Mercurial > p > roundup > code
annotate TAL/HTMLParser.py @ 1029:c3e391d9c4e9
more FieldStorage fun
| author | Richard Jones <richard@users.sourceforge.net> |
|---|---|
| date | Tue, 03 Sep 2002 07:42:38 +0000 |
| parents | bfd348432420 |
| children |
| rev | line source |
|---|---|
|
982
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
1 """A parser for HTML and XHTML.""" |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
2 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
3 # This file is based on sgmllib.py, but the API is slightly different. |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
4 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
5 # XXX There should be a way to distinguish between PCDATA (parsed |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
6 # character data -- the normal case), RCDATA (replaceable character |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
7 # data -- only char and entity references and end tags are special) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
8 # and CDATA (character data -- only end tags are special). |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
9 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
10 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
11 import markupbase |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
12 import re |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
13 import string |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
14 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
15 # Regular expressions used for parsing |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
16 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
17 interesting_normal = re.compile('[&<]') |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
18 interesting_cdata = re.compile(r'<(/|\Z)') |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
19 incomplete = re.compile('&[a-zA-Z#]') |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
20 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
21 entityref = re.compile('&([a-zA-Z][-.a-zA-Z0-9]*)[^a-zA-Z0-9]') |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
22 charref = re.compile('&#(?:[0-9]+|[xX][0-9a-fA-F]+)[^0-9a-fA-F]') |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
23 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
24 starttagopen = re.compile('<[a-zA-Z]') |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
25 piclose = re.compile('>') |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
26 endtagopen = re.compile('</') |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
27 commentclose = re.compile(r'--\s*>') |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
28 tagfind = re.compile('[a-zA-Z][-.a-zA-Z0-9:_]*') |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
29 attrfind = re.compile( |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
30 r'\s*([a-zA-Z_][-.:a-zA-Z_0-9]*)(\s*=\s*' |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
31 r'(\'[^\']*\'|"[^"]*"|[-a-zA-Z0-9./:;+*%?!&$\(\)_#=~]*))?') |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
32 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
33 locatestarttagend = re.compile(r""" |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
34 <[a-zA-Z][-.a-zA-Z0-9:_]* # tag name |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
35 (?:\s+ # whitespace before attribute name |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
36 (?:[a-zA-Z_][-.:a-zA-Z0-9_]* # attribute name |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
37 (?:\s*=\s* # value indicator |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
38 (?:'[^']*' # LITA-enclosed value |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
39 |\"[^\"]*\" # LIT-enclosed value |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
40 |[^'\">\s]+ # bare value |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
41 ) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
42 )? |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
43 ) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
44 )* |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
45 \s* # trailing whitespace |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
46 """, re.VERBOSE) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
47 endendtag = re.compile('>') |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
48 endtagfind = re.compile('</\s*([a-zA-Z][-.a-zA-Z0-9:_]*)\s*>') |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
49 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
50 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
51 class HTMLParseError(Exception): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
52 """Exception raised for all parse errors.""" |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
53 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
54 def __init__(self, msg, position=(None, None)): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
55 assert msg |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
56 self.msg = msg |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
57 self.lineno = position[0] |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
58 self.offset = position[1] |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
59 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
60 def __str__(self): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
61 result = self.msg |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
62 if self.lineno is not None: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
63 result = result + ", at line %d" % self.lineno |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
64 if self.offset is not None: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
65 result = result + ", column %d" % (self.offset + 1) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
66 return result |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
67 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
68 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
69 def _contains_at(s, sub, pos): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
70 return s[pos:pos+len(sub)] == sub |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
71 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
72 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
73 class HTMLParser(markupbase.ParserBase): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
74 """Find tags and other markup and call handler functions. |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
75 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
76 Usage: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
77 p = HTMLParser() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
78 p.feed(data) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
79 ... |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
80 p.close() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
81 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
82 Start tags are handled by calling self.handle_starttag() or |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
83 self.handle_startendtag(); end tags by self.handle_endtag(). The |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
84 data between tags is passed from the parser to the derived class |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
85 by calling self.handle_data() with the data as argument (the data |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
86 may be split up in arbitrary chunks). Entity references are |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
87 passed by calling self.handle_entityref() with the entity |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
88 reference as the argument. Numeric character references are |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
89 passed to self.handle_charref() with the string containing the |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
90 reference as the argument. |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
91 """ |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
92 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
93 CDATA_CONTENT_ELEMENTS = ("script", "style") |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
94 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
95 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
96 def __init__(self): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
