markdown/inlinepatterns.py

"""
INLINE PATTERNS
=============================================================================

Inline patterns such as *emphasis* are handled by means of auxiliary
objects, one per pattern.  Pattern objects must be instances of classes
that extend markdown.Pattern.  Each pattern object uses a single regular
expression and needs support the following methods:

    pattern.getCompiledRegExp() # returns a regular expression

    pattern.handleMatch(m) # takes a match object and returns
                           # an ElementTree element or just plain text

All of python markdown's built-in patterns subclass from Pattern,
but you can add additional patterns that don't.

Also note that all the regular expressions used by inline must
capture the whole block.  For this reason, they all start with
'^(.*)' and end with '(.*)!'.  In case with built-in expression
Pattern takes care of adding the "^(.*)" and "(.*)!".

Finally, the order in which regular expressions are applied is very
important - e.g. if we first replace http://.../ links with <a> tags
and _then_ try to replace inline html, we would end up with a mess.
So, we apply the expressions in the following order:

* escape and backticks have to go before everything else, so
  that we can preempt any markdown patterns by escaping them.

* then we handle auto-links (must be done before inline html)

* then we handle inline HTML.  At this point we will simply
  replace all inline HTML strings with a placeholder and add
  the actual HTML to a hash.

* then inline images (must be done before links)

* then bracketed links, first regular then reference-style

* finally we apply strong and emphasis
"""

from __future__ import absolute_import
from __future__ import unicode_literals
from . import util
from . import odict
import re
try:  # pragma: no cover
    from urllib.parse import urlparse, urlunparse
except ImportError:  # pragma: no cover
    from urlparse import urlparse, urlunparse
try:  # pragma: no cover
    from html import entities
except ImportError:  # pragma: no cover
    import htmlentitydefs as entities


def build_inlinepatterns(md_instance, **kwargs):
    """ Build the default set of inline patterns for Markdown. """
    inlinePatterns = odict.OrderedDict()
    inlinePatterns["backtick"] = BacktickPattern(BACKTICK_RE)
    inlinePatterns["escape"] = EscapePattern(ESCAPE_RE, md_instance)
    inlinePatterns["reference"] = ReferencePattern(REFERENCE_RE, md_instance)
    inlinePatterns["link"] = LinkPattern(LINK_RE, md_instance)
    inlinePatterns["image_link"] = ImagePattern(IMAGE_LINK_RE, md_instance)
    inlinePatterns["image_reference"] = ImageReferencePattern(
        IMAGE_REFERENCE_RE, md_instance
    )
    inlinePatterns["short_reference"] = ReferencePattern(
        SHORT_REF_RE, md_instance
    )
    inlinePatterns["autolink"] = AutolinkPattern(AUTOLINK_RE, md_instance)
    inlinePatterns["automail"] = AutomailPattern(AUTOMAIL_RE, md_instance)
    inlinePatterns["linebreak"] = SubstituteTagPattern(LINE_BREAK_RE, 'br')
    if md_instance.safeMode != 'escape':
        inlinePatterns["html"] = HtmlPattern(HTML_RE, md_instance)
    inlinePatterns["entity"] = HtmlPattern(ENTITY_RE, md_instance)
    inlinePatterns["not_strong"] = SimpleTextPattern(NOT_STRONG_RE)
    inlinePatterns["em_strong"] = DoubleTagPattern(EM_STRONG_RE, 'strong,em')
    inlinePatterns["strong_em"] = DoubleTagPattern(STRONG_EM_RE, 'em,strong')
    inlinePatterns["strong"] = SimpleTagPattern(STRONG_RE, 'strong')
    inlinePatterns["emphasis"] = SimpleTagPattern(EMPHASIS_RE, 'em')
    if md_instance.smart_emphasis:
        inlinePatterns["emphasis2"] = SimpleTagPattern(SMART_EMPHASIS_RE, 'em')
    else:
        inlinePatterns["emphasis2"] = SimpleTagPattern(EMPHASIS_2_RE, 'em')
    return inlinePatterns

"""
The actual regular expressions for patterns
-----------------------------------------------------------------------------
"""

NOBRACKET = r'[^\]\[]*'
BRK = (
    r'\[(' +
    (NOBRACKET + r'(\[')*6 +
    (NOBRACKET + r'\])*')*6 +
    NOBRACKET + r')\]'
)
NOIMG = r'(?<!\!)'

