xamplr-pp 1.0.0
Sign up to get free protection for your applications and to get access to all the features.
- data/LICENSE +504 -0
- data/README.rdoc +94 -0
- data/Rakefile +56 -0
- data/VERSION.yml +4 -0
- data/lib/xampl-pp-dtd.rb +126 -0
- data/lib/xampl-pp-wf.rb +1037 -0
- data/lib/xamplr-pp.rb +991 -0
- data/lib/xamplr-pp/ANNOUNCE.TXT +47 -0
- data/lib/xamplr-pp/LICENSE +504 -0
- data/lib/xamplr-pp/Makefile +122 -0
- data/lib/xamplr-pp/examples/parse-wf.rb +55 -0
- data/lib/xamplr-pp/examples/parse.rb +59 -0
- data/lib/xamplr-pp/license.inc +17 -0
- data/lib/xamplr-pp/saxdemo.rb +214 -0
- data/lib/xamplr-pp/saxish.rb +298 -0
- data/lib/xamplr-pp/saxishHandler.rb +58 -0
- data/lib/xamplr-pp/toys/chew.rb +62 -0
- data/lib/xamplr-pp/toys/chewMultibyte.rb +44 -0
- data/lib/xamplr-pp/toys/dump.rb +58 -0
- data/lib/xamplr-pp/xmlName.defn +67 -0
- data/lib/xamplr-pp/xpp.rb +908 -0
- data/lib/xamplr-pp/xppDeluxe.rb +49 -0
- data/lib/xamplr-pp/xppIter.rb +845 -0
- data/test/test_helper.rb +10 -0
- data/test/xamplr_pp_gem_test.rb +7 -0
- metadata +79 -0
@@ -0,0 +1,58 @@
|
|
1
|
+
#!/usr/local/bin/ruby
|
2
|
+
|
3
|
+
pattern = Regexp.compile(".", nil, 'u')
|
4
|
+
|
5
|
+
for filename in ARGV do
|
6
|
+
printf("file: %s\n", filename)
|
7
|
+
file = File.new(filename)
|
8
|
+
file.each do
|
9
|
+
| line |
|
10
|
+
p = 0
|
11
|
+
while true do
|
12
|
+
line.index(pattern, p)
|
13
|
+
if nil == $& then
|
14
|
+
break;
|
15
|
+
end
|
16
|
+
p += $&.length
|
17
|
+
$&.each_byte do
|
18
|
+
| c |
|
19
|
+
printf("%3x ", c)
|
20
|
+
end
|
21
|
+
printf("-- '%s'\n", $&);
|
22
|
+
end
|
23
|
+
end
|
24
|
+
end
|
25
|
+
|
26
|
+
def decode(s)
|
27
|
+
r = 0
|
28
|
+
$&.each_byte do
|
29
|
+
| c |
|
30
|
+
if c < 0x80 then
|
31
|
+
r += c
|
32
|
+
elsif
|
33
|
+
end
|
34
|
+
end
|
35
|
+
|
36
|
+
|
37
|
+
def encode(c)
|
38
|
+
if utf8encode then
|
39
|
+
if c < 0x80 then
|
40
|
+
@textBuffer << c
|
41
|
+
elsif c < 0x0800
|
42
|
+
@textBuffer << ((c >> 6) | 0xC0)
|
43
|
+
@textBuffer << (c & (0x3F | 0x80))
|
44
|
+
elsif c < 0x10000
|
45
|
+
@textBuffer << ((c >> 12) | 0xE0)
|
46
|
+
@textBuffer << ((c >> 6) & (0x3F | 0x80))
|
47
|
+
@textBuffer << (c & (0x3F | 0x80))
|
48
|
+
else
|
49
|
+
@textBuffer << ((c >> 18) | 0xF0)
|
50
|
+
@textBuffer << ((c >> 12) & (0x3F | 0x80))
|
51
|
+
@textBuffer << ((c >> 6) & (0x3F | 0x80))
|
52
|
+
@textBuffer << (c & (0x3F | 0x80))
|
53
|
+
end
|
54
|
+
else
|
55
|
+
@textBuffer << c
|
56
|
+
end
|
57
|
+
end
|
58
|
+
|
@@ -0,0 +1,67 @@
|
|
1
|
+
Name ::= (Letter | '_' | ':') (NameChar)*
|
2
|
+
NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | | CombiningChar | Extender
|
3
|
+
Letter ::= BaseChar | Ideographic
|
4
|
+
Ideographic ::= [#x4E00-#x9FA5] | #x3007 | [#x3021-#x3029]
|
5
|
+
WhiteSpace ::= x20 | x9 | xD | xA
|
6
|
+
Char := x09 | xA | xD | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF]
|
7
|
+
|
8
|
+
Digit ::= [#x0030-#x0039] [#x0660-#x0669] |[#x06F0-#x06F9]
|
9
|
+
|[#x0966-#x096F] |[#x09E6-#x09EF] |[#x0A66-#x0A6F] |[#x0AE6-#x0AEF]
|
10
|
+
|[#x0B66-#x0B6F] |[#x0BE7-#x0BEF] |[#x0C66-#x0C6F] |[#x0CE6-#x0CEF]
|
11
|
+
|[#x0D66-#x0D6F] |[#x0E50-#x0E59] |[#x0ED0-#x0ED9] |[#x0F20-#x0F29]
|
12
|
+
|
13
|
+
Extender ::= [#x0030-#x0039] |[#x0660-#x0669] |[#x06F0-#x06F9]
|
14
|
+
