Add "Railway Station alias" to improve searches

config/venue_normalization.js

@@ -0,0 +1,21 @@
+
+/**
+  expansion and normalization of venue names to allow for consistent searching and more effective synonym use
+**/
+
+var mapping = {
+  'railway': {
+    'station': {
+      'alt_suffixes': ['station'],
+      'suffix': 'Railway Station',
+    }
+  },
+  'station': {
+    'light_rail': {
+      'alt_suffixes': ['station', 'light rail'],
+      'suffix': 'light rail station',
+    }
+  }
+};
+
+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');
 
@@ -24,6 +27,7 @@ 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'}) );

stream/venue_normalization.js

@@ -0,0 +1,76 @@
+
+/**
+  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 key in mapping ){
+
+        // check each mapping key against document tags
+        if( !tags.hasOwnProperty( key ) ){ continue; }
+
+        // handle regular features
+        for( var feature in mapping[key] ){
+          if( tags[key] === feature ){
+	    var rule = mapping[key][feature];
+	    addAliases( name, rule, doc );
+          }
+	}
+      }
+    }
+
+    catch( e ){
+      peliasLogger.error( 'venue normalization error' );
+      peliasLogger.error( e.stack );
+      peliasLogger.error( JSON.stringify( doc, null, 2 ) );
+    }
+
+    return next( null, doc );
+
+  });
+
+};
+
+function addAliases( name, rule, doc ) {
+  if( rule.hasOwnProperty( 'alt_suffixes' ) ){
+    for( var suffix in rule.alt_suffixes ){
+      var suffix_name = rule.alt_suffixes[suffix];
+      if( name.toLowerCase().endsWith( ' ' + suffix_name ) ){
+	name = name.slice( 0, -suffix_name.length - 1 );
+      }
+    }
+  }
+
+  if( rule.hasOwnProperty( 'suffix' ) && name.length > 0 ){
+    doc.setNameAlias( 'default', name + ' ' + rule.suffix );
+  }
+}

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);