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 +1,49 @@
|
|
1
|
+
|
2
|
+
require "xpp"
|
3
|
+
|
4
|
+
class Xpp
|
5
|
+
def startDocument?
|
6
|
+
@type.equal? START_DOCUMENT
|
7
|
+
end
|
8
|
+
|
9
|
+
def endDocument?
|
10
|
+
@type.equal? END_DOCUMENT
|
11
|
+
end
|
12
|
+
|
13
|
+
def startElement?
|
14
|
+
@type.equal? START_ELEMENT
|
15
|
+
end
|
16
|
+
|
17
|
+
def endElement?
|
18
|
+
@type.equal? END_ELEMENT
|
19
|
+
end
|
20
|
+
|
21
|
+
def text?
|
22
|
+
@type.equal? TEXT
|
23
|
+
end
|
24
|
+
|
25
|
+
def cdata?
|
26
|
+
@type.equal? CDATA_SECTION
|
27
|
+
end
|
28
|
+
|
29
|
+
def entityRef?
|
30
|
+
@type.equal? ENTITY_REF
|
31
|
+
end
|
32
|
+
|
33
|
+
def ignorableWhitespace?
|
34
|
+
@type.equal? IGNORABLE_WHITESPACE
|
35
|
+
end
|
36
|
+
|
37
|
+
def processingInstruction?
|
38
|
+
@type.equal? PROCESSING_INSTRUCTION
|
39
|
+
end
|
40
|
+
|
41
|
+
def comment?
|
42
|
+
@type.equal? COMMENT
|
43
|
+
end
|
44
|
+
|
45
|
+
def doctype?
|
46
|
+
@type.equal? DOCTYPE
|
47
|
+
end
|
48
|
+
end
|
49
|
+
|
@@ -0,0 +1,845 @@
|
|
1
|
+
|
2
|
+
class Xpp
|
3
|
+
# XML 'event' types
|
4
|
+
START_DOCUMENT = 'START_DOCUMENT'
|
5
|
+
END_DOCUMENT = 'END_DOCUMENT'
|
6
|
+
START_ELEMENT = 'START_ELEMENT'
|
7
|
+
END_ELEMENT = 'END_ELEMENT'
|
8
|
+
TEXT = 'TEXT'
|
9
|
+
CDATA_SECTION = 'CDATA_SECTION'
|
10
|
+
ENTITY_REF = 'ENTITY_REF'
|
11
|
+
IGNORABLE_WHITESPACE = 'IGNORABLE_WHITESPACE'
|
12
|
+
PROCESSING_INSTRUCTION = 'PROCESSING_INSTRUCTION'
|
13
|
+
COMMENT = 'COMMENT'
|
14
|
+
DOCTYPE = 'DOCTYPE'
|
15
|
+
LOOK_FURTHER = 'LOOK_FURTHER'
|
16
|
+
|
17
|
+
# 'Features', acutally just processing options
|
18
|
+
attr :processNamespace, true
|
19
|
+
attr :reportNamespaceAttributes, true
|
20
|
+
|
21
|
+
# the entities that we will recognise
|
22
|
+
attr :entityMap, true
|
23
|
+
attr :unresolvedEntity
|
24
|
+
|
25
|
+
# some information about where we are
|
26
|
+
attr :line
|
27
|
+
attr :column
|
28
|
+
|
29
|
+
# is the current text whitespace?
|
30
|
+
attr :whitespace
|
31
|
+
|
32
|
+
# element information
|
33
|
+
attr :type
|
34
|
+
attr :emptyElement
|
35
|
+
attr :name
|
36
|
+
attr :qname
|
37
|
+
attr :namespace
|
38
|
+
attr :prefix
|
39
|
+
attr :attributeName
|
40
|
+
attr :attributeQName
|
41
|
+
attr :attributeNamespace
|
42
|
+
attr :attributePrefix
|
43
|
+
attr :attributeValue
|
44
|
+
|
45
|
+
attr :text
|
46
|
+
|
47
|
+
attr :elementNamespacePrefixStack
|
48
|
+
attr :elementNamespaceValueStack
|
49
|
+
attr :elementNamespaceDefaultStack
|
50
|
+
|
51
|
+
# open element information
|
52
|
+
attr :elementName
|
53
|
+
attr :elementQName
|
54
|
+
attr :elementNamespace
|
55
|
+
attr :elementPrefix
|
56
|
+
|
57
|
+
def startDocument?
