|
rpm-build |
d9acb6 |
"""External interface to the BeautifulSoup HTML parser.
|
|
rpm-build |
d9acb6 |
"""
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
__all__ = ["fromstring", "parse", "convert_tree"]
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
import re
|
|
rpm-build |
d9acb6 |
from lxml import etree, html
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
try:
|
|
rpm-build |
d9acb6 |
from bs4 import (
|
|
rpm-build |
d9acb6 |
BeautifulSoup, Tag, Comment, ProcessingInstruction, NavigableString,
|
|
rpm-build |
d9acb6 |
Declaration, Doctype)
|
|
rpm-build |
d9acb6 |
_DECLARATION_OR_DOCTYPE = (Declaration, Doctype)
|
|
rpm-build |
d9acb6 |
except ImportError:
|
|
rpm-build |
d9acb6 |
from BeautifulSoup import (
|
|
rpm-build |
d9acb6 |
BeautifulSoup, Tag, Comment, ProcessingInstruction, NavigableString,
|
|
rpm-build |
d9acb6 |
Declaration)
|
|
rpm-build |
d9acb6 |
_DECLARATION_OR_DOCTYPE = Declaration
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def fromstring(data, beautifulsoup=None, makeelement=None, **bsargs):
|
|
rpm-build |
d9acb6 |
"""Parse a string of HTML data into an Element tree using the
|
|
rpm-build |
d9acb6 |
BeautifulSoup parser.
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
Returns the root ``<html>`` Element of the tree.
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
You can pass a different BeautifulSoup parser through the
|
|
rpm-build |
d9acb6 |
`beautifulsoup` keyword, and a diffent Element factory function
|
|
rpm-build |
d9acb6 |
through the `makeelement` keyword. By default, the standard
|
|
rpm-build |
d9acb6 |
``BeautifulSoup`` class and the default factory of `lxml.html` are
|
|
rpm-build |
d9acb6 |
used.
|
|
rpm-build |
d9acb6 |
"""
|
|
rpm-build |
d9acb6 |
return _parse(data, beautifulsoup, makeelement, **bsargs)
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def parse(file, beautifulsoup=None, makeelement=None, **bsargs):
|
|
rpm-build |
d9acb6 |
"""Parse a file into an ElemenTree using the BeautifulSoup parser.
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
You can pass a different BeautifulSoup parser through the
|
|
rpm-build |
d9acb6 |
`beautifulsoup` keyword, and a diffent Element factory function
|
|
rpm-build |
d9acb6 |
through the `makeelement` keyword. By default, the standard
|
|
rpm-build |
d9acb6 |
``BeautifulSoup`` class and the default factory of `lxml.html` are
|
|
rpm-build |
d9acb6 |
used.
|
|
rpm-build |
d9acb6 |
"""
|
|
rpm-build |
d9acb6 |
if not hasattr(file, 'read'):
|
|
rpm-build |
d9acb6 |
file = open(file)
|
|
rpm-build |
d9acb6 |
root = _parse(file, beautifulsoup, makeelement, **bsargs)
|
|
rpm-build |
d9acb6 |
return etree.ElementTree(root)
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def convert_tree(beautiful_soup_tree, makeelement=None):
|
|
rpm-build |
d9acb6 |
"""Convert a BeautifulSoup tree to a list of Element trees.
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
Returns a list instead of a single root Element to support
|
|
rpm-build |
d9acb6 |
HTML-like soup with more than one root element.
