/lib/markdown.py

Large files files are truncated, but you can click here to view the full file

#!/usr/bin/env python

version = "1.7"
version_info = (1,7,0,"rc-2")
__revision__ = "$Rev: 72 $"

"""
Python-Markdown
===============

Converts Markdown to HTML.  Basic usage as a module:

    import markdown
    md = Markdown()
    html = md.convert(your_text_string)

See http://www.freewisdom.org/projects/python-markdown/ for more
information and instructions on how to extend the functionality of the
script.  (You might want to read that before you try modifying this
file.)

Started by [Manfred Stienstra](http://www.dwerg.net/).  Continued and
maintained  by [Yuri Takhteyev](http://www.freewisdom.org) and [Waylan
Limberg](http://achinghead.com/).

Contact: yuri [at] freewisdom.org
         waylan [at] gmail.com

License: GPL 2 (http://www.gnu.org/copyleft/gpl.html) or BSD

"""


import re, sys, codecs

from logging import getLogger, StreamHandler, Formatter, \
                    DEBUG, INFO, WARN, ERROR, CRITICAL


MESSAGE_THRESHOLD = CRITICAL


# Configure debug message logger (the hard way - to support python 2.3)
logger = getLogger('MARKDOWN')
logger.setLevel(DEBUG) # This is restricted by handlers later
console_hndlr = StreamHandler()
formatter = Formatter('%(name)s-%(levelname)s: "%(message)s"')
console_hndlr.setFormatter(formatter)
console_hndlr.setLevel(MESSAGE_THRESHOLD)
logger.addHandler(console_hndlr)


def message(level, text):
    ''' A wrapper method for logging debug messages. '''
    logger.log(level, text)


# --------------- CONSTANTS YOU MIGHT WANT TO MODIFY -----------------

TAB_LENGTH = 4            # expand tabs to this many spaces
ENABLE_ATTRIBUTES = True  # @id = xyz -> <... id="xyz">
SMART_EMPHASIS = 1        # this_or_that does not become this<i>or</i>that
HTML_REMOVED_TEXT = "[HTML_REMOVED]" # text used instead of HTML in safe mode

RTL_BIDI_RANGES = ( (u'\u0590', u'\u07FF'),
                    # from Hebrew to Nko (includes Arabic, Syriac and Thaana)
                    (u'\u2D30', u'\u2D7F'),
                    # Tifinagh
                    )

# Unicode Reference Table:
# 0590-05FF - Hebrew
# 0600-06FF - Arabic
# 0700-074F - Syriac
# 0750-077F - Arabic Supplement
# 0780-07BF - Thaana
# 07C0-07FF - Nko

BOMS = { 'utf-8': (codecs.BOM_UTF8, ),
         'utf-16': (codecs.BOM_UTF16_LE, codecs.BOM_UTF16_BE),
         #'utf-32': (codecs.BOM_UTF32_LE, codecs.BOM_UTF32_BE)
         }

def removeBOM(text, encoding):
    convert = isinstance(text, unicode)
    for bom in BOMS[encoding]:
        bom = convert and bom.decode(encoding) or bom
        if text.startswith(bom):
            return text.lstrip(bom)
    return text

# The following constant specifies the name used in the usage
# statement displayed for python versions lower than 2.3.  (With
# python2.3 and higher the usage statement is generated by optparse
# and uses the actual name of the executable called.)

EXECUTABLE_NAME_FOR_USAGE = "python markdown.py"
                    

# --------------- CONSTANTS YOU _SHOULD NOT_ HAVE TO CHANGE ----------

# a template for html placeholders
HTML_PLACEHOLDER_PREFIX = "qaodmasdkwaspemas"
HTML_PLACEHOLDER = HTML_PLACEHOLDER_PREFIX + "%dajkqlsmdqpakldnzsdfls"

BLOCK_LEVEL_ELEMENTS = ['p', 'div', 'blockquote', 'pre', 'table',
                        'dl', 'ol', 'ul', 'script', 'noscript',
                        'form', 'fieldset', 'iframe', 'math', 'ins',
                        'del', 'hr', 'hr/', 'style']

def isBlockLevel (tag):
    return ( (tag in BLOCK_LEVEL_ELEMENTS) or
             (tag[0] == 'h' and tag[1] in "0123456789") )

"""
======================================================================
========================== NANODOM ===================================
======================================================================

The three classes below implement some of the most basic DOM
methods.  I use this instead of minidom because I need a simpler
functionality and do not want to require additional libraries.

Importantly, NanoDom does not do normalization, which is what we
want. It also adds extra white space when converting DOM to string
"""

ENTITY_NORMALIZATION_EXPRESSIONS = [ (re.compile("&"), "&amp;"),
                                     (re.compile("<"), "&lt;"),
                                     (re.compile(">"), "&gt;")]

