Added Check on Kanji/Hiragana... conversion. Added containsClasses op…

…tion.

README.md

@@ -388,6 +388,9 @@ are given than existing values the overcounting values are ignored.
 
 ```
 
+### containsClasses
+Array of classes to find a specific table using this class. Default is 'null/undefined'.
+
 ## Known issues and limitations
 
 This module only supports parsing basic tables with a simple horizontal set of <th></th> headings and 
@@ -464,7 +467,7 @@ tabletojson.convertUrl(url)
 
 # Issues
 
-Right now the table needs to be "well formatted" to be convertable. Tables in Html pages with not be 
+Right now the table needs to be "well formatted" to be convertable. Tables in tables with not be 
 processed.
 
 ```html

lib/tabletojson.js

@@ -18,6 +18,7 @@ class tabletojson {
      * @param options.onlyColumns <Array> Array of column indices to be used. Overrides ignoreColumn [default=null]
      * @param options.ignoreHiddenRows Ignoring hidden rows [default=true]
      * @param options.headings <Array> Array of Strings to be used as headings [default=null]
+     * @param options.headings <Array> Array of classes to find a specific table [default=null]
      * @return {Object} Converted Object as an object literal
      */
     static convert(html, options) {
@@ -32,7 +33,8 @@ class tabletojson {
                 ignoreColumns: null,
                 onlyColumns: null,
                 ignoreHiddenRows: true,
-                headings: null
+                headings: null,
+                containsClasses: null
             },
             options
         );
@@ -50,7 +52,9 @@ class tabletojson {
 
         const $ = cheerio.load(html);
 
-        $('table').each(function(i, table) {
+        const additionalSelectors = options.containsClasses ? `.${options.containsClasses.join('.')}` : '';
+
+        $(`table${additionalSelectors}`).each(function(i, table) {
             const tableAsJson = [];
             const alreadySeen = {};
             // Get column headings
@@ -149,6 +153,12 @@ class tabletojson {
      * @param arg1.stripHtmlFromCells Strip HTML from cells [default=true]
      * @param arg1.stripHtml Strip off HTML [default=null] if set true stripHtmlFromHeadings and stripHtmlFromCells will also be true
      * @param arg1.forceIndexAsNumber Force the index to be used as number [default=false]
+     * @param arg1.countDuplicateHeadings If given a _<NUMBER> will be added to the duplicate key [default=false]
+     * @param arg1.ignoreColumns <Array> Array of column indices to ignored [default=null]
+     * @param arg1.onlyColumns <Array> Array of column indices to be used. Overrides ignoreColumn [default=null]
+     * @param arg1.ignoreHiddenRows Ignoring hidden rows [default=true]
+     * @param arg1.headings <Array> Array of Strings to be used as headings [default=null]
+     * @param arg1.headings <Array> Array of classes to find a specific table [default=null]
      * @param arg1.request Options to be passed to request object
      * @param arg2 Callback function to be called when the conversion finished
      * @return {Promise<*>} Promise containing the result

package.json

@@ -6,7 +6,7 @@
   },
   "name": "tabletojson",
   "description": "Converts HTML tables to JSON objects",
-  "version": "0.9.3",
+  "version": "0.9.4",
   "main": "./lib/tabletojson.js",
   "keywords": [
     "table2json",

test/tables.html

@@ -218,5 +218,46 @@ <h2>Table #6: Table for conversion using options 'ignoreColumns' and 'onlyColumn
     </tbody>
 </table>
 