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
You can pass a different Element factory through the `makeelement`
|
|
rpm-build |
d9acb6 |
keyword.
|
|
rpm-build |
d9acb6 |
"""
|
|
rpm-build |
d9acb6 |
root = _convert_tree(beautiful_soup_tree, makeelement)
|
|
rpm-build |
d9acb6 |
children = root.getchildren()
|
|
rpm-build |
d9acb6 |
for child in children:
|
|
rpm-build |
d9acb6 |
root.remove(child)
|
|
rpm-build |
d9acb6 |
return children
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
# helpers
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def _parse(source, beautifulsoup, makeelement, **bsargs):
|
|
rpm-build |
d9acb6 |
if beautifulsoup is None:
|
|
rpm-build |
d9acb6 |
beautifulsoup = BeautifulSoup
|
|
rpm-build |
d9acb6 |
if hasattr(beautifulsoup, "HTML_ENTITIES"): # bs3
|
|
rpm-build |
d9acb6 |
if 'convertEntities' not in bsargs:
|
|
rpm-build |
d9acb6 |
bsargs['convertEntities'] = 'html'
|
|
rpm-build |
d9acb6 |
if hasattr(beautifulsoup, "DEFAULT_BUILDER_FEATURES"): # bs4
|
|
rpm-build |
d9acb6 |
if 'features' not in bsargs:
|
|
rpm-build |
d9acb6 |
bsargs['features'] = 'html.parser' # use Python html parser
|
|
rpm-build |
d9acb6 |
tree = beautifulsoup(source, **bsargs)
|
|
rpm-build |
d9acb6 |
root = _convert_tree(tree, makeelement)
|
|
rpm-build |
d9acb6 |
# from ET: wrap the document in a html root element, if necessary
|
|
rpm-build |
d9acb6 |
if len(root) == 1 and root[0].tag == "html":
|
|
rpm-build |
d9acb6 |
return root[0]
|
|
rpm-build |
d9acb6 |
root.tag = "html"
|
|
rpm-build |
d9acb6 |
return root
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
_parse_doctype_declaration = re.compile(
|
|
rpm-build |
d9acb6 |
r'(?:\s|[
|
|
rpm-build |
d9acb6 |
r'(?:\s+PUBLIC)?(?:\s+(\'[^\']*\'|"[^"]*"))?'
|
|
rpm-build |
d9acb6 |
r'(?:\s+(\'[^\']*\'|"[^"]*"))?',
|
|
rpm-build |
d9acb6 |
re.IGNORECASE).match
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
class _PseudoTag:
|
|
rpm-build |
d9acb6 |
# Minimal imitation of BeautifulSoup.Tag
|
|
rpm-build |
d9acb6 |
def __init__(self, contents):
|
|
rpm-build |
d9acb6 |
self.name = 'html'
|
|
rpm-build |
d9acb6 |
self.attrs = []
|
|
rpm-build |
d9acb6 |
self.contents = contents
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def __iter__(self):
|
|
rpm-build |
d9acb6 |
return self.contents.__iter__()
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def _convert_tree(beautiful_soup_tree, makeelement):
|
|
rpm-build |
d9acb6 |
if makeelement is None:
|
|
rpm-build |
d9acb6 |
makeelement = html.html_parser.makeelement
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
# Split the tree into three parts:
|
|
rpm-build |
d9acb6 |
# i) everything before the root element: document type
|
|
rpm-build |
d9acb6 |
# declaration, comments, processing instructions, whitespace
|
|
rpm-build |
d9acb6 |
# ii) the root(s),
|
|
rpm-build |
d9acb6 |
# iii) everything after the root: comments, processing
|
|
rpm-build |
d9acb6 |
# instructions, whitespace
|
|
rpm-build |
d9acb6 |
first_element_idx = last_element_idx = None
|
|
rpm-build |
d9acb6 |
html_root = declaration = None
|
|
rpm-build |
d9acb6 |
for i, e in enumerate(beautiful_soup_tree):
|
|
rpm-build |
d9acb6 |
if isinstance(e, Tag):
|
|
rpm-build |
d9acb6 |
if first_element_idx is None:
|
|
rpm-build |
d9acb6 |
first_element_idx = i
|
|
rpm-build |
d9acb6 |
last_element_idx = i
|
|
rpm-build |
d9acb6 |
if html_root is None and e.name and e.name.lower() == 'html':
|
|
rpm-build |
d9acb6 |
html_root = e
|
|
rpm-build |
d9acb6 |
elif declaration is None and isinstance(e, _DECLARATION_OR_DOCTYPE):