|[#x0966-#x096F] |[#x09E6-#x09EF] |[#x0A66-#x0A6F]
|
15
|
+
|[#x0AE6-#x0AEF] |[#x0B66-#x0B6F] |[#x0BE7-#x0BEF]
|
16
|
+
|[#x0C66-#x0C6F] |[#x0CE6-#x0CEF] |[#x0D66-#x0D6F]
|
17
|
+
|[#x0E50-#x0E59] |[#x0ED0-#x0ED9] |[#x0F20-#x0F29]
|
18
|
+
CombiningChar ::=[#x0300-#x0345] |[#x0360-#x0361] |[#x0483-#x0486]
|
19
|
+
|[#x0591-#x05A1] |[#x05A3-#x05B9] |[#x05BB-#x05BD]
|
20
|
+
|#x05BF |[#x05C1-#x05C2] |#x05C4 |[#x064B-#x0652]
|
21
|
+
|#x0670 |[#x06D6-#x06DC] |[#x06DD-#x06DF] |[#x06E0-#x06E4]
|
22
|
+
|[#x06E7-#x06E8] |[#x06EA-#x06ED] |[#x0901-#x0903] |#x093C
|
23
|
+
|[#x093E-#x094C] |#x094D |[#x0951-#x0954] |[#x0962-#x0963]
|
24
|
+
|[#x0981-#x0983] |#x09BC |#x09BE |#x09BF |[#x09C0-#x09C4]
|
25
|
+
|[#x09C7-#x09C8] |[#x09CB-#x09CD] |#x09D7 |[#x09E2-#x09E3]
|
26
|
+
|#x0A02 |#x0A3C |#x0A3E |#x0A3F |[#x0A40-#x0A42]
|
27
|
+
|[#x0A47-#x0A48] |[#x0A4B-#x0A4D] |[#x0A70-#x0A71]
|
28
|
+
|[#x0A81-#x0A83] |#x0ABC |[#x0ABE-#x0AC5] |[#x0AC7-#x0AC9]
|
29
|
+
|[#x0ACB-#x0ACD] |[#x0B01-#x0B03] |#x0B3C |[#x0B3E-#x0B43]
|
30
|
+
|[#x0B47-#x0B48] |[#x0B4B-#x0B4D] |[#x0B56-#x0B57]
|
31
|
+
|[#x0B82-#x0B83] |[#x0BBE-#x0BC2] |[#x0BC6-#x0BC8]
|
32
|
+
|[#x0BCA-#x0BCD] |#x0BD7 |[#x0C01-#x0C03] |[#x0C3E-#x0C44]
|
33
|
+
|[#x0C46-#x0C48] |[#x0C4A-#x0C4D] |[#x0C55-#x0C56]
|
34
|
+
|[#x0C82-#x0C83] |[#x0CBE-#x0CC4] |[#x0CC6-#x0CC8]
|
35
|
+
|[#x0CCA-#x0CCD] |[#x0CD5-#x0CD6] |[#x0D02-#x0D03]
|
36
|
+
|[#x0D3E-#x0D43] |[#x0D46-#x0D48] |[#x0D4A-#x0D4D]
|
37
|
+
|#x0D57 |#x0E31 [#x0E34-#x0E3A] |[#x0E47-#x0E4E] |#x0EB1
|
38
|
+
|[#x0EB4-#x0EB9] |[#x0EBB-#x0EBC] |[#x0EC8-#x0ECD][#x0F18-#x0F19]
|
39
|
+
|#x0F35 |#x0F37 |#x0F39 |#x0F3E |#x0F3F |[#x0F71-#x0F84]
|
40
|
+
|[#x0F86-#x0F8B] |[#x0F90-#x0F95] |#x0F97 |[#x0F99-#x0FAD]
|
41
|
+
|[#x0FB1-#x0FB7] |#x0FB9 |[#x20D0-#x20DC] |#x20E1
|
42
|
+
|[#x302A-#x302F] |#x3099 |#x309A
|
43
|
+
BaseChar ::= [#x0300-#x0345] |[#x0360-#x0361] |[#x0483-#x0486] |[#x0591-#x05A1]
|
44
|
+
|[#x05A3-#x05B9] |[#x05BB-#x05BD] |#x05BF |[#x05C1-#x05C2]
|
45
|
+
|#x05C4 |[#x064B-#x0652] |#x0670 |[#x06D6-#x06DC]
|
46
|
+
|[#x06DD-#x06DF] |[#x06E0-#x06E4] |[#x06E7-#x06E8]
|
47
|
+
|[#x06EA-#x06ED] |[#x0901-#x0903] |#x093C |[#x093E-#x094C]
|
48
|
+
|#x094D |[#x0951-#x0954] |[#x0962-#x0963] |[#x0981-#x0983]
|
49
|
+
|#x09BC |#x09BE |#x09BF |[#x09C0-#x09C4] |[#x09C7-#x09C8]
|
50
|
+
|[#x09CB-#x09CD] |#x09D7 |[#x09E2-#x09E3] |#x0A02 |#x0A3C
|
51
|
+
|#x0A3E |#x0A3F |[#x0A40-#x0A42] |[#x0A47-#x0A48]
|
52
|
+
|[#x0A4B-#x0A4D] |[#x0A70-#x0A71] |[#x0A81-#x0A83]
|
53
|
+
|#x0ABC |[#x0ABE-#x0AC5] |[#x0AC7-#x0AC9] |[#x0ACB-#x0ACD]
|
54
|
+
|[#x0B01-#x0B03] |#x0B3C |[#x0B3E-#x0B43] |[#x0B47-#x0B48]
|
55
|
+
|[#x0B4B-#x0B4D] |[#x0B56-#x0B57] |[#x0B82-#x0B83]
|
56
|
+
|[#x0BBE-#x0BC2] |[#x0BC6-#x0BC8] |[#x0BCA-#x0BCD] |#x0BD7
|
57
|
+
|[#x0C01-#x0C03] |[#x0C3E-#x0C44] |[#x0C46-#x0C48]
|
58
|
+
|[#x0C4A-#x0C4D] |[#x0C55-#x0C56] |[#x0C82-#x0C83]
|
59
|
+
|[#x0CBE-#x0CC4] |[#x0CC6-#x0CC8] |[#x0CCA-#x0CCD]
|
60
|
+
|[#x0CD5-#x0CD6] |[#x0D02-#x0D03] |[#x0D3E-#x0D43]
|
61
|
+
|[#x0D46-#x0D48] |[#x0D4A-#x0D4D] |#x0D57 |#x0E31
|
62
|
+
|[#x0E34-#x0E3A] |[#x0E47-#x0E4E] |#x0EB1 |[#x0EB4-#x0EB9]
|
63
|
+
|[#x0EBB-#x0EBC] |[#x0EC8-#x0ECD]|[#x0F18-#x0F19] |#x0F35
|
64
|
+
|#x0F37 |#x0F39 |#x0F3E |#x0F3F |[#x0F71-#x0F84]
|
65
|
+
|[#x0F86-#x0F8B] |[#x0F90-#x0F95] |#x0F97 |[#x0F99-#x0FAD]
