Merge 749c3f2 into 0513414

CHANGES.rst

@@ -4,6 +4,9 @@ Changelog
 1.0b2 (unreleased)
 ------------------
 
+- Review text extraction.
+  [rodfersou]
+
 - To avoid displaying the 'Listen' button with an incorrect voice,
   the feature is now globally disabled by default at installation time.
   [hvelarde]

README.rst

@@ -94,3 +94,18 @@ A viewlet with a 'Listen' button will be displayed on objects with the feature e
     The Text-To-Speech feature enabled.
 
 You can pause/resume the reader at any time by selecting 'Pause'/'Resume'.
+
+How does it work
+----------------
+
+Text to speech extract the text of the page, and send the text to `ResponsiveVoice <http://responsivevoice.org/>`_ to play the text.
+
+By default, the text extraction need to ignore some elements that should not be played:
+
+* Blockquotes
+* IFrames
+* Image captions
+
+And there are other custom elements that should be ignored, to manage these custom elements we use a blacklist.
+
+The blacklist is filled into the package configlet and should be filled with one CSS selector by line that should be ignored.

src/collective/texttospeech/static/main.js

@@ -23,6 +23,52 @@ var MainView = (function() {
     this.$button.html(this.label_stopped);
     this.$button.attr('class', 'stopped');
   };
+  MainView.prototype.extract_text = function() {
+    var i, j, len, len2, ref, results, $el, text, selector, ignore;
+    // blacklist example
+    var blacklist = ['.slideshow-link'];
+    // create an array with the text extracted
+    results = [];
+    // http://stackoverflow.com/questions/4602431/what-is-the-most-efficient-way-to-get-leaf-nodes-with-jquery
+    ref = $('#content *:not(:has(*))');
+    for (i = 0, len = ref.length; i < len; i++) {
+      $el = $(ref[i]);
+      // ignore blacklist
+      ignore = false;
+      for (j = 0, len2 = blacklist.length; j < len2; j++) {
+        selector = blacklist[j];
+        if ($el.is(selector) || $el.parents(selector).length > 0) {
+          ignore = true;
+          break;
+        }
+      }
+      if (ignore) {
+        continue;
+      }
+      // ignore blockquotes
+      if ($el.is('blockquote') || $el.parents('blockquote').length > 0) {
+        continue;
+      }
+      // ignore image captions
+      if ($el.hasClass('image-caption')) {
+        continue;
+      }
+      // Remove extra spaces
+      text = $el.text().replace(/\s+/g, ' ').trim();
+      // ignore invisible elements or empty lines
+      if (!$el.is(':visible') || text.length === 0) {
+        continue;
+      }
+      // ensure there is a pause after every line adding a period
+      if (!/[.,;:!?—]$/.test(text)) {
+        text += '.';
+      }
+      results.push(text);
+    }
+    // join array with texts
+    console.log(results);
+    return results.join(' ');
+  };
   MainView.prototype.play_pause = function(e) {
     e.preventDefault();
     if (this.playing) {
@@ -39,8 +85,7 @@ var MainView = (function() {
       }
     } else {
       responsiveVoice.speak(
-        // remove spaces to avoid issues with some Firefox versions
-        $('#content').text().replace(/\s+/g, ' ').trim(),
+        this.extract_text(),
         this.voice, {
           onstart: $.proxy(this.onstart, this),
           onend: $.proxy(this.onend, this)