ENTITY_NORMALIZATION_EXPRESSIONS_SOFT = [ (re.compile("&(?!\#)"), "&amp;"),
                                     (re.compile("<"), "&lt;"),
                                     (re.compile(">"), "&gt;"),
                                     (re.compile("\""), "&quot;")]


def getBidiType(text):

    if not text: return None

    ch = text[0]

    if not isinstance(ch, unicode) or not ch.isalpha():
        return None

    else:

        for min, max in RTL_BIDI_RANGES:
            if ( ch >= min and ch <= max ):
                return "rtl"
        else:
            return "ltr"


class Document:

    def __init__ (self):
        self.bidi = "ltr"

    def appendChild(self, child):
        self.documentElement = child
        child.isDocumentElement = True
        child.parent = self
        self.entities = {}

    def setBidi(self, bidi):
        if bidi:
            self.bidi = bidi

    def createElement(self, tag, textNode=None):
        el = Element(tag)
        el.doc = self
        if textNode:
            el.appendChild(self.createTextNode(textNode))
        return el

    def createTextNode(self, text):
        node = TextNode(text)
        node.doc = self
        return node

    def createEntityReference(self, entity):
        if entity not in self.entities:
            self.entities[entity] = EntityReference(entity)
        return self.entities[entity]

    def createCDATA(self, text):
        node = CDATA(text)
        node.doc = self
        return node

    def toxml (self):
        return self.documentElement.toxml()

    def normalizeEntities(self, text, avoidDoubleNormalizing=False):

        if avoidDoubleNormalizing:
            regexps = ENTITY_NORMALIZATION_EXPRESSIONS_SOFT
        else:
            regexps = ENTITY_NORMALIZATION_EXPRESSIONS

        for regexp, substitution in regexps:
            text = regexp.sub(substitution, text)
        return text

    def find(self, test):
        return self.documentElement.find(test)

    def unlink(self):
        self.documentElement.unlink()
        self.documentElement = None


class CDATA:

    type = "cdata"

    def __init__ (self, text):
        self.text = text

    def handleAttributes(self):
        pass

    def toxml (self):
        return "<![CDATA[" + self.text + "]]>"

class Element:

    type = "element"

    def __init__ (self, tag):

        self.nodeName = tag
        self.attributes = []
        self.attribute_values = {}
        self.childNodes = []
        self.bidi = None
        self.isDocumentElement = False

    def setBidi(self, bidi):

        if bidi:

            orig_bidi = self.bidi

            if not self.bidi or self.isDocumentElement:
                # Once the bidi is set don't change it (except for doc element)
                self.bidi = bidi
                self.parent.setBidi(bidi)


    def unlink(self):
        for child in self.childNodes:
            if child.type == "element":
                child.unlink()
        self.childNodes = None

    def setAttribute(self, attr, value):
        if not attr in self.attributes:
            self.attributes.append(attr)

        self.attribute_values[attr] = value

    def insertChild(self, position, child):
        self.childNodes.insert(position, child)
        child.parent = self

    def removeChild(self, child):
        self.childNodes.remove(child)

    def replaceChild(self, oldChild, newChild):
        position = self.childNodes.index(oldChild)
        self.removeChild(oldChild)
        self.insertChild(position, newChild)

    def appendChild(self, child):
        self.childNodes.append(child)
        child.parent = self

    def handleAttributes(self):
        pass

    def find(self, test, depth=0):
        """ Returns a list of descendants that pass the test function """
        matched_nodes = []
        for child in self.childNodes:
            if test(child):
                matched_nodes.append(child)
            if child.type == "element":
                matched_nodes += child.find(test, depth+1)
        return matched_nodes

    def toxml(self):
        if ENABLE_ATTRIBUTES:
            for child in self.childNodes:
                child.handleAttributes()

        buffer = ""
        if self.nodeName in ['h1', 'h2', 'h3', 'h4']:
            buffer += "\n"
        elif self.nodeName in ['li']:
            buffer += "\n "

        # Process children FIRST, then do the attributes

        childBuffer = ""

        if self.childNodes or self.nodeName in ['blockquote']:
            childBuffer += ">"
            for child in self.childNodes:
                childBuffer += child.toxml()
            if self.nodeName == 'p':
                childBuffer += "\n"
            elif self.nodeName == 'li':
                childBuffer += "\n "
            childBuffer += "</%s>" % self.nodeName
        else:
            childBuffer += "/>"


            
        buffer += "<" + self.nodeName

        if self.nodeName in ['p', 'li', 'ul', 'ol',
                             'h1', 'h2', 'h3', 'h4', 'h5', 'h6']:

            if not self.attribute_values.has_key("dir"):
                if self.bidi:
                    bidi = self.bidi
                else:
                    bidi = self.doc.bidi
                    
                if bidi=="rtl":
                    self.setAttribute("dir", "rtl")
        
        for attr in self.attributes:
            value = self.attribute_values[attr]
            value = self.doc.normalizeEntities(value,
                                               avoidDoubleNormalizing=True)
            buffer += ' %s="%s"' % (attr, value)


        # Now let's actually append the children

        buffer += childBuffer

        if self.nodeName in ['p', 'br ', 'li', 'ul', 'ol',
                             'h1', 'h2', 'h3', 'h4'] :
            buffer += "\n"

        return buffer


class TextNode:

    type = "text"
    attrRegExp = re.compile(r'\{@([^\}]*)=([^\}]*)}') # {@id=123}

    def __init__ (self, text):
        self.value = text        

    def attributeCallback(self, match):

        self.parent.setAttribute(match.group(1), match.group(2))

    def handleAttributes(self):
        self.value = self.attrRegExp.sub(self.attributeCallback, self.value)

    def toxml(self):

        text = self.value

        self.parent.setBidi(getBidiType(text))
        
        if not text.startswith(HTML_PLACEHOLDER_PREFIX):
            if self.parent.nodeName == "p":
                text = text.replace("\n", "\n   ")
            elif (self.parent.nodeName == "li"
                  and self.parent.childNodes[0]==self):
                text = "\n     " + text.replace("\n", "\n     ")
        text = self.doc.normalizeEntities(text)
        return text


class EntityReference:

    type = "entity_ref"

    def __init__(self, entity):
        self.entity = entity

    def handleAttributes(self):
        pass

    def toxml(self):
        return "&" + self.entity + ";"


"""
======================================================================
========================== PRE-PROCESSORS ============================
======================================================================

Preprocessors munge source text before we start doing anything too
complicated.