|
|
rpm-build |
d9acb6 |
declaration = e
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
# For a nice, well-formatted document, the variable roots below is
|
|
rpm-build |
d9acb6 |
# a list consisting of a single <html> element. However, the document
|
|
rpm-build |
d9acb6 |
# may be a soup like '<meta><head><title>Hello</head><body>Hi
|
|
rpm-build |
d9acb6 |
# all<\p>'. In this example roots is a list containing meta, head
|
|
rpm-build |
d9acb6 |
# and body elements.
|
|
rpm-build |
d9acb6 |
if first_element_idx is None:
|
|
rpm-build |
d9acb6 |
pre_root = post_root = []
|
|
rpm-build |
d9acb6 |
roots = beautiful_soup_tree.contents
|
|
rpm-build |
d9acb6 |
else:
|
|
rpm-build |
d9acb6 |
pre_root = beautiful_soup_tree.contents[:first_element_idx]
|
|
rpm-build |
d9acb6 |
roots = beautiful_soup_tree.contents[first_element_idx:last_element_idx+1]
|
|
rpm-build |
d9acb6 |
post_root = beautiful_soup_tree.contents[last_element_idx+1:]
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
# Reorganize so that there is one <html> root...
|
|
rpm-build |
d9acb6 |
if html_root is not None:
|
|
rpm-build |
d9acb6 |
# ... use existing one if possible, ...
|
|
rpm-build |
d9acb6 |
i = roots.index(html_root)
|
|
rpm-build |
d9acb6 |
html_root.contents = roots[:i] + html_root.contents + roots[i+1:]
|
|
rpm-build |
d9acb6 |
else:
|
|
rpm-build |
d9acb6 |
# ... otherwise create a new one.
|
|
rpm-build |
d9acb6 |
html_root = _PseudoTag(roots)
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
convert_node = _init_node_converters(makeelement)
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
# Process pre_root
|
|
rpm-build |
d9acb6 |
res_root = convert_node(html_root)
|
|
rpm-build |
d9acb6 |
prev = res_root
|
|
rpm-build |
d9acb6 |
for e in reversed(pre_root):
|
|
rpm-build |
d9acb6 |
converted = convert_node(e)
|
|
rpm-build |
d9acb6 |
if converted is not None:
|
|
rpm-build |
d9acb6 |
prev.addprevious(converted)
|
|
rpm-build |
d9acb6 |
prev = converted
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
# ditto for post_root
|
|
rpm-build |
d9acb6 |
prev = res_root
|
|
rpm-build |
d9acb6 |
for e in post_root:
|
|
rpm-build |
d9acb6 |
converted = convert_node(e)
|
|
rpm-build |
d9acb6 |
if converted is not None:
|
|
rpm-build |
d9acb6 |
prev.addnext(converted)
|
|
rpm-build |
d9acb6 |
prev = converted
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
if declaration is not None:
|
|
rpm-build |
d9acb6 |
try:
|
|
rpm-build |
d9acb6 |
# bs4 provides full Doctype string
|
|
rpm-build |
d9acb6 |
doctype_string = declaration.output_ready()
|
|
rpm-build |
d9acb6 |
except AttributeError:
|
|
rpm-build |
d9acb6 |
doctype_string = declaration.string
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
match = _parse_doctype_declaration(doctype_string)
|
|
rpm-build |
d9acb6 |
if not match:
