You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
bazarr/libs/markdown/inlinepatterns.py

893 lines
29 KiB

"""
Python Markdown
A Python implementation of John Gruber's Markdown.
Documentation: https://python-markdown.github.io/
GitHub: https://github.com/Python-Markdown/markdown/
PyPI: https://pypi.org/project/Markdown/
Started by Manfred Stienstra (http://www.dwerg.net/).
Maintained for a few years by Yuri Takhteyev (http://www.freewisdom.org).
Currently maintained by Waylan Limberg (https://github.com/waylan),
Dmitry Shachnev (https://github.com/mitya57) and Isaac Muse (https://github.com/facelessuser).
Copyright 2007-2018 The Python Markdown Project (v. 1.7 and later)
Copyright 2004, 2005, 2006 Yuri Takhteyev (v. 0.2-1.6b)
Copyright 2004 Manfred Stienstra (the original version)
License: BSD (see LICENSE.md for details).
INLINE PATTERNS
=============================================================================
Inline patterns such as *emphasis* are handled by means of auxiliary
objects, one per pattern. Pattern objects must be instances of classes
that extend markdown.Pattern. Each pattern object uses a single regular
expression and needs support the following methods:
pattern.getCompiledRegExp() # returns a regular expression
pattern.handleMatch(m) # takes a match object and returns
# an ElementTree element or just plain text
All of python markdown's built-in patterns subclass from Pattern,
but you can add additional patterns that don't.
Also note that all the regular expressions used by inline must
capture the whole block. For this reason, they all start with
'^(.*)' and end with '(.*)!'. In case with built-in expression
Pattern takes care of adding the "^(.*)" and "(.*)!".
Finally, the order in which regular expressions are applied is very
important - e.g. if we first replace http://.../ links with <a> tags
and _then_ try to replace inline html, we would end up with a mess.
So, we apply the expressions in the following order:
* escape and backticks have to go before everything else, so
that we can preempt any markdown patterns by escaping them.
* then we handle auto-links (must be done before inline html)
* then we handle inline HTML. At this point we will simply
replace all inline HTML strings with a placeholder and add
the actual HTML to a hash.
* then inline images (must be done before links)
* then bracketed links, first regular then reference-style
* finally we apply strong and emphasis
"""
from . import util
from collections import namedtuple
import re
import xml.etree.ElementTree as etree
try: # pragma: no cover
from html import entities
except ImportError: # pragma: no cover
import htmlentitydefs as entities
def build_inlinepatterns(md, **kwargs):
""" Build the default set of inline patterns for Markdown. """
inlinePatterns = util.Registry()
inlinePatterns.register(BacktickInlineProcessor(BACKTICK_RE), 'backtick', 190)
inlinePatterns.register(EscapeInlineProcessor(ESCAPE_RE, md), 'escape', 180)
inlinePatterns.register(ReferenceInlineProcessor(REFERENCE_RE, md), 'reference', 170)
inlinePatterns.register(LinkInlineProcessor(LINK_RE, md), 'link', 160)
inlinePatterns.register(ImageInlineProcessor(IMAGE_LINK_RE, md), 'image_link', 150)
inlinePatterns.register(
ImageReferenceInlineProcessor(IMAGE_REFERENCE_RE, md), 'image_reference', 140
)
inlinePatterns.register(
ShortReferenceInlineProcessor(REFERENCE_RE, md), 'short_reference', 130
)
inlinePatterns.register(
ShortImageReferenceInlineProcessor(IMAGE_REFERENCE_RE, md), 'short_image_ref', 125
)
inlinePatterns.register(AutolinkInlineProcessor(AUTOLINK_RE, md), 'autolink', 120)
inlinePatterns.register(AutomailInlineProcessor(AUTOMAIL_RE, md), 'automail', 110)
inlinePatterns.register(SubstituteTagInlineProcessor(LINE_BREAK_RE, 'br'), 'linebreak', 100)
inlinePatterns.register(HtmlInlineProcessor(HTML_RE, md), 'html', 90)
inlinePatterns.register(HtmlInlineProcessor(ENTITY_RE, md), 'entity', 80)
inlinePatterns.register(SimpleTextInlineProcessor(NOT_STRONG_RE), 'not_strong', 70)
inlinePatterns.register(AsteriskProcessor(r'\*'), 'em_strong', 60)
inlinePatterns.register(UnderscoreProcessor(r'_'), 'em_strong2', 50)
return inlinePatterns
"""
The actual regular expressions for patterns
-----------------------------------------------------------------------------
"""
NOIMG = r'(?<!\!)'