|
66
|
+
|[#x0FB1-#x0FB7] |#x0FB9 |[#x20D0-#x20DC] |#x20E1
|
67
|
+
|[#x302A-#x302F] |#x3099 |#x309A
|
@@ -0,0 +1,908 @@
|
|
1
|
+
# xampl-pp : XML pull parser
|
2
|
+
# Copyright (C) 2002-2009 Bob Hutchison
|
3
|
+
#
|
4
|
+
# This library is free software; you can redistribute it and/or
|
5
|
+
# modify it under the terms of the GNU Lesser General Public
|
6
|
+
# License as published by the Free Software Foundation; either
|
7
|
+
# version 2.1 of the License, or (at your option) any later version.
|
8
|
+
#
|
9
|
+
# This library is distributed in the hope that it will be useful,
|
10
|
+
# but WITHOUT ANY WARRANTY; without even the implied warranty of
|
11
|
+
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
12
|
+
# #Lesser General Public License for more details.
|
13
|
+
#
|
14
|
+
# You should have received a copy of the GNU Lesser General Public
|
15
|
+
# License along with this library; if not, write to the Free Software
|
16
|
+
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
17
|
+
#
|
18
|
+
|
19
|
+
|
20
|
+
class Xpp
|
21
|
+
# XML 'event' types
|
22
|
+
START_DOCUMENT = 'START_DOCUMENT'
|
23
|
+
END_DOCUMENT = 'END_DOCUMENT'
|
24
|
+
START_ELEMENT = 'START_ELEMENT'
|
25
|
+
END_ELEMENT = 'END_ELEMENT'
|
26
|
+
TEXT = 'TEXT'
|
27
|
+
CDATA_SECTION = 'CDATA_SECTION'
|
28
|
+
ENTITY_REF = 'ENTITY_REF'
|
29
|
+
IGNORABLE_WHITESPACE = 'IGNORABLE_WHITESPACE'
|
30
|
+
PROCESSING_INSTRUCTION = 'PROCESSING_INSTRUCTION'
|
31
|
+
COMMENT = 'COMMENT'
|
32
|
+
DOCTYPE = 'DOCTYPE'
|
33
|
+
UNDECIDED_TYPE = 'UNDECIDED_TYPE'
|
34
|
+
|
35
|
+
# 'Features', acutally just processing options
|
36
|
+
attr :processNamespace, true
|
37
|
+
attr :reportNamespaceAttributes, true
|
38
|
+
attr :checkWellFormed, true
|
39
|
+
attr :utf8encode, true
|
40
|
+
|
41
|
+
# the entities that we will recognise
|
42
|
+
attr :entityMap, true
|
43
|
+
attr :unresolvedEntity
|
44
|
+
attr :resolver, true
|
45
|
+
|
46
|
+
# some information about where we are
|
47
|
+
attr :line
|
48
|
+
attr :column
|
49
|
+
|
50
|
+
# element information
|
51
|
+
attr :type
|
52
|
+
attr :emptyElement
|
53
|
+
attr :name
|
54
|
+
attr :qname
|
55
|
+
attr :namespace
|
56
|
+
attr :prefix
|
57
|
+
attr :attributeName
|
58
|
+
attr :attributeQName
|
59
|
+
attr :attributeNamespace
|
60
|
+
attr :attributePrefix
|
61
|
+
attr :attributeValue
|
62
|
+
|
63
|
+
attr :text
|
64
|
+
|
65
|
+
# These are not intended for general use (they are not part of the api)
|
66
|
+
|
67
|
+
# open element information
|
68
|
+
attr :elementName
|
69
|
+
attr :elementQName
|
70
|
+
attr :elementNamespace
|
71
|
+
attr :elementPrefix
|
72
|
+
|
73
|
+
# some pre-compiled patterns
|
74
|
+
attr :namePattern, true
|
75
|
+
attr :skipWhitespacePattern, true
|
76
|
+
|
77
|
+
attr :elementNamespacePrefixStack
|
78
|
+
attr :elementNamespaceValueStack
|
79
|
+
attr :elementNamespaceDefaultStack
|
80
|
+
|
81
|
+
public
|
82
|
+
def startDocument?
|
83
|
+
@type.equal? START_DOCUMENT
|
84
|
+
end
|
85
|
+
|
86
|
+
def endDocument?
|
87
|
+
@type.equal? END_DOCUMENT
|
88
|
+
end
|
89
|
+
|
90
|
+
def startElement?
|
91
|
+
@type.equal? START_ELEMENT
|
92
|
+
end
|
93
|
+
|
94
|
+
def endElement?
|
95
|
+
@type.equal? END_ELEMENT
|
96
|
+
end
|
97
|
+
|
98
|
+
def text?
|
99
|
+
@type.equal? TEXT
|
100
|
+
end
|
101
|
+
|
102
|
+
def cdata?
|
103
|
+
@type.equal? CDATA_SECTION
|
104
|
+
end
|
105
|
+
|
106
|
+
def entityRef?
|
107
|
+
@type.equal? ENTITY_REF
|
108
|
+
end
|
109
|
+
|
110
|
+
def ignorableWhitespace?
|
111
|
+
@type.equal? IGNORABLE_WHITESPACE
|
112
|
+
end
|
113
|
+
|
114
|
+
def whitespace?
|
115
|
+
nil == @text.index(@skipWhitespacePattern)
|
116
|
+
end
|
117
|
+
|
118
|
+
def processingInstruction?
|
119
|
+
@type.equal? PROCESSING_INSTRUCTION
|
120
|
+
end
|
121
|
+
|
122
|
+
def comment?
|
123
|
+
@type.equal? COMMENT
|
124
|
+
end
|
125
|
+
|
126
|
+
def doctype?
