#!/usr/bin/env python """ Python-Markdown =============== Converts Markdown to HTML. Basic usage as a module: import markdown md = Markdown() html = md.convert(your_text_string) See 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: * * License: [GPL 2](http://www.gnu.org/copyleft/gpl.html) or BSD """ version = "1.7" version_info = (1,7,0,"final") __revision__ = "$Rev$" import re, sys, codecs, copy from urlparse import urlparse, urlunparse 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 thisorthat 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): """ Used by `markdownFromFile` to remove a "byte order mark" from the begining of an utf-8, utf-16 or utf-32 encoded file. """ 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): """ Used by HTMLBlockPreprocessor to check if a given tag is a block level element. """ 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("&"), "&"), (re.compile("<"), "<"), (re.compile(">"), ">")] ENTITY_NORMALIZATION_EXPRESSIONS_SOFT = [ (re.compile("&(?!\#)"), "&"), (re.compile("<"), "<"), (re.compile(">"), ">"), (re.compile("\""), """)] def getBidiType(text): """ Get Bi-directional text type. Used by TextNode to determine text direction. """ 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: """ Document root of the NanoDom. An instance stores DOM elements as children. """ def __init__ (self): """ Create a NanoDom document. """ self.bidi = "ltr" self.stripTopLevelTags = True def appendChild(self, child): """ Add a dom element as a child of the document root. """ self.documentElement = child child.isDocumentElement = True child.parentNode = self self.entities = {} def setBidi(self, bidi): """ Set text direction (right-left or left-right).""" if bidi: self.bidi = bidi def createElement(self, tag, textNode=None): """ Given a tag or textNode, return a dom element. """ el = Element(tag) el.doc = self if textNode: el.appendChild(self.createTextNode(textNode)) return el def createTextNode(self, text, type="text"): """ Return given text as a TextNode. """ node = TextNode(text) node.doc = self node.type = type return node def createEntityReference(self, entity): """ Return an html entitry reference (i.e.: `&`). """ if entity not in self.entities: self.entities[entity] = EntityReference(entity) return self.entities[entity] def createCDATA(self, text): """ Return the given text as a CDATA node. """ node = CDATA(text) node.doc = self return node def toxml (self): """ Convert document to xml and return a string. """ xml = self.documentElement.toxml() if self.stripTopLevelTags: xml = xml.strip()[23:-7] + "\n" return xml def normalizeEntities(self, text, avoidDoubleNormalizing=False): """ Return the given text as an html entity (i.e.: `<` => `>`). """ 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 a list of descendants that pass the test function """ return self.documentElement.find(test) def unlink(self): """ Cleanup: Remove all children from the document. """ self.documentElement.unlink() self.documentElement = None class CDATA: """ CDATA node type of NanoDom. """ type = "cdata" def __init__ (self, text): """ Create a CDATA node with given text. """ self.text = text def handleAttributes(self): """ Not implemented for CDATA node type. """ pass def toxml (self): """ Return CDATA node as a string. """ return "" class Element: """ Element node type of Nanodom. All html tags would most likely be represented as Elements. """ type = "element" def __init__ (self, tag): """ Create an Element node instance. """ self.nodeName = tag self.attributes = [] self.attribute_values = {} self.childNodes = [] self.bidi = None self.isDocumentElement = False def setBidi(self, bidi): """ Set text direction (i.e.: right-left or left-right). """ 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.parentNode.setBidi(bidi) def unlink(self): """ Cleanup: Remove all children of the Element. """ for child in self.childNodes: if child.type == "element": child.unlink() self.childNodes = None def setAttribute(self, attr, value): """ Assign an html/xml attribute to the Element (i.e.: id, class, href). """ if not attr in self.attributes: self.attributes.append(attr) self.attribute_values[attr] = value def insertChild(self, position, child): """ Insert a child Element