fix(document-generator): fix for duplicate documents sharing the same ID

a very old PR introduced functionality to generate version of the same document for each hierarchy present.
this results in the document being overridden in elasticsearch with the latter.

reverts #87

Author: missinglink

src/hierarchyFinder.js

@@ -1,13 +1,3 @@
-module.exports = {};
-
-var hasName = function(r) {
-  return r.name;
-};
-
-var isDefined = function(r) {
-  return r;
-};
-
 /*
   This function finds all the WOF records associated with a hierarchy
 
@@ -26,24 +16,20 @@ var isDefined = function(r) {
   ]
 
   lastly, filter out any hierarchy elements that are undefined or w/o a name
-
 */
-function resolveHierarchy(wofRecords, hierarchy) {
-  return Object.keys(hierarchy).map(function(key) {
-    return wofRecords[hierarchy[key]];
-  }).filter(isDefined).filter(hasName);
-}
 
 /*
  This function returns all the resolved hierarchies for a wofRecord.  Each
  wofRecord can have multiple hierarchies, so resolve them by looking up the
- referenced wofRecord in the big collection of wofRecords.
+ referenced wofRecord in the big collection of parentRecords.
 */
-module.exports = function(wofRecords) {
-  return function(wofRecord) {
-    return wofRecord.hierarchies.reduce(function(resolvedHierarchies, hierarchy) {
-      resolvedHierarchies.push(resolveHierarchy(wofRecords, hierarchy));
-      return resolvedHierarchies;
-    }, []);
+module.exports = (parentRecords) => {
+  return (wofRecord) => {
+    return wofRecord.hierarchies.map(hierarchy => {
+      return Object.values(hierarchy)
+        .map(parentId => parentRecords[parentId])
+        .filter(Boolean)
+        .filter(r => r.name);
+    });
   };
 };

src/peliasDocGenerators.js

@@ -151,17 +151,15 @@ function setupDocument(record, hierarchy) {
 
 }
 
-module.exports.create = function(hierarchy_finder) {
-  return through2.obj(function(record, enc, next) {
+module.exports.create = (hierarchy_finder) => {
+  return through2.obj(function(record, _enc, next) {
     // if there are no hierarchies, then just return the doc as-is
-    var hierarchies = hierarchy_finder(record);
+    const hierarchies = hierarchy_finder(record);
 
     try {
-      if (hierarchies && hierarchies.length > 0) {
-        hierarchies.forEach(function(hierarchy) {
-          this.push(setupDocument(record, hierarchy));
-        }, this);
-
+      // only use the first hierarchy when multiple hierachies exist
+      if (Array.isArray(hierarchies) && hierarchies.length > 0) {
+        this.push(setupDocument(record, hierarchies[0]));
       } else {
         this.push(setupDocument(record));
       }

test/peliasDocGeneratorsTest.js

@@ -579,7 +579,7 @@ tape('create', function(test) {
 
   });
 
-  test.test('a document should be created for each available hierarchy', function(t) {
+  test.test('a single document should be created when multiple hierarchies exist', function(t) {
     var wofRecords = {
       1: {
         id: 1,
@@ -619,11 +619,6 @@ tape('create', function(test) {
         .setCentroid({ lat: 12.121212, lon: 21.212121 })
         .addParent( 'neighbourhood', 'neighbourhood name', '1')
         .addParent( 'country', 'country name 1', '2'),
-      new Document( 'whosonfirst', 'neighbourhood', '1')
-        .setName('default', 'neighbourhood name')
-        .setCentroid({ lat: 12.121212, lon: 21.212121 })
-        .addParent( 'neighbourhood', 'neighbourhood name', '1')
-        .addParent( 'country', 'country name 2', '3')
     ];
 
     var hierarchies_finder = function() {