From 9643e2f2149c5b76526559621452b2373b831492 Mon Sep 17 00:00:00 2001
From: Martin Edenhofer <me@edenhofer.de>
Date: Wed, 5 Aug 2015 11:40:50 +0200
Subject: [PATCH] Improved performance of html utils (do replacement in a copy
 of sting, replace it later once), added word markup removal.

---
 .../app/lib/app_post/utils.js.coffee          |  66 +++++++----
 public/assets/tests/html-utils.js             | 108 +++++++++---------
 2 files changed, 102 insertions(+), 72 deletions(-)

diff --git a/app/assets/javascripts/app/lib/app_post/utils.js.coffee b/app/assets/javascripts/app/lib/app_post/utils.js.coffee
index a3096c1bc..064d7bf6a 100644
--- a/app/assets/javascripts/app/lib/app_post/utils.js.coffee
+++ b/app/assets/javascripts/app/lib/app_post/utils.js.coffee
@@ -100,47 +100,62 @@ class App.Utils
 
   # textWithoutTags = App.Utils.htmlRemoveTags( html )
   @htmlRemoveTags: (html) ->
+    htmlTmp = $( '<div>' + html.html() + '</div>' )
+
+    # remove comments
+    @_removeComments( htmlTmp )
+
+    # remove work markup
+    htmlTmp = @_removeWordMarkup( htmlTmp )
 
     # remove tags, keep content
-    html.find('div, span, p, li, ul, ol, a, b, u, i, label, small, strong, strike, pre, code, center, blockquote, form, textarea, font, address, table, thead, tbody, tr, td, h1, h2, h3, h4, h5, h6').replaceWith( ->
+    htmlTmp.find('div, span, p, li, ul, ol, a, b, u, i, label, small, strong, strike, pre, code, center, blockquote, form, textarea, font, address, table, thead, tbody, tr, td, h1, h2, h3, h4, h5, h6').replaceWith( ->
       $(@).contents()
     )
 
     # remove tags & content
-    html.find('div, span, p, li, ul, ol, a, b, u, i, label, small, strong, strike, pre, code, center, blockquote, form, textarea, font, table, thead, tbody, tr, td, h1, h2, h3, h4, h5, h6, br, hr, img, input, select, button, style, applet, canvas, script, frame, iframe').remove()
+    htmlTmp.find('div, span, p, li, ul, ol, a, b, u, i, label, small, strong, strike, pre, code, center, blockquote, form, textarea, font, table, thead, tbody, tr, td, h1, h2, h3, h4, h5, h6, br, hr, img, input, select, button, style, applet, embed, noframes, canvas, script, frame, iframe').remove()
 
-    html
+    html.html(htmlTmp)
 
   # htmlOnlyWithRichtext = App.Utils.htmlRemoveRichtext( html )
   @htmlRemoveRichtext: (html) ->
+    htmlTmp = $( '<div>' + html.html() + '</div>' )
 
     # remove comments
-    @_removeComments( html )
+    @_removeComments( htmlTmp )
 
     # remove style and class
-    @_removeAttributes( html )
+    @_removeAttributes( htmlTmp )
+
+    # remove work markup
+    htmlTmp = @_removeWordMarkup( htmlTmp )
 
     # remove tags, keep content
-    html.find('li, ul, ol, a, b, u, i, label, small, strong, strike, pre, code, center, blockquote, form, textarea, font, address, table, thead, tbody, tr, td, h1, h2, h3, h4, h5, h6').replaceWith( ->
+    htmlTmp.find('li, ul, ol, a, b, u, i, label, small, strong, strike, pre, code, center, blockquote, form, textarea, font, address, table, thead, tbody, tr, td, h1, h2, h3, h4, h5, h6').replaceWith( ->
       $(@).contents()
     )
 