+<h2>Table #7: Table using Kanji / Hiragana</h2>
+<table>
+    <tbody>
+    <tr>
+        <th>Kanji</th>
+        <th>Hiragana</th>
+        <th>Katakana</th>
+        <th>Rōmaji</th>
+        <th>English</th>
+    </tr>
+    <tr>
+        <td><span lang="ja-Hani" title="Japanese language text">私</span></td>
+        <td><span lang="ja-Hira" title="Japanese language text">わたし</span></td>
+        <td><span lang="ja-Kana" title="Japanese language text">ワタシ</span></td>
+        <td><i>watashi</i></td>
+        <td>I, me</td>
+    </tr>
+    <tr>
+        <td><span lang="ja-Hani" title="Japanese language text">金魚</span></td>
+        <td><span lang="ja-Hira" title="Japanese language text">きんぎょ</span></td>
+        <td><span lang="ja-Kana" title="Japanese language text">キンギョ</span></td>
+        <td><i>kingyo</i></td>
+        <td>goldfish</td>
+    </tr>
+    <tr>
+        <td><span lang="ja-Hani" title="Japanese language text">煙草</span> or <span lang="ja-Hani" title="Japanese language text">莨</span></td>
+        <td><span lang="ja-Hira" title="Japanese language text">たばこ</span></td>
+        <td><span lang="ja-Kana" title="Japanese language text">タバコ</span></td>
+        <td><i>tabako</i></td>
+        <td>tobacco, cigarette</td>
+    </tr>
+    <tr>
+        <td><span lang="ja-Hani" title="Japanese language text">東京</span></td>
+        <td><span lang="ja-Hira" title="Japanese language text">とうきょう</span></td>
+        <td><span lang="ja-Kana" title="Japanese language text">トウキョウ</span></td>
+        <td><i>tōkyō</i></td>
+        <td><a href="/wiki/Tokyo" title="Tokyo">Tokyo</a>, literally meaning "eastern capital"</td>
+    </tr>
+    </tbody>
+</table>
+
 </body>
 </html>

test/test-tabletojson-local.js

@@ -479,4 +479,24 @@ describe('TableToJSON Local', function() {
 
         table[2].A.should.equal('Bill');
     });
+
+    // FIX/TEST: https://github.com/maugenst/tabletojson/issues/19
+    it('Test to check conversion and handling of Kanji, Hiragana, Katakana and latin texts', async function() {
+        const converted = await tabletojson.convert(html);
+        converted.should.be.ok();
+
+        const table = converted[6];
+
+        _.has(table[0], 'Kanji').should.be.true();
+        _.has(table[0], 'Hiragana').should.be.true();
+        _.has(table[0], 'Katakana').should.be.true();
+        _.has(table[0], 'Rōmaji').should.be.true();
+        _.has(table[0], 'English').should.be.true();
+
+        table[0]['Kanji'].should.equal('私');
+        table[0]['Hiragana'].should.equal('わたし');
+        table[0]['Katakana'].should.equal('ワタシ');
+        table[0]['Rōmaji'].should.equal('watashi');
+        table[0]['English'].should.equal('I, me');
+    });
 });

test/test-tabletojson-remote.js

@@ -2,6 +2,7 @@
 
 require('should');
 const config = require('config');
+const _ = require('lodash');
 const tabletojson = require('../lib/tabletojson');
 
 describe('TableToJSON Remote', function() {
@@ -63,6 +64,31 @@ describe('TableToJSON Remote', function() {
         });
     });
 
+    it('Get table from wikipedia containing Kanji, Hiragana, Katakana and latin texts', async function() {
+        const converted = await tabletojson.convertUrl('https://en.wikipedia.org/wiki/Japanese_writing_system', {
+            containsClasses: ['wikitable'],
+            request: {
+                proxy: config.get('request.proxy')
+            }
+        });
+
+        converted.should.be.ok();
+        const table = converted[0];
+        (table instanceof Array).should.be.true();
+
+        _.has(table[0], 'Kanji').should.be.true();
+        _.has(table[0], 'Hiragana').should.be.true();
+        _.has(table[0], 'Katakana').should.be.true();
+        _.has(table[0], 'Rōmaji').should.be.true();
+        _.has(table[0], 'English').should.be.true();
+
+        table[0]['Kanji'].should.equal('私');
+        table[0]['Hiragana'].should.equal('わたし');
+        table[0]['Katakana'].should.equal('ワタシ');
+        table[0]['Rōmaji'].should.equal('watashi');
+        table[0]['English'].should.equal('I, me');
+    });
+
     it.skip('Try to get a table from a nonexisting domain', async function() {
         try {
             await tabletojson.convertUrl('https://www.klhsfljkag.com/ydasdadad/adsaakhjg/jahsgajhvas.html');