There are two types of preprocessors: TextPreprocessor and Preprocessor.

"""


class TextPreprocessor:
    '''
    TextPreprocessors are run before the text is broken into lines.
    
    Each TextPreprocessor implements a "run" method that takes a pointer to a
    text string of the document, modifies it as necessary and returns
    either the same pointer or a pointer to a new string.  
    
    TextPreprocessors must extend markdown.TextPreprocessor.
    '''

    def run(self, text):
        pass


class Preprocessor:
    '''
    Preprocessors are run after the text is broken into lines.

    Each preprocessor implements a "run" method that takes a pointer to a
    list of lines of the document, modifies it as necessary and returns
    either the same pointer or a pointer to a new list.  
    
    Preprocessors must extend markdown.Preprocessor.
    '''

    def run(self, lines):
        pass
 

class HtmlBlockPreprocessor(TextPreprocessor):
    """Removes html blocks from the source text and stores it."""
    
    def _get_left_tag(self, block):
        return block[1:].replace(">", " ", 1).split()[0].lower()


    def _get_right_tag(self, left_tag, block):
        return block.rstrip()[-len(left_tag)-2:-1].lower()

    def _equal_tags(self, left_tag, right_tag):
        
        if left_tag == 'div' or left_tag[0] in ['?', '@', '%']: # handle PHP, etc.
            return True
        if ("/" + left_tag) == right_tag:
            return True
        if (right_tag == "--" and left_tag == "--"):
            return True
        elif left_tag == right_tag[1:] \
            and right_tag[0] != "<":
            return True
        else:
            return False

    def _is_oneliner(self, tag):
        return (tag in ['hr', 'hr/'])

    
    def run(self, text):

        new_blocks = []
        text = text.split("\n\n")
        
        items = []
        left_tag = ''
        right_tag = ''
        in_tag = False # flag
        
        for block in text:
            if block.startswith("\n"):
                block = block[1:]

            if not in_tag:

                if block.startswith("<"):
                    
                    left_tag = self._get_left_tag(block)
                    right_tag = self._get_right_tag(left_tag, block)

                    if not (isBlockLevel(left_tag) \
                        or block[1] in ["!", "?", "@", "%"]):
                        new_blocks.append(block)
                        continue

                    if self._is_oneliner(left_tag):
                        new_blocks.append(block.strip())
                        continue
                        
                    if block[1] == "!":
                        # is a comment block
                        left_tag = "--"
                        right_tag = self._get_right_tag(left_tag, block)
                        # keep checking conditions below and maybe just append
                        
                    if block.rstrip().endswith(">") \
                        and self._equal_tags(left_tag, right_tag):
                        new_blocks.append(
                            self.stash.store(block.strip()))
                        continue
                    else: #if not block[1] == "!":
                        # if is block level tag and is not complete
                        items.append(block.strip())
                        in_tag = True
                        continue

                new_blocks.append(block)

            else:
                items.append(block.strip())
                
                right_tag = self._get_right_tag(left_tag, block)
                
                if self._equal_tags(left_tag, right_tag):
                    # if find closing tag
                    in_tag = False
                    new_blocks.append(
                        self.stash.store('\n\n'.join(items)))
                    items = []

        if items:
            new_blocks.append(self.stash.store('\n\n'.join(items)))
            new_blocks.append('\n')
            
        return "\n\n".join(new_blocks)

HTML_BLOCK_PREPROCESSOR = HtmlBlockPreprocessor()


class HeaderPreprocessor(Preprocessor):

    """
       Replaces underlined headers with hashed headers to avoid
       the nead for lookahead later.
    """

    def run (self, lines):

        i = -1
        while i+1 < len(lines):
            i = i+1
            if not lines[i].strip():
                continue

            if lines[i].startswith("#"):
                lines.insert(i+1, "\n")

            if (i+1 <= len(lines)
                  and lines[i+1]
                  and lines[i+1][0] in ['-', '=']):

                underline = lines[i+1].strip()

                if underline == "="*len(underline):
                    lines[i] = "# " + lines[i].strip()
                    lines[i+1] = ""
                elif underline == "-"*len(underline):
                    lines[i] = "## " + lines[i].strip()
                    lines[i+1] = ""

        return lines

HEADER_PREPROCESSOR = HeaderPreprocessor()


class LinePreprocessor(Preprocessor):
    """Deals with HR lines (needs to be done before processing lists)"""

    blockquote_re = re.compile(r'^(> )+')

    def run (self, lines):
        for i in range(len(lines)):
            prefix = ''
            m = self.blockquote_re.search(lines[i])
            if m : prefix = m.group(0)
            if self._isLine(lines[i][len(prefix):]):
                lines[i] = prefix + self.stash.store("<hr />", safe=True)
        return lines

    def _isLine(self, block):
        """Determines if a block should be replaced with an <HR>"""
        if block.startswith("    "): return 0  # a code block
        text = "".join([x for x in block if not x.isspace()])
        if len(text) <= 2:
            return 0
        for pattern in ['isline1', 'isline2', 'isline3']:
            m = RE.regExp[pattern].match(text)
            if (m and m.group(1)):
                return 1
        else:
            return 0