|
|
rpm-build |
d9acb6 |
# Something is wrong if we end up in here. Since soupparser should
|
|
rpm-build |
d9acb6 |
# tolerate errors, do not raise Exception, just let it pass.
|
|
rpm-build |
d9acb6 |
pass
|
|
rpm-build |
d9acb6 |
else:
|
|
rpm-build |
d9acb6 |
external_id, sys_uri = match.groups()
|
|
rpm-build |
d9acb6 |
docinfo = res_root.getroottree().docinfo
|
|
rpm-build |
d9acb6 |
# strip quotes and update DOCTYPE values (any of None, '', '...')
|
|
rpm-build |
d9acb6 |
docinfo.public_id = external_id and external_id[1:-1]
|
|
rpm-build |
d9acb6 |
docinfo.system_url = sys_uri and sys_uri[1:-1]
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
return res_root
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def _init_node_converters(makeelement):
|
|
rpm-build |
d9acb6 |
converters = {}
|
|
rpm-build |
d9acb6 |
ordered_node_types = []
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def converter(*types):
|
|
rpm-build |
d9acb6 |
def add(handler):
|
|
rpm-build |
d9acb6 |
for t in types:
|
|
rpm-build |
d9acb6 |
converters[t] = handler
|
|
rpm-build |
d9acb6 |
ordered_node_types.append(t)
|
|
rpm-build |
d9acb6 |
return handler
|
|
rpm-build |
d9acb6 |
return add
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def find_best_converter(node):
|
|
rpm-build |
d9acb6 |
for t in ordered_node_types:
|
|
rpm-build |
d9acb6 |
if isinstance(node, t):
|
|
rpm-build |
d9acb6 |
return converters[t]
|
|
rpm-build |
d9acb6 |
return None
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def convert_node(bs_node, parent=None):
|
|
rpm-build |
d9acb6 |
# duplicated in convert_tag() below
|
|
rpm-build |
d9acb6 |
try:
|
|
rpm-build |
d9acb6 |
handler = converters[type(bs_node)]
|
|
rpm-build |
d9acb6 |
except KeyError:
|
|
rpm-build |
d9acb6 |
handler = converters[type(bs_node)] = find_best_converter(bs_node)
|
|
rpm-build |
d9acb6 |
if handler is None:
|
|
rpm-build |
d9acb6 |
return None
|
|
rpm-build |
d9acb6 |
return handler(bs_node, parent)
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def map_attrs(bs_attrs):
|
|
rpm-build |
d9acb6 |
if isinstance(bs_attrs, dict): # bs4
|
|
rpm-build |
d9acb6 |
attribs = {}
|
|
rpm-build |
d9acb6 |
for k, v in bs_attrs.items():
|
|
rpm-build |
d9acb6 |
if isinstance(v, list):
|
|
rpm-build |
d9acb6 |
v = " ".join(v)
|
|
rpm-build |
d9acb6 |
attribs[k] = unescape(v)
|
|
rpm-build |
d9acb6 |
else:
|
|
rpm-build |
d9acb6 |
attribs = dict((k, unescape(v)) for k, v in bs_attrs)
|
|
rpm-build |
d9acb6 |
return attribs
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def append_text(parent, text):
|
|
rpm-build |
d9acb6 |
if len(parent) == 0:
|
|
rpm-build |
d9acb6 |
parent.text = (parent.text or '') + text
|
|
rpm-build |
d9acb6 |
else:
|
|
rpm-build |
d9acb6 |
parent[-1].tail = (parent[-1].tail or '') + text
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
# converters are tried in order of their definition
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
@converter(Tag, _PseudoTag)
|
|
rpm-build |
d9acb6 |
def convert_tag(bs_node, parent):
|
|
rpm-build |
d9acb6 |
attrs = bs_node.attrs
|
|
rpm-build |
d9acb6 |
if parent is not None:
|
|
rpm-build |
d9acb6 |
attribs = map_attrs(attrs) if attrs else None
|
|
rpm-build |
d9acb6 |