|
127
|
+
@type.equal? DOCTYPE
|
128
|
+
end
|
129
|
+
|
130
|
+
def input=(v)
|
131
|
+
if nil == v then
|
132
|
+
@input = nil
|
133
|
+
@inputBuffer = nil
|
134
|
+
@inputBufferLength = 0
|
135
|
+
@column = 0
|
136
|
+
@line = 0
|
137
|
+
elsif v.kind_of? String then
|
138
|
+
@input = nil
|
139
|
+
@inputBuffer = v
|
140
|
+
@inputBufferLength = v.length
|
141
|
+
@line = 1
|
142
|
+
@column = 0
|
143
|
+
elsif v.kind_of? IO then
|
144
|
+
@input = v
|
145
|
+
@inputBuffer = nil
|
146
|
+
@inputBufferLength = 0
|
147
|
+
@column = 0
|
148
|
+
@line = 0
|
149
|
+
else
|
150
|
+
raise "illegalInput"
|
151
|
+
end
|
152
|
+
@nextInputBuffer = nil
|
153
|
+
@textBuffer = ''
|
154
|
+
|
155
|
+
@elementNamespacePrefixStack = []
|
156
|
+
@elementNamespaceValueStack = []
|
157
|
+
@elementNamespaceDefaultStack = []
|
158
|
+
|
159
|
+
@elementName = []
|
160
|
+
@elementQName = []
|
161
|
+
@elementNamespace = []
|
162
|
+
@elementPrefix = []
|
163
|
+
|
164
|
+
@type = START_DOCUMENT
|
165
|
+
@unresolvedEntity = false
|
166
|
+
|
167
|
+
@name = nil
|
168
|
+
@namespace = nil
|
169
|
+
|
170
|
+
@attributeName = []
|
171
|
+
@attributeQName = []
|
172
|
+
@attributeNamespace = []
|
173
|
+
@attributePrefix = []
|
174
|
+
@attributeValue = []
|
175
|
+
|
176
|
+
@emptyElement = false
|
177
|
+
|
178
|
+
@haveRoot = false
|
179
|
+
|
180
|
+
initInput
|
181
|
+
end
|
182
|
+
|
183
|
+
def initInput
|
184
|
+
# override this if needed
|
185
|
+
end
|
186
|
+
|
187
|
+
def nextEvent
|
188
|
+
begin
|
189
|
+
@type = END_DOCUMENT
|
190
|
+
if (nil == @inputBuffer) and (nil == @input) then
|
191
|
+
#if END_DOCUMENT == @type then
|
192
|
+
return @type
|
193
|
+
#end
|
194
|
+
#raise "no input defined"
|
195
|
+
end
|
196
|
+
|
197
|
+
@unresolvedEntity = false
|
198
|
+
|
199
|
+
@text = nil
|
200
|
+
|
201
|
+
parseNextEvent
|
202
|
+
|
203
|
+
return @type
|
204
|
+
rescue Exception => message
|
205
|
+
#print message.backtrace.join("\n")
|
206
|
+
if nil != @inputBuffer then
|
207
|
+
message = sprintf("parse error: '%s' -- String input, line %d, column %d", message, @line, @column)
|
208
|
+
elsif nil != @input then
|
209
|
+
if @input.kind_of? File then
|
210
|
+
message = sprintf("parse error: '%s' -- file '%s', line %d, column %d", message, @input.path, @line, @column)
|
211
|
+
else
|
212
|
+
message = sprintf("parse error: '%s' -- unnamed IO stream, line %d, column %d", message, @line, @column)
|
213
|
+
end
|
214
|
+
else
|
215
|
+
message = sprintf("parse error: '%s' -- unknown source, line %d, column %d", message, @line, @column)
|
216
|
+
end
|
217
|
+
raise message
|
218
|
+
end
|
219
|
+
end
|
220
|
+
|
221
|
+
private
|
222
|
+
def initialize
|
223
|
+
self.processNamespace = true
|
224
|
+
self.reportNamespaceAttributes = false
|
225
|
+
self.checkWellFormed = true
|
226
|
+
self.utf8encode = true
|
227
|
+
|
228
|
+
self.input = nil
|
229
|
+
|
230
|
+
self.entityMap = {"amp"=>"&",
|
231
|
+
"apos"=>"'",
|
232
|
+
"gt"=>">",
|
233
|
+
"lt"=>"<",
|
234
|
+
"quot"=>"\""}
|
235
|
+
self.namePattern = Regexp.compile(/[^\x00-\x20=\/>\`\.\~\!\@\#\$\%\^\&\*\(\)\+\=\]\[\{\}\\\|\;\'\"\,\<\>\/\?][^\x00-\x20=\/>\`\!\@\#\$\%\^\&\*\(\)\+\=\]\[\{\}\\\|\;\'\"\,\<\>\/\?]*/, nil, 'u')
|
236
|
+
self.skipWhitespacePattern = Regexp.compile(/[^\s]+|\x00/, nil, 'u')
|
237
|
+
end
|
238
|
+
|
239
|
+
def getMoreInput
|
240
|
+
@column = 0
|
241
|
+
@line += 1
|
242
|
+
if nil == @input then
|
243
|
+
@inputBuffer = nil
|
244
|
+
@inputBufferLength = -1
|
245
|
+
return nil
|
246
|
+
end
|
247
|
+
@inputBuffer = @nextInputBuffer
|
248
|
+
if nil == @inputBuffer then
|
249
|
+
@inputBuffer = @input.gets
|
250
|
+
@column = 0
|
251
|
+
if nil == @inputBuffer then
|
252
|
+
@inputBufferLength = -1
|
253
|
+
return nil
|
254
|
+
end
|
255
|
+
end
|
256
|
+
@inputBufferLength = @inputBuffer.length
|
257
|
+
@nextInputBuffer = @input.gets
|
258
|
+
end
|
259
|
+
|
260
|
+
def expect(e)
|
261
|
+
c = read
|
262
|
+
if (nil == c) or (c != e) then
|
263
|
+
msg = sprintf("unexpectedChar:: expect '%s' got '%s' in %s", (''<<e), (''<<c), caller[0])
|
264
|
+
raise msg
|
265
|
+
end
|
266
|
+
return c
|
267
|
+
end
|
268
|
+
|
269
|
+
def read
|
270
|
+
if (nil == @inputBuffer) or (@inputBufferLength <= @column) then
|
271
|
+
getMoreInput
|
272
|
+
end
|
273
|
+
|
274
|
+
if nil != @inputBuffer then
|
275
|
+
c = @inputBuffer[@column]
|
276
|
+
@column += 1
|
277
|
+
return c
|
278
|
+
else
|
279
|
+
return nil
|
280
|
+
end
|
281
|
+
end
|
282
|
+
|
283
|
+
def peekAt0
|
284
|
+
if nil == @inputBuffer then
