Mercurial > cpython-lru_cache-weakref
annotate Lib/HTMLParser.py @ 19283:b865cd5c1925 legacy-trunk
Added comments before recently added/assigned slots in the type object,
so the backward compatibility issues will be easier to understand. I only
added comments indicating additions and assignments back to Python 2.0.
| author | Fred Drake <fdrake@acm.org> |
|---|---|
| date | Wed, 15 Aug 2001 18:32:33 +0000 |
| parents | 00935031b373 |
| children | c403af003884 |
| rev | line source |
|---|---|
|
18984
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
1 """A parser for HTML and XHTML.""" |
|
17853
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
2 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
3 # This file is based on sgmllib.py, but the API is slightly different. |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
4 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
5 # XXX There should be a way to distinguish between PCDATA (parsed |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
6 # character data -- the normal case), RCDATA (replaceable character |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
7 # data -- only char and entity references and end tags are special) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
8 # and CDATA (character data -- only end tags are special). |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
9 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
10 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
11 import re |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
12 import string |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
13 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
14 # Regular expressions used for parsing |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
15 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
16 interesting_normal = re.compile('[&<]') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
17 interesting_cdata = re.compile(r'<(/|\Z)') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
18 incomplete = re.compile('&([a-zA-Z][-.a-zA-Z0-9]*|#[0-9]*)?') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
19 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
20 entityref = re.compile('&([a-zA-Z][-.a-zA-Z0-9]*)[^a-zA-Z0-9]') |
|
18984
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
21 charref = re.compile('&#(?:[0-9]+|[xX][0-9a-fA-F]+)[^0-9a-fA-F]') |
|
17853
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
22 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
23 starttagopen = re.compile('<[a-zA-Z]') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
24 piopen = re.compile(r'<\?') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
25 piclose = re.compile('>') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
26 endtagopen = re.compile('</') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
27 declopen = re.compile('<!') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
28 special = re.compile('<![^<>]*>') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
29 commentopen = re.compile('<!--') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
30 commentclose = re.compile(r'--\s*>') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
31 tagfind = re.compile('[a-zA-Z][-.a-zA-Z0-9:_]*') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
32 attrfind = re.compile( |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
33 r'\s*([a-zA-Z_][-.:a-zA-Z_0-9]*)(\s*=\s*' |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
34 r'(\'[^\']*\'|"[^"]*"|[-a-zA-Z0-9./:;+*%?!&$\(\)_#=~]*))?') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
35 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
36 locatestarttagend = re.compile(r""" |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
37 <[a-zA-Z][-.a-zA-Z0-9:_]* # tag name |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
38 (?:\s+ # whitespace before attribute name |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
39 (?:[a-zA-Z_][-.:a-zA-Z0-9_]* # attribute name |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
40 (?:\s*=\s* # value indicator |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
41 (?:'[^']*' # LITA-enclosed value |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
42 |\"[^\"]*\" # LIT-enclosed value |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
43 |[^'\">\s]+ # bare value |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
44 ) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
45 )? |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
46 ) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
47 )* |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
48 \s* # trailing whitespace |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
49 """, re.VERBOSE) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
50 endstarttag = re.compile(r"\s*/?>") |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
51 endendtag = re.compile('>') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
52 endtagfind = re.compile('</\s*([a-zA-Z][-.a-zA-Z0-9:_]*)\s*>') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
53 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
54 declname = re.compile(r'[a-zA-Z][-_.a-zA-Z0-9]*\s*') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
55 declstringlit = re.compile(r'(\'[^\']*\'|"[^"]*")\s*') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
56 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
57 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
58 class HTMLParseError(Exception): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
59 """Exception raised for all parse errors.""" |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
60 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
61 def __init__(self, msg, position=(None, None)): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
62 assert msg |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
63 self.msg = msg |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
64 self.lineno = position[0] |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
65 self.offset = position[1] |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
66 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
67 def __str__(self): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
68 result = self.msg |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
69 if self.lineno is not None: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
70 result = result + ", at line %d" % self.lineno |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
71 if self.offset is not None: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
72 result = result + ", column %d" % (self.offset + 1) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
73 return result |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
74 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
75 |
|
18984
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
76 class HTMLParser: |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
77 """Find tags and other markup and call handler functions. |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
78 |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
79 Usage: |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
80 p = HTMLParser() |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
81 p.feed(data) |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
82 ... |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
83 p.close() |
|
17932
ce7f0e380830
Removed incorrect comment left over from sgmllib.py.
