Merge pull request #1216 from pelias/libpostal_aus_unit_numbers

libpostal aus unit numbers

controller/libpostal.js

@@ -17,7 +17,8 @@ var field_mapping = {
   state_district: 'county',
   state:          'state',
   postcode:       'postalcode',
-  country:        'country'
+  country:        'country',
+  unit:           'unit',
 };
 
 // This controller calls the hosted libpostal service and converts the response
@@ -136,19 +137,37 @@ function patchBuggyResponses(response){
     });
   }
 
+  // generate an index to avoid multiple iterations over the response array
+  let idx = {};
+  response.forEach((res, pos) => idx[res.label] = _.assign({ _pos: pos }, res));
+
   // known bug where the street name is only a directional, in this case we will merge it
   // with the subsequent element.
   // note: the bug only affects diagonals, not N,S,E,W
   // https://github.com/OpenTransitTools/trimet-mod-pelias/issues/20#issuecomment-417732128
-  for( var i=0; i<response.length-1; i++ ){ // dont bother checking the last element
-    if( 'road' !== response[i].label ){ continue; }
-    if( 'string' !== typeof response[i].value ){ continue; }
-    if( 2 !== response[i].value.length ){ continue; }
-    if( DIAGONAL_DIRECTIONALS.includes( response[i].value.toLowerCase() ) ){
-      if( 'string' !== typeof response[i+1].value ){ continue; }
-      response[i].value += ' ' + response[i+1].value; // merge elements
-      response.splice(i+1, 1); // remove merged element
-      break;
+  if( response.length > 1 ){
+    let road = _.get(idx, 'road');
+    if( _.isPlainObject(road) && _.isString(road.value) && road.value.length === 2 ){
+      if( DIAGONAL_DIRECTIONALS.includes( road.value.toLowerCase() ) ){
+        let subsequentElement = response[road._pos+1];
+        if( _.isString(subsequentElement.value) ){
+          response[road._pos].value += ' ' + subsequentElement.value; // merge elements
+          response.splice(road._pos+1, 1); // remove merged element
+        }
+      }
+    }
+  }
+
+  // known bug where Australian unit numbers are incorrectly included in the house_number label
+  // note: in the case where a 'unit' label already exists, do nothing.
+  // https://github.com/pelias/pelias/issues/753
+  let unit = _.get(idx, 'unit');
+  let house_number = _.get(idx, 'house_number');
+  if( _.isPlainObject(house_number) && !_.isPlainObject(unit) && _.isString(house_number.value) ){
+    let split = _.trim(_.trim(house_number.value),'/').split('/');
+    if( split.length === 2 ){
+      response[house_number._pos].value = _.trim(split[1]); // second part (house number)
+      response.push({ label: 'unit', value: _.trim(split[0]) }); // first part (unit number)
     }
   }
 

test/unit/controller/libpostal.js

@@ -396,6 +396,250 @@ module.exports.tests.bug_fixes = (test, common) => {
 
   });
 
