"""Beautiful Soup
Elixir and Tonic
"The Screen-Scraper's Friend"
http://www.crummy.com/software/BeautifulSoup/
Beautiful Soup parses a (possibly invalid) XML or HTML document into a
tree representation. It provides methods and Pythonic idioms that make
it easy to navigate, search, and modify the tree.
A well-formed XML/HTML document yields a well-formed data
structure. An ill-formed XML/HTML document yields a correspondingly
ill-formed data structure. If your document is only locally
well-formed, you can use this library to find and process the
well-formed part of it. The BeautifulSoup class
Beautiful Soup works with Python 2.2 and up. It has no external
dependencies, but you'll have more success at converting data to UTF-8
if you also install these three packages:
* chardet, for auto-detecting character encodings
http://chardet.feedparser.org/
* cjkcodecs and iconv_codec, which add more encodings to the ones supported
by stock Python.
http://cjkpython.i18n.org/
Beautiful Soup defines classes for two main parsing strategies:
* BeautifulStoneSoup, for parsing XML, SGML, or your domain-specific
language that kind of looks like XML.
* BeautifulSoup, for parsing run-of-the-mill HTML code, be it valid
or invalid. This class has web browser-like heuristics for
obtaining a sensible parse tree in the face of common HTML errors.
Beautiful Soup also defines a class (UnicodeDammit) for autodetecting
the encoding of an HTML or XML document, and converting it to
Unicode. Much of this code is taken from Mark Pilgrim's Universal Feed Parser.
For more than you ever wanted to know about Beautiful Soup, see the
documentation:
http://www.crummy.com/software/BeautifulSoup/documentation.html
"""
from __future__ import generators
__author__ = "Leonard Richardson (leonardr@segfault.org)"
__version__ = "3.0.4"
__copyright__ = "Copyright (c) 2004-2007 Leonard Richardson"
__license__ = "PSF"
from sgmllib import SGMLParser, SGMLParseError
import codecs
import types
import re
import sgmllib
try:
from htmlentitydefs import name2codepoint
except ImportError:
name2codepoint = {}
#This hack makes Beautiful Soup able to parse XML with namespaces
sgmllib.tagfind = re.compile('[a-zA-Z][-_.:a-zA-Z0-9]*')
DEFAULT_OUTPUT_ENCODING = "utf-8"
# First, the classes that represent markup elements.
class PageElement:
"""Contains the navigational information for some part of the page
(either a tag or a piece of text)"""
def setup(self, parent=None, previous=None):
"""Sets up the initial relations between this element and
other elements."""
self.parent = parent
self.previous = previous
self.next = None
self.previousSibling = None
self.nextSibling = None
if self.parent and self.parent.contents:
self.previousSibling = self.parent.contents[-1]
self.previousSibling.nextSibling = self
def replaceWith(self, replaceWith):
oldParent = self.parent
myIndex = self.parent.contents.index(self)
if hasattr(replaceWith, 'parent') and replaceWith.parent == self.parent:
# We're replacing this element with one of its siblings.
index = self.parent.contents.index(replaceWith)
if index and index < myIndex:
# Furthermore, it comes before this element. That
# means that when we extract it, the index of this
# element will change.
myIndex = myIndex - 1
self.extract()
oldParent.insert(myIndex, replaceWith)
def extract(self):
"""Destructively rips this element out of the tree."""
if self.parent:
try:
self.parent.contents.remove(self)
except ValueError:
pass
#Find the two elements that would be next to each other if
#this element (and any children) hadn't been parsed. Connect
#the two.
lastChild = self._lastRecursiveChild()
nextElement = lastChild.next
if self.previous:
self.previous.next = nextElement
if nextElement:
nextElement.previous = self.previous
self.previous = None
lastChild.next = None
self.parent = None
if self.previousSibling:
self.previousSibling.nextSibling = self.nextSibling
if self.nextSibling:
self.nextSibling.previousSibling = self.previousSibling
self.previousSibling = self.nextSibling = None
def _lastRecursiveChild(self):
"Finds the last element beneath this object to be parsed."