Guido van Rossum <guido@python.org>
parents:
17853
diff
changeset
|
84 |
|
18984
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
85 Start tags are handled by calling self.handle_starttag() or |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
86 self.handle_startendtag(); end tags by self.handle_endtag(). The |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
87 data between tags is passed from the parser to the derived class |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
88 by calling self.handle_data() with the data as argument (the data |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
89 may be split up in arbitrary chunks). Entity references are |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
90 passed by calling self.handle_entityref() with the entity |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
91 reference as the argument. Numeric character references are |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
92 passed to self.handle_charref() with the string containing the |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
93 reference as the argument. |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
94 """ |
|
17853
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
95 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
96 CDATA_CONTENT_ELEMENTS = ("script", "style") |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
97 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
98 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
99 def __init__(self): |
|
18984
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
100 """Initialize and reset this instance.""" |
|
17853
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
101 self.reset() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
102 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
103 def reset(self): |
|
18984
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
104 """Reset this instance. Loses all unprocessed data.""" |
|
17853
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
105 self.rawdata = '' |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
106 self.stack = [] |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
107 self.lasttag = '???' |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
108 self.lineno = 1 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
109 self.offset = 0 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
110 self.interesting = interesting_normal |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
111 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
112 def feed(self, data): |
|
18984
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
113 """Feed data to the parser. |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
114 |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
115 Call this as often as you want, with as little or as much text |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
116 as you want (may include '\n'). |
|
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
117 """ |
|
17853
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
118 self.rawdata = self.rawdata + data |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
119 self.goahead(0) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
120 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
121 def close(self): |
|
18984
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
122 """Handle any buffered data.""" |
|
17853
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
123 self.goahead(1) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
124 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
125 # Internal -- update line number and offset. This should be |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
126 # called for each piece of data exactly once, in order -- in other |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
127 # words the concatenation of all the input strings to this |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
128 # function should be exactly the entire input. |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
129 def updatepos(self, i, j): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
130 if i >= j: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
131 return j |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
132 rawdata = self.rawdata |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
133 nlines = string.count(rawdata, "\n", i, j) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
134 if nlines: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
135 self.lineno = self.lineno + nlines |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
136 pos = string.rindex(rawdata, "\n", i, j) # Should not fail |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
137 self.offset = j-(pos+1) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
138 else: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
139 self.offset = self.offset + j-i |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
140 return j |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
141 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
142 def getpos(self): |
|
18984
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
143 """Return current line number and offset.""" |
|
17853
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
144 return self.lineno, self.offset |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
145 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
146 __starttag_text = None |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
147 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
148 def get_starttag_text(self): |
|
18984
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
149 """Return full source of start tag: '<...>'.""" |
|
17853
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
150 return self.__starttag_text |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
151 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
152 def set_cdata_mode(self): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
153 self.interesting = interesting_cdata |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
154 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
155 def clear_cdata_mode(self): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
156 self.interesting = interesting_normal |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
157 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
158 # Internal -- handle data as far as reasonable. May leave state |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
159 # and data to be processed by a subsequent call. If 'end' is |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
160 # true, force handling all data as if followed by EOF marker. |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
161 def goahead(self, end): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
162 rawdata = self.rawdata |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
163 i = 0 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
164 n = len(rawdata) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
165 while i < n: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
166 match = self.interesting.search(rawdata, i) # < or & |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
167 if match: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
168 j = match.start() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
169 else: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
170 j = n |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
171 if i < j: self.handle_data(rawdata[i:j]) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
172 i = self.updatepos(i, j) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
173 if i == n: break |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
174 if rawdata[i] == '<': |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
175 if starttagopen.match(rawdata, i): # < + letter |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
176 k = self.parse_starttag(i) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
177 elif endtagopen.match(rawdata, i): # </ |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
178 k = self.parse_endtag(i) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
179 if k >= 0: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
180 self.clear_cdata_mode() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
181 elif commentopen.match(rawdata, i): # <!-- |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
182 k = self.parse_comment(i) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
183 elif piopen.match(rawdata, i): # <? |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
184 k = self.parse_pi(i) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
185 elif declopen.match(rawdata, i): # <! |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
186 k = self.parse_declaration(i) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
187 else: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
188 if i < n-1: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
189 raise HTMLParseError( |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
190 "invalid '<' construct: %s" % `rawdata[i:i+2]`, |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
191 self.getpos()) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
192 k = -1 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
193 if k < 0: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
194 if end: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
195 raise HTMLParseError("EOF in middle of construct", |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
196 self.getpos()) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
197 break |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
198 i = self.updatepos(i, k) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
199 elif rawdata[i] == '&': |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
200 match = charref.match(rawdata, i) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
201 if match: |
|
18984
00935031b373
Change some comments into docstrings.