+  test('bug fix: correctly parse australian-style unit numbers', t => {
+    const service = (req, callback) => {
+      const response = [
+        {
+          'label': 'house_number',
+          'value': '11/1015'
+        },
+        {
+          'label': 'road',
+          'value': 'nudgee road'
+        },
+        {
+          'label': 'suburb',
+          'value': 'banyo'
+        },
+        {
+          'label': 'postcode',
+          'value': '4014'
+        },
+        {
+          'label': 'state',
+          'value': 'qld'
+        }
+      ];
+
+      callback(null, response);
+    };
+
+    const controller = libpostal(service, () => true);
+
+    const req = {
+      clean: {
+        text: 'original query'
+      },
+      errors: []
+    };
+
+    controller(req, undefined, () => {
+      t.deepEquals(req, {
+        clean: {
+          text: 'original query',
+          parser: 'libpostal',
+          parsed_text: {
+            unit: '11',
+            number: '1015',
+            street: 'nudgee road',
+            neighbourhood: 'banyo',
+            postalcode: '4014',
+            state: 'qld'
+          }
+        },
+        errors: []
+      }, 'req should not have been modified');
+
+      t.end();
+
+    });
+
+  });
+
+  test('bug fix: correctly parse australian-style unit numbers - with plus', t => {
+    const service = (req, callback) => {
+      const response = [
+        {
+          'label': 'house_number',
+          'value': '2+3/32'
+        },
+        {
+          'label': 'road',
+          'value': 'dixon street'
+        },
+        {
+          'label': 'suburb',
+          'value': 'strathpine'
+        },
+        {
+          'label': 'postcode',
+          'value': '4500'
+        },
+        {
+          'label': 'state',
+          'value': 'qld'
+        }
+      ];
+
+      callback(null, response);
+    };
+
+    const controller = libpostal(service, () => true);
+
+    const req = {
+      clean: {
+        text: 'original query'
+      },
+      errors: []
+    };
+
+    controller(req, undefined, () => {
+      t.deepEquals(req, {
+        clean: {
+          text: 'original query',
+          parser: 'libpostal',
+          parsed_text: {
+            unit: '2+3',
+            number: '32',
+            street: 'dixon street',
+            neighbourhood: 'strathpine',
+            postalcode: '4500',
+            state: 'qld'
+          }
+        },
+        errors: []
+      }, 'req should not have been modified');
+
+      t.end();
+
+    });
+
+  });
+
+  test('bug fix: correctly parse australian-style unit numbers - with unit spelled out', t => {
+    const service = (req, callback) => {
+      const response = [
+        {
+          'label': 'house_number',
+          'value': 'unit 3 /30'
+        },
+        {
+          'label': 'road',
+          'value': 'dan rees street'
+        },
+        {
+          'label': 'suburb',
+          'value': 'wallsend'
+        },
+        {
+          'label': 'postcode',
+          'value': '2287'
+        },
+        {
+          'label': 'state',
+          'value': 'nsw'
+        }
+      ];
+
+      callback(null, response);
+    };
+
+    const controller = libpostal(service, () => true);
+
+    const req = {
+      clean: {
+        text: 'original query'
+      },
+      errors: []
+    };
+
+    controller(req, undefined, () => {
+      t.deepEquals(req, {
+        clean: {
+          text: 'original query',
+          parser: 'libpostal',
+          parsed_text: {
+            unit: 'unit 3',
+            number: '30',
+            street: 'dan rees street',
+            neighbourhood: 'wallsend',
+            postalcode: '2287',
+            state: 'nsw'
+          }
+        },
+        errors: []
+      }, 'req should not have been modified');
+
+      t.end();
+
+    });
+
+  });
+
+  test('bug fix: correctly parse australian-style unit numbers - no-op if "unit" already assigned', t => {
+    const service = (req, callback) => {
+      const response = [
+        {
+          'label': 'unit',
+          'value': '99'
+        },
+        {
+          'label': 'house_number',
+          'value': '11/1015'
+        },
+        {
+          'label': 'road',
+          'value': 'nudgee road'
+        },
+        {
+          'label': 'suburb',
+          'value': 'banyo'
+        },
+        {
+          'label': 'postcode',
+          'value': '4014'
+        },
+        {
+          'label': 'state',
+          'value': 'qld'
+        }
+      ];
+
+      callback(null, response);
+    };
+
+    const controller = libpostal(service, () => true);
+
+    const req = {
+      clean: {
+        text: 'original query'
+      },
+      errors: []
+    };
+
+    controller(req, undefined, () => {
+      t.deepEquals(req, {
+        clean: {
+          text: 'original query',
+          parser: 'libpostal',
+          parsed_text: {
+            unit: '99',
+            number: '11/1015',
+            street: 'nudgee road',
+            neighbourhood: 'banyo',
+            postalcode: '4014',
+            state: 'qld'
+          }
+        },
+        errors: []
+      }, 'req should not have been modified');
+
+      t.end();
+
+    });
+
+  });
+
 };
 
 module.exports.all = (tape, common) => {