annotate roundup/cgi/TAL/HTMLParser.py @ 5265:63868084b8bb

Python 2 and 3 support. Convert Exception to BaseException. TAL and PageTemplate code.
author John Rouillard <rouilj@ieee.org>
date Mon, 04 Sep 2017 21:58:16 -0400
parents 8c2402a78bb0
children 12fe83f90f0d
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
1 """A parser for HTML and XHTML."""
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
2
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
3 # This file is based on sgmllib.py, but the API is slightly different.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
4
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
5 # XXX There should be a way to distinguish between PCDATA (parsed
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
6 # character data -- the normal case), RCDATA (replaceable character
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
7 # data -- only char and entity references and end tags are special)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
8 # and CDATA (character data -- only end tags are special).
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
9
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
10
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
11 import markupbase
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
12 import re
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
13
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
14 # Regular expressions used for parsing
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
15
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
16 interesting_normal = re.compile('[&<]')
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
17 interesting_cdata = re.compile(r'<(/|\Z)')
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
18 incomplete = re.compile('&[a-zA-Z#]')
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
19
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
20 entityref = re.compile('&([a-zA-Z][-.a-zA-Z0-9]*)[^a-zA-Z0-9]')
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
21 charref = re.compile('&#(?:[0-9]+|[xX][0-9a-fA-F]+)[^0-9a-fA-F]')
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
22
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
23 starttagopen = re.compile('<[a-zA-Z]')
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
24 piclose = re.compile('>')
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
25 endtagopen = re.compile('</')
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
26 commentclose = re.compile(r'--\s*>')
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
27 tagfind = re.compile('[a-zA-Z][-.a-zA-Z0-9:_]*')
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
28 attrfind = re.compile(
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
29 r'\s*([a-zA-Z_][-.:a-zA-Z_0-9]*)(\s*=\s*'
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
30 r'(\'[^\']*\'|"[^"]*"|[-a-zA-Z0-9./:;+*%?!&$\(\)_#=~]*))?')
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
31
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
32 locatestarttagend = re.compile(r"""
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
33 <[a-zA-Z][-.a-zA-Z0-9:_]* # tag name
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
34 (?:\s+ # whitespace before attribute name
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
35 (?:[a-zA-Z_][-.:a-zA-Z0-9_]* # attribute name
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
36 (?:\s*=\s* # value indicator
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
37 (?:'[^']*' # LITA-enclosed value
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
38 |\"[^\"]*\" # LIT-enclosed value
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
39 |[^'\">\s]+ # bare value
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
40 )
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
41 )?
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
42 )
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
43 )*
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
44 \s* # trailing whitespace
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
45 """, re.VERBOSE)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
46 endendtag = re.compile('>')
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
47 endtagfind = re.compile('</\s*([a-zA-Z][-.a-zA-Z0-9:_]*)\s*>')
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
48
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
49
5265
63868084b8bb Python 2 and 3 support. Convert Exception to BaseException. TAL and
John Rouillard <rouilj@ieee.org>
parents: 2348
diff changeset
50 class HTMLParseError(BaseException):
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
51 """Exception raised for all parse errors."""
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
52
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
53 def __init__(self, msg, position=(None, None)):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
54 assert msg
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
55 self.msg = msg
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
56 self.lineno = position[0]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
57 self.offset = position[1]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
58
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
59 def __str__(self):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
60 result = self.msg
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
61 if self.lineno is not None:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
62 result = result + ", at line %d" % self.lineno
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
63 if self.offset is not None:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
64 result = result + ", column %d" % (self.offset + 1)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
65 return result
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
66
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
67
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
68 def _contains_at(s, sub, pos):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
69 return s[pos:pos+len(sub)] == sub
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
70
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
71
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
72 class HTMLParser(markupbase.ParserBase):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
73 """Find tags and other markup and call handler functions.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
74
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
75 Usage:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
76 p = HTMLParser()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
77 p.feed(data)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
78 ...
