security: use custom writer for RST rendering to prevent injection of javascript: tags.

rhodecode/lib/markup_renderer.py

@@ -1,461 +1,488 b''
 # -*- coding: utf-8 -*-
 
 # Copyright (C) 2011-2017 RhodeCode GmbH
 #
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU Affero General Public License, version 3
 # (only), as published by the Free Software Foundation.
 #
 # 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 Affero General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 # This program is dual-licensed. If you wish to learn more about the
 # RhodeCode Enterprise Edition, including its added features, Support services,
 # and proprietary license terms, please see https://rhodecode.com/licenses/
 
 
 """
 Renderer for markup languages with ability to parse using rst or markdown
 """
 
 import re
 import os
 import lxml
 import logging
 import urlparse
 
 from mako.lookup import TemplateLookup
 from mako.template import Template as MakoTemplate
 
 from docutils.core import publish_parts
 from docutils.parsers.rst import directives
+from docutils import writers
+from docutils.writers import html4css1
 import markdown
 
 from rhodecode.lib.markdown_ext import GithubFlavoredMarkdownExtension
 from rhodecode.lib.utils2 import (
     safe_str, safe_unicode, md5_safe, MENTIONS_REGEX)
 
 log = logging.getLogger(__name__)
 
 # default renderer used to generate automated comments
 DEFAULT_COMMENTS_RENDERER = 'rst'
 
 
+class CustomHTMLTranslator(writers.html4css1.HTMLTranslator):
+    """
+    Custom HTML Translator used for sandboxing potential
+    JS injections in ref links
+    """
+
+    def visit_reference(self, node):
+        if 'refuri' in node.attributes:
+            refuri = node['refuri']
+            if ':' in refuri:
+                prefix, link = refuri.lstrip().split(':', 1)
+                if prefix == 'javascript':
+                    # we don't allow javascript type of refs...
+                    node['refuri'] = 'javascript:alert("SandBoxedJavascript")'
+
+        # old style class requires this...
+        return html4css1.HTMLTranslator.visit_reference(self, node)
+
+
+class RhodeCodeWriter(writers.html4css1.Writer):
+    def __init__(self):
+        writers.Writer.__init__(self)
+        self.translator_class = CustomHTMLTranslator
+
+
 def relative_links(html_source, server_path):
     if not html_source:
         return html_source
 
     try:
         doc = lxml.html.fromstring(html_source)
     except Exception:
         return html_source
 
     for el in doc.cssselect('img, video'):
         src = el.attrib['src']
         if src:
             el.attrib['src'] = relative_path(src, server_path)
 
     for el in doc.cssselect('a:not(.gfm)'):
         src = el.attrib['href']
         if src:
             el.attrib['href'] = relative_path(src, server_path)
 
     return lxml.html.tostring(doc)
 
 
 def relative_path(path, request_path, is_repo_file=None):
     """
     relative link support, path is a rel path, and request_path is current
     server path (not absolute)
 
     e.g.
 
     path = '../logo.png'
     request_path= '/repo/files/path/file.md'
     produces: '/repo/files/logo.png'
     """
     # TODO(marcink): unicode/str support ?
     # maybe=> safe_unicode(urllib.quote(safe_str(final_path), '/:'))
 
     def dummy_check(p):
         return True  # assume default is a valid file path
 
     is_repo_file = is_repo_file or dummy_check
     if not path:
         return request_path
 
     path = safe_unicode(path)
     request_path = safe_unicode(request_path)
 
     if path.startswith((u'data:', u'javascript:', u'#', u':')):
         # skip data, anchor, invalid links
         return path
 
     is_absolute = bool(urlparse.urlparse(path).netloc)
     if is_absolute:
         return path
 
     if not request_path:
         return path
 
     if path.startswith(u'/'):
         path = path[1:]
 
     if path.startswith(u'./'):
         path = path[2:]
 
     parts = request_path.split('/')
     # compute how deep we need to traverse the request_path
     depth = 0
 
     if is_repo_file(request_path):
         # if request path is a VALID file, we use a relative path with
         # one level up
         depth += 1
 
     while path.startswith(u'../'):
         depth += 1
         path = path[3:]
 
     if depth > 0:
         parts = parts[:-depth]
 
     parts.append(path)
     final_path = u'/'.join(parts).lstrip(u'/')
 
     return u'/' + final_path
 
 
 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)
     JUPYTER_PAT = re.compile(r'\.(ipynb)$', re.IGNORECASE)
     PLAIN_PAT = re.compile(r'^readme$', re.IGNORECASE)
 
     extensions = ['codehilite', 'extra', 'def_list', 'sane_lists']
     markdown_renderer = markdown.Markdown(
         extensions, safe_mode=True, enable_attributes=False)
 
     markdown_renderer_flavored = markdown.Markdown(
         extensions + [GithubFlavoredMarkdownExtension()], safe_mode=True,
         enable_attributes=False)
 