LINE_PREPROCESSOR = LinePreprocessor()


class ReferencePreprocessor(Preprocessor):
    ''' 
    Removes reference definitions from the text and stores them for later use.
    '''

    def run (self, lines):

        new_text = [];
        for line in lines:
            m = RE.regExp['reference-def'].match(line)
            if m:
                id = m.group(2).strip().lower()
                t = m.group(4).strip()  # potential title
                if not t:
                    self.references[id] = (m.group(3), t)
                elif (len(t) >= 2
                      and (t[0] == t[-1] == "\""
                           or t[0] == t[-1] == "\'"
                           or (t[0] == "(" and t[-1] == ")") ) ):
                    self.references[id] = (m.group(3), t[1:-1])
                else:
                    new_text.append(line)
            else:
                new_text.append(line)

        return new_text #+ "\n"

REFERENCE_PREPROCESSOR = ReferencePreprocessor()

"""
======================================================================
========================== 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, doc) - takes a match object and returns
                                a NanoDom node (as a part of the provided
                                doc) or None

All of python markdown's built-in patterns subclass from Patter,
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
"""

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

BACKTICK_RE = r'\`([^\`]*)\`'                    # `e= m*c^2`
DOUBLE_BACKTICK_RE =  r'\`\`(.*)\`\`'            # ``e=f("`")``
ESCAPE_RE = r'\\(.)'                             # \<
EMPHASIS_RE = r'\*([^\*]*)\*'                    # *emphasis*
STRONG_RE = r'\*\*(.*)\*\*'                      # **strong**
STRONG_EM_RE = r'\*\*\*([^_]*)\*\*\*'            # ***strong***

if SMART_EMPHASIS:
    EMPHASIS_2_RE = r'(?<!\S)_(\S[^_]*)_'        # _emphasis_
else:
    EMPHASIS_2_RE = r'_([^_]*)_'                 # _emphasis_

STRONG_2_RE = r'__([^_]*)__'                     # __strong__
STRONG_EM_2_RE = r'___([^_]*)___'                # ___strong___

LINK_RE = NOIMG + BRK + r'\s*\(([^\)]*)\)'               # [text](url)
LINK_ANGLED_RE = NOIMG + BRK + r'\s*\(<([^\)]*)>\)'      # [text](<url>)
IMAGE_LINK_RE = r'\!' + BRK + r'\s*\(([^\)]*)\)' # ![alttxt](http://x.com/)
REFERENCE_RE = NOIMG + BRK+ r'\s*\[([^\]]*)\]'           # [Google][3]
IMAGE_REFERENCE_RE = r'\!' + BRK + '\s*\[([^\]]*)\]' # ![alt text][2]
NOT_STRONG_RE = r'( \* )'                        # stand-alone * or _
AUTOLINK_RE = r'<(http://[^>]*)>'                # <http://www.123.com>
AUTOMAIL_RE = r'<([^> \!]*@[^> ]*)>'               # <me@example.com>
#HTML_RE = r'(\<[^\>]*\>)'                        # <...>
HTML_RE = r'(\<[a-zA-Z/][^\>]*\>)'               # <...>
ENTITY_RE = r'(&[\#a-zA-Z0-9]*;)'                # &amp;
LINE_BREAK_RE = r'  \n'                     # two spaces at end of line
LINE_BREAK_2_RE = r'  $'                    # two spaces at end of text

class Pattern:

    def __init__ (self, pattern):
        self.pattern = pattern
        self.compiled_re = re.compile("^(.*)%s(.*)$" % pattern, re.DOTALL)

    def getCompiledRegExp (self):
        return self.compiled_re

BasePattern = Pattern # for backward compatibility

class SimpleTextPattern (Pattern):

    def handleMatch(self, m, doc):
        return doc.createTextNode(m.group(2))

class SimpleTagPattern (Pattern):

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

    def handleMatch(self, m, doc):
        el = doc.createElement(self.tag)
        el.appendChild(doc.createTextNode(m.group(2)))
        return el

class SubstituteTagPattern (SimpleTagPattern):

    def handleMatch (self, m, doc):
        return doc.createElement(self.tag)

class BacktickPattern (Pattern):

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

    def handleMatch(self, m, doc):
        el = doc.createElement(self.tag)
        text = m.group(2).strip()
        #text = text.replace("&", "&amp;")
        el.appendChild(doc.createTextNode(text))
        return el


class DoubleTagPattern (SimpleTagPattern): 

    def handleMatch(self, m, doc):
        tag1, tag2 = self.tag.split(",")
        el1 = doc.createElement(tag1)
        el2 = doc.createElement(tag2)
        el1.appendChild(el2)
        el2.appendChild(doc.createTextNode(m.group(2)))
        return el1


class HtmlPattern (Pattern):

    def handleMatch (self, m, doc):
        rawhtml = m.group(2)
        inline = True
        place_holder = self.stash.store(rawhtml)
        return doc.createTextNode(place_holder)


class LinkPattern (Pattern):

    def handleMatch(self, m, doc):
        el = doc.createElement('a')
        el.appendChild(doc.createTextNode(m.group(2)))
        parts = m.group(9).split('"')
        # We should now have [], [href], or [href, title]
        if parts:
            el.setAttribute('href', parts[0].strip())
        else:
            el.setAttribute('href', "")
        if len(parts) > 1:
            # we also got a title
            title = '"' + '"'.join(parts[1:]).strip()
            title = dequote(title) #.replace('"', "&quot;")
            el.setAttribute('title', title)
        return el