lastChild = self
while hasattr(lastChild, 'contents') and lastChild.contents:
lastChild = lastChild.contents[-1]
return lastChild
def insert(self, position, newChild):
if (isinstance(newChild, basestring)
or isinstance(newChild, unicode)) \
and not isinstance(newChild, NavigableString):
newChild = NavigableString(newChild)
position = min(position, len(self.contents))
if hasattr(newChild, 'parent') and newChild.parent != None:
# We're 'inserting' an element that's already one
# of this object's children.
if newChild.parent == self:
index = self.find(newChild)
if index and index < position:
# Furthermore we're moving it further down the
# list of this object's children. That means that
# when we extract this element, our target index
# will jump down one.
position = position - 1
newChild.extract()
newChild.parent = self
previousChild = None
if position == 0:
newChild.previousSibling = None
newChild.previous = self
else:
previousChild = self.contents[position-1]
newChild.previousSibling = previousChild
newChild.previousSibling.nextSibling = newChild
newChild.previous = previousChild._lastRecursiveChild()
if newChild.previous:
newChild.previous.next = newChild
newChildsLastElement = newChild._lastRecursiveChild()
if position >= len(self.contents):
newChild.nextSibling = None
parent = self
parentsNextSibling = None
while not parentsNextSibling:
parentsNextSibling = parent.nextSibling
parent = parent.parent
if not parent: # This is the last element in the document.
break
if parentsNextSibling:
newChildsLastElement.next = parentsNextSibling
else:
newChildsLastElement.next = None
else:
nextChild = self.contents[position]
newChild.nextSibling = nextChild
if newChild.nextSibling:
newChild.nextSibling.previousSibling = newChild
newChildsLastElement.next = nextChild
if newChildsLastElement.next:
newChildsLastElement.next.previous = newChildsLastElement
self.contents.insert(position, newChild)
def findNext(self, name=None, attrs={}, text=None, **kwargs):
"""Returns the first item that matches the given criteria and
appears after this Tag in the document."""
return self._findOne(self.findAllNext, name, attrs, text, **kwargs)
def findAllNext(self, name=None, attrs={}, text=None, limit=None,
**kwargs):
"""Returns all items that match the given criteria and appear
before after Tag in the document."""
return self._findAll(name, attrs, text, limit, self.nextGenerator)
def findNextSibling(self, name=None, attrs={}, text=None, **kwargs):
"""Returns the closest sibling to this Tag that matches the
given criteria and appears after this Tag in the document."""
return self._findOne(self.findNextSiblings, name, attrs, text,
**kwargs)
def findNextSiblings(self, name=None, attrs={}, text=None, limit=None,
**kwargs):
"""Returns the siblings of this Tag that match the given
criteria and appear after this Tag in the document."""
return self._findAll(name, attrs, text, limit,
self.nextSiblingGenerator, **kwargs)
fetchNextSiblings = findNextSiblings # Compatibility with pre-3.x
def findPrevious(self, name=None, attrs={}, text=None, **kwargs):
"""Returns the first item that matches the given criteria and
appears before this Tag in the document."""
return self._findOne(self.findAllPrevious, name, attrs, text, **kwargs)
def findAllPrevious(self, name=None, attrs={}, text=None, limit=None,
**kwargs):
"""Returns all items that match the given criteria and appear
before this Tag in the document."""
return self._findAll(name, attrs, text, limit, self.previousGenerator,
**kwargs)
fetchPrevious = findAllPrevious # Compatibility with pre-3.x
def findPreviousSibling(self, name=None, attrs={}, text=None, **kwargs):
"""Returns the closest sibling to this Tag that matches the
given criteria and appears before this Tag in the document."""
return self._findOne(self.findPreviousSiblings, name, attrs, text,
**kwargs)
def findPreviousSiblings(self, name=None, attrs={}, text=None,
limit=None, **kwargs):
"""Returns the siblings of this Tag that match the given
criteria and appear before this Tag in the document."""
return self._findAll(name, attrs, text, limit,
self.previousSiblingGenerator, **kwargs)
fetchPreviousSiblings = findPreviousSiblings # Compatibility with pre-3.x
def findParent(self, name=None, attrs={}, **kwargs):
"""Returns the closest parent of this Tag that matches the given
criteria."""
# NOTE: We can't use _findOne because findParents takes a different
# set of arguments.
r = None
l = self.findParents(name, attrs, 1)
if l:
r = l[0]
return r
def findParents(self, name=None, attrs={}, limit=None, **kwargs):
"""Returns the parents of this Tag that match the given
criteria."""
return self._findAll(name, attrs, None, limit, self.parentGenerator,
**kwargs)
fetchParents = findParents # Compatibility with pre-3.x
#These methods do the real heavy lifting.
def _findOne(self, method, name, attrs, text, **kwargs):
r = None
l = method(name, attrs, text, 1, **kwargs)
if l:
r = l[0]
return r
def _findAll(self, name, attrs, text, limit, generator, **kwargs):
"Iterates over a generator looking for things that match."