97 """Initialize and reset this instance.""" |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
98 self.reset() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
99 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
100 def reset(self): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
101 """Reset this instance. Loses all unprocessed data.""" |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
102 self.rawdata = '' |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
103 self.stack = [] |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
104 self.lasttag = '???' |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
105 self.interesting = interesting_normal |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
106 markupbase.ParserBase.reset(self) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
107 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
108 def feed(self, data): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
109 """Feed data to the parser. |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
110 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
111 Call this as often as you want, with as little or as much text |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
112 as you want (may include '\n'). |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
113 """ |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
114 self.rawdata = self.rawdata + data |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
115 self.goahead(0) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
116 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
117 def close(self): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
118 """Handle any buffered data.""" |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
119 self.goahead(1) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
120 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
121 def error(self, message): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
122 raise HTMLParseError(message, self.getpos()) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
123 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
124 __starttag_text = None |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
125 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
126 def get_starttag_text(self): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
127 """Return full source of start tag: '<...>'.""" |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
128 return self.__starttag_text |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
129 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
130 cdata_endtag = None |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
131 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
132 def set_cdata_mode(self, endtag=None): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
133 self.cdata_endtag = endtag |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
134 self.interesting = interesting_cdata |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
135 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
136 def clear_cdata_mode(self): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
137 self.cdata_endtag = None |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
138 self.interesting = interesting_normal |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
139 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
140 # Internal -- handle data as far as reasonable. May leave state |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
141 # and data to be processed by a subsequent call. If 'end' is |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
142 # true, force handling all data as if followed by EOF marker. |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
143 def goahead(self, end): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
144 rawdata = self.rawdata |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
145 i = 0 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
146 n = len(rawdata) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
147 while i < n: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
148 match = self.interesting.search(rawdata, i) # < or & |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
149 if match: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
150 j = match.start() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
151 else: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
152 j = n |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
153 if i < j: self.handle_data(rawdata[i:j]) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
154 i = self.updatepos(i, j) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
155 if i == n: break |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
156 if rawdata[i] == '<': |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
157 if starttagopen.match(rawdata, i): # < + letter |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
158 k = self.parse_starttag(i) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
159 elif endtagopen.match(rawdata, i): # </ |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
160 k = self.parse_endtag(i) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
161 elif _contains_at(rawdata, "<!--", i): # <!-- |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
162 k = self.parse_comment(i) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
163 elif _contains_at(rawdata, "<!", i): # <! |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
164 k = self.parse_declaration(i) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
165 elif _contains_at(rawdata, "<?", i): # <? |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
166 k = self.parse_pi(i) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
167 elif _contains_at(rawdata, "<?", i): # <! |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
168 k = self.parse_declaration(i) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
169 elif (i + 1) < n: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
170 self.handle_data("<") |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
171 k = i + 1 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
172 else: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
173 break |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
174 if k < 0: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
175 if end: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
176 self.error("EOF in middle of construct") |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
177 break |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
178 i = self.updatepos(i, k) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
179 elif rawdata[i:i+2] == "&#": |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
180 match = charref.match(rawdata, i) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
181 if match: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
182 name = match.group()[2:-1] |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
183 self.handle_charref(name) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
184 k = match.end() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
185 if rawdata[k-1] != ';': |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
186 k = k - 1 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
187 i = self.updatepos(i, k) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
188 continue |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
189 else: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
190 break |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
191 elif rawdata[i] == '&': |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
192 match = entityref.match(rawdata, i) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
193 if match: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
194 name = match.group(1) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
195 self.handle_entityref(name) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
196 k = match.end() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
197 if rawdata[k-1] != ';': |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
198 k = k - 1 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
199 i = self.updatepos(i, k) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
200 continue |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