|
285
|
+
getMoreInput
|
286
|
+
end
|
287
|
+
if @column < @inputBufferLength then
|
288
|
+
return @inputBuffer[@column]
|
289
|
+
else
|
290
|
+
if (nil != @nextInputBuffer) and (0 < @nextInputBuffer.length) then
|
291
|
+
return @nextInputBuffer[0]
|
292
|
+
else
|
293
|
+
return nil
|
294
|
+
end
|
295
|
+
end
|
296
|
+
end
|
297
|
+
|
298
|
+
def peekAt1
|
299
|
+
if nil == @inputBuffer then
|
300
|
+
getMoreInput
|
301
|
+
end
|
302
|
+
if (@column + 1) < @inputBufferLength then
|
303
|
+
return @inputBuffer[@column + 1]
|
304
|
+
else
|
305
|
+
if @column < @inputBufferLength then
|
306
|
+
if (nil != @nextInputBuffer) and (0 < @nextInputBuffer.length) then
|
307
|
+
return @nextInputBuffer[0]
|
308
|
+
else
|
309
|
+
return nil
|
310
|
+
end
|
311
|
+
else
|
312
|
+
if (nil != @nextInputBuffer) and (1 < @nextInputBuffer.length) then
|
313
|
+
return @nextInputBuffer[1]
|
314
|
+
else
|
315
|
+
return nil
|
316
|
+
end
|
317
|
+
end
|
318
|
+
end
|
319
|
+
end
|
320
|
+
|
321
|
+
def parseNextEvent
|
322
|
+
@attributeName.clear
|
323
|
+
@attributeQName.clear
|
324
|
+
@attributeNamespace.clear
|
325
|
+
@attributePrefix.clear
|
326
|
+
@attributeValue.clear
|
327
|
+
|
328
|
+
if @emptyElement then
|
329
|
+
# the last event was an empty start element like <start/>
|
330
|
+
@type = END_ELEMENT
|
331
|
+
@emptyElement = false
|
332
|
+
return
|
333
|
+
end
|
334
|
+
|
335
|
+
@prefix = nil
|
336
|
+
@name = nil
|
337
|
+
@qname = nil
|
338
|
+
@namespace = nil
|
339
|
+
@type = peekType
|
340
|
+
|
341
|
+
@textBuffer = ''
|
342
|
+
|
343
|
+
case @type
|
344
|
+
when END_DOCUMENT
|
345
|
+
# nothing to do
|
346
|
+
when ENTITY_REF
|
347
|
+
parseEntity
|
348
|
+
@text = @textBuffer
|
349
|
+
@textBuffer = ''
|
350
|
+
when START_ELEMENT
|
351
|
+
parseStartElement
|
352
|
+
when END_ELEMENT
|
353
|
+
parseEndElement
|
354
|
+
when TEXT
|
355
|
+
parseText(?<, false)
|
356
|
+
@text = @textBuffer
|
357
|
+
@textBuffer = ''
|
358
|
+
if 0 == @elementName.length then
|
359
|
+
if nil == @text.index(@skipWhitespacePattern) then
|
360
|
+
@type = IGNORABLE_WHITESPACE
|
361
|
+
end
|
362
|
+
end
|
363
|
+
else
|
364
|
+
@type = parseUndecided
|
365
|
+
end
|
366
|
+
|
367
|
+
end
|
368
|
+
|
369
|
+
def parseUndecided
|
370
|
+
# this could be a comment, processing instruction, or CDATA section
|
371
|
+
expect ?<
|
372
|
+
|
373
|
+
demand = nil
|
374
|
+
delimiter = nil # **first** (not last) character in delimting string
|
375
|
+
|
376
|
+
@text = @textBuffer
|
377
|
+
@textBuffer = ''
|
378
|
+
|
379
|
+
c = read
|
380
|
+
if ?? == c then
|
381
|
+
result = PROCESSING_INSTRUCTION
|
382
|
+
demand = nil
|
383
|
+
delimiter = ??
|
384
|
+
elsif ?! == c then
|
385
|
+
cc = peekAt0
|
386
|
+
if ?- == cc then
|
387
|
+
result = COMMENT
|
388
|
+
demand = '--'
|
389
|
+
delimiter = ?-
|
390
|
+
elsif ?[ == cc then
|
391
|
+
result = CDATA_SECTION
|
392
|
+
demand = '[CDATA['
|
393
|
+
delimiter = ?]
|
394
|
+
else
|
395
|
+
result = DOCTYPE
|
396
|
+
demand = 'DOCTYPE'
|
397
|
+
delimiter = nil
|
398
|
+
end
|
399
|
+
else
|
400
|
+
# this should never happen because we'll get an illegal name execption
|
401
|
+
# first
|
402
|
+
raise "illegal <{c}"
|
403
|
+
end
|
404
|
+
|
405
|
+
if nil != demand then
|
406
|
+
demand.each_byte do
|
407
|
+
| d |
|
408
|
+
expect d
|
409
|
+
end
|
410
|
+
end
|
411
|
+
|
412
|
+
if DOCTYPE == result then
|
413
|
+
parseDoctype
|
414
|
+
else
|
415
|
+
if ?? == delimiter then
|
416
|
+
s = Regexp.escape "?>"
|
417
|
+
inc = 2
|
418
|
+
else
|
419
|
+
s = Regexp.escape "" << delimiter << delimiter << '>'
|
420
|
+
inc = 3
|
421
|
+
end
|
422
|
+
regex = /#{s}/u
|
423
|
+
p = findOneOfThese(regex)
|
424
|
+
@text = @textBuffer
|
425
|
+
if nil != p then
|
426
|
+
@column += inc
|
427
|
+
end
|
428
|
+
end
|
429
|
+
|
430
|
+
return result
|
431
|
+
end
|
432
|
+
|
433
|
+
def parseDoctype
|
434
|
+
depth = 1
|
435
|
+
quoted = false
|
436
|
+
delimiter = nil
|
437
|
+
entityDefinitionText = ''
|
438
|
+
|
439
|
+
@text = ''
|
440
|
+
|
441
|
+
while true do
|
442
|
+
c = read
|
443
|
+
case c
|
444
|
+
when ?', ?" # for the sake of vim '
|
445
|
+
if quoted then
|
446
|
+
if c == delimiter then
|
447
|
+
quoted = false
|
448
|
+
delimiter = nil
|
449
|
+
end
|
450
|
+
else
|
451
|
+
quoted = true
|
452
|
+
delimiter = c
|
453
|
+
end
|
454
|
+
when ?<
|
455
|
+
if not quoted then
|
456
|
+
if (?! == peekAt0) and (?- == peekAt1) then
|
457
|
+
#this is looking like a comment
|
458
|
+
@text << c
|
459
|
+
@text << (expect ?!)