class ImagePattern (Pattern):

    def handleMatch(self, m, doc):
        el = doc.createElement('img')
        src_parts = m.group(9).split()
        if src_parts:
            el.setAttribute('src', src_parts[0])
        else:
            el.setAttribute('src', "")
        if len(src_parts) > 1:
            el.setAttribute('title', dequote(" ".join(src_parts[1:])))
        if ENABLE_ATTRIBUTES:
            text = doc.createTextNode(m.group(2))
            el.appendChild(text)
            text.handleAttributes()
            truealt = text.value
            el.childNodes.remove(text)
        else:
            truealt = m.group(2)
        el.setAttribute('alt', truealt)
        return el

class ReferencePattern (Pattern):

    def handleMatch(self, m, doc):

        if m.group(9):
            id = m.group(9).lower()
        else:
            # if we got something like "[Google][]"
            # we'll use "google" as the id
            id = m.group(2).lower()

        if not self.references.has_key(id): # ignore undefined refs
            return None
        href, title = self.references[id]
        text = m.group(2)
        return self.makeTag(href, title, text, doc)

    def makeTag(self, href, title, text, doc):
        el = doc.createElement('a')
        el.setAttribute('href', href)
        if title:
            el.setAttribute('title', title)
        el.appendChild(doc.createTextNode(text))
        return el


class ImageReferencePattern (ReferencePattern):

    def makeTag(self, href, title, text, doc):
        el = doc.createElement('img')
        el.setAttribute('src', href)
        if title:
            el.setAttribute('title', title)
        el.setAttribute('alt', text)
        return el


class AutolinkPattern (Pattern):

    def handleMatch(self, m, doc):
        el = doc.createElement('a')
        el.setAttribute('href', m.group(2))
        el.appendChild(doc.createTextNode(m.group(2)))
        return el

class AutomailPattern (Pattern):

    def handleMatch(self, m, doc):
        el = doc.createElement('a')
        email = m.group(2)
        if email.startswith("mailto:"):
            email = email[len("mailto:"):]
        for letter in email:
            entity = doc.createEntityReference("#%d" % ord(letter))
            el.appendChild(entity)
        mailto = "mailto:" + email
        mailto = "".join(['&#%d;' % ord(letter) for letter in mailto])
        el.setAttribute('href', mailto)
        return el

ESCAPE_PATTERN          = SimpleTextPattern(ESCAPE_RE)
NOT_STRONG_PATTERN      = SimpleTextPattern(NOT_STRONG_RE)

BACKTICK_PATTERN        = BacktickPattern(BACKTICK_RE)
DOUBLE_BACKTICK_PATTERN = BacktickPattern(DOUBLE_BACKTICK_RE)
STRONG_PATTERN          = SimpleTagPattern(STRONG_RE, 'strong')
STRONG_PATTERN_2        = SimpleTagPattern(STRONG_2_RE, 'strong')
EMPHASIS_PATTERN        = SimpleTagPattern(EMPHASIS_RE, 'em')
EMPHASIS_PATTERN_2      = SimpleTagPattern(EMPHASIS_2_RE, 'em')

STRONG_EM_PATTERN       = DoubleTagPattern(STRONG_EM_RE, 'strong,em')
STRONG_EM_PATTERN_2     = DoubleTagPattern(STRONG_EM_2_RE, 'strong,em')

LINE_BREAK_PATTERN      = SubstituteTagPattern(LINE_BREAK_RE, 'br ')
LINE_BREAK_PATTERN_2    = SubstituteTagPattern(LINE_BREAK_2_RE, 'br ')

LINK_PATTERN            = LinkPattern(LINK_RE)
LINK_ANGLED_PATTERN     = LinkPattern(LINK_ANGLED_RE)
IMAGE_LINK_PATTERN      = ImagePattern(IMAGE_LINK_RE)
IMAGE_REFERENCE_PATTERN = ImageReferencePattern(IMAGE_REFERENCE_RE)
REFERENCE_PATTERN       = ReferencePattern(REFERENCE_RE)

HTML_PATTERN            = HtmlPattern(HTML_RE)
ENTITY_PATTERN          = HtmlPattern(ENTITY_RE)

AUTOLINK_PATTERN        = AutolinkPattern(AUTOLINK_RE)
AUTOMAIL_PATTERN        = AutomailPattern(AUTOMAIL_RE)


"""
======================================================================
========================== POST-PROCESSORS ===========================
======================================================================

Markdown also allows post-processors, which are similar to
preprocessors in that they need to implement a "run" method. However,
they are run after core processing.

There are two types of post-processors: Postprocessor and TextPostprocessor
"""


class Postprocessor:
    '''
    Postprocessors are run before the dom it converted back into text.
    
    Each Postprocessor implements a "run" method that takes a pointer to a
    NanoDom document, modifies it as necessary and returns a NanoDom 
    document.
    
    Postprocessors must extend markdown.Postprocessor.

    There are currently no standard post-processors, but the footnote
    extension uses one.
    '''

    def run(self, dom):
        pass



class TextPostprocessor:
    '''
    TextPostprocessors are run after the dom it converted back into text.
    
    Each TextPostprocessor implements a "run" method that takes a pointer to a
    text string, modifies it as necessary and returns a text string.
    
