annotate roundup/cgi/TAL/HTMLTALParser.py @ 8200:f5d6dc28d571

chore(ruff): remove unreachable line
author John Rouillard <rouilj@ieee.org>
date Wed, 11 Dec 2024 14:17:01 -0500
parents 23b8e6067f7c
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
1 ##############################################################################
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
2 #
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
3 # Copyright (c) 2001, 2002 Zope Corporation and Contributors.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
4 # All Rights Reserved.
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
5 #
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
6 # This software is subject to the provisions of the Zope Public License,
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
7 # Version 2.0 (ZPL). A copy of the ZPL should accompany this distribution.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
8 # THIS SOFTWARE IS PROVIDED "AS IS" AND ANY AND ALL EXPRESS OR IMPLIED
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
9 # WARRANTIES ARE DISCLAIMED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
10 # WARRANTIES OF TITLE, MERCHANTABILITY, AGAINST INFRINGEMENT, AND FITNESS
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
11 # FOR A PARTICULAR PURPOSE.
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
12 #
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
13 ##############################################################################
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
14 """
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
15 Parse HTML and compile to TALInterpreter intermediate code.
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
16 """
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
17
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
18 import sys
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
19
5388
d26921b851c3 Python 3 preparation: make relative imports explicit.
Joseph Myers <jsm@polyomino.org.uk>
parents: 5381
diff changeset
20 from .TALGenerator import TALGenerator
d26921b851c3 Python 3 preparation: make relative imports explicit.
Joseph Myers <jsm@polyomino.org.uk>
parents: 5381
diff changeset
21 from .HTMLParser import HTMLParser, HTMLParseError
d26921b851c3 Python 3 preparation: make relative imports explicit.
Joseph Myers <jsm@polyomino.org.uk>
parents: 5381
diff changeset
22 from .TALDefs import \
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
23 ZOPE_METAL_NS, ZOPE_TAL_NS, ZOPE_I18N_NS, METALError, TALError, I18NError
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
24
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
25 BOOLEAN_HTML_ATTRS = [
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
26 # List of Boolean attributes in HTML that may be given in
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
27 # minimized form (e.g. <img ismap> rather than <img ismap="">)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
28 # From http://www.w3.org/TR/xhtml1/#guidelines (C.10)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
29 "compact", "nowrap", "ismap", "declare", "noshade", "checked",
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
30 "disabled", "readonly", "multiple", "selected", "noresize",
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
31 "defer"
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
32 ]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
33
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
34 EMPTY_HTML_TAGS = [
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
35 # List of HTML tags with an empty content model; these are
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
36 # rendered in minimized form, e.g. <img />.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
37 # From http://www.w3.org/TR/xhtml1/#dtds
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
38 "base", "meta", "link", "hr", "br", "param", "img", "area",
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
39 "input", "col", "basefont", "isindex", "frame",
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 PARA_LEVEL_HTML_TAGS = [
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
43 # List of HTML elements that close open paragraph-level elements
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
44 # and are themselves paragraph-level.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
45 "h1", "h2", "h3", "h4", "h5", "h6", "p",
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
46 ]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
47
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
48 BLOCK_CLOSING_TAG_MAP = {
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
49 "tr": ("tr", "td", "th"),
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
50 "td": ("td", "th"),
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
51 "th": ("td", "th"),
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
52 "li": ("li",),
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
53 "dd": ("dd", "dt"),
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
54 "dt": ("dd", "dt"),
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
55 }
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
56
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
57 BLOCK_LEVEL_HTML_TAGS = [
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
58 # List of HTML tags that denote larger sections than paragraphs.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
59 "blockquote", "table", "tr", "th", "td", "thead", "tfoot", "tbody",
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
60 "noframe", "ul", "ol", "li", "dl", "dt", "dd", "div",
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
61 ]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
62
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
63 TIGHTEN_IMPLICIT_CLOSE_TAGS = (PARA_LEVEL_HTML_TAGS
5395
23b8e6067f7c Python 3 preparation: update calls to dict methods.