Fred Drake <fdrake@acm.org>
parents:
17934
diff
changeset
|
202 name = match.group()[2:-1] |
|
17853
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
203 self.handle_charref(name) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
204 k = match.end() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
205 if rawdata[k-1] != ';': |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
206 k = k-1 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
207 i = self.updatepos(i, k) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
208 continue |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
209 match = entityref.match(rawdata, i) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
210 if match: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
211 name = match.group(1) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
212 self.handle_entityref(name) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
213 k = match.end() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
214 if rawdata[k-1] != ';': |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
215 k = k-1 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
216 i = self.updatepos(i, k) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
217 continue |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
218 if incomplete.match(rawdata, i): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
219 if end: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
220 raise HTMLParseError( |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
221 "EOF in middle of entity or char ref", |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
222 self.getpos()) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
223 return -1 # incomplete |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
224 raise HTMLParseError("'&' not part of entity or char ref", |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
225 self.getpos()) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
226 else: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
227 assert 0, "interesting.search() lied" |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
228 # end while |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
229 if end and i < n: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
230 self.handle_data(rawdata[i:n]) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
231 i = self.updatepos(i, n) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
232 self.rawdata = rawdata[i:] |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
233 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
234 # Internal -- parse comment, return end or -1 if not terminated |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
235 def parse_comment(self, i): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
236 rawdata = self.rawdata |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
237 assert rawdata[i:i+4] == '<!--', 'unexpected call to parse_comment()' |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
238 match = commentclose.search(rawdata, i+4) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
239 if not match: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
240 return -1 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
241 j = match.start() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
242 self.handle_comment(rawdata[i+4: j]) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
243 j = match.end() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
244 return j |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
245 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
246 # Internal -- parse declaration. |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
247 def parse_declaration(self, i): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
248 # This is some sort of declaration; in "HTML as |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
249 # deployed," this should only be the document type |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
250 # declaration ("<!DOCTYPE html...>"). |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
251 rawdata = self.rawdata |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
252 j = i + 2 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
253 assert rawdata[i:j] == "<!", "unexpected call to parse_declaration" |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
254 if rawdata[j:j+1] in ("-", ""): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
255 # Start of comment followed by buffer boundary, |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
256 # or just a buffer boundary. |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
257 return -1 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
258 # in practice, this should look like: ((name|stringlit) S*)+ '>' |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
259 n = len(rawdata) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
260 while j < n: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
261 c = rawdata[j] |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
262 if c == ">": |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
263 # end of declaration syntax |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
264 self.handle_decl(rawdata[i+2:j]) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
265 return j + 1 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
266 if c in "\"'": |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
267 m = declstringlit.match(rawdata, j) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
268 if not m: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
269 return -1 # incomplete |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
270 j = m.end() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
271 elif c in "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ": |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
272 m = declname.match(rawdata, j) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
273 if not m: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
274 return -1 # incomplete |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
275 j = m.end() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
276 else: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
277 raise HTMLParseError( |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
278 "unexpected char in declaration: %s" % `rawdata[j]`, |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
279 self.getpos()) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
280 return -1 # incomplete |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
281 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
282 # Internal -- parse processing instr, return end or -1 if not terminated |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
283 def parse_pi(self, i): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
284 rawdata = self.rawdata |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
285 assert rawdata[i:i+2] == '<?', 'unexpected call to parse_pi()' |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
286 match = piclose.search(rawdata, i+2) # > |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
287 if not match: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
288 return -1 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
289 j = match.start() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
290 self.handle_pi(rawdata[i+2: j]) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
291 j = match.end() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
292 return j |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
293 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
294 # Internal -- handle starttag, return end or -1 if not terminated |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
295 def parse_starttag(self, i): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
296 self.__starttag_text = None |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
297 endpos = self.check_for_whole_start_tag(i) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
298 if endpos < 0: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
299 return endpos |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
300 rawdata = self.rawdata |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
301 self.__starttag_text = rawdata[i:endpos] |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
302 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
303 # Now parse the data between i+1 and j into a tag and attrs |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
304 attrs = [] |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
305 match = tagfind.match(rawdata, i+1) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
306 assert match, 'unexpected call to parse_starttag()' |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
307 k = match.end() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
308 self.lasttag = tag = string.lower(rawdata[i+1:k]) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
309 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
310 while k < endpos: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
311 m = attrfind.match(rawdata, k) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
312 if not m: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
313 break |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
314 attrname, rest, attrvalue = m.group(1, 2, 3) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
315 if not rest: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
316 attrvalue = None |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
317 elif attrvalue[:1] == '\'' == attrvalue[-1:] or \ |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
318 attrvalue[:1] == '"' == attrvalue[-1:]: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
319 attrvalue = attrvalue[1:-1] |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
320 attrvalue = self.unescape(attrvalue) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
