'conditions': [
+            ['railway', 'station']
+        ],
+        'synonyms': [
+            'train station',
+            'station'
+        ],
+        'suffix': 'railway station'
+    },
+    {
+        'conditions': [
+            ['public_transport', 'station'],
+            ['station', 'light_rail']
+        ],
+        'synonyms': [
+            'light rail'
+        ],
+        'suffix': 'light rail station'
+    },
+    {
+        'conditions': [
+            ['public_transport', 'stop_position'],
+            ['ferry', 'yes']
+        ],
+        'synonyms': [
+            'ferry',
+            'wharf',
+        ],
+        'suffix': 'ferry terminal'
+    },
+    {
+        'conditions': [
+            ['amenity', 'parking']
+        ],
+        'synonyms': [
+            'garage',
+            'car park',
+            'parking',
+        ],
+        'suffix': 'car park'
+    }
+];
+
+/*{
+  'railway': {
+    'station': {
+      'alt_suffixes': ['station'],
+      'suffix': 'Railway Station',
+    }
+  },
+  'station': {
+    'light_rail': {
+      'alt_suffixes': ['station', 'light rail'],
+      'suffix': 'light rail station',
+    }
+  },
+  'public_transport': {
+
+  }
+};*/
+
+module.exports = mapping;

stream/importPipeline.js

@@ -1,9 +1,11 @@
 var categoryDefaults = require('../config/category_map');
+var venueNormalizations = require('../config/venue_normalization');
 
 var streams = {};
 
 streams.config = {
-  categoryDefaults: categoryDefaults
+  categoryDefaults: categoryDefaults,
+  venueNormalizations: venueNormalizations
 };
 
 streams.pbfParser = require('./multiple_pbfs').create;
@@ -13,6 +15,7 @@ streams.tagMapper = require('./tag_mapper');
 streams.adminLookup = require('pelias-wof-admin-lookup').create;
 streams.addressExtractor = require('./address_extractor');
 streams.categoryMapper = require('./category_mapper');
+streams.venueNormalization = require('./venue_normalization');
 streams.dbMapper = require('pelias-model').createDocumentMapperStream;
 streams.elasticsearch = require('pelias-dbclient');
 
@@ -24,9 +27,10 @@ streams.import = function(){
     .pipe( streams.addressExtractor() )
     .pipe( streams.blacklistStream() )
     .pipe( streams.categoryMapper( categoryDefaults ) )
+    .pipe( streams.venueNormalization( venueNormalizations ) )
     .pipe( streams.adminLookup() )
     .pipe( streams.dbMapper() )
     .pipe( streams.elasticsearch({name: 'openstreetmap'}) );
 };
 
-module.exports = streams;
+module.exports = streams;

stream/venue_normalization.js

@@ -0,0 +1,87 @@
+
+/**
+  The venue normalization is similar to the category mapper
+  It's designed to add standardized aliases for different venue types to allow easier searching'
+**/
+
+var through = require('through2');
+var peliasLogger = require( 'pelias-logger' ).get( 'openstreetmap' );
+
+module.exports = function( mapping ){
+
+  return through.obj( function( doc, enc, next ){
+
+    try {
+
+      // do not normalize addresses
+      if( doc.getType().match('address') ){
+        return next( null, doc );
+      }
+
+      // skip records with no tags
+      var tags = doc.getMeta('tags');
+      if( !tags ){
+        return next( null, doc );
+      }
+
+      var name = doc.getName( 'default' );
+      if ( !name ) {
+        name = doc.getNameAliases( 'default' );
+      }
+      if ( !name || name.legnth === 0 ){
+        return next( null, doc );
+      }
+
+      // iterate over mapping
+      for( var idx in mapping ){
+        var process = true;
+        var rule = mapping[idx];
+
+        loop_conditions:
+        for( var condition in rule.conditions ) {
+          var cond = rule.conditions[condition];
+
+          if( !tags.hasOwnProperty( cond[0] ) ) {
+            process = false;
+            break loop_conditions;
+          }
+
+          if ( cond.length === 2 && cond[1] !== tags[ cond[0] ] )
+          {
+            process = false;
+            break loop_conditions;
+          }
+        }
+
+        if (!process) {
+          continue;
+        }
+
+        var current_name = name;
+
+        if( rule.hasOwnProperty( 'synonyms' ) ) {
+          for( var synonym_idx in rule.synonyms ) {
+            var synonym = rule.synonyms[synonym_idx];
+            if( current_name.toLowerCase().endsWith( ' ' + synonym ) ) {
+              current_name = current_name.slice( 0, -synonym.length - 1 );
+            }
+          }
+        }
+
+        if( rule.hasOwnProperty( 'suffix' ) && current_name.length > 0 ){
+          doc.setNameAlias( 'default', current_name + ' ' + rule.suffix );
+        }
+      }
+    }
+
+    catch( e ){
+      peliasLogger.error( 'venue normalization error' );
+      peliasLogger.error( e.stack );
+      peliasLogger.error( JSON.stringify( doc, null, 2 ) );
+    }
+
+    return next( null, doc );
+
+  });
+
+};

test/end-to-end.js

@@ -51,6 +51,7 @@ streams.pbfParser()
   .pipe( streams.tagMapper() )
   .pipe( streams.addressExtractor() )
   .pipe( streams.categoryMapper( streams.config.categoryDefaults ) )
+  .pipe( streams.venueNormalization( streams.config.venueNormalizations ) )
   .pipe( model.createDocumentMapperStream() )
   .pipe( sink.obj(function (doc) {
     results.push(doc);