if isinstance(name, SoupStrainer):
strainer = name
else:
# Build a SoupStrainer
strainer = SoupStrainer(name, attrs, text, **kwargs)
results = ResultSet(strainer)
g = generator()
while True:
try:
i = g.next()
except StopIteration:
break
if i:
found = strainer.search(i)
if found:
results.append(found)
if limit and len(results) >= limit:
break
return results
#These Generators can be used to navigate starting from both
#NavigableStrings and Tags.
def nextGenerator(self):
i = self
while i:
i = i.next
yield i
def nextSiblingGenerator(self):
i = self
while i:
i = i.nextSibling
yield i
def previousGenerator(self):
i = self
while i:
i = i.previous
yield i
def previousSiblingGenerator(self):
i = self
while i:
i = i.previousSibling
yield i
def parentGenerator(self):
i = self
while i:
i = i.parent
yield i
# Utility methods
def substituteEncoding(self, str, encoding=None):
encoding = encoding or "utf-8"
return str.replace("%SOUP-ENCODING%", encoding)
def toEncoding(self, s, encoding=None):
"""Encodes an object to a string in some encoding, or to Unicode.
."""
if isinstance(s, unicode):
if encoding:
s = s.encode(encoding)
elif isinstance(s, str):
if encoding:
s = s.encode(encoding)
else:
s = unicode(s)
else:
if encoding:
s = self.toEncoding(str(s), encoding)
else:
s = unicode(s)
return s
class NavigableString(unicode, PageElement):
def __getattr__(self, attr):
"""text.string gives you text. This is for backwards
compatibility for Navigable*String, but for CData* it lets you
get the string without the CData wrapper."""
if attr == 'string':
return self
else:
raise AttributeError, "'%s' object has no attribute '%s'" % (self.__class__.__name__, attr)
def __unicode__(self):
return self.__str__(None)
def __str__(self, encoding=DEFAULT_OUTPUT_ENCODING):
if encoding:
return self.encode(encoding)
else:
return self
class CData(NavigableString):
def __str__(self, encoding=DEFAULT_OUTPUT_ENCODING):
return "" % NavigableString.__str__(self, encoding)
class ProcessingInstruction(NavigableString):
def __str__(self, encoding=DEFAULT_OUTPUT_ENCODING):
output = self
if "%SOUP-ENCODING%" in output:
output = self.substituteEncoding(output, encoding)
return "%s?>" % self.toEncoding(output, encoding)
class Comment(NavigableString):
def __str__(self, encoding=DEFAULT_OUTPUT_ENCODING):
return "" % NavigableString.__str__(self, encoding)
class Declaration(NavigableString):
def __str__(self, encoding=DEFAULT_OUTPUT_ENCODING):
return "" % NavigableString.__str__(self, encoding)
class Tag(PageElement):
"""Represents a found HTML tag with its attributes and contents."""
XML_SPECIAL_CHARS_TO_ENTITIES = { "'" : "squot",
'"' : "quote",
"&" : "amp",
"<" : "lt",
">" : "gt" }
def __init__(self, parser, name, attrs=None, parent=None,
previous=None):
"Basic constructor."
# We don't actually store the parser object: that lets extracted
# chunks be garbage-collected
self.parserClass = parser.__class__
self.isSelfClosing = parser.isSelfClosingTag(name)
self.name = name
if attrs == None:
attrs = []
self.attrs = attrs
self.contents = []
self.setup(parent, previous)
self.hidden = False
self.containsSubstitutions = False
def get(self, key, default=None):
"""Returns the value of the 'key' attribute for the tag, or
the value given for 'default' if it doesn't have that
attribute."""
return self._getAttrMap().get(key, default)
def has_key(self, key):
return self._getAttrMap().has_key(key)
def __getitem__(self, key):
"""tag[key] returns the value of the 'key' attribute for the tag,
and throws an exception if it's not there."""
return self._getAttrMap()[key]
def __iter__(self):
"Iterating over a tag iterates over its contents."
return iter(self.contents)
def __len__(self):
"The length of a tag is the length of its list of contents."
return len(self.contents)
def __contains__(self, x):
return x in self.contents
def __nonzero__(self):
"A tag is non-None even if it has no contents."
return True
def __setitem__(self, key, value):
"""Setting tag[key] sets the value of the 'key' attribute for the
tag."""