# `e=f()` or ``e=f("`")``
BACKTICK_RE = r'(?<!\\)(`+)(.+?)(?<!`)\2(?!`)'

# \<
ESCAPE_RE = r'\\(.)'

# *emphasis*
EMPHASIS_RE = r'(\*)([^\*]+)\2'

# **strong**
STRONG_RE = r'(\*{2}|_{2})(.+?)\2'

# ***strongem*** or ***em*strong**
EM_STRONG_RE = r'(\*|_)\2{2}(.+?)\2(.*?)\2{2}'

# ***strong**em*
STRONG_EM_RE = r'(\*|_)\2{2}(.+?)\2{2}(.*?)\2'

# _smart_emphasis_
SMART_EMPHASIS_RE = r'(?<!\w)(_)(?!_)(.+?)(?<!_)\2(?!\w)'

# _emphasis_
EMPHASIS_2_RE = r'(_)(.+?)\2'

# [text](url) or [text](<url>) or [text](url "title")
LINK_RE = NOIMG + BRK + \
    r'''\(\s*(<.*?>|((?:(?:\(.*?\))|[^\(\)]))*?)\s*((['"])(.*?)\12\s*)?\)'''

# ![alttxt](http://x.com/) or ![alttxt](<http://x.com/>)
IMAGE_LINK_RE = r'\!' + BRK + r'\s*\((<.*?>|([^")]+"[^"]*"|[^\)]*))\)'

# [Google][3]
REFERENCE_RE = NOIMG + BRK + r'\s?\[([^\]]*)\]'

# [Google]
SHORT_REF_RE = NOIMG + r'\[([^\]]+)\]'

# ![alt text][2]
IMAGE_REFERENCE_RE = r'\!' + BRK + '\s?\[([^\]]*)\]'

# stand-alone * or _
NOT_STRONG_RE = r'((^| )(\*|_)( |$))'

# <http://www.123.com>
AUTOLINK_RE = r'<((?:[Ff]|[Hh][Tt])[Tt][Pp][Ss]?://[^>]*)>'

# <me@example.com>
AUTOMAIL_RE = r'<([^> \!]*@[^> ]*)>'

# <...>
HTML_RE = r'(\<([a-zA-Z/][^\>]*?|\!--.*?--)\>)'

# &amp;
ENTITY_RE = r'(&[\#a-zA-Z0-9]*;)'

# two spaces at end of line
LINE_BREAK_RE = r'  \n'


def dequote(string):
    """Remove quotes from around a string."""
    if ((string.startswith('"') and string.endswith('"')) or
       (string.startswith("'") and string.endswith("'"))):
        return string[1:-1]
    else:
        return string


ATTR_RE = re.compile("\{@([^\}]*)=([^\}]*)}")  # {@id=123}


def handleAttributes(text, parent):
    """Set values of an element based on attribute definitions ({@id=123})."""
    def attributeCallback(match):
        parent.set(match.group(1), match.group(2).replace('\n', ' '))
    return ATTR_RE.sub(attributeCallback, text)


"""
The pattern classes
-----------------------------------------------------------------------------
"""


class Pattern(object):
    """Base class that inline patterns subclass. """

    def __init__(self, pattern, markdown_instance=None):
        """
        Create an instant of an inline pattern.

        Keyword arguments:

        * pattern: A regular expression that matches a pattern

        """
        self.pattern = pattern
        self.compiled_re = re.compile("^(.*?)%s(.*?)$" % pattern,
                                      re.DOTALL | re.UNICODE)

        # Api for Markdown to pass safe_mode into instance
        self.safe_mode = False
        if markdown_instance:
            self.markdown = markdown_instance

    def getCompiledRegExp(self):
        """ Return a compiled regular expression. """
        return self.compiled_re

    def handleMatch(self, m):
        """Return a ElementTree element from the given match.

        Subclasses should override this method.

        Keyword arguments:

        * m: A re match object containing a match of the pattern.