     # remove tags & content
-    html.find('li, ul, ol, a, b, u, i, label, small, strong, strike, pre, code, center, blockquote, form, textarea, font, address, table, thead, tbody, tr, td, h1, h2, h3, h4, h5, h6, hr, img, input, select, button, style, applet, canvas, script, frame, iframe').remove()
+    htmlTmp.find('li, ul, ol, a, b, u, i, label, small, strong, strike, pre, code, center, blockquote, form, textarea, font, address, table, thead, tbody, tr, td, h1, h2, h3, h4, h5, h6, hr, img, input, select, button, style, applet, embed, noframes, canvas, script, frame, iframe').remove()
 
-    html
+    html.html(htmlTmp)
 
   # cleanHtmlWithRichText = App.Utils.htmlCleanup( html )
   @htmlCleanup: (html) ->
+    htmlTmp = $( '<div>' + html.html() + '</div>' )
 
     # remove comments
-    @_removeComments( html )
+    @_removeComments( htmlTmp )
 
     # remove style and class
-    @_removeAttributes( html )
+    @_removeAttributes( htmlTmp )
+
+    # remove work markup
+    htmlTmp = @_removeWordMarkup( htmlTmp )
 
     # remove tags, keep content
-    html.find('a, font, small, time').replaceWith( ->
+    htmlTmp.find('a, font, small, time').replaceWith( ->
       $(@).contents()
     )
 
@@ -148,31 +163,32 @@ class App.Utils
     # New type of the tag
     replacementTag = 'div';
 
-    # Replace all a tags with the type of replacementTag
-    html.find('h1, h2, h3, h4, h5, h6, textarea').each( ->
+    # Replace all x tags with the type of replacementTag
+    htmlTmp.find('h1, h2, h3, h4, h5, h6, textarea').each( ->
       outer = this.outerHTML;
 
       # Replace opening tag
-      regex = new RegExp('<' + this.tagName, 'i');
-      newTag = outer.replace(regex, '<' + replacementTag);
+      regex = new RegExp('<' + this.tagName, 'i')
+      newTag = outer.replace(regex, '<' + replacementTag)
 
       # Replace closing tag
-      regex = new RegExp('</' + this.tagName, 'i');
-      newTag = newTag.replace(regex, '</' + replacementTag);
+      regex = new RegExp('</' + this.tagName, 'i')
+      newTag = newTag.replace(regex, '</' + replacementTag)
 
-      $(@).replaceWith(newTag);
+      $(@).replaceWith(newTag)
     )
 
     # remove tags & content
-    html.find('form, font, hr, img, input, select, button, style, applet, canvas, script, frame, iframe').remove()
+    htmlTmp.find('form, font, hr, img, input, select, button, style, applet, embed, noframes, canvas, script, frame, iframe').remove()
 
-    html
+    html.html(htmlTmp)
 
   @_removeAttributes: (html) ->
-    html.find('div, span, p, li, ul, ol, a, b, u, i, label, small, strong, strike, pre, code, center, blockquote, h1, h2, h3, h4, h5, h6')
+    html.find('*')
       .removeAttr( 'style' )
       .removeAttr( 'class' )
       .removeAttr( 'title' )
+      .removeAttr( 'lang' )
     html
 
   @_removeComments: (html) ->
@@ -182,6 +198,14 @@ class App.Utils
     )
     html
 
+  @_removeWordMarkup: (html) ->
+    htmlTmp = html.get(0).outerHTML
+    regex = new RegExp('<(/w|w)\:[A-Za-z]{3}>')
+    htmlTmp = htmlTmp.replace(regex, '')
+    regex = new RegExp('<(/o|o)\:[A-Za-z]{1}>')
+    htmlTmp = htmlTmp.replace(regex, '')
+    $(htmlTmp)
+
   # signatureNeeded = App.Utils.signatureCheck( message, signature )
   @signatureCheck: (message, signature) ->
     messageText   = $( '<div>' + message + '</div>' ).text().trim()
diff --git a/public/assets/tests/html-utils.js b/public/assets/tests/html-utils.js
index 18e42afc1..b603ee1b0 100644
--- a/public/assets/tests/html-utils.js
+++ b/public/assets/tests/html-utils.js
@@ -212,43 +212,47 @@ test( "htmlEscape", function() {
 test( "htmlRemoveTags", function() {
 
   var source = "<div>test</div>"
-  var should = "test"
+  var should = "<div>test</div>"
   var result = App.Utils.htmlRemoveTags( $(source) )
   equal( result.html(), should, source )
 