Joseph Myers <jsm@polyomino.org.uk>
parents: 5388
diff changeset
64 + list(BLOCK_CLOSING_TAG_MAP.keys()))
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
65
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
66
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
67 class NestingError(HTMLParseError):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
68 """Exception raised when elements aren't properly nested."""
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
69
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
70 def __init__(self, tagstack, endtag, position=(None, None)):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
71 self.endtag = endtag
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
72 if tagstack:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
73 if len(tagstack) == 1:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
74 msg = ('Open tag <%s> does not match close tag </%s>'
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
75 % (tagstack[0], endtag))
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
76 else:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
77 msg = ('Open tags <%s> do not match close tag </%s>'
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
78 % ('>, <'.join(tagstack), endtag))
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
79 else:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
80 msg = 'No tags are open to match </%s>' % endtag
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
81 HTMLParseError.__init__(self, msg, position)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
82
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
83 class EmptyTagError(NestingError):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
84 """Exception raised when empty elements have an end tag."""
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
85
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
86 def __init__(self, tag, position=(None, None)):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
87 self.tag = tag
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
88 msg = 'Close tag </%s> should be removed' % tag
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
89 HTMLParseError.__init__(self, msg, position)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
90
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
91 class OpenTagError(NestingError):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
92 """Exception raised when a tag is not allowed in another tag."""
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
93
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
94 def __init__(self, tagstack, tag, position=(None, None)):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
95 self.tag = tag
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
96 msg = 'Tag <%s> is not allowed in <%s>' % (tag, tagstack[-1])
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
97 HTMLParseError.__init__(self, msg, position)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
98
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
99 class HTMLTALParser(HTMLParser):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
100
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
101 # External API
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
102
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
103 def __init__(self, gen=None):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
104 HTMLParser.__init__(self)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
105 if gen is None:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
106 gen = TALGenerator(xml=0)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
107 self.gen = gen
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
108 self.tagstack = []
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
109 self.nsstack = []
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
110 self.nsdict = {'tal': ZOPE_TAL_NS,
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
111 'metal': ZOPE_METAL_NS,
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
112 'i18n': ZOPE_I18N_NS,
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
113 }
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
114
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
115 def parseFile(self, file):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
116 f = open(file)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
117 data = f.read()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
118 f.close()
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
119 try:
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
120 self.parseString(data)
5248
198b6e810c67 Use Python-3-compatible 'as' syntax for except statements
Eric S. Raymond <esr@thyrsus.com>
parents: 2348
diff changeset
121 except TALError as e:
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
122 e.setFile(file)
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
123 raise
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
124
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
125 def parseString(self, data):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
126 self.feed(data)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
127 self.close()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
128 while self.tagstack:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
129 self.implied_endtag(self.tagstack[-1], 2)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
130 assert self.nsstack == [], self.nsstack
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
131
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
132 def getCode(self):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
133 return self.gen.getCode()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
134
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
135 def getWarnings(self):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
136 return ()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
137
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
138 # Overriding HTMLParser methods
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
139
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
140 def handle_starttag(self, tag, attrs):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
141 self.close_para_tags(tag)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
142 self.scan_xmlns(attrs)
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
143 tag, attrlist, taldict, metaldict, i18ndict \
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
144 = self.process_ns(tag, attrs)
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
145 if tag in EMPTY_HTML_TAGS and taldict.get("content"):
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
146 raise TALError(
5377
12fe83f90f0d Python 3 preparation: use repr() instead of ``.
Joseph Myers <jsm@polyomino.org.uk>
parents: 5248
diff changeset
147 "empty HTML tags cannot use tal:content: %s" % repr(tag),
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
148 self.getpos())
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
149 self.tagstack.append(tag)
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
150 self.gen.emitStartElement(tag, attrlist, taldict, metaldict, i18ndict,
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
151 self.getpos())
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
152 if tag in EMPTY_HTML_TAGS:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
153 self.implied_endtag(tag, -1)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
154
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
155 def handle_startendtag(self, tag, attrs):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
156 self.close_para_tags(tag)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
157 self.scan_xmlns(attrs)
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
158 tag, attrlist, taldict, metaldict, i18ndict \
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
159 = self.process_ns(tag, attrs)
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
160 if taldict.get("content"):
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
161 if tag in EMPTY_HTML_TAGS:
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
162 raise TALError(
5377
12fe83f90f0d Python 3 preparation: use repr() instead of ``.