at the given position. """ self.childNodes.insert(position, child) child.parentNode = self def removeChild(self, child): """ Remove the given child from the Element. """ self.childNodes.remove(child) def replaceChild(self, newChild, oldChild): """ Replace an old child Element with a new child Element. """ position = self.childNodes.index(oldChild) self.removeChild(oldChild) self.insertChild(position, newChild) def appendChild(self, child): """ Append a new child Element to the end of the child Elements. """ self.childNodes.append(child) child.parentNode = self def handleAttributes(self): """ Not implemented for Element node type. """ 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): """ Return the Element and all children as a string. """ 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 += "" % 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: """ A Text node type of the NanoDom. """ type = "text" attrRegExp = re.compile(r'\{@([^\}]*)=([^\}]*)}') # {@id=123} def __init__ (self, text): """ Create a TextNode with the given text. """ self.value = text def attributeCallback(self, match): """ Regex callback method to set attribute on parent. """ self.parentNode.setAttribute(match.group(1), match.group(2)) def handleAttributes(self): """ Parse and assign attributes to the parent Element. """ self.value = self.attrRegExp.sub(self.attributeCallback, self.value) def toxml(self): """ Return the TextNode as a string. """ text = self.value self.parentNode.setBidi(getBidiType(text)) if not text.startswith(HTML_PLACEHOLDER_PREFIX): if self.parentNode.nodeName == "p": text = text.replace("\n", "\n ") elif (self.parentNode.nodeName == "li" and self.parentNode.childNodes[0]==self): text = "\n " + text.replace("\n", "\n ") text = self.doc.normalizeEntities(text) return text class EntityReference: """ EntityReference node type of NanoDom. """ type = "entity_ref" def __init__(self, entity): """ Create an EntityReference of the given entity. """ self.entity = entity def handleAttributes(self): """ Not implemented for EntityReference. """ pass def toxml(self): """ Return the EntityReference as a string. """ 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): """ Each subclass of TextPreprocessor should override the `run` method, which takes the document text as a single string and returns the (possibly modified) document as a single string. """ 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): """ Each subclass of Preprocessor should override the `run` method, which takes the document as a list of strings split by newlines and returns the (possibly modified) list of lines. """ pass class HtmlBlockPreprocessor(TextPreprocessor): """ Remove html blocks from the source text and store them for later retrieval. """ 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): """ Find and remove raw html from 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): """ Replace underlined headers with hashed headers to avoid the nead for lookahead later. """ def run (self, lines): """ Find and replace underlined headers. """ 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): """ Convert HR lines to raw html and store (needs to be done before processing lists). """ blockquote_re = re.compile(r'^(> )+') def run (self, lines): """ Find a store HR 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("
", safe=True) return lines def _isLine(self, block): """Determine if a block should be replaced with an
""" 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): """ Remove reference definitions from the text and store them for later use. """ def run (self, lines): """ Remove and store reference defs. """ 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 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 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'(?\)' # [text]() 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'<((?:f|ht)tps?://[^>]*)>' # AUTOMAIL_RE = r'<([^> \!]*@[^> ]*)>' # #HTML_RE = r'(\<[^\>]*\>)' # <...> HTML_RE = r'(\<[a-zA-Z/][^\>]*\>)' # <...> ENTITY_RE = r'(&[\#a-zA-Z0-9]*;)' # & LINE_BREAK_RE = r' \n' # two spaces at end of line LINE_BREAK_2_RE = r' $' # two spaces at end of text class Pattern: """Base class that inline patterns subclass. """ def __init__ (self, pattern): """ Create an instant of an inline pattern. Keyword arguments: * pattern: A regular expression that matches a pattern """ self.pattern = pattern self.compiled_re = re.compile("^(.