Merge pull request pelias#1454 from pelias/libpostal_bugfixes

patch libpostal output for post-directionals
davetha · Jun 26, 2020 · 56b6bdb · 56b6bdb
2 parents a3fc0b6 + f503118
commit 56b6bdb
Show file tree

Hide file tree

Showing 2 changed files with 82 additions and 11 deletions.
diff --git a/controller/libpostal.js b/controller/libpostal.js
@@ -155,18 +155,35 @@ function patchBuggyResponses(response){
   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.
+  // known bug where the road name directional is parsed into an adjacent label, in
+  // this case we will merge it with the road label.
   // note: the bug only affects diagonals, not N,S,E,W
   // https://github.com/OpenTransitTools/trimet-mod-pelias/issues/20#issuecomment-417732128
-  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( subsequentElement && _.isString(subsequentElement.value) ){
-          response[road._pos].value += ' ' + subsequentElement.value; // merge elements
-          response.splice(road._pos+1, 1); // remove merged element
+  if (_.size(response) > 1) {
+
+    // find the road label
+    const road = _.get(idx, 'road');
+    if (_.isPlainObject(road) && _.isString(road.value)) {
+
+      // find the label which directly follows the road label
+      const next = _.nth(response, road._pos + 1);
+      if (_.isPlainObject(next) && _.isString(next.value)) {
+
+        // detect the two different occurrences of the bug
+        if (
+          // where the road label contains the directional
+          // eg. { 'label':'road', 'value':'nw' }, { 'label':'suburb', 'value':'foo st' }
+          (_.size(road.value) === 2 && DIAGONAL_DIRECTIONALS.includes(road.value.toLowerCase())) ||
+
+          // or where the following label contains the directional
+          // eg. { 'label':'road', 'value':'foo st' }, { 'label':'suburb', 'value':'nw' }
+          (_.size(next.value) === 2 && DIAGONAL_DIRECTIONALS.includes(next.value.toLowerCase()))
+        ) {
+          // concatenate the two labels
+          response[road._pos].value += ` ${next.value}`;
+
+          // remove the obsolete label
+          _.pullAt(response, road._pos + 1);
         }
       }
     }

diff --git a/test/unit/controller/libpostal.js b/test/unit/controller/libpostal.js
@@ -304,7 +304,7 @@ module.exports.tests.success_conditions = (test, common) => {
 };
 
 module.exports.tests.bug_fixes = (test, common) => {
-  test('bug fix: incorrect parsing of diagonal directionals', t => {
+  test('bug fix: incorrect parsing of diagonal directionals - pre-directional', t => {
     const service = (req, callback) => {
       const response =[
         {
@@ -357,6 +357,60 @@ module.exports.tests.bug_fixes = (test, common) => {
 
   });
 
+  test('bug fix: incorrect parsing of diagonal directionals - post-directional', t => {
+    const service = (req, callback) => {
+      // 1125 Couch St NW, Portland
+      const response = [
+        {
+          'label': 'house_number',
+          'value': '1125'
+        },
+        {
+          'label': 'road',
+          'value': 'couch st'
+        },
+        {
+          'label': 'suburb',
+          'value': 'nw'
+        },
+        {
+          'label': 'city',
+          'value': 'portland'
+        }
+      ];
+
+      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: {
+            housenumber: '1125',
+            street: 'couch st nw',
+            city: 'portland'
+          }
+        },
+        errors: []
+      }, 'req should not have been modified');
+
+      t.end();
+
+    });
+
+  });
+
   test('bug fix: incorrect parsing of diagonal directionals - no subsequent element', t => {
     const service = (req, callback) => {
       const response = [