# `e=f()` or ``e=f("`")``
BACKTICK_RE = r'(?:(?<!\\)((?:\\{2})+)(?=`+)|(?<!\\)(`+)(.+?)(?<!`)\2(?!`))'
# \<
ESCAPE_RE = r'\\(.)'
# *emphasis*
EMPHASIS_RE = r'(\*)([^\*]+)\1'
# **strong**
STRONG_RE = r'(\*{2})(.+?)\1'
# __smart__strong__
SMART_STRONG_RE = r'(?<!\w)(_{2})(?!_)(.+?)(?<!_)\1(?!\w)'
# _smart_emphasis_
SMART_EMPHASIS_RE = r'(?<!\w)(_)(?!_)(.+?)(?<!_)\1(?!\w)'
# __strong _em__
SMART_STRONG_EM_RE = r'(?<!\w)(\_)\1(?!\1)(.+?)(?<!\w)\1(?!\1)(.+?)\1{3}(?!\w)'
# ***strongem*** or ***em*strong**
EM_STRONG_RE = r'(\*)\1{2}(.+?)\1(.*?)\1{2}'
# ___strongem___ or ___em_strong__
EM_STRONG2_RE = r'(_)\1{2}(.+?)\1(.*?)\1{2}'
# ***strong**em*
STRONG_EM_RE = r'(\*)\1{2}(.+?)\1{2}(.*?)\1'
# ___strong__em_
STRONG_EM2_RE = r'(_)\1{2}(.+?)\1{2}(.*?)\1'
# **strong*em***
STRONG_EM3_RE = r'(\*)\1(?!\1)([^*]+?)\1(?!\1)(.+?)\1{3}'
# [text](url) or [text](<url>) or [text](url "title")
LINK_RE = NOIMG + r'\['
# ![alttxt](http://x.com/) or ![alttxt](<http://x.com/>)
IMAGE_LINK_RE = r'\!\['
# [Google][3]
REFERENCE_RE = LINK_RE
# ![alt text][2]
IMAGE_REFERENCE_RE = IMAGE_LINK_RE
# stand-alone * or _
NOT_STRONG_RE = r'((^|\s)(\*|_)(\s|$))'
# <http://www.123.com>
AUTOLINK_RE = r'<((?:[Ff]|[Hh][Tt])[Tt][Pp][Ss]?://[^<>]*)>'
# <me@example.com>
AUTOMAIL_RE = r'<([^<> !]+@[^@<> ]+)>'
# <...>
HTML_RE = r'(<(\/?[a-zA-Z][^<>@ ]*( [^<>]*)?|!--(?:(?!<!--|-->).)*--)>)'
# "&#38;" (decimal) or "&#x26;" (hex) or "&amp;" (named)
ENTITY_RE = r'(&(?:\#[0-9]+|\#x[0-9a-fA-F]+|[a-zA-Z0-9]+);)'
# two spaces at end of line
LINE_BREAK_RE = r' \n'
def dequote(string):
"""Remove quotes from around a string."""
if ((string.startswith('"') and string.endswith('"')) or
(string.startswith("'") and string.endswith("'"))):
return string[1:-1]
else:
return string
class EmStrongItem(namedtuple('EmStrongItem', ['pattern', 'builder', 'tags'])):
"""Emphasis/strong pattern item."""