*?)%s(.*?)$" % pattern, re.DOTALL) # Api for Markdown to pass safe_mode into instance self.safe_mode = False def getCompiledRegExp (self): """ Return a compiled regular expression. """ return self.compiled_re def handleMatch(self, m, doc): """ Return a NanoDom element from the given match. Subclasses should override this method. Keyword arguments: * m: A re match object containing a match of the pattern. * doc: An instance of a NanoDom Document. """ pass BasePattern = Pattern # for backward compatibility class SimpleTextPattern (Pattern): """ Return a simple TextNode of group(2) of a Pattern. """ def handleMatch(self, m, doc): return doc.createTextNode(m.group(2)) class SimpleTagPattern (Pattern): """ Return NanoDom Element of type `tag` with a child TextNode of group(2) of a 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): """ Return a NanoDom ELement of type `tag` with no children. """ def handleMatch (self, m, doc): return doc.createElement(self.tag) class BacktickPattern (Pattern): """ Return a NanoDom `` Element containing the matching text. """ 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("&", "&") el.appendChild(doc.createTextNode(text)) return el class DoubleTagPattern (SimpleTagPattern): """ Return a TextNode nested in tag2 nested in tag1. Usefull for strong emphasis etc. """ 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): """ Store raw inline html and return a placeholder. """ 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): """ Return a NanoDom link Element from the given match. """ 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', self.sanatize_url(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('"', """) el.setAttribute('title', title) return el def sanatize_url(self, url): """ Sanitize a url against xss attacks in "safe_mode". Rather than specifically blacklisting `javascript:alert("XSS")` and all its aliases (see ), 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`. """ locless_schemes = ['', 'mailto', 'news'] scheme, netloc, path, params, query, fragment = url = urlparse(url) safe_url = False if netloc != '' or scheme in locless_schemes: safe_url = True for part in url[2:]: if ":" in part: safe_url = False if self.safe_mode and not safe_url: return '' else: return urlunparse(url) class ImagePattern(LinkPattern): """ Return a NanoDom img Element from the given match. """ def handleMatch(self, m, doc): el = doc.createElement('img') src_parts = m.group(9).split() if src_parts: el.setAttribute('src', self.sanatize_url(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(LinkPattern): """ Match to a stored reference and return a NanoDom link Element. """ 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', self.sanatize_url(href)) if title: el.setAttribute('title', title) el.appendChild(doc.createTextNode(text)) return el class ImageReferencePattern (ReferencePattern): """ Match to a stored reference and return a NanoDom img Element. """ def makeTag(self, href, title, text, doc): el = doc.createElement('img') el.setAttribute('src', self.sanatize_url(href)) if title: el.setAttribute('title', title) el.setAttribute('alt', text) return el class AutolinkPattern (Pattern): """ Return a link Element given an autolink (``). """ 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): """ Return a mailto link Element given an automail link (``). """ 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): """ Subclasses of Postprocessor should implement a `run` method, which takes a NanoDOm document and returns a (possably modified) NanoDom document. """ 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): """ Subclasses of TextPostprocessor should implement a `run` method, which takes the html document as a single text string and returns a (possably modified) string. """ pass class RawHtmlTextPostprocessor(TextPostprocessor): """ Restore raw html to the document. """ def __init__(self): pass def run(self, text): """ Iterate over html stash and restore "safe" html. """ 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("

