1049
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
1 """Shared support for scanning document type declarations in HTML and XHTML."""
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
2
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
3 import re
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
4 import string
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
5
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
6 _declname_match = re.compile(r'[a-zA-Z][-_.a-zA-Z0-9]*\s*').match
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
7 _declstringlit_match = re.compile(r'(\'[^\']*\'|"[^"]*")\s*').match
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
8
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
9 del re
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
10
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
11
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
12 class ParserBase:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
13 """Parser base class which provides some common support methods used
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
14 by the SGML/HTML and XHTML parsers."""
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
15
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
16 def reset(self):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
17 self.lineno = 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
18 self.offset = 0
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
19
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
20 def getpos(self):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
21 """Return current line number and offset."""
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
22 return self.lineno, self.offset
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
23
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
24 # Internal -- update line number and offset. This should be
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
25 # called for each piece of data exactly once, in order -- in other
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
26 # words the concatenation of all the input strings to this
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
27 # function should be exactly the entire input.
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
28 def updatepos(self, i, j):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
29 if i >= j:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
30 return j
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
31 rawdata = self.rawdata
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
32 nlines = string.count(rawdata, "\n", i, j)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
33 if nlines:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
34 self.lineno = self.lineno + nlines
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
35 pos = string.rindex(rawdata, "\n", i, j) # Should not fail
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
36 self.offset = j-(pos+1)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
37 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
38 self.offset = self.offset + j-i
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
39 return j
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
40
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
41 _decl_otherchars = ''
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
42
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
43 # Internal -- parse declaration (for use by subclasses).
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
44 def parse_declaration(self, i):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
45 # This is some sort of declaration; in "HTML as
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
46 # deployed," this should only be the document type
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
47 # declaration ("<!DOCTYPE html...>").
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
48 rawdata = self.rawdata
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
49 import sys
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
50 j = i + 2
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
51 assert rawdata[i:j] == "<!", "unexpected call to parse_declaration"
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
52 if rawdata[j:j+1] in ("-", ""):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
53 # Start of comment followed by buffer boundary,
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
54 # or just a buffer boundary.
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
55 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
56 # in practice, this should look like: ((name|stringlit) S*)+ '>'
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
57 n = len(rawdata)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
58 decltype, j = self._scan_name(j, i)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
59 if j < 0:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
60 return j
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
61 if decltype == "doctype":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
62 self._decl_otherchars = ''
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
63 while j < n:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
64 c = rawdata[j]
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
65 if c == ">":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
66 # end of declaration syntax
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
67 data = rawdata[i+2:j]
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
68 if decltype == "doctype":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
69 self.handle_decl(data)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
70 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
71 self.unknown_decl(data)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
72 return j + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
73 if c in "\"'":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
74 m = _declstringlit_match(rawdata, j)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
75 if not m:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
76 return -1 # incomplete
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
77 j = m.end()
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
78 elif c in "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
79 name, j = self._scan_name(j, i)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
80 elif c in self._decl_otherchars:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
81 j = j + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
82 elif c == "[":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
83 if decltype == "doctype":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
84 j = self._parse_doctype_subset(j + 1, i)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
85 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
86 self.error("unexpected '[' char in declaration")
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
87 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
88 self.error(
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
89 "unexpected %s char in declaration" % `rawdata[j]`)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
90 if j < 0:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
91 return j
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
92 return -1 # incomplete
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
93
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
94 # Internal -- scan past the internal subset in a <!DOCTYPE declaration,
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
95 # returning the index just past any whitespace following the trailing ']'.
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
96 def _parse_doctype_subset(self, i, declstartpos):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
97 rawdata = self.rawdata
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
98 n = len(rawdata)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
99 j = i
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
100 while j < n:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
101 c = rawdata[j]
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
102 if c == "<":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
103 s = rawdata[j:j+2]
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
104 if s == "<":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
105 # end of buffer; incomplete
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
106 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
107 if s != "<!":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
108 self.updatepos(declstartpos, j + 1)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
109 self.error("unexpected char in internal subset (in %s)"
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
110 % `s`)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
111 if (j + 2) == n:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
112 # end of buffer; incomplete
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
113 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
114 if (j + 4) > n:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
115 # end of buffer; incomplete
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
116 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
117 if rawdata[j:j+4] == "<!--":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
118 j = self.parse_comment(j, report=0)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
119 if j < 0:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
120 return j
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
121 continue
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
122 name, j = self._scan_name(j + 2, declstartpos)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
123 if j == -1:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
124 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
125 if name not in ("attlist", "element", "entity", "notation"):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
126 self.updatepos(declstartpos, j + 2)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
127 self.error(
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
128 "unknown declaration %s in internal subset" % `name`)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
129 # handle the individual names
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
130 meth = getattr(self, "_parse_doctype_" + name)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
131 j = meth(j, declstartpos)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
132 if j < 0:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
133 return j
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
134 elif c == "%":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
135 # parameter entity reference
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
136 if (j + 1) == n:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
137 # end of buffer; incomplete
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
138 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
139 s, j = self._scan_name(j + 1, declstartpos)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
140 if j < 0:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
141 return j
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
142 if rawdata[j] == ";":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
143 j = j + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
144 elif c == "]":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
145 j = j + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
146 while j < n and rawdata[j] in string.whitespace:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
147 j = j + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
148 if j < n:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