"""
The pattern classes
-----------------------------------------------------------------------------
"""
class Pattern: # pragma: no cover
"""Base class that inline patterns subclass. """
ANCESTOR_EXCLUDES = tuple()
def __init__(self, pattern, md=None):
"""
Create an instant of an inline pattern.
Keyword arguments:
* pattern: A regular expression that matches a pattern
"""
self.pattern = pattern
self.compiled_re = re.compile(r"^(.*?)%s(.*)$" % pattern,
re.DOTALL | re.UNICODE)
self.md = md
@property
@util.deprecated("Use 'md' instead.")
def markdown(self):
# TODO: remove this later
return self.md
def getCompiledRegExp(self):
""" Return a compiled regular expression. """
return self.compiled_re
def handleMatch(self, m):
"""Return a ElementTree element from the given match.
Subclasses should override this method.
Keyword arguments:
* m: A re match object containing a match of the pattern.
"""
pass # pragma: no cover
def type(self):
""" Return class name, to define pattern type """
return self.__class__.__name__
def unescape(self, text):
""" Return unescaped text given text with an inline placeholder. """
try:
stash = self.md.treeprocessors['inline'].stashed_nodes
except KeyError: # pragma: no cover
return text
def get_stash(m):
id = m.group(1)
if id in stash:
value = stash.get(id)
if isinstance(value, str):
return value
else:
# An etree Element - return text content only
return ''.join(value.itertext())
return util.INLINE_PLACEHOLDER_RE.sub(get_stash, text)
class InlineProcessor(Pattern):
"""
Base class that inline patterns subclass.
This is the newer style inline processor that uses a more
efficient and flexible search approach.
"""
def __init__(self, pattern, md=None):
"""
Create an instant of an inline pattern.
Keyword arguments:
* pattern: A regular expression that matches a pattern
"""
self.pattern = pattern
self.compiled_re = re.compile(pattern, re.DOTALL | re.UNICODE)
# Api for Markdown to pass safe_mode into instance
self.safe_mode = False
self.md = md
def handleMatch(self, m, data):
"""Return a ElementTree element from the given match and the
start and end index of the matched text.
If `start` and/or `end` are returned as `None`, it will be
assumed that the processor did not find a valid region of text.
Subclasses should override this method.
Keyword arguments:
* m: A re match object containing a match of the pattern.
* data: The buffer current under analysis
Returns:
* el: The ElementTree element, text or None.
* start: The start of the region that has been matched or None.
* end: The end of the region that has been matched or None.
"""
pass # pragma: no cover
class SimpleTextPattern(Pattern): # pragma: no cover
""" Return a simple text of group(2) of a Pattern. """
def handleMatch(self, m):
return m.group(2)
class SimpleTextInlineProcessor(InlineProcessor):
""" Return a simple text of group(1) of a Pattern. """
def handleMatch(self, m, data):
return m.group(1), m.start(0), m.end(0)
class EscapeInlineProcessor(InlineProcessor):
""" Return an escaped character. """
def handleMatch(self, m, data):
char = m.group(1)
if char in self.md.ESCAPED_CHARS:
return '{}{}{}'.format(util.STX, ord(char), util.ETX), m.start(0), m.end(0)
else:
return None, m.start(0), m.end(0)
class SimpleTagPattern(Pattern): # pragma: no cover
"""
Return element of type `tag` with a text attribute of group(3)
of a Pattern.
"""
def __init__(self, pattern, tag):
Pattern.__init__(self, pattern)
self.tag = tag
def handleMatch(self, m):
el = etree.Element(self.tag)
el.text = m.group(3)
return el
class SimpleTagInlineProcessor(InlineProcessor):
"""
Return element of type `tag` with a text attribute of group(2)
of a Pattern.
