libpostal changes

- updated for libpostal interface
- added tests for text_parser

query/text_parser.js

@@ -1,42 +1,18 @@
-
 var logger = require('pelias-logger').get('api');
-var placeTypes = require('../helper/placeTypes');
-
-/*
-This list should only contain admin fields we are comfortable matching in the case
-when we can't identify parts of an address. This shouldn't contain fields like country_a
-or postalcode because we should only try to match those when we're sure that's what they are.
- */
-var adminFields = placeTypes.concat([
-  'region_a'
-]);
-
-/**
-  @todo: refactor me
-**/
 
 // all the address parsing logic
 function addParsedVariablesToQueryVariables( parsed_text, vs ){
+  // ==== add parsed matches [address components] ====
 
-  // is it a street address?
+  // query - Mexitaly, Sunoco, Lowes
-  // var isStreetAddress = parsed_text.hasOwnProperty('number') && parsed_text.hasOwnProperty('street');
+  if (parsed_text.hasOwnProperty('query')) {
-  // if( isStreetAddress ){
+    vs.var('input:query', parsed_text.query);
-  //   vs.var( 'input:name', parsed_text.number + ' ' + parsed_text.street );
+  }
-  // }
-  //
-  // // ?
-  // else if( parsed_text.admin_parts ) {
-  //   vs.var( 'input:name', parsed_text.name );
-  // }
-  //
-  // // ?
-  // else {
-  //   logger.warn( 'chaos monkey asks: what happens now?' );
-  //   logger.warn( parsed_text );
-  //   try{ throw new Error(); } catch(e){ logger.warn( e.stack ); } // print a stack trace
-  // }
 
-  // ==== add parsed matches [address components] ====
+  // categories - restaurants, hotels, bars
+  if (parsed_text.hasOwnProperty('category')) {
+    vs.var('input:category', parsed_text.category);
+  }
 
   // house number
   if( parsed_text.hasOwnProperty('number') ){
@@ -48,6 +24,16 @@ function addParsedVariablesToQueryVariables( parsed_text, vs ){
     vs.var( 'input:street', parsed_text.street );
   }
 
+  // neighbourhood
+  if (parsed_text.hasOwnProperty('neighbourhood')) {
+    vs.var( 'input:neighbourhood', parsed_text.neighbourhood);
+  }
+
+  // borough
+  if (parsed_text.hasOwnProperty('borough')) {
+    vs.var( 'input:borough', parsed_text.borough);
+  }
+
   // postal code
   if( parsed_text.hasOwnProperty('postalcode') ){
     vs.var( 'input:postcode', parsed_text.postalcode );
@@ -57,43 +43,24 @@ function addParsedVariablesToQueryVariables( parsed_text, vs ){
 
   // city
   if( parsed_text.hasOwnProperty('city') ){
-    vs.var( 'input:county', parsed_text.city );
+    vs.var( 'input:locality', parsed_text.city );
+  }
+
+  // county
+  if( parsed_text.hasOwnProperty('county') ){
+    vs.var( 'input:county', parsed_text.county );
   }
 
   // state
   if( parsed_text.hasOwnProperty('state') ){
-    vs.var( 'input:region_a', parsed_text.state );
+    vs.var( 'input:region', parsed_text.state );
   }
 
   // country
   if( parsed_text.hasOwnProperty('country') ){
-    vs.var( 'input:country_a', parsed_text.country );
+    vs.var( 'input:country', parsed_text.country );
   }
 
-  // ==== deal with the 'leftover' components ====
-  // @todo: clean up this code
-
-  // a concept called 'leftovers' which is just 'admin_parts' /or 'regions'.
-  // var leftoversString = '';
-  // if( parsed_text.hasOwnProperty('admin_parts') ){
-  //   leftoversString = parsed_text.admin_parts;
-  // }
-  // else if( parsed_text.hasOwnProperty('regions') ){
-  //   leftoversString = parsed_text.regions.join(' ');
-  // }
-  //
-  // // if we have 'leftovers' then assign them to any fields which
-  // // currently don't have a value assigned.
-  // if( leftoversString.length ){
-  //
-  //   // cycle through fields and set fields which
-  //   // are still currently unset
-  //   adminFields.forEach( function( key ){
-  //     if( !vs.isset( 'input:' + key ) ){
-  //       vs.var( 'input:' + key, leftoversString );
-  //     }
-  //   });
-  // }
 }
 
 module.exports = addParsedVariablesToQueryVariables;

test/unit/query/text_parser.js

@@ -0,0 +1,78 @@
+var VariableStore = require('pelias-query').Vars;
+var text_parser = require('../../../query/text_parser');
+
+module.exports.tests = {};
+
+module.exports.tests.interface = function(test, common) {
+  test('valid interface', function(t) {
+    t.equal(typeof text_parser, 'function', 'valid function');
+    t.end();
+  });
+};
+
+module.exports.tests.query = function(test, common) {
+  test('parsed_text without properties should leave vs properties unset', function(t) {
+    var parsed_text = {};
+    var vs = new VariableStore();
+
+    text_parser(parsed_text, vs);
+
+    t.false(vs.isset('input:query'));
+    t.false(vs.isset('input:category'));
+    t.false(vs.isset('input:housenumber'));
+    t.false(vs.isset('input:street'));
+    t.false(vs.isset('input:neighbourhood'));
+    t.false(vs.isset('input:borough'));
+    t.false(vs.isset('input:postcode'));
+    t.false(vs.isset('input:locality'));
+    t.false(vs.isset('input:county'));
+    t.false(vs.isset('input:region'));
+    t.false(vs.isset('input:country'));
+    t.end();
+
+  });
+
+  test('parsed_text without properties should leave vs properties unset', function(t) {
+    var parsed_text = {
+      query: 'query value',
+      category: 'category value',
+      number: 'number value',
+      street: 'street value',
+      neighbourhood: 'neighbourhood value',
+      borough: 'borough value',
+      postalcode: 'postalcode value',
+      city: 'city value',
+      county: 'county value',
+      state: 'state value',
+      country: 'country value'
+    };
+    var vs = new VariableStore();
+
+    text_parser(parsed_text, vs);
+
+    t.equals(vs.var('input:query').toString(), 'query value');
+    t.equals(vs.var('input:category').toString(), 'category value');
+    t.equals(vs.var('input:housenumber').toString(), 'number value');
+    t.equals(vs.var('input:street').toString(), 'street value');
+    t.equals(vs.var('input:neighbourhood').toString(), 'neighbourhood value');
+    t.equals(vs.var('input:borough').toString(), 'borough value');
+    t.equals(vs.var('input:postcode').toString(), 'postalcode value');
+    t.equals(vs.var('input:locality').toString(), 'city value');
+    t.equals(vs.var('input:county').toString(), 'county value');
+    t.equals(vs.var('input:region').toString(), 'state value');
+    t.equals(vs.var('input:country').toString(), 'country value');
+    t.end();
+
+  });
+
+};
+
+module.exports.all = function (tape, common) {
+  function test(name, testFunction) {
+    return tape('text_parser ' + name, testFunction);
+  }
+
+  for( var testCase in module.exports.tests ){
+    module.exports.tests[testCase](test, common);
+  }
+};