|
460
|
+
@text << (expect ?-)
|
461
|
+
c = read
|
462
|
+
if ?- == c then
|
463
|
+
@text << ?-
|
464
|
+
regex = /-->/u
|
465
|
+
p = findOneOfThese(regex)
|
466
|
+
@text << @textBuffer
|
467
|
+
@textBuffer = ''
|
468
|
+
@text << (expect ?-)
|
469
|
+
@text << (expect ?-)
|
470
|
+
c = (expect ?>)
|
471
|
+
else
|
472
|
+
depth += 1
|
473
|
+
entityDefinitionText = ''
|
474
|
+
end
|
475
|
+
else
|
476
|
+
depth += 1
|
477
|
+
entityDefinitionText = ''
|
478
|
+
end
|
479
|
+
end
|
480
|
+
when ?>
|
481
|
+
if not quoted then
|
482
|
+
depth -= 1
|
483
|
+
#check right here for an entity definition!!!
|
484
|
+
entityDefinitionText = ''
|
485
|
+
if 0 == depth then
|
486
|
+
return
|
487
|
+
end
|
488
|
+
end
|
489
|
+
when nil
|
490
|
+
raise sprintf("unexpected EOF in DOCTYPE (depth: %d, quoted: %s)", depth, quoted)
|
491
|
+
end
|
492
|
+
@text << c
|
493
|
+
entityDefinitionText << c
|
494
|
+
end
|
495
|
+
end
|
496
|
+
|
497
|
+
def peekType
|
498
|
+
c = peekAt0
|
499
|
+
case c
|
500
|
+
when nil, 0
|
501
|
+
return END_DOCUMENT
|
502
|
+
when ?&
|
503
|
+
return ENTITY_REF
|
504
|
+
when ?<
|
505
|
+
case peekAt1
|
506
|
+
when ?/
|
507
|
+
return END_ELEMENT
|
508
|
+
when ?[
|
509
|
+
return CDATA_SECTION
|
510
|
+
when ??, ?!
|
511
|
+
return UNDECIDED_TYPE
|
512
|
+
else
|
513
|
+
return START_ELEMENT
|
514
|
+
end
|
515
|
+
else
|
516
|
+
return TEXT
|
517
|
+
end
|
518
|
+
end
|
519
|
+
|
520
|
+
def encode(c)
|
521
|
+
if utf8encode then
|
522
|
+
if c < 0x80 then
|
523
|
+
@textBuffer << c
|
524
|
+
elsif c < 0x0800
|
525
|
+
@textBuffer << ((c >> 6) | 0xC0)
|
526
|
+
@textBuffer << (c & (0x3F | 0x80))
|
527
|
+
elsif c < 0x10000
|
528
|
+
@textBuffer << ((c >> 12) | 0xE0)
|
529
|
+
@textBuffer << ((c >> 6) & (0x3F | 0x80))
|
530
|
+
@textBuffer << (c & (0x3F | 0x80))
|
531
|
+
else
|
532
|
+
@textBuffer << ((c >> 18) | 0xF0)
|
533
|
+
@textBuffer << ((c >> 12) & (0x3F | 0x80))
|
534
|
+
@textBuffer << ((c >> 6) & (0x3F | 0x80))
|
535
|
+
@textBuffer << (c & (0x3F | 0x80))
|
536
|
+
end
|
537
|
+
else
|
538
|
+
@textBuffer << c
|
539
|
+
end
|
540
|
+
end
|
541
|
+
|
542
|
+
def parseEntity
|
543
|
+
expect ?&
|
544
|
+
|
545
|
+
@name = ''
|
546
|
+
while true do
|
547
|
+
c = read
|
548
|
+
if ?; == c then
|
549
|
+
break
|
550
|
+
end
|
551
|
+
if nil == c then
|
552
|
+
raise "unexpectedEOF"
|
553
|
+
end
|
554
|
+
@name << c
|
555
|
+
end
|
556
|
+
|
557
|
+
if ?\# == @name[0] then
|
558
|
+
if ?x == @name[1] then
|
559
|
+
c = @name[2..@name.length].hex
|
560
|
+
else
|
561
|
+
c = @name[1..@name.length].to_i
|
562
|
+
end
|
563
|
+
encode(c)
|
564
|
+
else
|
565
|
+
value = entityMap[@name]
|
566
|
+
if nil != value then
|
567
|
+
@textBuffer << value
|
568
|
+
else
|
569
|
+
if nil != @resolver then
|
570
|
+
value = @resolver.resolve(@name)
|
571
|
+
end
|
572
|
+
|
573
|
+
if nil != value then
|
574
|
+
@textBuffer << value
|
575
|
+
else
|
576
|
+
@unresolvedEntity = true
|
577
|
+
end
|
578
|
+
end
|
579
|
+
end
|
580
|
+
end
|
581
|
+
|
582
|
+
def parseStartElement
|
583
|
+
expect ?<
|
584
|
+
skipWhitespace
|
585
|
+
|
586
|
+
@qname = readName
|
587
|
+
@textBuffer = ''
|
588
|
+
|
589
|
+
while true do
|
590
|
+
skipWhitespace
|
591
|
+
c = peekAt0
|
592
|
+
if nil == c then
|
593
|
+
raise "unexpectedEOF"
|
594
|
+
end
|
595
|
+
if ?/ == c then
|
596
|
+
@emptyElement = true
|
597
|
+
|
598
|
+
read
|
599
|
+
skipWhitespace
|
600
|
+
expect ?>
|
601
|
+
|
602
|
+
break
|
603
|
+
end
|
604
|
+
if ?> == c then
|
605
|
+
@emptyElement = false
|
606
|
+
read
|
607
|
+
break
|
608
|
+
end
|
609
|
+
|
610
|
+
aName = readName
|
611
|
+
@textBuffer = ''
|
612
|
+
if (nil == aName) or (0 == aName.length) then
|
613
|
+
raise "name expected (start element)"
|
614
|
+
end
|
615
|
+
|
616
|
+
skipWhitespace
|
617
|
+
expect ?=
|
618
|
+
skipWhitespace
|
619
|
+
|
620
|
+
delimiter = read
|
621
|
+
if ?' == delimiter then # for vim: '
|
622
|
+
value = parseText(?', true) # for vim: '
|
623
|
+
