Merge lp:~openerp-dev/openobject-server/saas-3-opw-605315-acl into lp:~openerp/openobject-server/saas-3

[Anaël Closson (openerp)]

The html editor modifies mako instruction not inside html tags.
This keep modifies html sanitisation so that
 - comments are kept in the sanitisation process
 - they are considered as inside html tags by the html editor, hence not modified

This is particularly needed when a template generates rows in a html table, the processing instructions should be added around "<tr></tr>". Any modifications in the template before sending the e-mail break it.

[Martin Trigaux (OpenERP) (mat-openerp)]

LGMT, merged in saas-3

revno: 5117 [merge]
revision-id: <email address hidden>

review: Approve

=== modified file 'openerp/tools/mail.py'
--- openerp/tools/mail.py	2014-02-19 14:06:17 +0000
+++ openerp/tools/mail.py	2014-04-15 14:45:57 +0000
@@ -45,7 +45,7 @@
 tags_to_remove = ['html', 'body', 'font']
 
 # allow new semantic HTML5 tags
-allowed_tags = clean.defs.tags | frozenset('article section header footer hgroup nav aside figure main'.split())
+allowed_tags = clean.defs.tags | frozenset('article section header footer hgroup nav aside figure main'.split() + [etree.Comment])
 safe_attrs = clean.defs.safe_attrs | frozenset(
     ['style',
      'data-oe-model', 'data-oe-id', 'data-oe-field', 'data-oe-type', 'data-oe-expression', 'data-oe-translate', 'data-oe-nodeid',
@@ -70,6 +70,8 @@
         'forms': True,              # remove form tags
         'remove_unknown_tags': False,
         'allow_tags': allowed_tags,
+        'comments': False,
+        'processing_instructions' : False
     }
     if etree.LXML_VERSION >= (2, 3, 1):
         # kill_tags attribute has been added in version 2.3.1
@@ -339,10 +341,14 @@
     overlength_section_count = 0
     cur_char_nbr = 0
     for node in root.iter():
+        # comments do not need processing
+        if node.tag == etree.Comment:
+            continue
         # do not take into account multiple spaces that are displayed as max 1 space in html
         node_text = ' '.join((node.text and node.text.strip(' \t\r\n') or '').split())
 
         # root: try to tag the client used to write the html
+        # note: bug in node.get(value, default) for HtmlComments, default never returned
         if 'WordSection1' in node.get('class', '') or 'MsoNormal' in node.get('class', ''):
             root.set('msoffice', '1')
         if 'SkyDrivePlaceholder' in node.get('class', '') or 'SkyDrivePlaceholder' in node.get('id', ''):
@@ -657,4 +663,4 @@
                 # sometimes returns emails without at least '@'. The '@'
                 # is strictly required in RFC2822's `addr-spec`.
                 if addr[1]
-                if '@' in addr[1]]
 \ No newline at end of file
+                if '@' in addr[1]]