    TextPostprocessors must extend markdown.TextPostprocessor.
    '''

    def run(self, text):
        pass


class RawHtmlTextPostprocessor(TextPostprocessor):

    def __init__(self):
        pass

    def run(self, text):
        for i in range(self.stash.html_counter):
            html, safe  = self.stash.rawHtmlBlocks[i]
            if self.safeMode and not safe:
                if str(self.safeMode).lower() == 'escape':
                    html = self.escape(html)
                elif str(self.safeMode).lower() == 'remove':
                    html = ''
                else:
                    html = HTML_REMOVED_TEXT
                                   
            text = text.replace("<p>%s\n</p>" % (HTML_PLACEHOLDER % i),
                              html + "\n")
            text =  text.replace(HTML_PLACEHOLDER % i, html)
        return text

    def escape(self, html):
        ''' Basic html escaping '''
        html = html.replace('&', '&amp;')
        html = html.replace('<', '&lt;')
        html = html.replace('>', '&gt;')
        return html.replace('"', '&quot;')

RAWHTMLTEXTPOSTPROCESSOR = RawHtmlTextPostprocessor()

"""
======================================================================
========================== MISC AUXILIARY CLASSES ====================
======================================================================
"""

class HtmlStash:
    """This class is used for stashing HTML objects that we extract
        in the beginning and replace with place-holders."""

    def __init__ (self):
        self.html_counter = 0 # for counting inline html segments
        self.rawHtmlBlocks=[]

    def store(self, html, safe=False):
        """Saves an HTML segment for later reinsertion.  Returns a
           placeholder string that needs to be inserted into the
           document.

           @param html: an html segment
           @param safe: label an html segment as safe for safemode
           @param inline: label a segmant as inline html
           @returns : a placeholder string """
        self.rawHtmlBlocks.append((html, safe))
        placeholder = HTML_PLACEHOLDER % self.html_counter
        self.html_counter += 1
        return placeholder


class BlockGuru:

    def _findHead(self, lines, fn, allowBlank=0):

        """Functional magic to help determine boundaries of indented
           blocks.

           @param lines: an array of strings
           @param fn: a function that returns a substring of a string
                      if the string matches the necessary criteria
           @param allowBlank: specifies whether it's ok to have blank
                      lines between matching functions
           @returns: a list of post processes items and the unused
                      remainder of the original list"""

        items = []
        item = -1

        i = 0 # to keep track of where we are

        for line in lines:

            if not line.strip() and not allowBlank:
                return items, lines[i:]

            if not line.strip() and allowBlank:
                # If we see a blank line, this _might_ be the end
                i += 1

                # Find the next non-blank line
                for j in range(i, len(lines)):
                    if lines[j].strip():
                        next = lines[j]
                        break
                else:
                    # There is no more text => this is the end
                    break

                # Check if the next non-blank line is still a part of the list

                part = fn(next)

                if part:
                    items.append("")
                    continue
                else:
                    break # found end of the list

            part = fn(line)

            if part:
                items.append(part)
                i += 1
                continue
            else:
                return items, lines[i:]
        else:
            i += 1

        return items, lines[i:]


    def detabbed_fn(self, line):
        """ An auxiliary method to be passed to _findHead """
        m = RE.regExp['tabbed'].match(line)
        if m:
            return m.group(4)
        else:
            return None


    def detectTabbed(self, lines):

        return self._findHead(lines, self.detabbed_fn,
                              allowBlank = 1)


def print_error(string):
    """Print an error string to stderr"""
    sys.stderr.write(string +'\n')


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

"""
======================================================================
========================== CORE MARKDOWN =============================
======================================================================

This stuff is ugly, so if you are thinking of extending the syntax,
see first if you can do it via pre-processors, post-processors,
inline patterns or a combination of the three.
"""

class CorePatterns:
    """This class is scheduled for removal as part of a refactoring
        effort."""

    patterns = {
        'header':          r'(#*)([^#]*)(#*)', # # A title
        'reference-def':   r'(\ ?\ ?\ ?)\[([^\]]*)\]:\s*([^ ]*)(.*)',
                           # [Google]: http://www.google.com/
        'containsline':    r'([-]*)$|^([=]*)', # -----, =====, etc.
        'ol':              r'[ ]{0,3}[\d]*\.\s+(.*)', # 1. text
        'ul':              r'[ ]{0,3}[*+-]\s+(.*)', # "* text"
        'isline1':         r'(\**)', # ***
        'isline2':         r'(\-*)', # ---
        'isline3':         r'(\_*)', # ___
        'tabbed':          r'((\t)|(    ))(.*)', # an indented line
        'quoted':          r'> ?(.*)', # a quoted block ("> ...")
    }

    def __init__ (self):

        self.regExp = {}
        for key in self.patterns.keys():
            self.regExp[key] = re.compile("^%s$" % self.patterns[key],
                                          re.DOTALL)

        self.regExp['containsline'] = re.compile(r'^([-]*)$|^([=]*)$', re.M)

RE = CorePatterns()


class Markdown:
    """ Markdown formatter class for creating an html document from
        Markdown text """


    def __init__(self, source=None,  # depreciated
                 extensions=[],
                 extension_configs=None,
                 safe_mode = False):
        """Creates a new Markdown instance.