201 match = incomplete.match(rawdata, i) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
202 if match: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
203 # match.group() will contain at least 2 chars |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
204 rest = rawdata[i:] |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
205 if end and match.group() == rest: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
206 self.error("EOF in middle of entity or char ref") |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
207 # incomplete |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
208 break |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
209 elif (i + 1) < n: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
210 # not the end of the buffer, and can't be confused |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
211 # with some other construct |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
212 self.handle_data("&") |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
213 i = self.updatepos(i, i + 1) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
214 else: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
215 break |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
216 else: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
217 assert 0, "interesting.search() lied" |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
218 # end while |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
219 if end and i < n: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
220 self.handle_data(rawdata[i:n]) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
221 i = self.updatepos(i, n) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
222 self.rawdata = rawdata[i:] |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
223 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
224 # Internal -- parse comment, return end or -1 if not terminated |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
225 def parse_comment(self, i, report=1): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
226 rawdata = self.rawdata |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
227 assert rawdata[i:i+4] == '<!--', 'unexpected call to parse_comment()' |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
228 match = commentclose.search(rawdata, i+4) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
229 if not match: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
230 return -1 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
231 if report: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
232 j = match.start() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
233 self.handle_comment(rawdata[i+4: j]) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
234 j = match.end() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
235 return j |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
236 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
237 # Internal -- parse processing instr, return end or -1 if not terminated |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
238 def parse_pi(self, i): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
239 rawdata = self.rawdata |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
240 assert rawdata[i:i+2] == '<?', 'unexpected call to parse_pi()' |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
241 match = piclose.search(rawdata, i+2) # > |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
242 if not match: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
243 return -1 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
244 j = match.start() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
245 self.handle_pi(rawdata[i+2: j]) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
246 j = match.end() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
247 return j |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
248 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
249 # Internal -- handle starttag, return end or -1 if not terminated |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
250 def parse_starttag(self, i): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
251 self.__starttag_text = None |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
252 endpos = self.check_for_whole_start_tag(i) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
253 if endpos < 0: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
254 return endpos |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
255 rawdata = self.rawdata |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
256 self.__starttag_text = rawdata[i:endpos] |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
257 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
258 # Now parse the data between i+1 and j into a tag and attrs |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
259 attrs = [] |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
260 match = tagfind.match(rawdata, i+1) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
261 assert match, 'unexpected call to parse_starttag()' |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
262 k = match.end() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
263 self.lasttag = tag = string.lower(rawdata[i+1:k]) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
264 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
265 while k < endpos: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
266 m = attrfind.match(rawdata, k) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
267 if not m: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
268 break |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
269 attrname, rest, attrvalue = m.group(1, 2, 3) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
270 if not rest: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
271 attrvalue = None |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
272 elif attrvalue[:1] == '\'' == attrvalue[-1:] or \ |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
273 attrvalue[:1] == '"' == attrvalue[-1:]: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
274 attrvalue = attrvalue[1:-1] |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
275 attrvalue = self.unescape(attrvalue) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
276 attrs.append((string.lower(attrname), attrvalue)) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
277 k = m.end() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
278 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
279 end = string.strip(rawdata[k:endpos]) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
280 if end not in (">", "/>"): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
281 lineno, offset = self.getpos() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
282 if "\n" in self.__starttag_text: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
283 lineno = lineno + string.count(self.__starttag_text, "\n") |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
284 offset = len(self.__starttag_text) \ |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
285 - string.rfind(self.__starttag_text, "\n") |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
286 else: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
287 offset = offset + len(self.__starttag_text) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
288 self.error("junk characters in start tag: %s" |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
289 % `rawdata[k:endpos][:20]`) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
290 if end[-2:] == '/>': |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
291 # XHTML-style empty tag: <span attr="value" /> |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
292 self.handle_startendtag(tag, attrs) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
293 else: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
294 self.handle_starttag(tag, attrs) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
295 if tag in self.CDATA_CONTENT_ELEMENTS: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
296 self.set_cdata_mode(tag) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
297 return endpos |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
298 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
299 # Internal -- check to see if we have a complete starttag; return end |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
300 # or -1 if incomplete. |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