"""
def __init__(self, pattern, tag):
InlineProcessor.__init__(self, pattern)
self.tag = tag
def handleMatch(self, m, data): # pragma: no cover
el = etree.Element(self.tag)
el.text = m.group(2)
return el, m.start(0), m.end(0)
class SubstituteTagPattern(SimpleTagPattern): # pragma: no cover
""" Return an element of type `tag` with no children. """
def handleMatch(self, m):
return etree.Element(self.tag)
class SubstituteTagInlineProcessor(SimpleTagInlineProcessor):
""" Return an element of type `tag` with no children. """
def handleMatch(self, m, data):
return etree.Element(self.tag), m.start(0), m.end(0)
class BacktickInlineProcessor(InlineProcessor):
""" Return a `<code>` element containing the matching text. """
def __init__(self, pattern):
InlineProcessor.__init__(self, pattern)
self.ESCAPED_BSLASH = '{}{}{}'.format(util.STX, ord('\\'), util.ETX)
self.tag = 'code'
def handleMatch(self, m, data):
if m.group(3):
el = etree.Element(self.tag)
el.text = util.AtomicString(util.code_escape(m.group(3).strip()))
return el, m.start(0), m.end(0)
else:
return m.group(1).replace('\\\\', self.ESCAPED_BSLASH), m.start(0), m.end(0)
class DoubleTagPattern(SimpleTagPattern): # pragma: no cover
"""Return a ElementTree element nested in tag2 nested in tag1.
Useful for strong emphasis etc.
"""
def handleMatch(self, m):
tag1, tag2 = self.tag.split(",")
el1 = etree.Element(tag1)
el2 = etree.SubElement(el1, tag2)
el2.text = m.group(3)
if len(m.groups()) == 5:
el2.tail = m.group(4)
return el1
class DoubleTagInlineProcessor(SimpleTagInlineProcessor):
"""Return a ElementTree element nested in tag2 nested in tag1.
Useful for strong emphasis etc.
"""
def handleMatch(self, m, data): # pragma: no cover
tag1, tag2 = self.tag.split(",")
el1 = etree.Element(tag1)
el2 = etree.SubElement(el1, tag2)
el2.text = m.group(2)
if len(m.groups()) == 3:
el2.tail = m.group(3)
return el1, m.start(0), m.end(0)
class HtmlInlineProcessor(InlineProcessor):
""" Store raw inline html and return a placeholder. """
def handleMatch(self, m, data):
rawhtml = self.unescape(m.group(1))
place_holder = self.md.htmlStash.store(rawhtml)
return place_holder, m.start(0), m.end(0)
def unescape(self, text):
""" Return unescaped text given text with an inline placeholder. """
try:
stash = self.md.treeprocessors['inline'].stashed_nodes
except KeyError: # pragma: no cover
return text
def get_stash(m):
id = m.group(1)
value = stash.get(id)
if value is not None:
try:
return self.md.serializer(value)
except Exception:
return r'\%s' % value
return util.INLINE_PLACEHOLDER_RE.sub(get_stash, text)
class AsteriskProcessor(InlineProcessor):
"""Emphasis processor for handling strong and em matches inside asterisks."""
PATTERNS = [
EmStrongItem(re.compile(EM_STRONG_RE, re.DOTALL | re.UNICODE), 'double', 'strong,em'),
EmStrongItem(re.compile(STRONG_EM_RE, re.DOTALL | re.UNICODE), 'double', 'em,strong'),
EmStrongItem(re.compile(STRONG_EM3_RE, re.DOTALL | re.UNICODE), 'double2', 'strong,em'),
EmStrongItem(re.compile(STRONG_RE, re.DOTALL | re.UNICODE), 'single', 'strong'),
EmStrongItem(re.compile(EMPHASIS_RE, re.DOTALL | re.UNICODE), 'single', 'em')
]
def build_single(self, m, tag, idx):
"""Return single tag."""
el1 = etree.Element(tag)
text = m.group(2)
self.parse_sub_patterns(text, el1, None, idx)
return el1
def build_double(self, m, tags, idx):
"""Return double tag."""