           @param source: The text in Markdown format. Depreciated!
           @param extensions: A list if extensions.
           @param extension-configs: Configuration setting for extensions.
           @param safe_mode: Disallow raw html. """

        self.source = source
        if source is not None:
            message(WARN, "The `source` arg of Markdown.__init__() is depreciated and will be removed in the future. Use `instance.convert(source)` instead.")
        self.safeMode = safe_mode
        self.blockGuru = BlockGuru()
        self.registeredExtensions = []
        self.stripTopLevelTags = 1
        self.docType = ""

        self.textPreprocessors = [HTML_BLOCK_PREPROCESSOR]

        self.preprocessors = [HEADER_PREPROCESSOR,
                              LINE_PREPROCESSOR,
                              # A footnote preprocessor will
                              # get inserted here
                              REFERENCE_PREPROCESSOR]


        self.postprocessors = [] # a footnote postprocessor will get
                                 # inserted later

        self.textPostprocessors = [# a footnote postprocessor will get
                                   # inserted here
                                   RAWHTMLTEXTPOSTPROCESSOR]

        self.prePatterns = []
        

        self.inlinePatterns = [DOUBLE_BACKTICK_PATTERN,
                               BACKTICK_PATTERN,
                               ESCAPE_PATTERN,
                               REFERENCE_PATTERN,
                               LINK_ANGLED_PATTERN,
                               LINK_PATTERN,
                               IMAGE_LINK_PATTERN,
			                   IMAGE_REFERENCE_PATTERN,
			                   AUTOLINK_PATTERN,
                               AUTOMAIL_PATTERN,
                               LINE_BREAK_PATTERN_2,
                               LINE_BREAK_PATTERN,
                               HTML_PATTERN,
                               ENTITY_PATTERN,
                               NOT_STRONG_PATTERN,
                               STRONG_EM_PATTERN,
                               STRONG_EM_PATTERN_2,
                               STRONG_PATTERN,
                               STRONG_PATTERN_2,
                               EMPHASIS_PATTERN,
                               EMPHASIS_PATTERN_2
                               # The order of the handlers matters!!!
                               ]

        self.registerExtensions(extensions = extensions,
                                configs = extension_configs)

        self.reset()


    def registerExtensions(self, extensions, configs):

        if not configs:
            configs = {}

        for ext in extensions:

            extension_module_name = "mdx_" + ext

            try:
                module = __import__(extension_module_name)

            except:
                message(CRITICAL,
                        "couldn't load extension %s (looking for %s module)"
                        % (ext, extension_module_name) )
            else:

                if configs.has_key(ext):
                    configs_for_ext = configs[ext]
                else:
                    configs_for_ext = []
                extension = module.makeExtension(configs_for_ext)    
                extension.extendMarkdown(self, globals())




    def registerExtension(self, extension):
        """ This gets called by the extension """
        self.registeredExtensions.append(extension)

    def reset(self):
        """Resets all state variables so that we can start
            with a new text."""
        self.references={}
        self.htmlStash = HtmlStash()

        HTML_BLOCK_PREPROCESSOR.stash = self.htmlStash
        LINE_PREPROCESSOR.stash = self.htmlStash
        REFERENCE_PREPROCESSOR.references = self.references
        HTML_PATTERN.stash = self.htmlStash
        ENTITY_PATTERN.stash = self.htmlStash
        REFERENCE_PATTERN.references = self.references
        IMAGE_REFERENCE_PATTERN.references = self.references
        RAWHTMLTEXTPOSTPROCESSOR.stash = self.htmlStash
        RAWHTMLTEXTPOSTPROCESSOR.safeMode = self.safeMode

        for extension in self.registeredExtensions:
            extension.reset()


    def _transform(self):
        """Transforms the Markdown text into a XHTML body document

           @returns: A NanoDom Document """

        # Setup the document

        self.doc = Document()
        self.top_element = self.doc.createElement("span")
        self.top_element.appendChild(self.doc.createTextNode('\n'))
        self.top_element.setAttribute('class', 'markdown')
        self.doc.appendChild(self.top_element)

        # Fixup the source text
        text = self.source
        text = text.replace("\r\n", "\n").replace("\r", "\n")
        text += "\n\n"
        text = text.expandtabs(TAB_LENGTH)

        # Split into lines and run the preprocessors that will work with
        # self.lines

        self.lines = text.split("\n")

        # Run the pre-processors on the lines
        for prep in self.preprocessors :
            self.lines = prep.run(self.lines)

        # Create a NanoDom tree from the lines and attach it to Document


        buffer = []
        for line in self.lines:
            if line.startswith("#"):
                self._processSection(self.top_element, buffer)
                buffer = [line]
            else:
                buffer.append(line)
        self._processSection(self.top_element, buffer)
        
        #self._processSection(self.top_element, self.lines)

        # Not sure why I put this in but let's leave it for now.
        self.top_element.appendChild(self.doc.createTextNode('\n'))

        # Run the post-processors
        for postprocessor in self.postprocessors:
            postprocessor.run(self.doc)

        return self.doc


    def _processSection(self, parent_elem, lines,
                        inList = 0, looseList = 0):

        """Process a section of a source document, looking for high
           level structural elements like lists, block quotes, code
           segments, html blocks, etc.  Some those then get stripped
           of their high level markup (e.g. get unindented) and the
           lower-level markup is processed recursively.