+  source = "<div>test<!-- some comment --></div>"
+  should = "<div>test</div>"
+  result = App.Utils.htmlRemoveTags( $(source) )
+  equal( result.html(), should, source )
+
   source = "<a href=\"some_link\">some link to somewhere</a>"
-  should = "some link to somewhere"
+  should = "<div>some link to somewhere</div>"
   result = App.Utils.htmlRemoveTags( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><a href=\"some_link\">some link to somewhere</a></div>"
-  should = "some link to somewhere"
+  should = "<div>some link to somewhere</div>"
   result = App.Utils.htmlRemoveTags( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><a href=\"some_link\">some link to somewhere</a><input value=\"should not be shown\"></div>"
-  should = "some link to somewhere"
+  should = "<div>some link to somewhere</div>"
   result = App.Utils.htmlRemoveTags( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><a href=\"some_link\">some link to somewhere</a> <div><hr></div> <span>123</span> <img src=\"some_image\"/></div>"
-  should = "some link to somewhere  123 "
+  should = "<div>some link to somewhere  123 </div>"
   result = App.Utils.htmlRemoveTags( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><form class=\"xxx\">test 123</form></div>"
-  should = "test 123"
+  should = "<div>test 123</div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><textarea class=\"xxx\">test 123</textarea></div>"
-  should = "test 123"
+  should = "<div>test 123</div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><font size=\"3\" color=\"red\">This is some text!</font></div>"
-  //should = "<div>This is some text!</div>"
-  should = "This is some text!"
+  should = "<div>This is some text!</div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 });
@@ -257,145 +261,147 @@ test( "htmlRemoveTags", function() {
 test( "htmlRemoveRichtext", function() {
 
   var source = "<div><!--test comment--><a href=\"test\">test</a></div>"
-  var should = "test"
+  var should = "<div>test</div>"
   var result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
+  source = "<div><!--[if !supportLists]--><span lang=\"DE\">1.1.1<span>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</span></span><!--[endif]--><span lang=\"DE\">Description</span></div>"
+  should = "<div><span>1.1.1<span>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</span></span><span>Description</span></div>"
+  result = App.Utils.htmlRemoveRichtext( $(source) )
+  equal( result.html(), should, source )
+
   source = "<a href=\"some_link\">some link to somewhere</a>"
-  should = "some link to somewhere"
+  should = "<div>some link to somewhere</div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><a href=\"some_link\"></a> test </div>"
-  should = " test "
+  should = "<div> test </div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><b></b> test </div>"
-  should = " test "
+  should = "<div> test </div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><div><b></b> test </div></div>"
-  should = "<div> test </div>"
+  should = "<div><div> test </div></div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><div><b></b> test <input value=\"should not be shown\"></div></div>"
-  should = "<div> test </div>"
+  should = "<div><div> test </div></div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><div><b></b> test </div><span>123</span></div>"
-  should = "<div> test </div><span>123</span>"
+  should = "<div><div> test </div><span>123</span></div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
-  source = "<div><div class=\"xxx\"><b></b> test </div></div>"
-  should = "<div> test </div>"
+  source = "<div><div class=\"xxx\" title=\"some title\" lang=\"en\"><b></b> test </div></div>"
+  should = "<div><div> test </div></div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><textarea class=\"xxx\"> test </textarea></div>"
-  //should = "<div> test </div>"
-  should = " test "
+  should = "<div> test </div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><br></div>"
-  should = "<br>"
-  result = App.Utils.htmlRemoveRichtext( $(source) )
-  equal( result.html(), should, source )
-
-  source = "<div><div class=\"xxx\"><br></div></div>"
   should = "<div><br></div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
+  source = "<div><div class=\"xxx\"><br></div></div>"
+  should = "<div><div><br></div></div>"
+  result = App.Utils.htmlRemoveRichtext( $(source) )
+  equal( result.html(), should, source )
+
   source = "<div><form class=\"xxx\">test 123</form></div>"
-  //should = "<div>test 123</div>"
-  should = "test 123"
+  should = "<div>test 123</div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><font size=\"3\" color=\"red\">This is some text!</font></div>"
-  //should = "<div>This is some text!</div>"
-  should = "This is some text!"
+  should = "<div>This is some text!</div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
-
 });
 