elsif ?" == delimiter then # for vim: "
|
624
|
+
value = parseText(?", true) # for vim: "
|
625
|
+
else
|
626
|
+
raise "invalidDelimiter"
|
627
|
+
end
|
628
|
+
|
629
|
+
@textBuffer = ''
|
630
|
+
|
631
|
+
# skip the end delimiter
|
632
|
+
read
|
633
|
+
|
634
|
+
if processNamespace then
|
635
|
+
@attributeQName.push aName
|
636
|
+
else
|
637
|
+
@attributeName.push aName
|
638
|
+
@attributeQName.push aName
|
639
|
+
@attributeNamespace.push nil
|
640
|
+
@attributePrefix.push nil
|
641
|
+
end
|
642
|
+
@attributeValue.push value
|
643
|
+
|
644
|
+
end
|
645
|
+
|
646
|
+
if processNamespace then
|
647
|
+
handleNamespaces
|
648
|
+
else
|
649
|
+
@name = @qname
|
650
|
+
@namespace = nil
|
651
|
+
@prefix = nil
|
652
|
+
end
|
653
|
+
|
654
|
+
if not @emptyElement then
|
655
|
+
if @checkWellFormed and (0 == @elementName.length) then
|
656
|
+
if @haveRoot then
|
657
|
+
raise "unexpected element"
|
658
|
+
end
|
659
|
+
@haveRoot = true
|
660
|
+
end
|
661
|
+
@elementName.push @name
|
662
|
+
@elementQName.push @qname
|
663
|
+
@elementNamespace.push @namespace
|
664
|
+
@elementPrefix.push @prefix
|
665
|
+
end
|
666
|
+
end
|
667
|
+
|
668
|
+
def parseEndElement
|
669
|
+
if 0 == @elementName.length then
|
670
|
+
raise "elementStackEmpty"
|
671
|
+
end
|
672
|
+
|
673
|
+
expect ?<
|
674
|
+
expect ?/
|
675
|
+
skipWhitespace
|
676
|
+
|
677
|
+
@qname = readName
|
678
|
+
startQName = @elementQName.pop
|
679
|
+
if @qname != startQName then
|
680
|
+
raise sprintf("unexpectedEndElement wanted '%s' found '%s'", startQName, @qname)
|
681
|
+
end
|
682
|
+
skipWhitespace
|
683
|
+
expect ?>
|
684
|
+
|
685
|
+
@name = @elementName.pop
|
686
|
+
@prefix = @elementPrefix.pop
|
687
|
+
@namespace = @elementNamespace.pop
|
688
|
+
|
689
|
+
@elementNamespacePrefixStack.pop
|
690
|
+
@elementNamespaceValueStack.pop
|
691
|
+
@elementNamespaceDefaultStack.pop
|
692
|
+
end
|
693
|
+
|
694
|
+
def readName
|
695
|
+
@textBuffer = ''
|
696
|
+
matchThis(@namePattern)
|
697
|
+
if 0 == @textBuffer.length then
|
698
|
+
raise "name expected (readName)"
|
699
|
+
end
|
700
|
+
return @textBuffer
|
701
|
+
end
|
702
|
+
|
703
|
+
def handleNamespaces
|
704
|
+
# This is called by parseStartElement to deal with namespaces. Updates knows
|
705
|
+
# name spaces based on the attributes in this start element. Then sets up
|
706
|
+
# the namespaces for this element itself (i.e. process the qname).
|
707
|
+
|
708
|
+
i = 0
|
709
|
+
|
710
|
+
defaultNamespace = @elementNamespaceDefaultStack.last
|
711
|
+
|
712
|
+
qnames = @attributeQName
|
713
|
+
@attributeQName = []
|
714
|
+
values = @attributeValue
|
715
|
+
@attributeValue = []
|
716
|
+
|
717
|
+
prefixList = []
|
718
|
+
valueList = []
|
719
|
+
|
720
|
+
while i < qnames.length do
|
721
|
+
qname = qnames[i]
|
722
|
+
value = values[i]
|
723
|
+
i += 1
|
724
|
+
|
725
|
+
if 'xmlns' == qname then
|
726
|
+
prefix = 'xmlns'
|
727
|
+
name = nil
|
728
|
+
namespace = lookupNamespace prefix
|
729
|
+
defaultNamespace = value
|
730
|
+
else
|
731
|
+
pieces = qname.split(':', 2)
|
732
|
+
if 2 == pieces.length then
|
733
|
+
namespace = value
|
734
|
+
prefix = pieces[0]
|
735
|
+
name = pieces[1]
|
736
|
+
|
737
|
+
if 0 == prefix.length then
|
738
|
+
raise "illegalEmptyAtributePrefix"
|
739
|
+
end
|
740
|
+
if 0 == name.length then
|
741
|
+
raise "illegalEmptyAttributeName"
|
742
|
+
end
|
743
|
+
else
|
744
|
+
# this is a un-prefixed non-xmlns attribute
|
745
|
+
@attributeQName.push qname
|
746
|
+
@attributeName.push qname
|
747
|
+
@attributeNamespace.push nil
|
748
|
+
@attributePrefix.push nil
|
749
|
+
@attributeValue.push value
|
750
|
+
|
751
|
+
next
|
752
|
+
end
|
753
|
+
end
|
754
|
+
|
755
|
+
# only prefixed attributes beyond here
|
756
|
+
|
757
|
+
if nil == namespace then
|
758
|
+
raise "illegalEmptyNamespace"
|
759
|
+
end
|
760
|
+
|
761
|
+
if "xmlns" != prefix then
|
762
|
+
anyQualifiedAttributes = true
|
763
|
+
|
764
|
+
@attributeQName.push qname
|
765
|
