searching 'ticket:1,2' shows a blank page with UnicodeDecodeError

[matobaa <matobaa+trac-hacks@…>]

click search on mainnav, input ticket:1,2 in a textbox then click 'search' shows a blank page with log follows:

  File "c:\trac\1.4\.venv\lib\site-packages\trac\templates\theme.html", line 128, in block "body"
    # block content
  File "c:\trac\1.4\.venv\lib\site-packages\trac\search\templates\search.html", line 79, in block "content"
    _('Quickjump to %(name)s', name=quickjump.name)|safe}</a>
  File "c:\trac\1.4\.venv\lib\site-packages\jinja2\filters.py", line 887, in do_mark_safe
    return Markup(value)
  File "c:\trac\1.4\.venv\lib\site-packages\markupsafe\__init__.py", line 74, in __new__
    return text_type.__new__(cls, base)
UnicodeDecodeError: 'ascii' codec can't decode byte 0xe2 in position 26: ordinal not in range(128)

System Information
User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/76.0.3809.132 Safari/537.36

Package	Version
Trac	1.4
Genshi	(not installed, some old plugins may not work as expected)
Jinja2	2.10.1
pysqlite	2.6.0
Python	2.7.16 (v2.7.16:413a49145e, Mar 4 2019, 01:37:19) [MSC v.1500 64 bit (AMD64)]
setuptools	41.2.0
SQLite	3.14.2
jQuery	1.12.4
jQuery UI	1.12.1
jQuery Timepicker	1.6.3


Configuration

database	sqlite:db/trac.db

[Jun Omae]

Good catch! Thanks.

Reproduced it if Babel is not installed. Unable to reproduce if Babel is installed. Also, visiting search:?q=ticket:1,2&noquickjump=1 link would directly lead the issue.

[Jun Omae]

We should use tag_ rather than _ because quickjump.name in search.html is a Fragment instance.

trac/search/templates/search.html

@@ -76 +76 @@
           # if quickjump:
           <dt id="quickjump">
             <a href="${quickjump.href}">${
-              _('Quickjump to %(name)s', name=quickjump.name)|safe}</a>
+              tag_('Quickjump to %(name)s', name=quickjump.name)}</a>
           </dt>
           <dd>${quickjump.description}</dd>
           # endif

Another considering, _ function returns inconsistent instance with/without Babel. It would be good that the _ function always returns a unicode instance.

Without Babel

>>> from trac.util.html import tag
>>> from trac.util.translation import _, tag_, has_babel
>>> has_babel
False
>>> _('Quickjump to %(name)s', name=tag.em(u'ticket:1,\u200B2'))
'Quickjump to <em>ticket:1,\xe2\x80\x8b2</em>'   # ==> bytes instance
>>> tag_('Quickjump to %(name)s', name=tag.em(u'ticket:1,\u200B2')).__html__()
Markup(u'Quickjump to <em>ticket:1,\u200b2</em>')

With Babel

>>> from trac.util.html import tag
>>> from trac.util.translation import _, tag_, has_babel, make_activable
>>> has_babel
True
>>> make_activable(lambda: 'en_US')
>>> _('Quickjump to %(name)s', name=tag.em(u'ticket:1,\u200B2'))
u'Quickjump to <em>ticket:1,\u200b2</em>'  # ==> unicode instance
>>> tag_('Quickjump to %(name)s', name=tag.em(u'ticket:1,\u200B2')).__html__()
Markup(u'Quickjump to <em>ticket:1,\u200b2</em>')

[ff@…]

I got the same error when searching for something which looks like an email address (a@b.c will do, leaving something away not). Changing the line to use tag_ solved the problem for me.

[Jun Omae]

Fixed in [17215] and merged in [17216].

Note that the issue which _ function has inconsistent behavior with/without babel package still exists.

[Jun Omae]

#13227 was closed as a duplicate.