def testHtmlToMessageWithBlockTags(self):
msg = tr_html.HtmlToMessage(
'Hello<p>Howdie<img alt="bingo" src="image.gif">', True)
result = msg.GetPresentableContent()
self.failUnless(
result == 'HelloBEGIN_PARAGRAPHHowdieBEGIN_BLOCKbingoEND_BLOCK')
msg = tr_html.HtmlToMessage(
'Hello<p>Howdie<input type="button" value="bingo">', True)
result = msg.GetPresentableContent()
self.failUnless(
result == 'HelloBEGIN_PARAGRAPHHowdieBEGIN_BLOCKbingoEND_BLOCK')
def Placeholderize(self, text):
'''Creates a tclib.Message object from 'text', attempting to recognize
a few different formats of text that can be automatically placeholderized
(HTML code, printf-style format strings, and FormatMessage-style format
strings).
'''
try:
# First try HTML placeholderizing.
# TODO(joi) Allow use of non-TotalRecall flavors of HTML placeholderizing
msg = tr_html.HtmlToMessage(text, True)
for item in msg.GetContent():
if not isinstance(item, types.StringTypes):
return msg # Contained at least one placeholder, so we're done
# HTML placeholderization didn't do anything, so try to find printf or
# FormatMessage format specifiers and change them into placeholders.
msg = tclib.Message()
parts = _FORMAT_SPECIFIER.split(text)
todo_counter = 1 # We make placeholder IDs 'TODO_0001' etc.
for part in parts:
if _FORMAT_SPECIFIER.match(part):
msg.AppendPlaceholder(
tclib.Placeholder('TODO_%04d' % todo_counter, part,
'TODO'))
todo_counter += 1
elif part != '':
msg.AppendText(part)
if self.role_model and len(
parts) > 1: # there are TODO placeholders
role_model_msg = self.role_model.UberClique(
).BestCliqueByOriginalText(msg.GetRealContent(), '')
if role_model_msg:
# replace wholesale to get placeholder names and examples
msg = role_model_msg
return msg
except:
print 'Exception processing message with text "%s"' % text
raise
def testHtmlToMessageRegressions(self):
msg = tr_html.HtmlToMessage(' - ', True)
result = msg.GetPresentableContent()
self.failUnless(result == ' - ')
def testTranslateableHtmlToMessage(self):
msg = tr_html.HtmlToMessage(
'Hello <b>[USERNAME]</b>, <how> <i>are</i> you?')
pres = msg.GetPresentableContent()
self.failUnless(pres ==
'Hello BEGIN_BOLDX_USERNAME_XEND_BOLD, '
'<how> BEGIN_ITALICareEND_ITALIC you?')
msg = tr_html.HtmlToMessage('<b>Hello</b><I>Hello</I><b>Hello</b>')
pres = msg.GetPresentableContent()
self.failUnless(pres ==
'BEGIN_BOLD_1HelloEND_BOLD_1BEGIN_ITALICHelloEND_ITALIC'
'BEGIN_BOLD_2HelloEND_BOLD_2')
# Check that nesting (of the <font> tags) is handled correctly - i.e. that
# the closing placeholder numbers match the opening placeholders.
msg = tr_html.HtmlToMessage(
'''<font size=-1><font color=#FF0000>Update!</font> '''
'''<a href='http://desktop.google.com/whatsnew.html?hl=[$~LANG~$]'>'''
'''New Features</a>: Now search PDFs, MP3s, Firefox web history, and '''
'''more</font>''')
pres = msg.GetPresentableContent()
self.failUnless(pres ==
'BEGIN_FONT_1BEGIN_FONT_2Update!END_FONT_2 BEGIN_LINK'
'New FeaturesEND_LINK: Now search PDFs, MP3s, Firefox '
'web history, and moreEND_FONT_1')
msg = tr_html.HtmlToMessage('''<a href='[$~URL~$]'><b>[NUM][CAT]</b></a>''')
pres = msg.GetPresentableContent()
self.failUnless(pres == 'BEGIN_LINKBEGIN_BOLDX_NUM_XX_CAT_XEND_BOLDEND_LINK')
msg = tr_html.HtmlToMessage(
'''<font size=-1><a class=q onClick='return window.qs?qs(this):1' '''
'''href='http://[WEBSERVER][SEARCH_URI]'>Desktop</a></font> '''
''' ''')
pres = msg.GetPresentableContent()
self.failUnless(pres ==
'''BEGIN_FONTBEGIN_LINKDesktopEND_LINKEND_FONTSPACE''')
msg = tr_html.HtmlToMessage(
'''<br><br><center><font size=-2>©2005 Google </font></center>''', 1)
pres = msg.GetPresentableContent()
self.failUnless(pres ==
u'BEGIN_BREAK_1BEGIN_BREAK_2BEGIN_CENTERBEGIN_FONT\xa92005'
u' Google END_FONTEND_CENTER')
msg = tr_html.HtmlToMessage(
''' - <a class=c href=[$~CACHE~$]>Cached</a>''')
pres = msg.GetPresentableContent()
self.failUnless(pres ==
' - BEGIN_LINKCachedEND_LINK')
# Check that upper-case tags are handled correctly.
msg = tr_html.HtmlToMessage(
'''You can read the <A HREF='http://desktop.google.com/privacypolicy.'''
'''html?hl=[LANG_CODE]'>Privacy Policy</A> and <A HREF='http://desktop'''
'''.google.com/privacyfaq.html?hl=[LANG_CODE]'>Privacy FAQ</A> online.''')
pres = msg.GetPresentableContent()
self.failUnless(pres ==
'You can read the BEGIN_LINK_1Privacy PolicyEND_LINK_1 and '
'BEGIN_LINK_2Privacy FAQEND_LINK_2 online.')
# Check that tags with linebreaks immediately preceding them are handled
# correctly.
msg = tr_html.HtmlToMessage(
'''You can read the
<A HREF='http://desktop.google.com/privacypolicy.html?hl=[LANG_CODE]'>Privacy Policy</A>
and <A HREF='http://desktop.google.com/privacyfaq.html?hl=[LANG_CODE]'>Privacy FAQ</A> online.''')
pres = msg.GetPresentableContent()
self.failUnless(pres == '''You can read the
BEGIN_LINK_1Privacy PolicyEND_LINK_1
and BEGIN_LINK_2Privacy FAQEND_LINK_2 online.''')
