bloodhound-trac / trac / util /

# -*- coding: utf-8 -*-
# Copyright (C) 2003-2009 Edgewall Software
# All rights reserved.
# This software is licensed as described in the file COPYING, which
# you should have received as part of this distribution. The terms
# are also available at
# This software consists of voluntary contributions made by many
# individuals. For the exact contribution history, see the revision
# history and logs, available at

from HTMLParser import HTMLParser
import re

from genshi import Markup, HTML, escape, unescape
from genshi.core import stripentities, striptags, START, END
from genshi.builder import Element, ElementFactory, Fragment
from genshi.filters.html import HTMLSanitizer
from genshi.input import ParseError

__all__ = ['escape', 'unescape', 'html', 'plaintext', 'find_element',
           'TracHTMLSanitizer', 'Deuglifier', 'FormTokenInjector']

class TracHTMLSanitizer(HTMLSanitizer):
    """Sanitize HTML constructions which are potentially vector of
    phishing or XSS attacks, in user-supplied HTML.

    See also `genshi.HTMLSanitizer`_.

    .. _genshi.HTMLSanitizer:

    SAFE_CSS = frozenset([
        # CSS 3 properties <>
        'background', 'background-attachment', 'background-color',
        'background-image', 'background-position', 'background-repeat',
        'border', 'border-bottom', 'border-bottom-color',
        'border-bottom-style', 'border-bottom-width', 'border-collapse',
        'border-color', 'border-left', 'border-left-color',
        'border-left-style', 'border-left-width', 'border-right',
        'border-right-color', 'border-right-style', 'border-right-width',
        'border-spacing', 'border-style', 'border-top', 'border-top-color',
        'border-top-style', 'border-top-width', 'border-width', 'bottom',
        'caption-side', 'clear', 'clip', 'color', 'content',
        'counter-increment', 'counter-reset', 'cursor', 'direction', 'display',
        'empty-cells', 'float', 'font', 'font-family', 'font-size',
        'font-style', 'font-variant', 'font-weight', 'height', 'left',
        'letter-spacing', 'line-height', 'list-style', 'list-style-image',
        'list-style-position', 'list-style-type', 'margin', 'margin-bottom',
        'margin-left', 'margin-right', 'margin-top', 'max-height', 'max-width',
        'min-height', 'min-width', 'opacity', 'orphans', 'outline',
        'outline-color', 'outline-style', 'outline-width', 'overflow',
        'padding', 'padding-bottom', 'padding-left', 'padding-right',
        'padding-top', 'page-break-after', 'page-break-before',
        'page-break-inside', 'position', 'quotes', 'right', 'table-layout',
        'text-align', 'text-decoration', 'text-indent', 'text-transform',
        'top', 'unicode-bidi', 'vertical-align', 'visibility', 'white-space',
        'widows', 'width', 'word-spacing', 'z-index',

    def __init__(self, safe_schemes=HTMLSanitizer.SAFE_SCHEMES,
        safe_attrs = HTMLSanitizer.SAFE_ATTRS | frozenset(['style'])
        safe_schemes = frozenset(safe_schemes)
        super(TracHTMLSanitizer, self).__init__(safe_attrs=safe_attrs,
        self.safe_css = frozenset(safe_css)

    # IE6 <>
    _EXPRESSION_SEARCH = re.compile(
        u'[eE\uFF25\uFF45]'         # FULLWIDTH LATIN CAPITAL LETTER E
                                    # FULLWIDTH LATIN SMALL LETTER E
        u'[xX\uFF38\uFF58]'         # FULLWIDTH LATIN CAPITAL LETTER X
                                    # FULLWIDTH LATIN SMALL LETTER X
        u'[pP\uFF30\uFF50]'         # FULLWIDTH LATIN CAPITAL LETTER P
                                    # FULLWIDTH LATIN SMALL LETTER P
        u'[rR\u0280\uFF32\uFF52]'   # LATIN LETTER SMALL CAPITAL R
                                    # FULLWIDTH LATIN CAPITAL LETTER R
                                    # FULLWIDTH LATIN SMALL LETTER R
        u'[eE\uFF25\uFF45]'         # FULLWIDTH LATIN CAPITAL LETTER E
                                    # FULLWIDTH LATIN SMALL LETTER E
        u'[sS\uFF33\uFF53]{2}'      # FULLWIDTH LATIN CAPITAL LETTER S
                                    # FULLWIDTH LATIN SMALL LETTER S
        u'[iI\u026A\uFF29\uFF49]'   # LATIN LETTER SMALL CAPITAL I
                                    # FULLWIDTH LATIN CAPITAL LETTER I
                                    # FULLWIDTH LATIN SMALL LETTER I
        u'[oO\uFF2F\uFF4F]'         # FULLWIDTH LATIN CAPITAL LETTER O
                                    # FULLWIDTH LATIN SMALL LETTER O
        u'[nN\u0274\uFF2E\uFF4E]'   # LATIN LETTER SMALL CAPITAL N
                                    # FULLWIDTH LATIN CAPITAL LETTER N
                                    # FULLWIDTH LATIN SMALL LETTER N