        """
        pass  # pragma: no cover

    def type(self):
        """ Return class name, to define pattern type """
        return self.__class__.__name__

    def unescape(self, text):
        """ Return unescaped text given text with an inline placeholder. """
        try:
            stash = self.markdown.treeprocessors['inline'].stashed_nodes
        except KeyError:  # pragma: no cover
            return text

        def itertext(el):  # pragma: no cover
            ' Reimplement Element.itertext for older python versions '
            tag = el.tag
            if not isinstance(tag, util.string_type) and tag is not None:
                return
            if el.text:
                yield el.text
            for e in el:
                for s in itertext(e):
                    yield s
                if e.tail:
                    yield e.tail

        def get_stash(m):
            id = m.group(1)
            if id in stash:
                value = stash.get(id)
                if isinstance(value, util.string_type):
                    return value
                else:
                    # An etree Element - return text content only
                    return ''.join(itertext(value))
        return util.INLINE_PLACEHOLDER_RE.sub(get_stash, text)


class SimpleTextPattern(Pattern):
    """ Return a simple text of group(2) of a Pattern. """
    def handleMatch(self, m):
        return m.group(2)


class EscapePattern(Pattern):
    """ Return an escaped character. """

    def handleMatch(self, m):
        char = m.group(2)
        if char in self.markdown.ESCAPED_CHARS:
            return '%s%s%s' % (util.STX, ord(char), util.ETX)
        else:
            return None


class SimpleTagPattern(Pattern):
    """
    Return element of type `tag` with a text attribute of group(3)
    of a Pattern.

    """
    def __init__(self, pattern, tag):
        Pattern.__init__(self, pattern)
        self.tag = tag

    def handleMatch(self, m):
        el = util.etree.Element(self.tag)
        el.text = m.group(3)
        return el


class SubstituteTagPattern(SimpleTagPattern):
    """ Return an element of type `tag` with no children. """
    def handleMatch(self, m):
        return util.etree.Element(self.tag)


class BacktickPattern(Pattern):
    """ Return a `<code>` element containing the matching text. """
    def __init__(self, pattern):
        Pattern.__init__(self, pattern)
        self.tag = "code"

    def handleMatch(self, m):
        el = util.etree.Element(self.tag)
        el.text = util.AtomicString(m.group(3).strip())
        return el


class DoubleTagPattern(SimpleTagPattern):
    """Return a ElementTree element nested in tag2 nested in tag1.

    Useful for strong emphasis etc.

    """
    def handleMatch(self, m):
        tag1, tag2 = self.tag.split(",")
        el1 = util.etree.Element(tag1)
        el2 = util.etree.SubElement(el1, tag2)
        el2.text = m.group(3)
        if len(m.groups()) == 5:
            el2.tail = m.group(4)
        return el1


class HtmlPattern(Pattern):
    """ Store raw inline html and return a placeholder. """
    def handleMatch(self, m):
        rawhtml = self.unescape(m.group(2))
        place_holder = self.markdown.htmlStash.store(rawhtml)
        return place_holder

    def unescape(self, text):
        """ Return unescaped text given text with an inline placeholder. """
        try:
            stash = self.markdown.treeprocessors['inline'].stashed_nodes
        except KeyError:  # pragma: no cover
            return text

        def get_stash(m):
            id = m.group(1)
            value = stash.get(id)
            if value is not None:
                try:
                    return self.markdown.serializer(value)
                except:
                    return '\%s' % value

        return util.INLINE_PLACEHOLDER_RE.sub(get_stash, text)


class LinkPattern(Pattern):
    """ Return a link element from the given match. """
    def handleMatch(self, m):
        el = util.etree.Element("a")
        el.text = m.group(2)
        title = m.group(13)
        href = m.group(9)

        if href:
            if href[0] == "<":
                href = href[1:-1]
            el.set("href", self.sanitize_url(self.unescape(href.strip())))
        else:
            el.set("href", "")

        if title:
            title = dequote(self.unescape(title))
            el.set("title", title)
        return el

    def sanitize_url(self, url):
        """
        Sanitize a url against xss attacks in "safe_mode".

        Rather than specifically blacklisting `javascript:alert("XSS")` and all
        its aliases (see <http://ha.ckers.org/xss.html>), we whitelist known
        safe url formats. Most urls contain a network location, however some
        are known not to (i.e.: mailto links). Script urls do not contain a
        location. Additionally, for `javascript:...`, the scheme would be
        "javascript" but some aliases will appear to `urlparse()` to have no
        scheme. On top of that relative links (i.e.: "foo/bar.html") have no
        scheme. Therefore we must check "path", "parameters", "query" and
        "fragment" for any literal colons. We don't check "scheme" for colons
        because it *should* never have any and "netloc" must allow the form:
        `username:password@host:port`.