149 if rawdata[j] == ">":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
150 return j
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
151 self.updatepos(declstartpos, j)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
152 self.error("unexpected char after internal subset")
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
153 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
154 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
155 elif c in string.whitespace:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
156 j = j + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
157 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
158 self.updatepos(declstartpos, j)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
159 self.error("unexpected char %s in internal subset" % `c`)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
160 # end of buffer reached
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
161 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
162
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
163 # Internal -- scan past <!ELEMENT declarations
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
164 def _parse_doctype_element(self, i, declstartpos):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
165 rawdata = self.rawdata
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
166 n = len(rawdata)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
167 name, j = self._scan_name(i, declstartpos)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
168 if j == -1:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
169 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
170 # style content model; just skip until '>'
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
171 if '>' in rawdata[j:]:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
172 return string.find(rawdata, ">", j) + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
173 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
174
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
175 # Internal -- scan past <!ATTLIST declarations
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
176 def _parse_doctype_attlist(self, i, declstartpos):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
177 rawdata = self.rawdata
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
178 name, j = self._scan_name(i, declstartpos)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
179 c = rawdata[j:j+1]
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
180 if c == "":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
181 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
182 if c == ">":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
183 return j + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
184 while 1:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
185 # scan a series of attribute descriptions; simplified:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
186 # name type [value] [#constraint]
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
187 name, j = self._scan_name(j, declstartpos)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
188 if j < 0:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
189 return j
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
190 c = rawdata[j:j+1]
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
191 if c == "":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
192 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
193 if c == "(":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
194 # an enumerated type; look for ')'
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
195 if ")" in rawdata[j:]:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
196 j = string.find(rawdata, ")", j) + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
197 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
198 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
199 while rawdata[j:j+1] in string.whitespace:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
200 j = j + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
201 if not rawdata[j:]:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
202 # end of buffer, incomplete
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
203 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
204 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
205 name, j = self._scan_name(j, declstartpos)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
206 c = rawdata[j:j+1]
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
207 if not c:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
208 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
209 if c in "'\"":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
210 m = _declstringlit_match(rawdata, j)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
211 if m:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
212 j = m.end()
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
213 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
214 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
215 c = rawdata[j:j+1]
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
216 if not c:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
217 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
218 if c == "#":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
219 if rawdata[j:] == "#":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
220 # end of buffer
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
221 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
222 name, j = self._scan_name(j + 1, declstartpos)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
223 if j < 0:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
224 return j
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
225 c = rawdata[j:j+1]
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
226 if not c:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
227 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
228 if c == '>':
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
229 # all done
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
230 return j + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
231
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
232 # Internal -- scan past <!NOTATION declarations
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
233 def _parse_doctype_notation(self, i, declstartpos):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
234 name, j = self._scan_name(i, declstartpos)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
235 if j < 0:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
236 return j
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
237 rawdata = self.rawdata
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
238 while 1:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
239 c = rawdata[j:j+1]
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
240 if not c:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
241 # end of buffer; incomplete
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
242 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
243 if c == '>':
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
244 return j + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
245 if c in "'\"":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
246 m = _declstringlit_match(rawdata, j)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
247 if not m:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
248 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
249 j = m.end()
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
250 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
251 name, j = self._scan_name(j, declstartpos)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
252 if j < 0:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
253 return j
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
254
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
255 # Internal -- scan past <!ENTITY declarations
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
256 def _parse_doctype_entity(self, i, declstartpos):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
257 rawdata = self.rawdata
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
258 if rawdata[i:i+1] == "%":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
259 j = i + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
260 while 1:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
261 c = rawdata[j:j+1]
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
262 if not c:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
263 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
264 if c in string.whitespace:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
265 j = j + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
266 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
267 break
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
268 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
269 j = i
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
270 name, j = self._scan_name(j, declstartpos)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
271 if j < 0:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
272 return j
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
273 while 1:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
274 c = self.rawdata[j:j+1]
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
275 if not c:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
276 return -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
277 if c in "'\"":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
278 m = _declstringlit_match(rawdata, j)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
279 if m:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
280 j = m.end()
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
281 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
282 return -1 # incomplete
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
283 elif c == ">":
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
284 return j + 1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
285 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
286 name, j = self._scan_name(j, declstartpos)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
287 if j < 0:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
288 return j
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
289
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
290 # Internal -- scan a name token and the new position and the token, or
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
291 # return -1 if we've reached the end of the buffer.
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
292 def _scan_name(self, i, declstartpos):
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
293 rawdata = self.rawdata
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
294 n = len(rawdata)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
295 if i == n:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
296 return None, -1
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
297 m = _declname_match(rawdata, i)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
298 if m:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
299 s = m.group()
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
300 name = string.strip(s)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
301 if (i + len(s)) == n:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
302 return None, -1 # end of buffer
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
303 return string.lower(name), m.end()
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
304 else:
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
305 self.updatepos(declstartpos, i)
|
Richard Jones <richard@users.sourceforge.net>
parents:
diff
changeset
|
306 self.error("expected name token", self.getpos())
|