%s\n

" % (HTML_PLACEHOLDER % i), html + "\n") text = text.replace(HTML_PLACEHOLDER % i, html) return text def escape(self, html): """ Basic html escaping """ html = html.replace('&', '&') html = html.replace('<', '<') html = html.replace('>', '>') return html.replace('"', '"') 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): """ Create a HtmlStash. """ 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. Keyword arguments: * html: an html segment * safe: label an html segment as safe for safemode Returns : a placeholder string """ self.rawHtmlBlocks.append((html, safe)) placeholder = HTML_PLACEHOLDER % self.html_counter self.html_counter += 1 return placeholder class BlockGuru: """ Parse document for block level constructs (paragraphs, lists, etc.).""" def _findHead(self, lines, fn, allowBlank=0): """ Functional magic to help determine boundaries of indented blocks. Keyword arguments: * lines: an array of strings * fn: a function that returns a substring of a string if the string matches the necessary criteria * 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): """ Find indented text and remove indent before further proccesing. """ return self._findHead(lines, self.detabbed_fn, allowBlank = 1) 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 class InlineStash: def __init__(self): self.prefix = "k@J!}" self.suffix = "lL5Qt" self._nodes = {} self.phLength = 4 + len(self.prefix) + len(self.suffix) def _genPlaceholder(self): hash = "%s%04d%s" % (self.prefix, len(self._nodes), self.suffix) return hash def isin(self, placeholder): return self._nodes.has_key(placeholder) def get(self, placeholder): return self._nodes.get(placeholder) def add(self, node): pholder = self._genPlaceholder() self._nodes[pholder] = node return pholder """ ====================================================================== ========================== CORE MARKDOWN ============================= ====================================================================== This stuff is hard, 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'(#{1,6})[ \t]*(.*?)[ \t]*(#*)', # # A title 'reference-def': r'(\ ?\ ?\ ?)\[([^\]]*)\]:\s*([^ ]*)(.*)', # [Google]: http://www.google.com/ 'containsline': r'([-]*)$|^([=]*)', # -----, =====, etc. 'ol': r'[ ]{0,3}[\d]*\.\s+(.*)', # 1. text 'ul': r'[ ]{0,3}[*+-]\s+(.*)', # "* text" 'isline1': r'(\**)', # *** 'isline2': r'(\-*)', # --- 'isline3': r'(\_*)', # ___ 'tabbed': r'((\t)|( ))(.*)', # an indented line 'quoted': r'[ ]{0,2}> ?(.*)', # a quoted block ("> ...") } def __init__ (self): self.regExp = {} for key in self.patterns.keys(): self.regExp[key] = re.compile("^%s$" % 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={}, safe_mode = False): """ Creates a new Markdown instance. Keyword arguments: * source: The text in Markdown format. Depreciated! * extensions: A list of extensions. If they are of type string, the module mdx_name.py will be loaded. If they are a subclass of markdown.Extension, they will be used as-is. * extension-configs: Configuration setting for extensions. * safe_mode: Disallow raw html. One of "remove", "replace" or "escape". """ 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.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 = [] # temporarily disabled patterns '''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,''' '''self.inlinePatterns = [LINK_PATTERN, STRONG_PATTERN, STRONG_PATTERN_2, EMPHASIS_PATTERN, EMPHASIS_PATTERN_2 # The order of the handlers matters!!! ]''' 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.inlineStash = InlineStash() self.registerExtensions(extensions = extensions, configs = extension_configs) self.reset() def registerExtensions(self, extensions, configs): """ Register extensions with this instance of Markdown. Keyword aurguments: * extensions: A list of extensions, which can either be strings or objects. See the docstring on Markdown. * configs: A dictionary mapping module names to config options. """ for ext in extensions: if isinstance(ext, basestring): ext = load_extension(ext, configs.get(ext, [])) elif hasattr(ext, 'extendMarkdown'): # Looks like an Extension. # Nothing to do here. pass else: message(ERROR, "Incorrect type! Extension '%s' is " "neither a string or an Extension." %(repr(ext))) continue ext.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() self.inlineStash = InlineStash() 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() for pattern in self.inlinePatterns: pattern.safe_mode = self.safeMode def _transform(self): """Transform 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) # Split into lines and run the preprocessors that will work with # self.lines self.lines = self.source.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. Keyword arguments: * parent_elem: A NanoDom element to which the content will be added. * lines: a list of lines * 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) h.appendChild(self.doc.createTextNode(m.group(2).strip(), "inline")) 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

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) el.appendChild(self.doc.createTextNode("\n".join(paragraph), "inline")) #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. Keyword arguments: * parent_elem: A dom element to which the content will be added * lines: a list of lines * inList: a level Returns: None """ ul = self.doc.createElement(tag) # ul might actually be '

    ' 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. Keyword arguments: * parent_elem: DOM element to which the content will be added * lines: a list of lines * inList: a level Returns: None """ dequoted = [] i = 0 blank_line = False # allow one blank line between paragraphs for line in lines: m = RE.regExp['quoted'].match(line) if m: dequoted.append(m.group(1)) i += 1 blank_line = False elif not blank_line and line.strip() != '': dequoted.append(line) i += 1 elif not blank_line and line.strip() == '': dequoted.append(line) i += 1 blank_line = True else: break blockquote = self.doc.createElement('blockquote') parent_elem.appendChild(blockquote) self._processSection(blockquote, dequoted, inList) self._processSection(parent_elem, lines[i:], inList) def _processCodeBlock(self, parent_elem, lines, inList): """ Given a list of document lines starting with a code block finds the end of the block, puts it into the dom verbatim wrapped in ("
    ") and recursively processes the
        the remainder of the text file.