    # IE6 <>
    #     7) Particular bit of Unicode characters
    _URL_FINDITER = re.compile(

    def sanitize_css(self, text):
        decls = []
        text = self._strip_css_comments(self._replace_unicode_escapes(text))
        for decl in filter(None, text.split(';')):
            decl = decl.strip()
            if not decl:
                prop, value = decl.split(':', 1)
            except ValueError:
            if not self.is_safe_css(prop.strip().lower(), value.strip()):
            is_evil = False
            if self._EXPRESSION_SEARCH(decl):
                is_evil = True
            for match in self._URL_FINDITER(decl):
                if not self.is_safe_uri(
                    is_evil = True
            if not is_evil:
        return decls

    def __call__(self, stream):
        """Remove input type="password" elements from the stream
        suppress = False
        for kind, data, pos in super(TracHTMLSanitizer, self).__call__(stream):
            if kind is START:
                tag, attrs = data
                if (tag == 'input' and
                    attrs.get('type', '').lower() == 'password'):
                    suppress = True
                    yield kind, data, pos
            elif kind is END:
                if not suppress:
                    yield kind, data, pos
                suppress = False
                yield kind, data, pos

    def is_safe_css(self, prop, value):
        """Determine whether the given css property declaration is to be 
        considered safe for inclusion in the output.
        if prop not in self.safe_css:
            return False
        # Position can be used for phishing, 'static' excepted
        if prop == 'position':
            return value.lower() == 'static'
        # Negative margins can be used for phishing
        if prop.startswith('margin'):
            return '-' not in value
        return True

    _NORMALIZE_NEWLINES = re.compile(r'\r\n').sub
    _UNICODE_ESCAPE = re.compile(

    def _replace_unicode_escapes(self, text):
        def _repl(match):
            t =
            if t:
                code = int(t, 16)
                chr = unichr(code)
                if code <= 0x1f:
                    # replace space character because IE ignores control
                    # characters
                    chr = ' '
                elif chr == '\\':
                    chr = r'\\'
                return chr
            t =
            if t == '\\':
                return r'\\'
                return t
        return self._UNICODE_ESCAPE(_repl,
                                    self._NORMALIZE_NEWLINES('\n', text))

    _CSS_COMMENTS = re.compile(r'/\*.*?\*/').sub

    def _strip_css_comments(self, text):
        """Replace comments with space character instead of superclass which
        removes comments to avoid problems when nested comments.
        return self._CSS_COMMENTS(' ', text)

class Deuglifier(object):
    """Help base class used for cleaning up HTML riddled with ``<FONT
    COLOR=...>`` tags and replace them with appropriate ``<span

    The subclass must define a `rules()` static method returning a
    list of regular expression fragments, each defining a capture
    group in which the name will be reused for the span's class. Two
    special group names, ``font`` and ``endfont`` are used to emit
    ``<span>`` and ``</span>``, respectively.
    def __new__(cls):
        self = object.__new__(cls)
        if not hasattr(cls, '_compiled_rules'):
            cls._compiled_rules = re.compile('(?:%s)' % '|'.join(cls.rules()))
        self._compiled_rules = cls._compiled_rules
        return self
    def format(self, indata):
        return re.sub(self._compiled_rules, self.replace, indata)

    def replace(self, fullmatch):
        for mtype, match in fullmatch.groupdict().items():
            if match:
                if mtype == 'font':
                    return '<span>'
                elif mtype == 'endfont':
                    return '</span>'
                return '<span class="code-%s">' % mtype

class FormTokenInjector(HTMLParser):
    """Identify and protect forms from CSRF attacks.

    This filter works by adding a input type=hidden field to POST forms.
    def __init__(self, form_token, out):
        self.out = out
        self.token = form_token

    def handle_starttag(self, tag, attrs):
        if tag.lower() == 'form':
            for name, value in attrs:
                if name.lower() == 'method' and value.lower() == 'post':
                    self.out.write('<input type="hidden" name="__FORM_TOKEN"'
                                   ' value="%s"/>' % self.token)
    def handle_startendtag(self, tag, attrs):
    def handle_charref(self, name):
        self.out.write('&#%s;' % name)

    def handle_entityref(self, name):
        self.out.write('&%s;' % name)

    def handle_comment(self, data):
        self.out.write('<!--%s-->' % data)

    def handle_decl(self, data):
        self.out.write('<!%s>' % data)

    def handle_pi(self, data):
        self.out.write('<?%s?>' % data)

    def handle_data(self, data):

    def handle_endtag(self, tag):
        self.out.write('</' + tag + '>')

class TransposingElementFactory(ElementFactory):
    """A `genshi.builder.ElementFactory` which applies `func` to the
    named attributes before creating a `genshi.builder.Element`.

    def __init__(self, func, namespace=None):
        ElementFactory.__init__(self, namespace=namespace)
        self.func = func

    def __getattr__(self, name):
        return ElementFactory.__getattr__(self, self.func(name))

html = TransposingElementFactory(str.lower)

def plaintext(text, keeplinebreaks=True):
    """Extract the text elements from (X)HTML content

    :param text: `unicode` or `genshi.builder.Fragment`
    :param keeplinebreaks: optionally keep linebreaks
    if isinstance(text, Fragment):
        text = text.generate().render('text', encoding=None)
        text = stripentities(striptags(text))
    if not keeplinebreaks:
        text = text.replace(u'\n', u' ')
    return text

def find_element(frag, attr=None, cls=None):
    """Return the first element in the fragment having the given attribute or
    class, using a preorder depth-first search.
    if isinstance(frag, Element):
        if attr is not None and attr in frag.attrib:
            return frag
        if cls is not None and cls in frag.attrib.get('class', '').split():
            return frag
    if isinstance(frag, Fragment):
        for child in frag.children:
            elt = find_element(child, attr, cls)
            if elt is not None:
                return elt

def expand_markup(stream, ctxt=None):
    """A Genshi stream filter for expanding `genshi.Markup` events.

    Note: Expansion may not be possible if the fragment is badly
    formed, or partial.
    for event in stream:
        if isinstance(event[1], Markup):
                for subevent in HTML(event[1]):
                    yield subevent
            except ParseError:
                yield event
            yield event