##// END OF EJS Templates
Add option to define custom lexers for custom extensions for code highlight in rcextension module
Add option to define custom lexers for custom extensions for code highlight in rcextension module

File last commit:

r3164:01cb7df1 beta
r3375:7000fc4a beta
Show More
markup_renderer.py
150 lines | 5.0 KiB | text/x-python | PythonLexer
# -*- coding: utf-8 -*-
"""
rhodecode.lib.markup_renderer
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Renderer for markup languages with ability to parse using rst or markdown
:created_on: Oct 27, 2011
:author: marcink
:copyright: (C) 2011-2012 Marcin Kuzminski <marcin@python-works.com>
:license: GPLv3, see COPYING for more details.
"""
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
import re
import logging
import traceback
from rhodecode.lib.utils2 import safe_unicode, MENTIONS_REGEX
log = logging.getLogger(__name__)
class MarkupRenderer(object):
RESTRUCTUREDTEXT_DISALLOWED_DIRECTIVES = ['include', 'meta', 'raw']
MARKDOWN_PAT = re.compile(r'md|mkdn?|mdown|markdown', re.IGNORECASE)
RST_PAT = re.compile(r're?st', re.IGNORECASE)
PLAIN_PAT = re.compile(r'readme', re.IGNORECASE)
def __detect_renderer(self, source, filename=None):
"""
runs detection of what renderer should be used for generating html
from a markup language
filename can be also explicitly a renderer name
:param source:
:param filename:
"""
if MarkupRenderer.MARKDOWN_PAT.findall(filename):
detected_renderer = 'markdown'
elif MarkupRenderer.RST_PAT.findall(filename):
detected_renderer = 'rst'
elif MarkupRenderer.PLAIN_PAT.findall(filename):
detected_renderer = 'rst'
else:
detected_renderer = 'plain'
return getattr(MarkupRenderer, detected_renderer)
def render(self, source, filename=None):
"""
Renders a given filename using detected renderer
it detects renderers based on file extension or mimetype.
At last it will just do a simple html replacing new lines with <br/>
:param file_name:
:param source:
"""
renderer = self.__detect_renderer(source, filename)
readme_data = renderer(source)
return readme_data
@classmethod
def plain(cls, source):
source = safe_unicode(source)
def urlify_text(text):
url_pat = re.compile(r'(http[s]?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]'
'|[!*\(\),]|(?:%[0-9a-fA-F][0-9a-fA-F]))+)')
def url_func(match_obj):
url_full = match_obj.groups()[0]
return '<a href="%(url)s">%(url)s</a>' % ({'url': url_full})
return url_pat.sub(url_func, text)
source = urlify_text(source)
return '<br />' + source.replace("\n", '<br />')
@classmethod
def markdown(cls, source, safe=True):
source = safe_unicode(source)
try:
import markdown as __markdown
return __markdown.markdown(source, ['codehilite', 'extra'])
except ImportError:
log.warning('Install markdown to use this function')
return cls.plain(source)
except Exception:
log.error(traceback.format_exc())
if safe:
return source
else:
raise
@classmethod
def rst(cls, source, safe=True):
source = safe_unicode(source)
try:
from docutils.core import publish_parts
from docutils.parsers.rst import directives
docutils_settings = dict([(alias, None) for alias in
cls.RESTRUCTUREDTEXT_DISALLOWED_DIRECTIVES])
docutils_settings.update({'input_encoding': 'unicode',
'report_level': 4})
for k, v in docutils_settings.iteritems():
directives.register_directive(k, v)
parts = publish_parts(source=source,
writer_name="html4css1",
settings_overrides=docutils_settings)
return parts['html_title'] + parts["fragment"]
except ImportError:
log.warning('Install docutils to use this function')
return cls.plain(source)
except Exception:
log.error(traceback.format_exc())
if safe:
return source
else:
raise
@classmethod
def rst_with_mentions(cls, source):
mention_pat = re.compile(MENTIONS_REGEX)
def wrapp(match_obj):
uname = match_obj.groups()[0]
return ' **@%(uname)s** ' % {'uname': uname}
mention_hl = mention_pat.sub(wrapp, source).strip()
return cls.rst(mention_hl)