301 def check_for_whole_start_tag(self, i): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
302 rawdata = self.rawdata |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
303 m = locatestarttagend.match(rawdata, i) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
304 if m: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
305 j = m.end() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
306 next = rawdata[j:j+1] |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
307 if next == ">": |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
308 return j + 1 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
309 if next == "/": |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
310 s = rawdata[j:j+2] |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
311 if s == "/>": |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
312 return j + 2 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
313 if s == "/": |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
314 # buffer boundary |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
315 return -1 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
316 # else bogus input |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
317 self.updatepos(i, j + 1) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
318 self.error("malformed empty start tag") |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
319 if next == "": |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
320 # end of input |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
321 return -1 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
322 if next in ("abcdefghijklmnopqrstuvwxyz=/" |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
323 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
324 # end of input in or before attribute value, or we have the |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
325 # '/' from a '/>' ending |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
326 return -1 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
327 self.updatepos(i, j) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
328 self.error("malformed start tag") |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
329 raise AssertionError("we should not get here!") |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
330 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
331 # Internal -- parse endtag, return end or -1 if incomplete |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
332 def parse_endtag(self, i): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
333 rawdata = self.rawdata |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
334 assert rawdata[i:i+2] == "</", "unexpected call to parse_endtag" |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
335 match = endendtag.search(rawdata, i+1) # > |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
336 if not match: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
337 return -1 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
338 j = match.end() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
339 match = endtagfind.match(rawdata, i) # </ + tag + > |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
340 if not match: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
341 self.error("bad end tag: %s" % `rawdata[i:j]`) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
342 tag = string.lower(match.group(1)) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
343 if ( self.cdata_endtag is not None |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
344 and tag != self.cdata_endtag): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
345 # Should be a mismatched end tag, but we'll treat it |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
346 # as text anyway, since most HTML authors aren't |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
347 # interested in the finer points of syntax. |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
348 self.handle_data(match.group(0)) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
349 else: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
350 self.handle_endtag(tag) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
351 self.clear_cdata_mode() |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
352 return j |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
353 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
354 # Overridable -- finish processing of start+end tag: <tag.../> |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
355 def handle_startendtag(self, tag, attrs): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
356 self.handle_starttag(tag, attrs) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
357 self.handle_endtag(tag) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
358 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
359 # Overridable -- handle start tag |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
360 def handle_starttag(self, tag, attrs): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
361 pass |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
362 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
363 # Overridable -- handle end tag |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
364 def handle_endtag(self, tag): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
365 pass |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
366 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
367 # Overridable -- handle character reference |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
368 def handle_charref(self, name): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
369 pass |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
370 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
371 # Overridable -- handle entity reference |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
372 def handle_entityref(self, name): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
373 pass |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
374 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
375 # Overridable -- handle data |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
376 def handle_data(self, data): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
377 pass |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
378 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
379 # Overridable -- handle comment |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
380 def handle_comment(self, data): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
381 pass |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
382 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
383 # Overridable -- handle declaration |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
384 def handle_decl(self, decl): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
385 pass |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
386 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
387 # Overridable -- handle processing instruction |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
388 def handle_pi(self, data): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
389 pass |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
390 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
391 def unknown_decl(self, data): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
392 self.error("unknown declaration: " + `data`) |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
393 |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
394 # Internal -- helper to remove special character quoting |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
395 def unescape(self, s): |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
396 if '&' not in s: |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
397 return s |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
398 s = string.replace(s, "<", "<") |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
399 s = string.replace(s, ">", ">") |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
400 s = string.replace(s, "'", "'") |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
401 s = string.replace(s, """, '"') |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
402 s = string.replace(s, "&", "&") # Must be last |
|
bfd348432420
Adding TAL to the dist
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
403 return s |