tag1, tag2 = tags.split(",")
el1 = etree.Element(tag1)
el2 = etree.Element(tag2)
text = m.group(2)
self.parse_sub_patterns(text, el2, None, idx)
el1.append(el2)
if len(m.groups()) == 3:
text = m.group(3)
self.parse_sub_patterns(text, el1, el2, idx)
return el1
def build_double2(self, m, tags, idx):
"""Return double tags (variant 2): `<strong>text <em>text</em></strong>`."""
tag1, tag2 = tags.split(",")
el1 = etree.Element(tag1)
el2 = etree.Element(tag2)
text = m.group(2)
self.parse_sub_patterns(text, el1, None, idx)
text = m.group(3)
el1.append(el2)
self.parse_sub_patterns(text, el2, None, idx)
return el1
def parse_sub_patterns(self, data, parent, last, idx):
"""
Parses sub patterns.
`data` (`str`):
text to evaluate.
`parent` (`etree.Element`):
Parent to attach text and sub elements to.
`last` (`etree.Element`):
Last appended child to parent. Can also be None if parent has no children.
`idx` (`int`):
Current pattern index that was used to evaluate the parent.
"""
offset = 0
pos = 0
length = len(data)
while pos < length:
# Find the start of potential emphasis or strong tokens
if self.compiled_re.match(data, pos):
matched = False
# See if the we can match an emphasis/strong pattern
for index, item in enumerate(self.PATTERNS):
# Only evaluate patterns that are after what was used on the parent
if index <= idx:
continue
m = item.pattern.match(data, pos)
if m:
# Append child nodes to parent
# Text nodes should be appended to the last
# child if present, and if not, it should
# be added as the parent's text node.
text = data[offset:m.start(0)]
if text:
if last is not None:
last.tail = text
else:
parent.text = text
el = self.build_element(m, item.builder, item.tags, index)
parent.append(el)
last = el
# Move our position past the matched hunk
offset = pos = m.end(0)
matched = True
if not matched:
# We matched nothing, move on to the next character
pos += 1
else:
# Increment position as no potential emphasis start was found.
pos += 1
# Append any leftover text as a text node.
text = data[offset:]
if text:
if last is not None:
last.tail = text
else:
parent.text = text
def build_element(self, m, builder, tags, index):
"""Element builder."""
if builder == 'double2':
return self.build_double2(m, tags, index)
elif builder == 'double':
return self.build_double(m, tags, index)
else:
return self.build_single(m, tags, index)
def handleMatch(self, m, data):
"""Parse patterns."""
el = None
start = None
end = None
for index, item in enumerate(self.PATTERNS):
m1 = item.pattern.match(data, m.start(0))
if m1:
start = m1.start(0)
end = m1.end(0)
el = self.build_element(m1, item.builder, item.tags, index)
break
return el, start, end
class UnderscoreProcessor(AsteriskProcessor):
"""Emphasis processor for handling strong and em matches inside underscores."""