self._getAttrMap()
self.attrMap[key] = value
found = False
for i in range(0, len(self.attrs)):
if self.attrs[i][0] == key:
self.attrs[i] = (key, value)
found = True
if not found:
self.attrs.append((key, value))
self._getAttrMap()[key] = value
def __delitem__(self, key):
"Deleting tag[key] deletes all 'key' attributes for the tag."
for item in self.attrs:
if item[0] == key:
self.attrs.remove(item)
#We don't break because bad HTML can define the same
#attribute multiple times.
self._getAttrMap()
if self.attrMap.has_key(key):
del self.attrMap[key]
def __call__(self, *args, **kwargs):
"""Calling a tag like a function is the same as calling its
findAll() method. Eg. tag('a') returns a list of all the A tags
found within this tag."""
return apply(self.findAll, args, kwargs)
def __getattr__(self, tag):
#print "Getattr %s.%s" % (self.__class__, tag)
if len(tag) > 3 and tag.rfind('Tag') == len(tag)-3:
return self.find(tag[:-3])
elif tag.find('__') != 0:
return self.find(tag)
def __eq__(self, other):
"""Returns true iff this tag has the same name, the same attributes,
and the same contents (recursively) as the given tag.
NOTE: right now this will return false if two tags have the
same attributes in a different order. Should this be fixed?"""
if not hasattr(other, 'name') or not hasattr(other, 'attrs') or not hasattr(other, 'contents') or self.name != other.name or self.attrs != other.attrs or len(self) != len(other):
return False
for i in range(0, len(self.contents)):
if self.contents[i] != other.contents[i]:
return False
return True
def __ne__(self, other):
"""Returns true iff this tag is not identical to the other tag,
as defined in __eq__."""
return not self == other
def __repr__(self, encoding=DEFAULT_OUTPUT_ENCODING):
"""Renders this tag as a string."""
return self.__str__(encoding)
def __unicode__(self):
return self.__str__(None)
def __str__(self, encoding=DEFAULT_OUTPUT_ENCODING,
prettyPrint=False, indentLevel=0):
"""Returns a string or Unicode representation of this tag and
its contents. To get Unicode, pass None for encoding.
NOTE: since Python's HTML parser consumes whitespace, this
method is not certain to reproduce the whitespace present in
the original string."""
encodedName = self.toEncoding(self.name, encoding)
attrs = []
if self.attrs:
for key, val in self.attrs:
fmt = '%s="%s"'
if isString(val):
if self.containsSubstitutions and '%SOUP-ENCODING%' in val:
val = self.substituteEncoding(val, encoding)
# The attribute value either:
#
# * Contains no embedded double quotes or single quotes.
# No problem: we enclose it in double quotes.
# * Contains embedded single quotes. No problem:
# double quotes work here too.
# * Contains embedded double quotes. No problem:
# we enclose it in single quotes.
# * Embeds both single _and_ double quotes. This
# can't happen naturally, but it can happen if
# you modify an attribute value after parsing
# the document. Now we have a bit of a
# problem. We solve it by enclosing the
# attribute in single quotes, and escaping any
# embedded single quotes to XML entities.
if '"' in val:
fmt = "%s='%s'"
# This can't happen naturally, but it can happen
# if you modify an attribute value after parsing.
if "'" in val:
val = val.replace("'", "&squot;")
# Now we're okay w/r/t quotes. But the attribute
# value might also contain angle brackets, or
# ampersands that aren't part of entities. We need
# to escape those to XML entities too.
val = re.sub("([<>]|&(?![^\s]+;))",
lambda x: "&" + self.XML_SPECIAL_CHARS_TO_ENTITIES[x.group(0)[0]] + ";",
val)
attrs.append(fmt % (self.toEncoding(key, encoding),
self.toEncoding(val, encoding)))
close = ''
closeTag = ''
if self.isSelfClosing:
close = ' /'
else:
closeTag = '%s>' % encodedName
indentTag, indentContents = 0, 0
if prettyPrint:
indentTag = indentLevel
space = (' ' * (indentTag-1))
indentContents = indentTag + 1
contents = self.renderContents(encoding, prettyPrint, indentContents)
if self.hidden:
s = contents
else:
s = []
attributeString = ''
if attrs:
attributeString = ' ' + ' '.join(attrs)
if prettyPrint:
s.append(space)
s.append('<%s%s%s>' % (encodedName, attributeString, close))
if prettyPrint:
s.append("\n")
s.append(contents)
if prettyPrint and contents and contents[-1] != "\n":
s.append("\n")
if prettyPrint and closeTag:
s.append(space)
s.append(closeTag)
if prettyPrint and closeTag and self.nextSibling:
s.append("\n")
s = ''.join(s)
return s
def prettify(self, encoding=DEFAULT_OUTPUT_ENCODING):
return self.__str__(encoding, True)
def renderContents(self, encoding=DEFAULT_OUTPUT_ENCODING,
prettyPrint=False, indentLevel=0):
"""Renders the contents of this tag as a string in the given
encoding. If encoding is None, returns a Unicode string.."""