Joseph Myers <jsm@polyomino.org.uk>
parents: 5248
diff changeset
163 "empty HTML tags cannot use tal:content: %s" % repr(tag),
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
164 self.getpos())
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
165 self.gen.emitStartElement(tag, attrlist, taldict, metaldict,
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
166 i18ndict, self.getpos())
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
167 self.gen.emitEndElement(tag, implied=-1)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
168 else:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
169 self.gen.emitStartElement(tag, attrlist, taldict, metaldict,
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
170 i18ndict, self.getpos(), isend=1)
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
171 self.pop_xmlns()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
172
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
173 def handle_endtag(self, tag):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
174 if tag in EMPTY_HTML_TAGS:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
175 # </img> etc. in the source is an error
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
176 raise EmptyTagError(tag, self.getpos())
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
177 self.close_enclosed_tags(tag)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
178 self.gen.emitEndElement(tag)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
179 self.pop_xmlns()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
180 self.tagstack.pop()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
181
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
182 def close_para_tags(self, tag):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
183 if tag in EMPTY_HTML_TAGS:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
184 return
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
185 close_to = -1
5381
0942fe89e82e Python 3 preparation: change "x.has_key(y)" to "y in x".
Joseph Myers <jsm@polyomino.org.uk>
parents: 5377
diff changeset
186 if tag in BLOCK_CLOSING_TAG_MAP:
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
187 blocks_to_close = BLOCK_CLOSING_TAG_MAP[tag]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
188 for i in range(len(self.tagstack)):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
189 t = self.tagstack[i]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
190 if t in blocks_to_close:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
191 if close_to == -1:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
192 close_to = i
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
193 elif t in BLOCK_LEVEL_HTML_TAGS:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
194 close_to = -1
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
195 elif tag in PARA_LEVEL_HTML_TAGS + BLOCK_LEVEL_HTML_TAGS:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
196 i = len(self.tagstack) - 1
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
197 while i >= 0:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
198 closetag = self.tagstack[i]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
199 if closetag in BLOCK_LEVEL_HTML_TAGS:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
200 break
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
201 if closetag in PARA_LEVEL_HTML_TAGS:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
202 if closetag != "p":
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
203 raise OpenTagError(self.tagstack, tag, self.getpos())
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
204 close_to = i
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
205 i = i - 1
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
206 if close_to >= 0:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
207 while len(self.tagstack) > close_to:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
208 self.implied_endtag(self.tagstack[-1], 1)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
209
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
210 def close_enclosed_tags(self, tag):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
211 if tag not in self.tagstack:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
212 raise NestingError(self.tagstack, tag, self.getpos())
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
213 while tag != self.tagstack[-1]:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
214 self.implied_endtag(self.tagstack[-1], 1)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
215 assert self.tagstack[-1] == tag
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
216
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
217 def implied_endtag(self, tag, implied):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
218 assert tag == self.tagstack[-1]
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
219 assert implied in (-1, 1, 2)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
220 isend = (implied < 0)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
221 if tag in TIGHTEN_IMPLICIT_CLOSE_TAGS:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
222 # Pick out trailing whitespace from the program, and
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
223 # insert the close tag before the whitespace.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
224 white = self.gen.unEmitWhitespace()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
225 else:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
226 white = None
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
227 self.gen.emitEndElement(tag, isend=isend, implied=implied)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
228 if white:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
229 self.gen.emitRawText(white)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
230 self.tagstack.pop()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
231 self.pop_xmlns()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
232
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
233 def handle_charref(self, name):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
234 self.gen.emitRawText("&#%s;" % name)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
235
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
236 def handle_entityref(self, name):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
237 self.gen.emitRawText("&%s;" % name)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
238
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
239 def handle_data(self, data):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
240 self.gen.emitRawText(data)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