PATTERNS = [
EmStrongItem(re.compile(EM_STRONG2_RE, re.DOTALL | re.UNICODE), 'double', 'strong,em'),
EmStrongItem(re.compile(STRONG_EM2_RE, re.DOTALL | re.UNICODE), 'double', 'em,strong'),
EmStrongItem(re.compile(SMART_STRONG_EM_RE, re.DOTALL | re.UNICODE), 'double2', 'strong,em'),
EmStrongItem(re.compile(SMART_STRONG_RE, re.DOTALL | re.UNICODE), 'single', 'strong'),
EmStrongItem(re.compile(SMART_EMPHASIS_RE, re.DOTALL | re.UNICODE), 'single', 'em')
]
class LinkInlineProcessor(InlineProcessor):
""" Return a link element from the given match. """
RE_LINK = re.compile(r'''\(\s*(?:(<[^<>]*>)\s*(?:('[^']*'|"[^"]*")\s*)?\))?''', re.DOTALL | re.UNICODE)
RE_TITLE_CLEAN = re.compile(r'\s')
def handleMatch(self, m, data):
text, index, handled = self.getText(data, m.end(0))
if not handled:
return None, None, None
href, title, index, handled = self.getLink(data, index)
if not handled:
return None, None, None
el = etree.Element("a")
el.text = text
el.set("href", href)
if title is not None:
el.set("title", title)
return el, m.start(0), index
def getLink(self, data, index):
"""Parse data between `()` of `[Text]()` allowing recursive `()`. """
href = ''
title = None
handled = False
m = self.RE_LINK.match(data, pos=index)
if m and m.group(1):
# Matches [Text](<link> "title")
href = m.group(1)[1:-1].strip()
if m.group(2):
title = m.group(2)[1:-1]
index = m.end(0)
handled = True
elif m:
# Track bracket nesting and index in string
bracket_count = 1
backtrack_count = 1
start_index = m.end()
index = start_index
last_bracket = -1
# Primary (first found) quote tracking.
quote = None
start_quote = -1
exit_quote = -1
ignore_matches = False
# Secondary (second found) quote tracking.
alt_quote = None
start_alt_quote = -1
exit_alt_quote = -1
# Track last character
last = ''
for pos in range(index, len(data)):
c = data[pos]
if c == '(':
# Count nested (
# Don't increment the bracket count if we are sure we're in a title.
if not ignore_matches:
bracket_count += 1
elif backtrack_count > 0:
backtrack_count -= 1
elif c == ')':
# Match nested ) to (
# Don't decrement if we are sure we are in a title that is unclosed.
if ((exit_quote != -1 and quote == last) or (exit_alt_quote != -1 and alt_quote == last)):
bracket_count = 0
elif not ignore_matches:
bracket_count -= 1
elif backtrack_count > 0:
backtrack_count -= 1
# We've found our backup end location if the title doesn't reslove.
if backtrack_count == 0:
last_bracket = index + 1
elif c in ("'", '"'):
# Quote has started
if not quote:
# We'll assume we are now in a title.
# Brackets are quoted, so no need to match them (except for the final one).
ignore_matches = True
backtrack_count = bracket_count
bracket_count = 1
start_quote = index + 1
quote = c
# Secondary quote (in case the first doesn't resolve): [text](link'"title")
elif c != quote and not alt_quote:
start_alt_quote = index + 1
alt_quote = c
# Update primary quote match
elif c == quote:
exit_quote = index + 1
# Update secondary quote match
elif alt_quote and c == alt_quote:
exit_alt_quote = index + 1
index += 1
# Link is closed, so let's break out of the loop
if bracket_count == 0:
# Get the title if we closed a title string right before link closed
if exit_quote >= 0 and quote == last:
href = data[start_index:start_quote - 1]
title = ''.join(data[start_quote:exit_quote - 1])
elif exit_alt_quote >= 0 and alt_quote == last:
href = data[start_index:start_alt_quote - 1]
title = ''.join(data[start_alt_quote:exit_alt_quote - 1])
else:
href = data[start_index:index - 1]
break
if c != ' ':
last = c
# We have a scenario: [test](link"notitle)
# When we enter a string, we stop tracking bracket resolution in the main counter,
# but we do keep a backup counter up until we discover where we might resolve all brackets
# if the title string fails to resolve.
if bracket_count != 0 and backtrack_count == 0:
href = data[start_index:last_bracket - 1]
index = last_bracket
bracket_count = 0
handled = bracket_count == 0
if title is not None:
title = self.RE_TITLE_CLEAN.sub(' ', dequote(self.unescape(title.strip())))
href = self.unescape(href).strip()
return href, title, index, handled
def getText(self, data, index):
"""Parse the content between `[]` of the start of an image or link
resolving nested square brackets.