321 attrs.append((string.lower(attrname), attrvalue)) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
322 k = m.end() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
323 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
324 end = string.strip(rawdata[k:endpos]) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
325 if end not in (">", "/>"): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
326 lineno, offset = self.getpos() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
327 if "\n" in self.__starttag_text: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
328 lineno = lineno + string.count(self.__starttag_text, "\n") |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
329 offset = len(self.__starttag_text) \ |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
330 - string.rfind(self.__starttag_text, "\n") |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
331 else: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
332 offset = offset + len(self.__starttag_text) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
333 raise HTMLParseError("junk characters in start tag: %s" |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
334 % `rawdata[k:endpos][:20]`, |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
335 (lineno, offset)) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
336 if end[-2:] == '/>': |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
337 # XHTML-style empty tag: <span attr="value" /> |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
338 self.handle_startendtag(tag, attrs) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
339 else: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
340 self.handle_starttag(tag, attrs) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
341 if tag in self.CDATA_CONTENT_ELEMENTS: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
342 self.set_cdata_mode() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
343 return endpos |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
344 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
345 # Internal -- check to see if we have a complete starttag; return end |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
346 # or -1 if incomplete. |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
347 def check_for_whole_start_tag(self, i): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
348 rawdata = self.rawdata |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
349 m = locatestarttagend.match(rawdata, i) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
350 if m: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
351 j = m.end() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
352 next = rawdata[j:j+1] |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
353 if next == ">": |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
354 return j + 1 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
355 if next == "/": |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
356 s = rawdata[j:j+2] |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
357 if s == "/>": |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
358 return j + 2 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
359 if s == "/": |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
360 # buffer boundary |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
361 return -1 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
362 # else bogus input |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
363 self.updatepos(i, j + 1) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
364 raise HTMLParseError("malformed empty start tag", |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
365 self.getpos()) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
366 if next == "": |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
367 # end of input |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
368 return -1 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
369 if next in ("abcdefghijklmnopqrstuvwxyz=/" |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
370 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
371 # end of input in or before attribute value, or we have the |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
372 # '/' from a '/>' ending |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
373 return -1 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
374 self.updatepos(i, j) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
375 raise HTMLParseError("malformed start tag", self.getpos()) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
376 raise AssertionError("we should not gt here!") |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
377 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
378 # Internal -- parse endtag, return end or -1 if incomplete |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
379 def parse_endtag(self, i): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
380 rawdata = self.rawdata |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
381 assert rawdata[i:i+2] == "</", "unexpected call to parse_endtag" |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
382 match = endendtag.search(rawdata, i+1) # > |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
383 if not match: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
384 return -1 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
385 j = match.end() |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
386 match = endtagfind.match(rawdata, i) # </ + tag + > |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
387 if not match: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
388 raise HTMLParseError("bad end tag: %s" % `rawdata[i:j]`, |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
389 self.getpos()) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
390 tag = match.group(1) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
391 self.handle_endtag(string.lower(tag)) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
392 return j |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
393 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
394 # Overridable -- finish processing of start+end tag: <tag.../> |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
395 def handle_startendtag(self, tag, attrs): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
396 self.handle_starttag(tag, attrs) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
397 self.handle_endtag(tag) |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
398 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
399 # Overridable -- handle start tag |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
400 def handle_starttag(self, tag, attrs): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
401 pass |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
402 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
403 # Overridable -- handle end tag |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
404 def handle_endtag(self, tag): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
405 pass |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
406 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
407 # Overridable -- handle character reference |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
408 def handle_charref(self, name): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
409 pass |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
410 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
411 # Overridable -- handle entity reference |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
412 def handle_entityref(self, name): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
413 pass |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
414 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
415 # Overridable -- handle data |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
416 def handle_data(self, data): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
417 pass |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
418 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
419 # Overridable -- handle comment |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
420 def handle_comment(self, data): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
421 pass |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
422 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
423 # Overridable -- handle declaration |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
424 def handle_decl(self, decl): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
425 pass |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
426 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
427 # Overridable -- handle processing instruction |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
428 def handle_pi(self, data): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
429 pass |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
430 |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
431 # Internal -- helper to remove special character quoting |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
432 def unescape(self, s): |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
433 if '&' not in s: |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
434 return s |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
435 s = string.replace(s, "<", "<") |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
436 s = string.replace(s, ">", ">") |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
437 s = string.replace(s, "'", "'") |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
438 s = string.replace(s, """, '"') |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
439 s = string.replace(s, "&", "&") # Must be last |
|
055fc046e080
A much improved HTML parser -- a replacement for sgmllib. The API is
Guido van Rossum <guido@python.org>
parents:
diff
changeset
|
440 return s |