        Keyword arguments:
        
        * parent_elem: DOM element to which the content will be added
        * lines: a list of lines
        * inList: a level
        
        Returns: None
        
        """

        detabbed, theRest = self.blockGuru.detectTabbed(lines)

        pre = self.doc.createElement('pre')
        code = self.doc.createElement('code')
        parent_elem.appendChild(pre)
        pre.appendChild(code)
        text = "\n".join(detabbed).rstrip()+"\n"
        #text = text.replace("&", "&")
        code.appendChild(self.doc.createTextNode(text))
        self._processSection(parent_elem, theRest, inList)        
        
    def _handleInline(self, data, patternIndex=0):
        """
        Processinf string with inline patterns and replasing it
        with placeholders

        Keyword arguments:
        
        * data: A line of Markdown text
        * patternIndex: The index of the inlinePattern to start with
        
        Return: String with placeholders. 
        
        """
        
        while patternIndex < len(self.inlinePatterns):
            
            data, matched = self._applyInline(self.inlinePatterns[patternIndex],
                                     data, patternIndex)
            if not matched:
                patternIndex += 1
        
        return data
        #return self._processPlaceholders(data)
   
    
    def _processPlaceholders(self, data):
        """
        Processes string with placeholders and generates DOM tree.
        
        * data: string with placeholders instead of DOM elements.

        Returns: NanoDOM Document object with applied inline patterns.
        """
 
        result = []
        prefix = self.inlineStash.prefix
        strartIndex = 0
        while data:
            
            index = data.find(prefix, strartIndex)
            if index != -1:
                
                phEndIndex = index + self.inlineStash.phLength
                placeHolder = data[index: phEndIndex]
                
                if self.inlineStash.isin(placeHolder):
                    
                    if index > 0:
                        
                        textNode = self.doc.createTextNode(data[strartIndex:index])
    
                        result.append(textNode)
                        
                    node = self.inlineStash.get(placeHolder)

                    if isinstance(node, Element):
        
                        for child in node.childNodes:
                            if isinstance(child, TextNode):
                                childResult = self._processPlaceholders(child.value)
                                pos = node.childNodes.index(child)
                                node.removeChild(child)
                                for newChild in childResult:
                                    node.insertChild(pos, newChild)
                                    
                        result.append(node)
                        
                    strartIndex = phEndIndex
   
                else:
            
                    strartIndex = index + len(prefix)
            else:
                text = self.doc.createTextNode(data[strartIndex:])
                result.append(text)
                data = ""
                    
        return result
        
   

    def _applyInline(self, pattern, data, patternIndex):
        """ 
        Given a pattern name, this function checks if the line
        fits the pattern, creates the necessary elements, adds it 
        to InlineStash, and returns string with placeholders,
        instead of DOM elements.
        
        Keyword arguments:
        
        * data: the text to be processed
        * pattern: the pattern to be checked
        * patternIndex: index of current pattern

        Returns: String with placeholders.
        """
        
        match = pattern.getCompiledRegExp().match(data)
 
        if not match:
            return data, False

        node = pattern.handleMatch(match, self.doc)
        
        if isinstance(node, Element):
            for child in node.childNodes:
                if isinstance(child, TextNode):
                    child.value = self._handleInline(child.value, patternIndex)
        
        
        pholder = self.inlineStash.add(node)

        return "%s%s%s" % (match.group(1), pholder, match.groups()[-1]), True
        
    
    def _processTree(self, el):
        """
        Processing NanoDOM markdown tree, and applying inline patterns
        
        Keyword arguments:
        
        * el - parent element of Document.

        Returns: NanoDOM Document object with applied inline patterns.
        """
        
        stack = [el]
        while stack:
            currElement = stack.pop()
            insertQueue = []
            for child in currElement.childNodes:
                
                if child.type == "inline":

                    lst = self._processPlaceholders(self._handleInline(
                                                    child.value))
             
                    pos = currElement.childNodes.index(child)
                    
                    insertQueue.append((pos, lst))
                    
                elif child.type == "element":
                    stack.append(child) 
                      
            for pos, lst in insertQueue:
                del currElement.childNodes[pos]
                for newChild in lst:
                    currElement.insertChild(pos, newChild)
                    pos += 1 

    def applyInlinePatterns(self, markdownTree):
        """
        Retrun NanoDOM markdown tree, with applied
        inline paterns
        
        Keyword arguments:
        
        * markdownTree: NanoDOM Document object, reppresenting Markdown tree.