res = etree.SubElement(parent, bs_node.name, attrib=attribs)
|
|
rpm-build |
d9acb6 |
else:
|
|
rpm-build |
d9acb6 |
attribs = map_attrs(attrs) if attrs else {}
|
|
rpm-build |
d9acb6 |
res = makeelement(bs_node.name, attrib=attribs)
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
for child in bs_node:
|
|
rpm-build |
d9acb6 |
# avoid double recursion by inlining convert_node(), see above
|
|
rpm-build |
d9acb6 |
try:
|
|
rpm-build |
d9acb6 |
handler = converters[type(child)]
|
|
rpm-build |
d9acb6 |
except KeyError:
|
|
rpm-build |
d9acb6 |
pass
|
|
rpm-build |
d9acb6 |
else:
|
|
rpm-build |
d9acb6 |
if handler is not None:
|
|
rpm-build |
d9acb6 |
handler(child, res)
|
|
rpm-build |
d9acb6 |
continue
|
|
rpm-build |
d9acb6 |
convert_node(child, res)
|
|
rpm-build |
d9acb6 |
return res
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
@converter(Comment)
|
|
rpm-build |
d9acb6 |
def convert_comment(bs_node, parent):
|
|
rpm-build |
d9acb6 |
res = html.HtmlComment(bs_node)
|
|
rpm-build |
d9acb6 |
if parent is not None:
|
|
rpm-build |
d9acb6 |
parent.append(res)
|
|
rpm-build |
d9acb6 |
return res
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
@converter(ProcessingInstruction)
|
|
rpm-build |
d9acb6 |
def convert_pi(bs_node, parent):
|
|
rpm-build |
d9acb6 |
if bs_node.endswith('?'):
|
|
rpm-build |
d9acb6 |
# The PI is of XML style () but BeautifulSoup
|
|
rpm-build |
d9acb6 |
# interpreted it as being SGML style (). Fix.
|
|
rpm-build |
d9acb6 |
bs_node = bs_node[:-1]
|
|
rpm-build |
d9acb6 |
res = etree.ProcessingInstruction(*bs_node.split(' ', 1))
|
|
rpm-build |
d9acb6 |
if parent is not None:
|
|
rpm-build |
d9acb6 |
parent.append(res)
|
|
rpm-build |
d9acb6 |
return res
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
@converter(NavigableString)
|
|
rpm-build |
d9acb6 |
def convert_text(bs_node, parent):
|
|
rpm-build |
d9acb6 |
if parent is not None:
|
|
rpm-build |
d9acb6 |
append_text(parent, unescape(bs_node))
|
|
rpm-build |
d9acb6 |
return None
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
return convert_node
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
# copied from ET's ElementSoup
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
try:
|
|
rpm-build |
d9acb6 |
from html.entities import name2codepoint # Python 3
|
|
rpm-build |
d9acb6 |
except ImportError:
|
|
rpm-build |
d9acb6 |
from htmlentitydefs import name2codepoint
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
handle_entities = re.compile(r"&(\w+);").sub
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
try:
|
|
rpm-build |
d9acb6 |
unichr
|
|
rpm-build |
d9acb6 |
except NameError:
|
|
rpm-build |
d9acb6 |
# Python 3
|
|
rpm-build |
d9acb6 |
unichr = chr
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def unescape(string):
|
|
rpm-build |
d9acb6 |
if not string:
|
|
rpm-build |
d9acb6 |
return ''
|
|
rpm-build |
d9acb6 |
# work around oddities in BeautifulSoup's entity handling
|
|
rpm-build |
d9acb6 |
def unescape_entity(m):
|
|
rpm-build |
d9acb6 |
try:
|
|
rpm-build |
d9acb6 |
return unichr(name2codepoint[m.group(1)])
|
|
rpm-build |
d9acb6 |
except KeyError:
|
|
rpm-build |
d9acb6 |
return m.group(0) # use as is
|
|
rpm-build |
d9acb6 |
return handle_entities(unescape_entity, string)
|