Merge 23847c4 into 0513414

CHANGES.rst

@@ -4,6 +4,10 @@ Changelog
 1.0b2 (unreleased)
 ------------------
 
+- Review text extraction. Now we have a field in the controlpanel to manage a blacklist
+  of css classes.
+  [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/browser/__init__.py

@@ -23,5 +23,7 @@ def enabled(self):
                 interface=ITextToSpeechControlPanel, name='globally_enabled')
         except InvalidParameterError:
             globally_enabled = False
+        except KeyError:
+            globally_enabled = False
 
         return globally_enabled

src/collective/texttospeech/browser/templates/texttospeech.pt

@@ -5,7 +5,8 @@
      data-label-paused="Resume"
      data-error-message="Could not load ResponsiveVoice library; Text-To-Speech feature is disabled or is not available."
      tal:attributes="data-voice view/voice;
-                     data-enabled view/enabled"
+                     data-enabled view/enabled;
+                     data-blacklist view/blacklist"
      i18n:attributes="data-label-stopped;
                       data-label-playing;
                       data-label-paused;

src/collective/texttospeech/browser/viewlets.py

@@ -2,6 +2,7 @@
 """Viewlets used on the package."""
 from collective.texttospeech.interfaces import ITextToSpeechControlPanel
 from plone import api
+from plone.api.exc import InvalidParameterError
 from plone.app.layout.viewlets.common import ViewletBase
 
 
@@ -27,3 +28,20 @@ def voice(self):
         return api.portal.get_registry_record(
             ITextToSpeechControlPanel.__identifier__ + '.voice'
         )
+
+    def blacklist(self):
+        record = dict(interface=ITextToSpeechControlPanel, name='css_class_blacklist')
+        try:
+            css_class_blacklist = api.portal.get_registry_record(**record)
+            if not css_class_blacklist:
+                css_class_blacklist = ()
+            classes = [c for c in css_class_blacklist]
+            # ignore blockquotes
+            classes.append('pullquote')
+            # ignore image captions
+            classes.append('image-caption')
+        except InvalidParameterError:
+            return ''
+        except KeyError:
+            return ''
+        return ','.join(classes)

src/collective/texttospeech/interfaces.py

@@ -38,3 +38,15 @@ class ITextToSpeechControlPanel(form.Schema):
         required=True,
         default=u'UK English Female',
     )
+
+    form.widget('css_class_blacklist', cols=25, rows=10)
+    css_class_blacklist = schema.Set(
+        title=_(u'CSS class blacklist'),
+        description=_(
+            u'A list of CSS class identifiers that Text-to-Speech will ignore. '
+            u'elements with "pullquote" or "image-caption" class directly applied to them, will be skipped.'
+        ),
+        required=False,
+        default=set([]),
+        value_type=schema.ASCIILine(title=_(u'CSS class')),
+    )

src/collective/texttospeech/static/main.js

@@ -1,12 +1,14 @@
 var MainView = (function() {
   function MainView() {
     this.$el = $('#viewlet-texttospeech');
+    this.$el.data('texttospeech', this);
     this.$button = $('#texttospeech-button', this.$el);
     this.$button.fadeIn();
     this.voice = this.$el.attr('data-voice');
     this.label_stopped = this.$el.attr('data-label-stopped');
     this.label_playing = this.$el.attr('data-label-playing');
     this.label_paused = this.$el.attr('data-label-paused');
+    this.blacklist = this.$el.attr('data-blacklist').split(',');
     this.playing = false;
     this.paused = true;
     this.$button.on('click', $.proxy(this.play_pause, this));
@@ -23,6 +25,56 @@ var MainView = (function() {
     this.$button.html(this.label_stopped);
     this.$button.attr('class', 'stopped');
   };
+  MainView.prototype.is_invisible = function($el) {
+    return $el.is(':visible') === false;
+  };
+  MainView.prototype.is_blacklisted = function($el) {
+    var i, len, selector;
+    var ignore = false;
+    for (i = 0, len = this.blacklist.length; i < len; i++) {
+      selector = '.' + this.blacklist[i];
+      if ($el.is(selector) || $el.parents(selector).length > 0) {
+        ignore = true;
+        break;
+      }
+    }
+    return ignore;
+  };
+  MainView.prototype.has_ending_punctuation = function(text) {
+    // regex test if the text end with one of these punctuations ". , ; : ! ? -"
+    return /[.,;:!?—]$/.test(text);
+  };
+  MainView.prototype.remove_extra_spaces = function(text) {
+    text = text.replace(/\s+/g, ' ');
+    text = text.trim();
+    return text
+  };
+  MainView.prototype.extract_text = function() {
+    var i, len, ref, results, $el, text;
+    // 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 = $(ref[i]);
+      if (this.is_invisible($el) || this.is_blacklisted($el)) {
+        continue;
+      }
+      text = $el.text();
+      text = this.remove_extra_spaces(text);
+      // ignore empty lines
+      if (text.length === 0) {
+        continue;
+      }
+      // ensure there is a pause after every line adding a period
+      if (!this.has_ending_punctuation(text)) {
+        text += '.';
+      }
+      results.push(text);
+    }
+    // join array with texts
+    return results.join(' ');
+  };
   MainView.prototype.play_pause = function(e) {
     e.preventDefault();
     if (this.playing) {
@@ -39,8 +91,7 @@ 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)

src/collective/texttospeech/tests/test_controlpanel.py

@@ -66,6 +66,10 @@ def test_voice_record_in_registry(self):
         self.assertTrue(hasattr(self.settings, 'voice'))
         self.assertEqual(self.settings.voice, u'UK English Female')
 
+    def test_css_class_blacklist_record_in_registry(self):
+        self.assertTrue(hasattr(self.settings, 'css_class_blacklist'))
+        self.assertEqual(self.settings.css_class_blacklist, set([]))
+
     def test_records_removed_on_uninstall(self):
         qi = self.portal['portal_quickinstaller']
         qi.uninstallProducts(products=[PROJECTNAME])
@@ -74,6 +78,7 @@ def test_records_removed_on_uninstall(self):
             ITextToSpeechControlPanel.__identifier__ + '.globally_enabled',
             ITextToSpeechControlPanel.__identifier__ + '.enabled_content_types',
             ITextToSpeechControlPanel.__identifier__ + '.voice',
+            ITextToSpeechControlPanel.__identifier__ + '.css_class_blacklist',
         ]
 
         for r in records:

src/collective/texttospeech/tests/test_upgrades.py

@@ -77,7 +77,7 @@ def test_profile_version(self):
 
     def test_registered_steps(self):
         steps = len(self.setup.listUpgrades(self.profile_id)[0])
-        self.assertEqual(steps, 3)
+        self.assertEqual(steps, 4)
 
     @unittest.skipIf(IS_PLONE_5, 'Upgrade step not supported under Plone 5')
     def test_update_library_condition(self):

src/collective/texttospeech/upgrades/v3/configure.zcml

@@ -25,6 +25,13 @@
         handler="..cook_javascript_resources"
         />
 
+    <genericsetup:upgradeDepends
+        title="Add new field to configlet"
+        description="Reload registration of configlet registry to add new field."
+        import_steps="plone.app.registry"
+        run_deps="false"
+        />
+
   </genericsetup:upgradeSteps>
 
 </configure>