s=[]
for c in self:
text = None
if isinstance(c, NavigableString):
text = c.__str__(encoding)
elif isinstance(c, Tag):
s.append(c.__str__(encoding, prettyPrint, indentLevel))
if text and prettyPrint:
text = text.strip()
if text:
if prettyPrint:
s.append(" " * (indentLevel-1))
s.append(text)
if prettyPrint:
s.append("\n")
return ''.join(s)
#Soup methods
def find(self, name=None, attrs={}, recursive=True, text=None,
**kwargs):
"""Return only the first child of this Tag matching the given
criteria."""
r = None
l = self.findAll(name, attrs, recursive, text, 1, **kwargs)
if l:
r = l[0]
return r
findChild = find
def findAll(self, name=None, attrs={}, recursive=True, text=None,
limit=None, **kwargs):
"""Extracts a list of Tag objects that match the given
criteria. You can specify the name of the Tag and any
attributes you want the Tag to have.
The value of a key-value pair in the 'attrs' map can be a
string, a list of strings, a regular expression object, or a
callable that takes a string and returns whether or not the
string matches for some custom definition of 'matches'. The
same is true of the tag name."""
generator = self.recursiveChildGenerator
if not recursive:
generator = self.childGenerator
return self._findAll(name, attrs, text, limit, generator, **kwargs)
findChildren = findAll
# Pre-3.x compatibility methods
first = find
fetch = findAll
def fetchText(self, text=None, recursive=True, limit=None):
return self.findAll(text=text, recursive=recursive, limit=limit)
def firstText(self, text=None, recursive=True):
return self.find(text=text, recursive=recursive)
#Utility methods
def append(self, tag):
"""Appends the given tag to the contents of this tag."""
self.contents.append(tag)
#Private methods
def _getAttrMap(self):
"""Initializes a map representation of this tag's attributes,
if not already initialized."""
if not getattr(self, 'attrMap'):
self.attrMap = {}
for (key, value) in self.attrs:
self.attrMap[key] = value
return self.attrMap
#Generator methods
def childGenerator(self):
for i in range(0, len(self.contents)):
yield self.contents[i]
raise StopIteration
def recursiveChildGenerator(self):
stack = [(self, 0)]
while stack:
tag, start = stack.pop()
if isinstance(tag, Tag):
for i in range(start, len(tag.contents)):
a = tag.contents[i]
yield a
if isinstance(a, Tag) and tag.contents:
if i < len(tag.contents) - 1:
stack.append((tag, i+1))
stack.append((a, 0))
break
raise StopIteration
# Next, a couple classes to represent queries and their results.
class SoupStrainer:
"""Encapsulates a number of ways of matching a markup element (tag or
text)."""
def __init__(self, name=None, attrs={}, text=None, **kwargs):
self.name = name
if isString(attrs):
kwargs['class'] = attrs
attrs = None
if kwargs:
if attrs:
attrs = attrs.copy()
attrs.update(kwargs)
else:
attrs = kwargs
self.attrs = attrs
self.text = text
def __str__(self):
if self.text:
return self.text
else:
return "%s|%s" % (self.name, self.attrs)
def searchTag(self, markupName=None, markupAttrs={}):
found = None
markup = None
if isinstance(markupName, Tag):
markup = markupName
markupAttrs = markup
callFunctionWithTagData = callable(self.name) \
and not isinstance(markupName, Tag)
if (not self.name) \
or callFunctionWithTagData \
or (markup and self._matches(markup, self.name)) \
or (not markup and self._matches(markupName, self.name)):
if callFunctionWithTagData:
match = self.name(markupName, markupAttrs)
else:
match = True
markupAttrMap = None
for attr, matchAgainst in self.attrs.items():
if not markupAttrMap:
if hasattr(markupAttrs, 'get'):
markupAttrMap = markupAttrs
else:
markupAttrMap = {}
for k,v in markupAttrs:
markupAttrMap[k] = v
attrValue = markupAttrMap.get(attr)
if not self._matches(attrValue, matchAgainst):
match = False
break
if match:
if markup:
found = markup
else:
found = markupName
return found
def search(self, markup):
#print 'looking for %s in %s' % (self, markup)
found = None
# If given a list of items, scan it for a text element that
# matches.
if isList(markup) and not isinstance(markup, Tag):
for element in markup:
if isinstance(element, NavigableString) \
and self.search(element):