     # extension together with weights. Lower is first means we control how
     # extensions are attached to readme names with those.
     PLAIN_EXTS = [
         # prefer no extension
         ('', 0),  # special case that renders READMES names without extension
         ('.text', 2), ('.TEXT', 2),
         ('.txt', 3), ('.TXT', 3)
     ]
 
     RST_EXTS = [
         ('.rst', 1), ('.rest', 1),
         ('.RST', 2), ('.REST', 2)
     ]
 
     MARKDOWN_EXTS = [
         ('.md', 1), ('.MD', 1),
         ('.mkdn', 2), ('.MKDN', 2),
         ('.mdown', 3), ('.MDOWN', 3),
         ('.markdown', 4), ('.MARKDOWN', 4)
     ]
 
     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.JUPYTER_PAT.findall(filename):
             detected_renderer = 'jupyter'
         elif MarkupRenderer.PLAIN_PAT.findall(filename):
             detected_renderer = 'plain'
         else:
             detected_renderer = 'plain'
 
         return getattr(MarkupRenderer, detected_renderer)
 
     @classmethod
     def renderer_from_filename(cls, filename, exclude):
         """
         Detect renderer markdown/rst from filename and optionally use exclude
         list to remove some options. This is mostly used in helpers.
         Returns None when no renderer can be detected.
         """
         def _filter(elements):
             if isinstance(exclude, (list, tuple)):
                 return [x for x in elements if x not in exclude]
             return elements
 
         if filename.endswith(
                 tuple(_filter([x[0] for x in cls.MARKDOWN_EXTS if x[0]]))):
             return 'markdown'
         if filename.endswith(tuple(_filter([x[0] for x in cls.RST_EXTS if x[0]]))):
             return 'rst'
 
         return None
 
     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 _flavored_markdown(cls, text):
         """
         Github style flavored markdown
 
         :param text:
         """
 
         # Extract pre blocks.
         extractions = {}
 
         def pre_extraction_callback(matchobj):
             digest = md5_safe(matchobj.group(0))
             extractions[digest] = matchobj.group(0)
             return "{gfm-extraction-%s}" % digest
         pattern = re.compile(r'<pre>.*?</pre>', re.MULTILINE | re.DOTALL)
         text = re.sub(pattern, pre_extraction_callback, text)
 
         # Prevent foo_bar_baz from ending up with an italic word in the middle.
         def italic_callback(matchobj):
             s = matchobj.group(0)
             if list(s).count('_') >= 2:
                 return s.replace('_', r'\_')
             return s
         text = re.sub(r'^(?! {4}|\t)\w+_\w+_\w[\w_]*', italic_callback, text)
 
         # Insert pre block extractions.
         def pre_insert_callback(matchobj):
             return '\n\n' + extractions[matchobj.group(1)]
         text = re.sub(r'\{gfm-extraction-([0-9a-f]{32})\}',
                       pre_insert_callback, text)
 
         return text
 
     @classmethod
     def urlify_text(cls, text):
         url_pat = re.compile(r'(http[s]?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]'
                              r'|[!*\(\),]|(?:%[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)
 
     @classmethod
     def plain(cls, source, universal_newline=True):
         source = safe_unicode(source)
         if universal_newline:
             newline = '\n'
             source = newline.join(source.splitlines())
 
         source = cls.urlify_text(source)
         return '<br />' + source.replace("\n", '<br />')
 
     @classmethod
     def markdown(cls, source, safe=True, flavored=True, mentions=False):
         # It does not allow to insert inline HTML. In presence of HTML tags, it
         # will replace them instead with [HTML_REMOVED]. This is controlled by
         # the safe_mode=True parameter of the markdown method.
 
         if flavored:
             markdown_renderer = cls.markdown_renderer_flavored
         else:
             markdown_renderer = cls.markdown_renderer
 
         if mentions:
             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()
             # we extracted mentions render with this using Mentions false
             return cls.markdown(mention_hl, safe=safe, flavored=flavored,
                                 mentions=False)
 
         source = safe_unicode(source)
         try:
             if flavored:
                 source = cls._flavored_markdown(source)
             return markdown_renderer.convert(source)
         except Exception:
             log.exception('Error when rendering Markdown')
             if safe:
                 log.debug('Fallback to render in plain mode')
                 return cls.plain(source)
             else:
                 raise
 