        """
        if not self.markdown.safeMode:
            # Return immediately bipassing parsing.
            return url

        try:
            scheme, netloc, path, params, query, fragment = url = urlparse(url)
        except ValueError:  # pragma: no cover
            # Bad url - so bad it couldn't be parsed.
            return ''

        locless_schemes = ['', 'mailto', 'news']
        allowed_schemes = locless_schemes + ['http', 'https', 'ftp', 'ftps']
        if scheme not in allowed_schemes:
            # Not a known (allowed) scheme. Not safe.
            return ''

        if netloc == '' and scheme not in locless_schemes:  # pragma: no cover
            # This should not happen. Treat as suspect.
            return ''

        for part in url[2:]:
            if ":" in part:
                # A colon in "path", "parameters", "query"
                # or "fragment" is suspect.
                return ''

        # Url passes all tests. Return url as-is.
        return urlunparse(url)


class ImagePattern(LinkPattern):
    """ Return a img element from the given match. """
    def handleMatch(self, m):
        el = util.etree.Element("img")
        src_parts = m.group(9).split()
        if src_parts:
            src = src_parts[0]
            if src[0] == "<" and src[-1] == ">":
                src = src[1:-1]
            el.set('src', self.sanitize_url(self.unescape(src)))
        else:
            el.set('src', "")
        if len(src_parts) > 1:
            el.set('title', dequote(self.unescape(" ".join(src_parts[1:]))))

        if self.markdown.enable_attributes:
            truealt = handleAttributes(m.group(2), el)
        else:
            truealt = m.group(2)

        el.set('alt', self.unescape(truealt))
        return el


class ReferencePattern(LinkPattern):
    """ Match to a stored reference and return link element. """

    NEWLINE_CLEANUP_RE = re.compile(r'[ ]?\n', re.MULTILINE)

    def handleMatch(self, m):
        try:
            id = m.group(9).lower()
        except IndexError:
            id = None
        if not id:
            # if we got something like "[Google][]" or "[Goggle]"
            # we'll use "google" as the id
            id = m.group(2).lower()

        # Clean up linebreaks in id
        id = self.NEWLINE_CLEANUP_RE.sub(' ', id)
        if id not in self.markdown.references:  # ignore undefined refs
            return None
        href, title = self.markdown.references[id]

        text = m.group(2)
        return self.makeTag(href, title, text)

    def makeTag(self, href, title, text):
        el = util.etree.Element('a')

        el.set('href', self.sanitize_url(href))
        if title:
            el.set('title', title)

        el.text = text
        return el


class ImageReferencePattern(ReferencePattern):
    """ Match to a stored reference and return img element. """
    def makeTag(self, href, title, text):
        el = util.etree.Element("img")
        el.set("src", self.sanitize_url(href))
        if title:
            el.set("title", title)

        if self.markdown.enable_attributes:
            text = handleAttributes(text, el)

        el.set("alt", self.unescape(text))
        return el


class AutolinkPattern(Pattern):
    """ Return a link Element given an autolink (`<http://example/com>`). """
    def handleMatch(self, m):
        el = util.etree.Element("a")
        el.set('href', self.unescape(m.group(2)))
        el.text = util.AtomicString(m.group(2))
        return el


class AutomailPattern(Pattern):
    """
    Return a mailto link Element given an automail link (`<foo@example.com>`).
    """
    def handleMatch(self, m):
        el = util.etree.Element('a')
        email = self.unescape(m.group(2))
        if email.startswith("mailto:"):
            email = email[len("mailto:"):]

        def codepoint2name(code):
            """Return entity definition by code, or the code if not defined."""
            entity = entities.codepoint2name.get(code)
            if entity:
                return "%s%s;" % (util.AMP_SUBSTITUTE, entity)
            else:
                return "%s#%d;" % (util.AMP_SUBSTITUTE, code)

        letters = [codepoint2name(ord(letter)) for letter in email]
        el.text = util.AtomicString(''.join(letters))

        mailto = "mailto:" + email
        mailto = "".join([util.AMP_SUBSTITUTE + '#%d;' %
                          ord(letter) for letter in mailto])
        el.set('href', mailto)
        return el