found = element
break
# If it's a Tag, make sure its name or attributes match.
# Don't bother with Tags if we're searching for text.
elif isinstance(markup, Tag):
if not self.text:
found = self.searchTag(markup)
# If it's text, make sure the text matches.
elif isinstance(markup, NavigableString) or \
isString(markup):
if self._matches(markup, self.text):
found = markup
else:
raise Exception, "I don't know how to match against a %s" \
% markup.__class__
return found
def _matches(self, markup, matchAgainst):
#print "Matching %s against %s" % (markup, matchAgainst)
result = False
if matchAgainst == True and type(matchAgainst) == types.BooleanType:
result = markup != None
elif callable(matchAgainst):
result = matchAgainst(markup)
else:
#Custom match methods take the tag as an argument, but all
#other ways of matching match the tag name as a string.
if isinstance(markup, Tag):
markup = markup.name
if markup and not isString(markup):
markup = unicode(markup)
#Now we know that chunk is either a string, or None.
if hasattr(matchAgainst, 'match'):
# It's a regexp object.
result = markup and matchAgainst.search(markup)
elif isList(matchAgainst):
result = markup in matchAgainst
elif hasattr(matchAgainst, 'items'):
result = markup.has_key(matchAgainst)
elif matchAgainst and isString(markup):
if isinstance(markup, unicode):
matchAgainst = unicode(matchAgainst)
else:
matchAgainst = str(matchAgainst)
if not result:
result = matchAgainst == markup
return result
class ResultSet(list):
"""A ResultSet is just a list that keeps track of the SoupStrainer
that created it."""
def __init__(self, source):
list.__init__([])
self.source = source
# Now, some helper functions.
def isList(l):
"""Convenience method that works with all 2.x versions of Python
to determine whether or not something is listlike."""
return hasattr(l, '__iter__') \
or (type(l) in (types.ListType, types.TupleType))
def isString(s):
"""Convenience method that works with all 2.x versions of Python
to determine whether or not something is stringlike."""
try:
return isinstance(s, unicode) or isintance(s, basestring)
except NameError:
return isinstance(s, str)
def buildTagMap(default, *args):
"""Turns a list of maps, lists, or scalars into a single map.
Used to build the SELF_CLOSING_TAGS, NESTABLE_TAGS, and
NESTING_RESET_TAGS maps out of lists and partial maps."""
built = {}
for portion in args:
if hasattr(portion, 'items'):
#It's a map. Merge it.
for k,v in portion.items():
built[k] = v
elif isList(portion):
#It's a list. Map each item to the default.
for k in portion:
built[k] = default
else:
#It's a scalar. Map it to the default.
built[portion] = default
return built
# Now, the parser classes.
class BeautifulStoneSoup(Tag, SGMLParser):
"""This class contains the basic parser and search code. It defines
a parser that knows nothing about tag behavior except for the
following:
You can't close a tag without closing all the tags it encloses.
That is, "
(No space between name of closing tag and tag close)
(Extraneous whitespace in declaration)
You can pass in a custom list of (RE object, replace method)
tuples to get Beautiful Soup to scrub your input the way you
want."""
self.parseOnlyThese = parseOnlyThese
self.fromEncoding = fromEncoding
self.smartQuotesTo = smartQuotesTo
self.convertEntities = convertEntities
if self.convertEntities:
# It doesn't make sense to convert encoded characters to
# entities even while you're converting entities to Unicode.
# Just convert it all to Unicode.
self.smartQuotesTo = None
self.instanceSelfClosingTags = buildTagMap(None, selfClosingTags)
SGMLParser.__init__(self)
if hasattr(markup, 'read'): # It's a file-type object.
markup = markup.read()
self.markup = markup
self.markupMassage = markupMassage
try:
self._feed()
except StopParsing:
pass
self.markup = None # The markup can now be GCed
def _feed(self, inDocumentEncoding=None):
# Convert the document to Unicode.
markup = self.markup
if isinstance(markup, unicode):
if not hasattr(self, 'originalEncoding'):
self.originalEncoding = None
else:
dammit = UnicodeDammit\
(markup, [self.fromEncoding, inDocumentEncoding],
smartQuotesTo=self.smartQuotesTo)
markup = dammit.unicode
self.originalEncoding = dammit.originalEncoding
if markup:
if self.markupMassage:
if not isList(self.markupMassage):
self.markupMassage = self.MARKUP_MASSAGE
for fix, m in self.markupMassage:
markup = fix.sub(m, markup)
self.reset()
SGMLParser.feed(self, markup)
# Close out any unfinished strings and close all the open tags.
self.endData()
while self.currentTag.name != self.ROOT_TAG_NAME:
self.popTag()
def __getattr__(self, methodName):
"""This method routes method call requests to either the SGMLParser
superclass or the Tag superclass, depending on the method name."""