        Returns: NanoDOM Document object.
        """
        
        self.doc = markdownTree
        
        el = markdownTree.documentElement
                
        self._processTree(el)
            
        return self.doc

        
    def markdownToTree(self, source=None):
        """
        Retrun NanoDOM markdown tree, without applying
        inline paterns
        
        Keyword arguments:
        
        * source: An ascii or unicode string of Markdown formated text.

        Returns: NanoDOM Document object.
        """
        if source is not None: #Allow blank string
            self.source = source
            
        if not self.source:
            return u""
        
        try:
            self.source = unicode(self.source)
        except UnicodeDecodeError:
            message(CRITICAL, 'UnicodeDecodeError: Markdown only accepts unicode or ascii  input.')
            return u""
        
        # Fixup the source text
        self.source = self.source.replace("\r\n", "\n").replace("\r", "\n")
        self.source += "\n\n"
        self.source = self.source.expandtabs(TAB_LENGTH)

        for pp in self.textPreprocessors:
            self.source = pp.run(self.source)
        
        markdownTree = self._transform()
        
        
        return markdownTree
        
        

    def convert (self, source=None):
        """
        Return the document in XHTML format.

        Keyword arguments:
        
        * source: An ascii or unicode string of Markdown formated text.

        Returns: A serialized XHTML body.

        """

        tree = self.markdownToTree(source)
 
        xml = self.applyInlinePatterns(tree).toxml()

        for pp in self.textPostprocessors:
            xml = pp.run(xml)

        return (self.docType + xml).strip()


    def __str__(self):
        ''' Report info about instance. Markdown always returns unicode. '''

        if self.source is None:
            status = 'in which no source text has been assinged.'
        else:
            status = 'which contains %d chars and %d line(s) of source.'%\
                     (len(self.source), self.source.count('\n')+1)
        return 'An instance of "%s" %s'% (self.__class__, status)

    __unicode__ = convert # markdown should always return a unicode string





# ====================================================================

def markdownFromFile(input = None,
                     output = None,
                     extensions = [],
                     encoding = None,
                     message_threshold = CRITICAL,
                     safe = False):
    """
    Convenience wrapper function that takes a filename as input.

    Used from the command-line, although may be useful in other situations. 
    Decodes the file using the provided encoding (defaults to utf-8), passes 
    the file content to markdown, and outputs the html to either the provided
    filename or stdout in the same encoding as the source file.

    **Note:** This is the only place that decoding and encoding takes place
    in Python-Markdown.

    Keyword arguments:

    * input: Name of source text file.
    * output: Name of output file. Writes to stdout if `None`.
    * extensions: A list of extension names (may contain config args).  
    * encoding: Encoding of input and output files. Defaults to utf-8.
    * message_threshold: Error reporting level.
    * safe_mode: Disallow raw html. One of "remove", "replace" or "escape".

    Returns: An HTML document as a string.

    """

    global console_hndlr
    console_hndlr.setLevel(message_threshold)

    message(DEBUG, "input file: %s" % input)

    if not encoding:
        encoding = "utf-8"

    input_file = codecs.open(input, mode="r", encoding=encoding)
    text = input_file.read()
    input_file.close()

    text = removeBOM(text, encoding)

    new_text = markdown(text, extensions, safe_mode = safe)

    if output:
        output_file = codecs.open(output, "w", encoding=encoding)
        output_file.write(new_text)
        output_file.close()

    else:
        sys.stdout.write(new_text.encode(encoding))

def markdown(text,
             extensions = [],
             safe_mode = False):
    """
    Convenience wrapper function for `Markdown` class.

    Useful in a typical use case. Initializes an instance of the `Markdown` 
    class, loads any extensions and runs the parser on the given text. 

    Keyword arguments:

    * text: An ascii or Unicode string of Markdown formatted text.
    * extensions: A list of extension names (may contain config args).  
    * safe_mode: Disallow raw html. One of "remove", "replace" or "escape".

    Returns: An HTML document as a string.