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
79 p.close()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
80
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
81 Start tags are handled by calling self.handle_starttag() or
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
82 self.handle_startendtag(); end tags by self.handle_endtag(). The
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
83 data between tags is passed from the parser to the derived class
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
84 by calling self.handle_data() with the data as argument (the data
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
85 may be split up in arbitrary chunks). Entity references are
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
86 passed by calling self.handle_entityref() with the entity
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
87 reference as the argument. Numeric character references are
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
88 passed to self.handle_charref() with the string containing the
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
89 reference as the argument.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
90 """
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
91
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
92 CDATA_CONTENT_ELEMENTS = ("script", "style")
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
93
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
94
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
95 def __init__(self):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
96 """Initialize and reset this instance."""
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
97 self.reset()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
98
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
99 def reset(self):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
100 """Reset this instance. Loses all unprocessed data."""
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
101 self.rawdata = ''
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
102 self.stack = []
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
103 self.lasttag = '???'
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
104 self.interesting = interesting_normal
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
105 markupbase.ParserBase.reset(self)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
106
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
107 def feed(self, data):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
108 """Feed data to the parser.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
109
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
110 Call this as often as you want, with as little or as much text
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
111 as you want (may include '\n').
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
112 """
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
113 self.rawdata = self.rawdata + data
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
114 self.goahead(0)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
115
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
116 def close(self):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
117 """Handle any buffered data."""
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
118 self.goahead(1)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
119
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
120 def error(self, message):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
121 raise HTMLParseError(message, self.getpos())
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
122
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
123 __starttag_text = None
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
124
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
125 def get_starttag_text(self):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
126 """Return full source of start tag: '<...>'."""
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
127 return self.__starttag_text
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
128
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
129 cdata_endtag = None
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
130
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
131 def set_cdata_mode(self, endtag=None):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
132 self.cdata_endtag = endtag
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
133 self.interesting = interesting_cdata
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
134
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
135 def clear_cdata_mode(self):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
136 self.cdata_endtag = None
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
137 self.interesting = interesting_normal
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
138
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
139 # Internal -- handle data as far as reasonable. May leave state
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
140 # and data to be processed by a subsequent call. If 'end' is
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
141 # true, force handling all data as if followed by EOF marker.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
142 def goahead(self, end):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
143 rawdata = self.rawdata
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
144 i = 0
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
145 n = len(rawdata)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
146 while i < n:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
147 match = self.interesting.search(rawdata, i) # < or &
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
148 if match:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
149 j = match.start()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
150 else:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
151 j = n
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
152 if i < j: self.handle_data(rawdata[i:j])
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
153 i = self.updatepos(i, j)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
154 if i == n: break
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
155 if rawdata[i] == '<':
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
156 if starttagopen.match(rawdata, i): # < + letter
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
157 k = self.parse_starttag(i)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
158 elif endtagopen.match(rawdata, i): # </
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
159 k = self.parse_endtag(i)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
160 elif _contains_at(rawdata, "<!--", i): # <!--
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
161 k = self.parse_comment(i)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
162 elif _contains_at(rawdata, "<!", i): # <!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
163 k = self.parse_declaration(i)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
164 elif _contains_at(rawdata, "<?", i): # <?
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
165 k = self.parse_pi(i)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
166 elif _contains_at(rawdata, "<?", i): # <!