 // htmlCleanup
 test( "htmlCleanup", function() {
 
   var source = "<div><!--test comment--><a href=\"test\">test</a></div>"
-  var should = "test"
+  var should = "<div>test</div>"
   var result = App.Utils.htmlCleanup( $(source) )
   equal( result.html(), should, source )
 
   source = "<a href=\"some_link\">some link to somewhere</a>"
-  should = "some link to somewhere"
-  result = App.Utils.htmlCleanup( $(source) )
-  equal( result.html(), should, source )
-
-  source = "<div><h1>some link to somewhere</h1></a>"
+  //should = "some link to somewhere"
   should = "<div>some link to somewhere</div>"
   result = App.Utils.htmlCleanup( $(source) )
   equal( result.html(), should, source )
 
+  source = "<div><h1>some link to somewhere</h1></div>"
+  should = "<div><div>some link to somewhere</div></div>"
+  result = App.Utils.htmlCleanup( $(source) )
+  equal( result.html(), should, source )
+
   source = "<div><small>some link to somewhere</small></a>"
-  //should = "<div>some link to somewhere</div>"
-  should = "some link to somewhere"
+  should = "<div>some link to somewhere</div>"
   result = App.Utils.htmlCleanup( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><time>some link to somewhere</time></a>"
-  //should = "<div>some link to somewhere</div>"
-  should = "some link to somewhere"
+  should = "<div>some link to somewhere</div>"
   result = App.Utils.htmlCleanup( $(source) )
   equal( result.html(), should, source )
 
-  source = "<div><h1>some link to somewhere</h1><p><hr></p></div>"
-  should = "<div>some link to somewhere</div><p></p><p></p>"
+  source = "<div><h1>some h1 for somewhere</h1><p><hr></p></div>"
+  should = "<div><div>some h1 for somewhere</div><p></p><p></p></div>"
   result = App.Utils.htmlCleanup( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><br></div>"
-  should = "<br>"
+  should = "<div><br></div>"
   result = App.Utils.htmlCleanup( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><div class=\"xxx\"><br></div></div>"
-  should = "<div><br></div>"
+  should = "<div><div><br></div></div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><form class=\"xxx\">test 123</form></div>"
   //should = "<div>test 123<br></div>"
-  should = "test 123"
+  should = "<div>test 123</div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><form class=\"xxx\">test 123</form> some other value</div>"
-  //should = "<div>ttest 123 some other value</div>"
-  should = "test 123 some other value"
+  should = "<div>test 123 some other value</div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><form class=\"xxx\">test 123</form> some other value<input value=\"should not be shown\"></div>"
-  //should = "<div>test 123 some other value</div>"
-  should = "test 123 some other value"
+  should = "<div>test 123 some other value</div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
   source = "<div><font size=\"3\" color=\"red\">This is some text!</font></div>"
-  //should = "<div>This is some text!</div>"
-  should = "This is some text!"
+  should = "<div>This is some text!</div>"
   result = App.Utils.htmlRemoveRichtext( $(source) )
   equal( result.html(), should, source )
 
+  source = "<div><p>some link to somewhere from word<w:sdt>abc</w:sdt></p><o:p></o:p></a>"
+  should = "<div><p>some link to somewhere from wordabc</p></div>"
+  result = App.Utils.htmlCleanup( $(source) )
+  equal( result.html(), should, source )
+
 });
 
 // wrap