+
@attributeName.push name
|
766
|
+
@attributeNamespace.push namespace
|
767
|
+
@attributePrefix.push prefix
|
768
|
+
@attributeValue.push value
|
769
|
+
else
|
770
|
+
if (nil != name) and ((nil == namespace) or (0 == namespace.length)) then
|
771
|
+
raise "illegalNamespace"
|
772
|
+
end
|
773
|
+
|
774
|
+
prefixList.push name
|
775
|
+
valueList.push value
|
776
|
+
|
777
|
+
if @reportNamespaceAttributes then
|
778
|
+
@attributeQName.push qname
|
779
|
+
@attributeName.push name
|
780
|
+
@attributeNamespace.push namespace
|
781
|
+
@attributePrefix.push prefix
|
782
|
+
@attributeValue.push value
|
783
|
+
end
|
784
|
+
end
|
785
|
+
|
786
|
+
end
|
787
|
+
|
788
|
+
@elementNamespacePrefixStack.push prefixList
|
789
|
+
@elementNamespaceValueStack.push valueList
|
790
|
+
@elementNamespaceDefaultStack.push defaultNamespace
|
791
|
+
|
792
|
+
if anyQualifiedAttributes then
|
793
|
+
# run over the attributes and make sure we have them qualified
|
794
|
+
for i in 0..(@attributeName.length-1) do
|
795
|
+
prefix = @attributePrefix[i]
|
796
|
+
|
797
|
+
if nil != prefix then
|
798
|
+
@attributeNamespace[i] = lookupNamespace prefix
|
799
|
+
end
|
800
|
+
end
|
801
|
+
end
|
802
|
+
|
803
|
+
# handle namespaces for the element name
|
804
|
+
pieces = @qname.split(':', 2)
|
805
|
+
if 2 == pieces.length then
|
806
|
+
@name = pieces[1]
|
807
|
+
@prefix = pieces[0]
|
808
|
+
@namespace = lookupNamespace @prefix
|
809
|
+
else
|
810
|
+
@name = @qname
|
811
|
+
@namespace = defaultNamespace
|
812
|
+
@prefix = nil
|
813
|
+
end
|
814
|
+
end
|
815
|
+
|
816
|
+
def lookupNamespace(prefix)
|
817
|
+
if nil == prefix then
|
818
|
+
raise "illegalPrefix"
|
819
|
+
end
|
820
|
+
if'xml' == prefix then
|
821
|
+
return 'http://www.w3.org/XML/1998/namespace'
|
822
|
+
end
|
823
|
+
if'xmlns' == prefix then
|
824
|
+
return 'http://www.w3.org/2000/xmlns/'
|
825
|
+
end
|
826
|
+
|
827
|
+
i = @elementNamespacePrefixStack.length - 1
|
828
|
+
while 0 <= i do
|
829
|
+
j = @elementNamespacePrefixStack[i].index(prefix)
|
830
|
+
if nil != j then
|
831
|
+
return @elementNamespaceValueStack[i][j]
|
832
|
+
end
|
833
|
+
|
834
|
+
i -= 1
|
835
|
+
end
|
836
|
+
raise sprintf("unknown Namespace Prefix '%s' [%s]", prefix, caller[0])
|
837
|
+
end
|
838
|
+
|
839
|
+
def parseText(delimiter, resolve)
|
840
|
+
s = "&" << delimiter
|
841
|
+
regex = /[#{s}]/u
|
842
|
+
c = findOneOfThese regex
|
843
|
+
while (nil != c) and (delimiter != c) do
|
844
|
+
if ?& == c then
|
845
|
+
if !resolve then
|
846
|
+
break
|
847
|
+
end
|
848
|
+
|
849
|
+
parseEntity
|
850
|
+
|
851
|
+
if @unresolvedEntity then
|
852
|
+
raise "unresolvedEntity"
|
853
|
+
end
|
854
|
+
else
|
855
|
+
c = read
|
856
|
+
@textBuffer << c
|
857
|
+
end
|
858
|
+
|
859
|
+
c = findOneOfThese regex
|
860
|
+
end
|
861
|
+
|
862
|
+
return @textBuffer
|
863
|
+
end
|
864
|
+
|
865
|
+
def skipWhitespace
|
866
|
+
while nil != @inputBuffer do
|
867
|
+
p = @inputBuffer.index(@skipWhitespacePattern, @column)
|
868
|
+
|
869
|
+
if nil != p then
|
870
|
+
if p != @column then
|
871
|
+
@column = p
|
872
|
+
end
|
873
|
+
return @inputBuffer[p]
|
874
|
+
end
|
875
|
+
getMoreInput
|
876
|
+
end
|
877
|
+
return nil
|
878
|
+
end
|
879
|
+
|
880
|
+
def matchThis(regex)
|
881
|
+
p = @inputBuffer.index(regex, @column)
|
882
|
+
if nil != $& then
|
883
|
+
@textBuffer << $&
|
884
|
+
@column += $&.length
|
885
|
+
else
|
886
|
+
@column = @inputBufferLength
|
887
|
+
end
|
888
|
+
end
|
889
|
+
|
890
|
+
def findOneOfThese(regex)
|
891
|
+
while nil != @inputBuffer do
|
892
|
+
p = @inputBuffer.index(regex, @column)
|
893
|
+
|
894
|
+
if nil != p then
|
895
|
+
if p != @column then
|
896
|
+
@textBuffer << @inputBuffer[@column..(p - 1)]
|
897
|
+
@column = p
|
898
|
+
end
|
899
|
+
return @inputBuffer[p]
|
900
|
+
else
|
901
|
+
@textBuffer << @inputBuffer[@column..-1]
|
902
|
+
end
|
903
|
+
getMoreInput
|
904
|
+
end
|
905
|
+
return nil
|
906
|
+
end
|
907
|
+
|
908
|
+
end
|