     @classmethod
     def rst(cls, source, safe=True, mentions=False):
         if mentions:
             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()
             # we extracted mentions render with this using Mentions false
             return cls.rst(mention_hl, safe=safe, mentions=False)
 
         source = safe_unicode(source)
         try:
             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",
+                                  writer=RhodeCodeWriter(),
                                   settings_overrides=docutils_settings)
 
             return parts['html_title'] + parts["fragment"]
         except Exception:
             log.exception('Error when rendering RST')
             if safe:
                 log.debug('Fallbacking to render in plain mode')
                 return cls.plain(source)
             else:
                 raise
 
     @classmethod
     def jupyter(cls, source, safe=True):
         from rhodecode.lib import helpers
 
         from traitlets.config import Config
         import nbformat
         from nbconvert import HTMLExporter
         from nbconvert.preprocessors import Preprocessor
 
         class CustomHTMLExporter(HTMLExporter):
             def _template_file_default(self):
                 return 'basic'
 
         class Sandbox(Preprocessor):
 
             def preprocess(self, nb, resources):
                 sandbox_text = 'SandBoxed(IPython.core.display.Javascript object)'
                 for cell in nb['cells']:
                     if safe and 'outputs' in cell:
                         for cell_output in cell['outputs']:
                             if 'data' in cell_output:
                                 if 'application/javascript' in cell_output['data']:
                                     cell_output['data']['text/plain'] = sandbox_text
                                     cell_output['data'].pop('application/javascript', None)
                 return nb, resources
 
         def _sanitize_resources(resources):
             """
             Skip/sanitize some of the CSS generated and included in jupyter
             so it doesn't messes up UI so much
             """
 
             # TODO(marcink): probably we should replace this with whole custom
             # CSS set that doesn't screw up, but jupyter generated html has some
             # special markers, so it requires Custom HTML exporter template with
             # _default_template_path_default, to achieve that
 
             # strip the reset CSS
             resources[0] = resources[0][resources[0].find('/*! Source'):]
             return resources
 
         def as_html(notebook):
             conf = Config()
             conf.CustomHTMLExporter.preprocessors = [Sandbox]
             html_exporter = CustomHTMLExporter(config=conf)
 
             (body, resources) = html_exporter.from_notebook_node(notebook)
             header = '<!-- ## IPYTHON NOTEBOOK RENDERING ## -->'
             js = MakoTemplate(r'''
             <!-- Load mathjax -->
                 <!-- MathJax configuration -->
                 <script type="text/x-mathjax-config">
                 MathJax.Hub.Config({
                     jax: ["input/TeX","output/HTML-CSS", "output/PreviewHTML"],
                     extensions: ["tex2jax.js","MathMenu.js","MathZoom.js", "fast-preview.js", "AssistiveMML.js", "[Contrib]/a11y/accessibility-menu.js"],
                     TeX: {
                         extensions: ["AMSmath.js","AMSsymbols.js","noErrors.js","noUndefined.js"]
                     },
                     tex2jax: {
                         inlineMath: [ ['$','$'], ["\\(","\\)"] ],
                         displayMath: [ ['$$','$$'], ["\\[","\\]"] ],
                         processEscapes: true,
                         processEnvironments: true
                     },
                     // Center justify equations in code and markdown cells. Elsewhere
                     // we use CSS to left justify single line equations in code cells.
                     displayAlign: 'center',
                     "HTML-CSS": {
                         styles: {'.MathJax_Display': {"margin": 0}},
                         linebreaks: { automatic: true },
                         availableFonts: ["STIX", "TeX"]
                     },
                     showMathMenu: false
                 });
                 </script>
                 <!-- End of mathjax configuration -->
                 <script src="${h.asset('js/src/math_jax/MathJax.js')}"></script>
             ''').render(h=helpers)
 
             css = '<style>{}</style>'.format(
                 ''.join(_sanitize_resources(resources['inlining']['css'])))
 
             body = '\n'.join([header, css, js, body])
             return body, resources
 
         notebook = nbformat.reads(source, as_version=4)
         (body, resources) = as_html(notebook)
         return body
 
 
 class RstTemplateRenderer(object):
 
     def __init__(self):
         base = os.path.abspath(os.path.dirname(os.path.dirname(__file__)))
         rst_template_dirs = [os.path.join(base, 'templates', 'rst_templates')]
         self.template_store = TemplateLookup(
             directories=rst_template_dirs,
             input_encoding='utf-8',
             imports=['from rhodecode.lib import helpers as h'])
 
     def _get_template(self, templatename):
         return self.template_store.get_template(templatename)
 
     def render(self, template_name, **kwargs):
         template = self._get_template(template_name)
         return template.render(**kwargs)