#print "__getattr__ called on %s.%s" % (self.__class__, methodName)
if methodName.find('start_') == 0 or methodName.find('end_') == 0 \
or methodName.find('do_') == 0:
return SGMLParser.__getattr__(self, methodName)
elif methodName.find('__') != 0:
return Tag.__getattr__(self, methodName)
else:
raise AttributeError
def isSelfClosingTag(self, name):
"""Returns true iff the given string is the name of a
self-closing tag according to this parser."""
return self.SELF_CLOSING_TAGS.has_key(name) \
or self.instanceSelfClosingTags.has_key(name)
def reset(self):
Tag.__init__(self, self, self.ROOT_TAG_NAME)
self.hidden = 1
SGMLParser.reset(self)
self.currentData = []
self.currentTag = None
self.tagStack = []
self.quoteStack = []
self.pushTag(self)
def popTag(self):
tag = self.tagStack.pop()
# Tags with just one string-owning child get the child as a
# 'string' property, so that soup.tag.string is shorthand for
# soup.tag.contents[0]
if len(self.currentTag.contents) == 1 and \
isinstance(self.currentTag.contents[0], NavigableString):
self.currentTag.string = self.currentTag.contents[0]
#print "Pop", tag.name
if self.tagStack:
self.currentTag = self.tagStack[-1]
return self.currentTag
def pushTag(self, tag):
#print "Push", tag.name
if self.currentTag:
self.currentTag.append(tag)
self.tagStack.append(tag)
self.currentTag = self.tagStack[-1]
def endData(self, containerClass=NavigableString):
if self.currentData:
currentData = ''.join(self.currentData)
if not currentData.strip():
if '\n' in currentData:
currentData = '\n'
else:
currentData = ' '
self.currentData = []
if self.parseOnlyThese and len(self.tagStack) <= 1 and \
(not self.parseOnlyThese.text or \
not self.parseOnlyThese.search(currentData)):
return
o = containerClass(currentData)
o.setup(self.currentTag, self.previous)
if self.previous:
self.previous.next = o
self.previous = o
self.currentTag.contents.append(o)
def _popToTag(self, name, inclusivePop=True):
"""Pops the tag stack up to and including the most recent
instance of the given tag. If inclusivePop is false, pops the tag
stack up to but *not* including the most recent instqance of
the given tag."""
#print "Popping to %s" % name
if name == self.ROOT_TAG_NAME:
return
numPops = 0
mostRecentTag = None
for i in range(len(self.tagStack)-1, 0, -1):
if name == self.tagStack[i].name:
numPops = len(self.tagStack)-i
break
if not inclusivePop:
numPops = numPops - 1
for i in range(0, numPops):
mostRecentTag = self.popTag()
return mostRecentTag
def _smartPop(self, name):
"""We need to pop up to the previous tag of this type, unless
one of this tag's nesting reset triggers comes between this
tag and the previous tag of this type, OR unless this tag is a
generic nesting trigger and another generic nesting trigger
comes between this tag and the previous tag of this type.
Examples:
FooBar
should pop to 'p', not 'b'.
Foo
* | * should pop to 'tr', not the first 'td'
"""
nestingResetTriggers = self.NESTABLE_TAGS.get(name)
isNestable = nestingResetTriggers != None
isResetNesting = self.RESET_NESTING_TAGS.has_key(name)
popTo = None
inclusive = True
for i in range(len(self.tagStack)-1, 0, -1):
p = self.tagStack[i]
if (not p or p.name == name) and not isNestable:
#Non-nestable tags get popped to the top or to their
#last occurance.
popTo = name
break
if (nestingResetTriggers != None
and p.name in nestingResetTriggers) \
or (nestingResetTriggers == None and isResetNesting
and self.RESET_NESTING_TAGS.has_key(p.name)):
#If we encounter one of the nesting reset triggers
#peculiar to this tag, or we encounter another tag
#that causes nesting to reset, pop up to but not
#including that tag.
popTo = p.name
inclusive = False
break
p = p.parent
if popTo:
self._popToTag(popTo, inclusive)
def unknown_starttag(self, name, attrs, selfClosing=0):
#print "Start tag %s: %s" % (name, attrs)
if self.quoteStack:
#This is not a real tag.