           @param parent_elem: A NanoDom element to which the content
                               will be added
           @param lines: a list of lines
           @param inList: a level
           @returns: None"""

        # Loop through lines until none left.
        while lines:

            # Check if this section starts with a list, a blockquote or
            # a code block

            processFn = { 'ul':     self._processUList,
                          'ol':     self._processOList,
                          'quoted': self._processQuote,
                          'tabbed': self._processCodeBlock}

            for regexp in ['ul', 'ol', 'quoted', 'tabbed']:
                m = RE.regExp[regexp].match(lines[0])
                if m:
                    processFn[regexp](parent_elem, lines, inList)
                    return

            # We are NOT looking at one of the high-level structures like
            # lists or blockquotes.  So, it's just a regular paragraph
            # (though perhaps nested inside a list or something else).  If
            # we are NOT inside a list, we just need to look for a blank
            # line to find the end of the block.  If we ARE inside a
            # list, however, we need to consider that a sublist does not
            # need to be separated by a blank line.  Rather, the following
            # markup is legal:
            #
            # * The top level list item
            #
            #     Another paragraph of the list.  This is where we are now.
            #     * Underneath we might have a sublist.
            #

            if inList:

                start, lines  = self._linesUntil(lines, (lambda line:
                                 RE.regExp['ul'].match(line)
                                 or RE.regExp['ol'].match(line)
                                                  or not line.strip()))

                self._processSection(parent_elem, start,
                                     inList - 1, looseList = looseList)
                inList = inList-1

            else: # Ok, so it's just a simple block

                paragraph, lines = self._linesUntil(lines, lambda line:
                                                     not line.strip())

                if len(paragraph) and paragraph[0].startswith('#'):
                    self._processHeader(parent_elem, paragraph)

                elif paragraph:
                    self._processParagraph(parent_elem, paragraph,
                                          inList, looseList)

            if lines and not lines[0].strip():
                lines = lines[1:]  # skip the first (blank) line


    def _processHeader(self, parent_elem, paragraph):
        m = RE.regExp['header'].match(paragraph[0])
        if m:
            level = len(m.group(1))
            h = self.doc.createElement("h%d" % level)
            parent_elem.appendChild(h)
            for item in self._handleInline(m.group(2).strip()):
                h.appendChild(item)
        else:
            message(CRITICAL, "We've got a problem header!")


    def _processParagraph(self, parent_elem, paragraph, inList, looseList):
        list = self._handleInline("\n".join(paragraph))

        if ( parent_elem.nodeName == 'li'
                and not (looseList or parent_elem.childNodes)):

            # If this is the first paragraph inside "li", don't
            # put <p> around it - append the paragraph bits directly
            # onto parent_elem
            el = parent_elem
        else:
            # Otherwise make a "p" element
            el = self.doc.createElement("p")
            parent_elem.appendChild(el)

        for item in list:
            el.appendChild(item)
 

    def _processUList(self, parent_elem, lines, inList):
        self._processList(parent_elem, lines, inList,
                         listexpr='ul', tag = 'ul')

    def _processOList(self, parent_elem, lines, inList):
        self._processList(parent_elem, lines, inList,
                         listexpr='ol', tag = 'ol')


    def _processList(self, parent_elem, lines, inList, listexpr, tag):
        """Given a list of document lines starting with a list item,
           finds the end of the list, breaks it up, and recursively
           processes each list item and the remainder of the text file.

           @param parent_elem: A dom element to which the content will be added
           @param lines: a list of lines
           @param inList: a level
           @returns: None"""

        ul = self.doc.createElement(tag)  # ul might actually be '<ol>'
        parent_elem.appendChild(ul)

        looseList = 0

        # Make a list of list items
        items = []
        item = -1

        i = 0  # a counter to keep track of where we are

        for line in lines: 

            loose = 0
            if not line.strip():
                # If we see a blank line, this _might_ be the end of the list
                i += 1
                loose = 1

                # Find the next non-blank line
                for j in range(i, len(lines)):
                    if lines[j].strip():
                        next = lines[j]
                        break
                else:
                    # There is no more text => end of the list
                    break

                # Check if the next non-blank line is still a part of the list
                if ( RE.regExp['ul'].match(next) or
                     RE.regExp['ol'].match(next) or 
                     RE.regExp['tabbed'].match(next) ):
                    # get rid of any white space in the line
                    items[item].append(line.strip())
                    looseList = loose or looseList
                    continue
                else:
                    break # found end of the list

            # Now we need to detect list items (at the current level)
            # while also detabing child elements if necessary

            for expr in ['ul', 'ol', 'tabbed']:

                m = RE.regExp[expr].match(line)
                if m:
                    if expr in ['ul', 'ol']:  # We are looking at a new item
                        #if m.group(1) :
                        # Removed the check to allow for a blank line
                        # at the beginning of the list item
                        items.append([m.group(1)])
                        item += 1
                    elif expr == 'tabbed':  # This line needs to be detabbed
                        items[item].append(m.group(4)) #after the 'tab'

                    i += 1
                    break
            else:
                items[item].append(line)  # Just regular continuation
                i += 1 # added on 2006.02.25
        else:
            i += 1

        # Add the dom elements
        for item in items:
            li = self.doc.createElement("li")
            ul.appendChild(li)

            self._processSection(li, item, inList + 1, looseList = looseList)

        # Process the remaining part of the section

        self._processSection(parent_elem, lines[i:], inList)


    def _linesUntil(self, lines, condition):
        """ A utility function to break a list of lines upon the
            first line that satisfied a condition.  The condition
            argument should be a predicate function.
            """

        i = -1
        for line in lines:
            i += 1
            if condition(line): break
        else:
            i += 1
        return lines[:i], lines[i:]

    def _processQuote(self, parent_elem, lines, inList):
        """Given a list of document lines starting with a quote finds
           the end of the quote, unindents it and recursively
           processes the body of the quote and the remainder of the
           text file.

    …