Gestion des mots contractés

lib/communeHelpers.js

@@ -1,5 +1,12 @@
 const { initFields, initFormat } = require('./helpers');
 
+const abbreviations = {
+  'st': 'saint',
+  'ste': 'sainte',
+  'cgne': 'campagne',
+};
+
+
 const initCommuneFields = initFields({
   default: ['nom', 'code', 'codeDepartement', 'codeRegion', 'codesPostaux', 'population'],
   base: ['nom', 'code'],
@@ -10,4 +17,4 @@ const initCommuneFormat = initFormat({
   defaultGeometry: 'centre',
 });
 
-module.exports = { initCommuneFields, initCommuneFormat };
+module.exports = { initCommuneFields, initCommuneFormat, abbreviations };

lib/communes.js

@@ -1,10 +1,12 @@
 const SearchableCollection = require('./searchableCollection');
+const { abbreviations } = require('./communeHelpers.js');
 
 const schema = {
   nom: {
     type: 'text',
     queryWith: 'nom',
     ref: 'code',
+    replacePatern: abbreviations,
     boost: {
       population: (commune, score) => {
         if (commune.population) {

lib/searchableCollection/indexes/text.js

@@ -1,4 +1,5 @@
 const normalizeString = require('../normalizeString');
+const replaceAbbreviations = require('../replaceAbbreviations');
 const lunr = require('lunr');
 const { clone, sortBy } = require('lodash');
 
@@ -7,6 +8,7 @@ class TextIndex {
     if (!key) throw new Error('key is required');
     this._key = key;
     this._boost = options.boost || {};
+    this._replacePatern = options.replacePatern || {};
     const refKey = this._refKey = options.ref || 'id';
     this._refIndex = new Map();
     this._index = lunr(function () {
@@ -31,6 +33,9 @@ class TextIndex {
 
   find(terms, options = {}) {
     let boosted = false;
+
+    terms = replaceAbbreviations(terms, this._replacePatern);
+
     const results = this._index.search(terms)
       .map(result => {
         const item = clone(this._refIndex.get(result.ref));

lib/searchableCollection/replaceAbbreviations.js

@@ -0,0 +1,17 @@
+const removeDiacritics = require('./removeDiacritics');
+
+function replaceAbbreviations(search, abbreviations) {
+  const terms = removeDiacritics(search)
+    .toLowerCase()
+    .replace(/"/g, '')
+    .replace(/-/g, ' ')
+    .split(' ');
+
+  if (terms.length <= 1) return search;
+
+  return terms
+    .map(token => token in abbreviations ? abbreviations[token] : token)
+    .join(' ');
+}
+
+module.exports = replaceAbbreviations;

test/replaceAbbreviations.js

@@ -0,0 +1,74 @@
+/* eslint-env mocha */
+const expect = require('expect.js');
+const replaceAbbreviations = require('../lib/searchableCollection/replaceAbbreviations');
+
+describe.only('replaceAbbreviations()', function () {
+  const abbreviations = {
+    'st': 'saint',
+    'ste': 'sainte',
+    'cgne': 'campagne',
+  };
+
+  beforeEach(done => {
+    done();
+  });
+
+  describe('Words separated by spaces', function () {
+    it('should replace patern', function () {
+      const str = '"st louis"';
+      const out = 'saint louis';
+
+      expect(replaceAbbreviations(str, abbreviations)).to.equal(out);
+    });
+
+    it('should replace patern', function () {
+      const str = '"marcilly la cgne"';
+      const out = 'marcilly la campagne';
+
+      expect(replaceAbbreviations(str, abbreviations)).to.equal(out);
+    });
+  });
+
+  describe('Words separated by dashes', function () {
+    it('should replace patern', function () {
+      const str = '"st-louis"';
+      const out = 'saint louis';
+
+      expect(replaceAbbreviations(str, abbreviations)).to.equal(out);
+    });
+
+    it('should replace patern', function () {
+      const str = '"marcilly-la-cgne"';
+      const out = 'marcilly la campagne';
+
+      expect(replaceAbbreviations(str, abbreviations)).to.equal(out);
+    });
+  });
+
+  describe('search contained only one word', () => {
+    it('should not replace patern', () => {
+      const str = '"st"';
+
+      expect(replaceAbbreviations(str, abbreviations)).to.equal(str);
+    });
+  });
+
+  describe('Patern is contained in a word', () => {
+    it('should not replace patern', () => {
+      const str = '"le stinx"';
+      const out = 'le stinx';
+
+      expect(replaceAbbreviations(str, abbreviations)).to.equal(out);
+    });
+  });
+
+  describe('Accents management', () => {
+    it('sould not replace the accent with a space', () => {
+      const str = '"St-Auban-sur-l\'Ouvèze"';
+      const out = 'saint auban sur l\'ouveze';
+
+      expect(replaceAbbreviations(str, abbreviations)).to.equal(out);
+    });
+  });
+
+});