"""
bracket_count = 1
text = []
for pos in range(index, len(data)):
c = data[pos]
if c == ']':
bracket_count -= 1
elif c == '[':
bracket_count += 1
index += 1
if bracket_count == 0:
break
text.append(c)
return ''.join(text), index, bracket_count == 0
class ImageInlineProcessor(LinkInlineProcessor):
""" Return a img element from the given match. """
def handleMatch(self, m, data):
text, index, handled = self.getText(data, m.end(0))
if not handled:
return None, None, None
src, title, index, handled = self.getLink(data, index)
if not handled:
return None, None, None
el = etree.Element("img")
el.set("src", src)
if title is not None:
el.set("title", title)
el.set('alt', self.unescape(text))
return el, m.start(0), index
class ReferenceInlineProcessor(LinkInlineProcessor):
""" Match to a stored reference and return link element. """
NEWLINE_CLEANUP_RE = re.compile(r'\s+', re.MULTILINE)
RE_LINK = re.compile(r'\s?\[([^\]]*)\]', re.DOTALL | re.UNICODE)
def handleMatch(self, m, data):
text, index, handled = self.getText(data, m.end(0))
if not handled:
return None, None, None
id, end, handled = self.evalId(data, index, text)
if not handled:
return None, None, None
# Clean up linebreaks in id
id = self.NEWLINE_CLEANUP_RE.sub(' ', id)
if id not in self.md.references: # ignore undefined refs
return None, m.start(0), end
href, title = self.md.references[id]
return self.makeTag(href, title, text), m.start(0), end
def evalId(self, data, index, text):
"""
Evaluate the id portion of [ref][id].
If [ref][] use [ref].
"""
m = self.RE_LINK.match(data, pos=index)
if not m:
return None, index, False
else:
id = m.group(1).lower()
end = m.end(0)
if not id:
id = text.lower()
return id, end, True
def makeTag(self, href, title, text):
el = etree.Element('a')
el.set('href', href)
if title:
el.set('title', title)
el.text = text
return el
class ShortReferenceInlineProcessor(ReferenceInlineProcessor):
"""Short form of reference: [google]. """
def evalId(self, data, index, text):
"""Evaluate the id from of [ref] """
return text.lower(), index, True
class ImageReferenceInlineProcessor(ReferenceInlineProcessor):
""" Match to a stored reference and return img element. """
def makeTag(self, href, title, text):
el = etree.Element("img")
el.set("src", href)
if title:
el.set("title", title)
el.set("alt", self.unescape(text))
return el
class ShortImageReferenceInlineProcessor(ImageReferenceInlineProcessor):
""" Short form of inage reference: ![ref]. """
def evalId(self, data, index, text):
"""Evaluate the id from of [ref] """
return text.lower(), index, True
class AutolinkInlineProcessor(InlineProcessor):
""" Return a link Element given an autolink (`<http://example/com>`). """
def handleMatch(self, m, data):
el = etree.Element("a")
el.set('href', self.unescape(m.group(1)))
el.text = util.AtomicString(m.group(1))
return el, m.start(0), m.end(0)
class AutomailInlineProcessor(InlineProcessor):
"""
Return a mailto link Element given an automail link (`<foo@example.com>`).
"""
def handleMatch(self, m, data):
el = etree.Element('a')
email = self.unescape(m.group(1))
if email.startswith("mailto:"):
email = email[len("mailto:"):]
def codepoint2name(code):
"""Return entity definition by code, or the code if not defined."""
entity = entities.codepoint2name.get(code)
if entity:
return "{}{};".format(util.AMP_SUBSTITUTE, entity)
else:
return "%s#%d;" % (util.AMP_SUBSTITUTE, code)
letters = [codepoint2name(ord(letter)) for letter in email]
el.text = util.AtomicString(''.join(letters))
mailto = "mailto:" + email
mailto = "".join([util.AMP_SUBSTITUTE + '#%d;' %
ord(letter) for letter in mailto])
el.set('href', mailto)
return el, m.start(0), m.end(0)