Merge 1e5b338 into 3dd4a35

lib/ignore.js

@@ -51,10 +51,12 @@ module.exports = {
             ['<!\\[CDATA\\[', '\\]\\]>']
         ];
 
-        comments.forEach(function(tag) {
-            const re = new RegExp(tag[0] + '[^]*?' + tag[1], 'gi');
-            text = text.replace(re, ' ');
-        });
+        for (const tag of comments) {
+            text = text.replace(
+                new RegExp(tag[0] + '[^]*?' + tag[1], 'gi'),
+                ' '
+            );
+        }
 
         return text;
     },
@@ -66,15 +68,15 @@ module.exports = {
      * @returns {text}
      */
     tags(text, tags) {
-        const bufTags = [];
-        tags.forEach(function(tag) {
-            bufTags.push(['<' + tag + '(\\s[^>]*?)?>', '</' + tag + '>']);
-        }, this);
+        for (const name of tags) {
+            const openingTags = '<' + name + '(\\s[^>]*?)?>';
+            const closingTags = '</' + name + '>';
 
-        bufTags.forEach(function(tag) {
-            const re = new RegExp(tag[0] + '[^]*?' + tag[1], 'gi');
-            text = text.replace(re, ' ');
-        });
+            text = text.replace(
+                new RegExp(openingTags + '[^]*?' + closingTags, 'gi'),
+                ' … ' // For repeated words. Example: the `code` the.
+            );
+        }
 
         return text;
     },
@@ -92,7 +94,7 @@ module.exports = {
             data = [data];
         }
 
-        Array.isArray(data) && data.forEach(function(re) {
+        Array.isArray(data) && data.forEach(re => {
             try {
                 if (typeof re === 'string') {
                     result.push(new RegExp(re, 'g'));

lib/report.js → lib/report/index.js

@@ -17,7 +17,7 @@ const buffer = [];
 module.exports = {
     addReports(names) {
         names.forEach(function(name) {
-            var moduleName = pth.extname(name) === '.js' ? name : './report/' + name;
+            const moduleName = pth.extname(name) === '.js' ? name : './report/' + name;
             if (reportNames[moduleName]) {
                 return;
             }

lib/yaspeller.js

@@ -3,7 +3,7 @@
 const async = require('async');
 const entities = require('entities');
 const fs = require('fs');
-const eyo = require('./eyo');
+const eyo = require('./plugin/eyo');
 const formatModule = require('./format');
 const ignore = require('./ignore');
 const isutf8 = require('isutf8');
@@ -40,6 +40,8 @@ function stripTags(html) {
 function checkText(originalText, callback, settings) {
     let text = originalText;
 
+    printDebug(`Original text: ${originalText}`);
+
     const apiSettings = Object.assign({}, settings);
     const format = formatModule.getFormat(text, apiSettings);
     const lang = apiSettings.lang || 'en,ru';
@@ -70,6 +72,7 @@ function checkText(originalText, callback, settings) {
     }
 
     text = prepareText(text, format);
+    printDebug(`Prepared text for API: ${text}`);
 
     const tasks = [];
     const texts = splitText(text);
@@ -83,7 +86,7 @@ function checkText(originalText, callback, settings) {
             apiFormat: apiSettings.format,
             lang: apiSettings.lang,
             options: apiSettings.options,
-            text: el.substring(0, 128)
+            text: el.substring(0, 128),
         });
 
         tasks.push(function(cb) {
@@ -100,6 +103,9 @@ function checkText(originalText, callback, settings) {
     async.parallelLimit(tasks, getMaxRequest(apiSettings), function(err, data) {
         const buf = mergeResults(data);
 
+        printDebug('Yandex.Speller API response:');
+        printDebug(JSON.stringify(buf, null, '  '));
+
         if (!buf.err && apiSettings.checkYo) {
             checkYo(text, buf.data);
         }
@@ -189,13 +195,13 @@ function checkFile(file, callback, settings) {
     settings = settings || {};
     settings.extname = pth.extname(file);
 
-    printDebug('get: ' + file);
+    printDebug('Get file: ' + file);
 
     if (fs.existsSync(file)) {
         if (fs.statSync(file).isFile()) {
             const buf = fs.readFileSync(file);
             if (isutf8(buf)) {
-                printDebug('post text -> Yandex.Speller API: ' + file);
+                printDebug('Post text → Yandex.Speller API: ' + file);
 
                 const startTime = Date.now();
                 checkText(buf.toString(), function(err, data, originalText) {
@@ -227,7 +233,7 @@ function checkUrl(url, callback, settings) {
     settings = settings || {};
     settings.extname = pth.extname(url);
 
-    printDebug('get: ' + url);
+    printDebug('Get url: ' + url);
 
     fetch(url)
         .then(response => {
@@ -265,7 +271,7 @@ function checkSitemap(url, commonCallback, settings, callback) {
 
     const results = [];
 
-    printDebug('get: ' + url);
+    printDebug('Get sitemap: ' + url);
 
     fetch(url)
         .then(res => {

test/test.api.js

@@ -18,7 +18,20 @@ describe('API', function() {
             assert.equal(err, false);
             assert.equal(data.data.length, 2);
             done();
-        }, {lang: 'ru', format: 'plain'});
+        }, {lang: 'ru', format: 'plain',  options: { findRepeatWords: true }});
+    });
+
+    it('checkFile markdown', function(done) {
+        yaspeller.checkFile('./test/texts/repeat_words.md', function(err, data) {
+            debug.setDebug(false);
+            assert.equal(err, false);
+            assert.equal(data.data.length, 0);
+            done();
+        }, {
+            lang: 'ru',
+            ignoreTags: ['code'],
+            options: { findRepeatWords: true }
+        });
     });
 
     it('checkFile with window 1251', function(done) {
@@ -111,7 +124,7 @@ describe('API', function() {
             assert.equal(err, false);
             assert.equal(data.length, 2);
             done();
-        }, {lang: 'ru', format: 'plain'});
+        }, {lang: 'ru', format: 'plain', options: { findRepeatWords: true }});
     });
 
     it('checkText > 10000 bytes', function(done) {

test/texts/repeat_words.md

@@ -0,0 +1 @@
+the `code` the