#print "<%s> is not real!" % name
attrs = ''.join(map(lambda(x, y): ' %s="%s"' % (x, y), attrs))
self.handle_data('<%s%s>' % (name, attrs))
return
self.endData()
if not self.isSelfClosingTag(name) and not selfClosing:
self._smartPop(name)
if self.parseOnlyThese and len(self.tagStack) <= 1 \
and (self.parseOnlyThese.text or not self.parseOnlyThese.searchTag(name, attrs)):
return
tag = Tag(self, name, attrs, self.currentTag, self.previous)
if self.previous:
self.previous.next = tag
self.previous = tag
self.pushTag(tag)
if selfClosing or self.isSelfClosingTag(name):
self.popTag()
if name in self.QUOTE_TAGS:
#print "Beginning quote (%s)" % name
self.quoteStack.append(name)
self.literal = 1
return tag
def unknown_endtag(self, name):
#print "End tag %s" % name
if self.quoteStack and self.quoteStack[-1] != name:
#This is not a real end tag.
#print "%s> is not real!" % name
self.handle_data('%s>' % name)
return
self.endData()
self._popToTag(name)
if self.quoteStack and self.quoteStack[-1] == name:
self.quoteStack.pop()
self.literal = (len(self.quoteStack) > 0)
def handle_data(self, data):
self.currentData.append(data)
def _toStringSubclass(self, text, subclass):
"""Adds a certain piece of text to the tree as a NavigableString
subclass."""
self.endData()
self.handle_data(text)
self.endData(subclass)
def handle_pi(self, text):
"""Handle a processing instruction as a ProcessingInstruction
object, possibly one with a %SOUP-ENCODING% slot into which an
encoding will be plugged later."""
if text[:3] == "xml":
text = "xml version='1.0' encoding='%SOUP-ENCODING%'"
self._toStringSubclass(text, ProcessingInstruction)
def handle_comment(self, text):
"Handle comments as Comment objects."
self._toStringSubclass(text, Comment)
def handle_charref(self, ref):
"Handle character references as data."
if self.convertEntities in [self.HTML_ENTITIES,
self.XML_ENTITIES]:
data = unichr(int(ref))
else:
data = '%s;' % ref
self.handle_data(data)
def handle_entityref(self, ref):
"""Handle entity references as data, possibly converting known
HTML entity references to the corresponding Unicode
characters."""
data = None
if self.convertEntities == self.HTML_ENTITIES or \
(self.convertEntities == self.XML_ENTITIES and \
self.XML_ENTITY_LIST.get(ref)):
try:
data = unichr(name2codepoint[ref])
except KeyError:
pass
if not data:
data = '&%s;' % ref
self.handle_data(data)
def handle_decl(self, data):
"Handle DOCTYPEs and the like as Declaration objects."
self._toStringSubclass(data, Declaration)
def parse_declaration(self, i):
"""Treat a bogus SGML declaration as raw data. Treat a CDATA
declaration as a CData object."""
j = None
if self.rawdata[i:i+9] == '', i)
if k == -1:
k = len(self.rawdata)
data = self.rawdata[i+9:k]
j = k+3
self._toStringSubclass(data, CData)
else:
try:
j = SGMLParser.parse_declaration(self, i)
except SGMLParseError:
toHandle = self.rawdata[i:]
self.handle_data(toHandle)
j = i + len(toHandle)
return j
class BeautifulSoup(BeautifulStoneSoup):
"""This parser knows the following facts about HTML:
* Some tags have no closing tag and should be interpreted as being
closed as soon as they are encountered.
* The text inside some tags (ie. 'script') may contain tags which
are not really part of the document and which should be parsed
as text, not tags. If you want to parse the text as tags, you can
always fetch it and parse it explicitly.
* Tag nesting rules:
Most tags can't be nested at all. For instance, the occurance of
a tag should implicitly close the previous tag. Para1 Para2 should be transformed into: Para1 Para2 Some tags can be nested arbitrarily. For instance, the occurance of a tag should _not_ implicitly close the previoustag. Alice said:Bob said:Blah should NOT be transformed into: Alice said:Bob said:Blah Some tags can be nested, but the nesting is reset by the interposition of other tags. For instance, a |