    """
    message(DEBUG, "in markdown.markdown(), received text:\n%s" % text)

    extensions = [load_extension(e) for e in extensions]

    md = Markdown(extensions=extensions,
                  safe_mode = safe_mode)

    return md.convert(text)
        

class Extension:
    """ Base class for extensions to subclass. """
    def __init__(self, configs = {}):
        """ 
        Create an instance of an Extention. 
        
        Keyword arguments:

        * configs: A dict of configuration setting used by an Extension.
        """
        self.config = configs

    def getConfig(self, key):
        """ Return a setting for the given key or an empty string. """
        if self.config.has_key(key):
            return self.config[key][0]
        else:
            return ""

    def getConfigInfo(self):
        """ Return all config settings as a list of tuples. """
        return [(key, self.config[key][1]) for key in self.config.keys()]

    def setConfig(self, key, value):
        """ Set a config setting for `key` with the given `value`. """
        self.config[key][0] = value


def load_extension(ext_name, configs = []):
    """ 
    Load extension by name, then return the module.
    
    The extension name may contain arguments as part of the string in the 
    following format:

        "extname(key1=value1,key2=value2)"
    
    Print an error message and exit on failure. 
    
    """

    # I am making the assumption that the order of config options
    # does not matter.
    configs = dict(configs)
    pos = ext_name.find("(") 
    if pos > 0:
        ext_args = ext_name[pos+1:-1]
        ext_name = ext_name[:pos]
        pairs = [x.split("=") for x in ext_args.split(",")]
        configs.update([(x.strip(), y.strip()) for (x, y) in pairs])

    extension_module_name = "mdx_" + ext_name

    try:
        module = __import__(extension_module_name)

    except:
        message(CRITICAL,
                "couldn't load extension %s (looking for %s module)"
                % (ext_name, extension_module_name) )
        sys.exit(1)

    return module.makeExtension(configs.items())    


OPTPARSE_WARNING = """
Python 2.3 or higher required for advanced command line options.
For lower versions of Python use:

      %s INPUT_FILE > OUTPUT_FILE
    
""" % EXECUTABLE_NAME_FOR_USAGE

def parse_options():
    """
    Define and parse `optparse` options for command-line usage.
    """

    try:
        optparse = __import__("optparse")
    except:
        if len(sys.argv) == 2:
            return {'input': sys.argv[1],
                    'output': None,
                    'message_threshold': CRITICAL,
                    'safe': False,
                    'extensions': [],
                    'encoding': None }

        else:
            print OPTPARSE_WARNING
            return None

    parser = optparse.OptionParser(usage="%prog INPUTFILE [options]")

    parser.add_option("-f", "--file", dest="filename",
                      help="write output to OUTPUT_FILE",
                      metavar="OUTPUT_FILE")
    parser.add_option("-e", "--encoding", dest="encoding",
                      help="encoding for input and output files",)
    parser.add_option("-q", "--quiet", default = CRITICAL,
                      action="store_const", const=60, dest="verbose",
                      help="suppress all messages")
    parser.add_option("-v", "--verbose",
                      action="store_const", const=INFO, dest="verbose",
                      help="print info messages")
    parser.add_option("-s", "--safe", dest="safe", default=False,
                      metavar="SAFE_MODE",
                      help="safe mode ('replace', 'remove' or 'escape'  user's HTML tag)")
    
    parser.add_option("--noisy",
                      action="store_const", const=DEBUG, dest="verbose",
                      help="print debug messages")
    parser.add_option("-x", "--extension", action="append", dest="extensions",
                      help = "load extension EXTENSION", metavar="EXTENSION")

    (options, args) = parser.parse_args()

    if not len(args) == 1:
        parser.print_help()
        return None
    else:
        input_file = args[0]

    if not options.extensions:
        options.extensions = []

    return {'input': input_file,
            'output': options.filename,
            'message_threshold': options.verbose,
            'safe': options.safe,
            'extensions': options.extensions,
            'encoding': options.encoding }

if __name__ == '__main__':
    """ Run Markdown from the command line. """

    options = parse_options()

    #if os.access(inFile, os.R_OK):

    if not options:
        sys.exit(0)
    
    markdownFromFile(**options)