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
167 k = self.parse_declaration(i)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
168 elif (i + 1) < n:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
169 self.handle_data("<")
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
170 k = i + 1
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
171 else:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
172 break
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
173 if k < 0:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
174 if end:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
175 self.error("EOF in middle of construct")
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
176 break
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
177 i = self.updatepos(i, k)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
178 elif rawdata[i:i+2] == "&#":
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
179 match = charref.match(rawdata, i)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
180 if match:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
181 name = match.group()[2:-1]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
182 self.handle_charref(name)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
183 k = match.end()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
184 if rawdata[k-1] != ';':
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
185 k = k - 1
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
186 i = self.updatepos(i, k)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
187 continue
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
188 else:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
189 break
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
190 elif rawdata[i] == '&':
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
191 match = entityref.match(rawdata, i)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
192 if match:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
193 name = match.group(1)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
194 self.handle_entityref(name)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
195 k = match.end()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
196 if rawdata[k-1] != ';':
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
197 k = k - 1
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
198 i = self.updatepos(i, k)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
199 continue
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
200 match = incomplete.match(rawdata, i)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
201 if match:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
202 # match.group() will contain at least 2 chars
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
203 rest = rawdata[i:]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
204 if end and match.group() == rest:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
205 self.error("EOF in middle of entity or char ref")
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
206 # incomplete
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
207 break
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
208 elif (i + 1) < n:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
209 # not the end of the buffer, and can't be confused
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
210 # with some other construct
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
211 self.handle_data("&")
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
212 i = self.updatepos(i, i + 1)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
213 else:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
214 break
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
215 else:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
216 assert 0, "interesting.search() lied"
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
217 # end while
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
218 if end and i < n:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
219 self.handle_data(rawdata[i:n])
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
220 i = self.updatepos(i, n)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
221 self.rawdata = rawdata[i:]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
222
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
223 # Internal -- parse comment, return end or -1 if not terminated
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
224 def parse_comment(self, i, report=1):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
225 rawdata = self.rawdata
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
226 assert rawdata[i:i+4] == '<!--', 'unexpected call to parse_comment()'
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
227 match = commentclose.search(rawdata, i+4)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
228 if not match:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
229 return -1
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
230 if report:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
231 j = match.start()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
232 self.handle_comment(rawdata[i+4: j])
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
233 j = match.end()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
234 return j
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
235
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
236 # Internal -- parse processing instr, return end or -1 if not terminated
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
237 def parse_pi(self, i):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
238 rawdata = self.rawdata
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
239 assert rawdata[i:i+2] == '<?', 'unexpected call to parse_pi()'
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
240 match = piclose.search(rawdata, i+2) # >
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
241 if not match:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
242 return -1
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
243 j = match.start()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
244 self.handle_pi(rawdata[i+2: j])
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
245 j = match.end()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
246 return j
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
247
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
248 # Internal -- handle starttag, return end or -1 if not terminated
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
249 def parse_starttag(self, i):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
250 self.__starttag_text = None
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
251 endpos = self.check_for_whole_start_tag(i)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
252 if endpos < 0:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
253 return endpos
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
254 rawdata = self.rawdata
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
255 self.__starttag_text = rawdata[i:endpos]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
256
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
257 # Now parse the data between i+1 and j into a tag and attrs
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
258 attrs = []
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
259 match = tagfind.match(rawdata, i+1)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
260 assert match, 'unexpected call to parse_starttag()'
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
261 k = match.end()
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
262 self.lasttag = tag = rawdata[i+1:k].lower()
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
263
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
264 while k < endpos:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
265 m = attrfind.match(rawdata, k)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
266 if not m:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
267 break
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
268 attrname, rest, attrvalue = m.group(1, 2, 3)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
269 if not rest:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
270 attrvalue = None
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
271 elif attrvalue[:1] == '\'' == attrvalue[-1:] or \
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
272 attrvalue[:1] == '"' == attrvalue[-1:]:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
273 attrvalue = attrvalue[1:-1]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
274 attrvalue = self.unescape(attrvalue)
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
275 attrs.append((attrname.lower(), attrvalue))
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
276 k = m.end()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
277
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
278 end = rawdata[k:endpos].strip()
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
279 if end not in (">", "/>"):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
280 lineno, offset = self.getpos()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
281 if "\n" in self.__starttag_text:
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
282 lineno = lineno + self.__starttag_text.count("\n")
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
283 offset = len(self.__starttag_text) \
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
284 - self.__starttag_text.rfind("\n")
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