241
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
242 def handle_comment(self, data):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
243 self.gen.emitRawText("<!--%s-->" % data)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
244
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
245 def handle_decl(self, data):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
246 self.gen.emitRawText("<!%s>" % data)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
247
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
248 def handle_pi(self, data):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
249 self.gen.emitRawText("<?%s>" % data)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
250
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
251 # Internal thingies
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
252
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
253 def scan_xmlns(self, attrs):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
254 nsnew = {}
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
255 for key, value in attrs:
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
256 if key.startswith("xmlns:"):
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
257 nsnew[key[6:]] = value
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
258 if nsnew:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
259 self.nsstack.append(self.nsdict)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
260 self.nsdict = self.nsdict.copy()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
261 self.nsdict.update(nsnew)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
262 else:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
263 self.nsstack.append(self.nsdict)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
264
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
265 def pop_xmlns(self):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
266 self.nsdict = self.nsstack.pop()
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
267
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
268 def fixname(self, name):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
269 if ':' in name:
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
270 prefix, suffix = name.split(':', 1)
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
271 if prefix == 'xmlns':
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
272 nsuri = self.nsdict.get(suffix)
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
273 if nsuri in (ZOPE_TAL_NS, ZOPE_METAL_NS, ZOPE_I18N_NS):
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
274 return name, name, prefix
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
275 else:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
276 nsuri = self.nsdict.get(prefix)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
277 if nsuri == ZOPE_TAL_NS:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
278 return name, suffix, 'tal'
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
279 elif nsuri == ZOPE_METAL_NS:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
280 return name, suffix, 'metal'
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
281 elif nsuri == ZOPE_I18N_NS:
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
282 return name, suffix, 'i18n'
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
283 return name, name, 0
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
284
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
285 def process_ns(self, name, attrs):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
286 attrlist = []
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
287 taldict = {}
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
288 metaldict = {}
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
289 i18ndict = {}
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
290 name, namebase, namens = self.fixname(name)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
291 for item in attrs:
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
292 key, value = item
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
293 key, keybase, keyns = self.fixname(key)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
294 ns = keyns or namens # default to tag namespace
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
295 if ns and ns != 'unknown':
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
296 item = (key, value, ns)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
297 if ns == 'tal':
5381
0942fe89e82e Python 3 preparation: change "x.has_key(y)" to "y in x".
Joseph Myers <jsm@polyomino.org.uk>
parents: 5377
diff changeset
298 if keybase in taldict:
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
299 raise TALError("duplicate TAL attribute " +
5377
12fe83f90f0d Python 3 preparation: use repr() instead of ``.
Joseph Myers <jsm@polyomino.org.uk>
parents: 5248
diff changeset
300 repr(keybase), self.getpos())
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
301 taldict[keybase] = value
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
302 elif ns == 'metal':
5381
0942fe89e82e Python 3 preparation: change "x.has_key(y)" to "y in x".
Joseph Myers <jsm@polyomino.org.uk>
parents: 5377
diff changeset
303 if keybase in metaldict:
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
304 raise METALError("duplicate METAL attribute " +
5377
12fe83f90f0d Python 3 preparation: use repr() instead of ``.
Joseph Myers <jsm@polyomino.org.uk>
parents: 5248
diff changeset
305 repr(keybase), self.getpos())
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
306 metaldict[keybase] = value
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
307 elif ns == 'i18n':
5381
0942fe89e82e Python 3 preparation: change "x.has_key(y)" to "y in x".
Joseph Myers <jsm@polyomino.org.uk>
parents: 5377
diff changeset
308 if keybase in i18ndict:
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
309 raise I18NError("duplicate i18n attribute " +
5377
12fe83f90f0d Python 3 preparation: use repr() instead of ``.
Joseph Myers <jsm@polyomino.org.uk>
parents: 5248
diff changeset
310 repr(keybase), self.getpos())
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
311 i18ndict[keybase] = value
1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
312 attrlist.append(item)
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
313 if namens in ('metal', 'tal'):
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
314 taldict['tal tag'] = namens
2348
8c2402a78bb0 beginning getting ZPT up to date: TAL first
Richard Jones <richard@users.sourceforge.net>
parents: 2005
diff changeset
315 return name, attrlist, taldict, metaldict, i18ndict

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