2
This module implements a VERY limited parser that finds <link> tags in
3
the head of HTML or XHTML documents and parses out their attributes
4
according to the OpenID spec. It is a liberal parser, but it requires
5
these things from the data in order to work:
7
* There must be an open <html> tag
9
* There must be an open <head> tag inside of the <html> tag
11
* Only <link>s that are found inside of the <head> tag are parsed
14
* The parser follows the OpenID specification in resolving the
15
attributes of the link tags. This means that the attributes DO NOT
16
get resolved as they would by an XML or HTML parser. In particular,
17
only certain entities get replaced, and href attributes do not get
18
resolved relative to a base URL.
20
From http://openid.net/specs.bml#linkrel:
22
* The openid.server URL MUST be an absolute URL. OpenID consumers
23
MUST NOT attempt to resolve relative URLs.
25
* The openid.server URL MUST NOT include entities other than &,
26
<, >, and ".
28
The parser ignores SGML comments and <![CDATA[blocks]]>. Both kinds of
29
quoting are allowed for attributes.
31
The parser deals with invalid markup in these ways:
33
* Tag names are not case-sensitive
35
* The <html> tag is accepted even when it is not at the top level
37
* The <head> tag is accepted even when it is not a direct child of
38
the <html> tag, but a <html> tag must be an ancestor of the <head>
41
* <link> tags are accepted even when they are not direct children of
42
the <head> tag, but a <head> tag must be an ancestor of the <link>
45
* If there is no closing tag for an open <html> or <head> tag, the
46
remainder of the document is viewed as being inside of the tag. If
47
there is no closing tag for a <link> tag, the link tag is treated
48
as a short tag. Exceptions to this rule are that <html> closes
49
<html> and <body> or <head> closes <head>
51
* Attributes of the <link> tag are not required to be quoted.
53
* In the case of duplicated attribute names, the attribute coming
54
last in the tag will be the value returned.
56
* Any text that does not parse as an attribute within a link tag will
57
be ignored. (e.g. <link pumpkin rel='openid.server' /> will ignore
60
* If there are more than one <html> or <head> tag, the parser only
61
looks inside of the first one.
63
* The contents of <script> tags are ignored entirely, except unclosed
64
<script> tags. Unclosed <script> tags are ignored.
66
* Any other invalid markup is ignored, including unclosed SGML
67
comments and unclosed <![CDATA[blocks.
70
__all__ = ['parseLinkAttrs']
74
flags = ( re.DOTALL # Match newlines with '.'
76
| re.VERBOSE # Allow comments and whitespace in patterns
77
| re.UNICODE # Make \b respect Unicode word boundaries
80
# Stuff to remove before we start looking for tags
81
removed_re = re.compile(r'''
91
# make sure script is not an XML namespace
99
# Starts with the tag name at a word boundary, where the tag name is
103
# All of the stuff up to a ">", hopefully attributes.
106
(?: # Match a short tag
115
(?: # One of the specified close tags
126
def tagMatcher(tag_name, *close_tags):
128
options = '|'.join((tag_name,) + close_tags)
129
closers = '(?:%s)' % (options,)
133
expr = tag_expr % locals()
134
return re.compile(expr, flags)
136
# Must contain at least an open html and an open head tag
137
html_find = tagMatcher('html')
138
head_find = tagMatcher('head', 'body')
139
link_find = re.compile(r'<link\b(?!:)', flags)
141
attr_find = re.compile(r'''
142
# Must start with a sequence of word-characters, followed by an equals sign
145
# Then either a quoted or unquoted attribute
148
# Match everything that\'s between matching quote marks
149
(?P<qopen>["\'])(?P<q_val>.*?)(?P=qopen)
152
# If the value is not quoted, match up to whitespace
153
(?P<unq_val>(?:[^\s<>/]|/(?!>))+)
161
# Entity replacement:
169
ent_replace = re.compile(r'&(%s);' % '|'.join(replacements.keys()))
171
"Replace the entities that are specified by OpenID"
172
return replacements.get(mo.group(1), mo.group())
174
def parseLinkAttrs(html):
175
"""Find all link tags in a string representing a HTML document and
176
return a list of their attributes.
178
@param html: the text to parse
179
@type html: str or unicode
181
@return: A list of dictionaries of attributes, one for each link tag
182
@rtype: [[(type(html), type(html))]]
184
stripped = removed_re.sub('', html)
185
html_mo = html_find.search(stripped)
186
if html_mo is None or html_mo.start('contents') == -1:
189
start, end = html_mo.span('contents')
190
head_mo = head_find.search(stripped, start, end)
191
if head_mo is None or head_mo.start('contents') == -1:
194
start, end = head_mo.span('contents')
195
link_mos = link_find.finditer(stripped, head_mo.start(), head_mo.end())
198
for link_mo in link_mos:
199
start = link_mo.start() + 5
201
for attr_mo in attr_find.finditer(stripped, start):
202
if attr_mo.lastgroup == 'end_link':
205
# Either q_val or unq_val must be present, but not both
206
# unq_val is a True (non-empty) value if it is present
207
attr_name, q_val, unq_val = attr_mo.group(
208
'attr_name', 'q_val', 'unq_val')
209
attr_val = ent_replace.sub(replaceEnt, unq_val or q_val)
211
link_attrs[attr_name] = attr_val
213
matches.append(link_attrs)
217
def relMatches(rel_attr, target_rel):
218
"""Does this target_rel appear in the rel_str?"""
220
rels = rel_attr.strip().split()
223
if rel == target_rel:
228
def linkHasRel(link_attrs, target_rel):
229
"""Does this link have target_rel as a relationship?"""
231
rel_attr = link_attrs.get('rel')
232
return rel_attr and relMatches(rel_attr, target_rel)
234
def findLinksRel(link_attrs_list, target_rel):
235
"""Filter the list of link attributes on whether it has target_rel
236
as a relationship."""
238
matchesTarget = lambda attrs: linkHasRel(attrs, target_rel)
239
return filter(matchesTarget, link_attrs_list)
241
def findFirstHref(link_attrs_list, target_rel):
242
"""Return the value of the href attribute for the first link tag
243
in the list that has target_rel as a relationship."""
245
matches = findLinksRel(link_attrs_list, target_rel)
249
return first.get('href')
251
class ParseError(ValueError):
252
"""Exception for errors in parsing the HTML text for OpenID
255
def openIDDiscover(html_text):
256
"""Parse OpenID settings out of the gived HTML text
259
# XXX: document interface
262
link_attrs = parseLinkAttrs(html_text)
264
server_url = findFirstHref(link_attrs, 'openid.server')
265
if server_url is None:
266
raise ParseError('No openid.server found')
268
delegate_url = findFirstHref(link_attrs, 'openid.delegate')
269
return delegate_url, server_url