285 else:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
286 offset = offset + len(self.__starttag_text)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
287 self.error("junk characters in start tag: %s"
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
288 % `rawdata[k:endpos][:20]`)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
289 if end[-2:] == '/>':
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
290 # XHTML-style empty tag: <span attr="value" />
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
291 self.handle_startendtag(tag, attrs)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
292 else:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
293 self.handle_starttag(tag, attrs)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
294 if tag in self.CDATA_CONTENT_ELEMENTS:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
295 self.set_cdata_mode(tag)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
296 return endpos
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
297
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
298 # Internal -- check to see if we have a complete starttag; return end
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
299 # or -1 if incomplete.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
300 def check_for_whole_start_tag(self, i):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
301 rawdata = self.rawdata
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
302 m = locatestarttagend.match(rawdata, i)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
303 if m:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
304 j = m.end()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
305 next = rawdata[j:j+1]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
306 if next == ">":
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
307 return j + 1
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
308 if next == "/":
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
309 s = rawdata[j:j+2]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
310 if s == "/>":
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
311 return j + 2
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
312 if s == "/":
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
313 # buffer boundary
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
314 return -1
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
315 # else bogus input
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
316 self.updatepos(i, j + 1)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
317 self.error("malformed empty start tag")
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
318 if next == "":
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
319 # end of input
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
320 return -1
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
321 if next in ("abcdefghijklmnopqrstuvwxyz=/"
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
322 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
323 # end of input in or before attribute value, or we have the
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
324 # '/' from a '/>' ending
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
325 return -1
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
326 self.updatepos(i, j)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
327 self.error("malformed start tag")
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
328 raise AssertionError("we should not get here!")
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
329
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
330 # Internal -- parse endtag, return end or -1 if incomplete
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
331 def parse_endtag(self, i):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
332 rawdata = self.rawdata
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
333 assert rawdata[i:i+2] == "</", "unexpected call to parse_endtag"
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
334 match = endendtag.search(rawdata, i+1) # >
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
335 if not match:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
336 return -1
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
337 j = match.end()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
338 match = endtagfind.match(rawdata, i) # </ + tag + >
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
339 if not match:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
340 self.error("bad end tag: %s" % `rawdata[i:j]`)
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
341 tag = match.group(1).lower()
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
342 if ( self.cdata_endtag is not None
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
343 and tag != self.cdata_endtag):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
344 # Should be a mismatched end tag, but we'll treat it
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
345 # as text anyway, since most HTML authors aren't
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
346 # interested in the finer points of syntax.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
347 self.handle_data(match.group(0))
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
348 else:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
349 self.handle_endtag(tag)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
350 self.clear_cdata_mode()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
351 return j
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
352
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
353 # Overridable -- finish processing of start+end tag: <tag.../>
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
354 def handle_startendtag(self, tag, attrs):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
355 self.handle_starttag(tag, attrs)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
356 self.handle_endtag(tag)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
357
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
358 # Overridable -- handle start tag
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
359 def handle_starttag(self, tag, attrs):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
360 pass
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
361
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
362 # Overridable -- handle end tag
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
363 def handle_endtag(self, tag):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
364 pass
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
365
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
366 # Overridable -- handle character reference
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
367 def handle_charref(self, name):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
368 pass
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
369
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
370 # Overridable -- handle entity reference
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
371 def handle_entityref(self, name):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
372 pass
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
373
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
374 # Overridable -- handle data
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
375 def handle_data(self, data):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
376 pass
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
377
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
378 # Overridable -- handle comment
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
379 def handle_comment(self, data):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
380 pass
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
381
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
382 # Overridable -- handle declaration
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
383 def handle_decl(self, decl):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
384 pass
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
385
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
386 # Overridable -- handle processing instruction
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
387 def handle_pi(self, data):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
388 pass
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
389
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
390 def unknown_decl(self, data):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
391 self.error("unknown declaration: " + `data`)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
392
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
393 # Internal -- helper to remove special character quoting
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
394 def unescape(self, s):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
395 if '&' not in s:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
396 return s
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
397 s = s.replace("&lt;", "<")
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
398 s = s.replace("&gt;", ">")
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
399 s = s.replace("&apos;", "'")
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
400 s = s.replace("&quot;", '"')
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
401 s = s.replace("&amp;", "&") # Must be last
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
402 return s

Roundup Issue Tracker: http://roundup-tracker.org/