|
58
|
+
@type.equal? START_DOCUMENT
|
59
|
+
end
|
60
|
+
|
61
|
+
def endDocument?
|
62
|
+
@type.equal? END_DOCUMENT
|
63
|
+
end
|
64
|
+
|
65
|
+
def startElement?
|
66
|
+
@type.equal? START_ELEMENT
|
67
|
+
end
|
68
|
+
|
69
|
+
def endElement?
|
70
|
+
@type.equal? END_ELEMENT
|
71
|
+
end
|
72
|
+
|
73
|
+
def text?
|
74
|
+
@type.equal? TEXT
|
75
|
+
end
|
76
|
+
|
77
|
+
def cdata?
|
78
|
+
@type.equal? CDATA_SECTION
|
79
|
+
end
|
80
|
+
|
81
|
+
def entityRef?
|
82
|
+
@type.equal? ENTITY_REF
|
83
|
+
end
|
84
|
+
|
85
|
+
def ignorableWhitespace?
|
86
|
+
@type.equal? IGNORABLE_WHITESPACE
|
87
|
+
end
|
88
|
+
|
89
|
+
def processingInstruction?
|
90
|
+
@type.equal? PROCESSING_INSTRUCTION
|
91
|
+
end
|
92
|
+
|
93
|
+
def comment?
|
94
|
+
@type.equal? COMMENT
|
95
|
+
end
|
96
|
+
|
97
|
+
def doctype?
|
98
|
+
@type.equal? DOCTYPE
|
99
|
+
end
|
100
|
+
|
101
|
+
def lookFurther?
|
102
|
+
@type.equal? LOOK_FURTHER
|
103
|
+
end
|
104
|
+
|
105
|
+
#def whitespace?
|
106
|
+
#
|
107
|
+
# What is this all about -- FIX
|
108
|
+
#if text? or ignorableWhitespace? or cdata? then
|
109
|
+
#return @whitespace
|
110
|
+
#end
|
111
|
+
#raise "illegal type for whitespace query"
|
112
|
+
#end
|
113
|
+
|
114
|
+
def input=(v)
|
115
|
+
if nil == v then
|
116
|
+
@input = nil
|
117
|
+
@inputBuffer = nil
|
118
|
+
@inputBufferLength = 0
|
119
|
+
elsif v.kind_of? String then
|
120
|
+
@input = nil
|
121
|
+
@inputBuffer = v
|
122
|
+
@inputBufferLength = v.length
|
123
|
+
elsif v.kind_of? IO then
|
124
|
+
@input = v
|
125
|
+
@inputBuffer = nil
|
126
|
+
@inputBufferLength = 0
|
127
|
+
else
|
128
|
+
raise "illegalInput"
|
129
|
+
end
|
130
|
+
@nextInputBuffer = nil
|
131
|
+
@inputBufferPosition = 0
|
132
|
+
@textBuffer = ''
|
133
|
+
|
134
|
+
@line = 1
|
135
|
+
@column = 0
|
136
|
+
@compactNewLine = false
|
137
|
+
|
138
|
+
@elementNamespacePrefixStack = []
|
139
|
+
@elementNamespaceValueStack = []
|
140
|
+
@elementNamespaceDefaultStack = []
|
141
|
+
|
142
|
+
@elementName = []
|
143
|
+
@elementQName = []
|
144
|
+
@elementNamespace = []
|
145
|
+
@elementPrefix = []
|
146
|
+
|
147
|
+
@whitespace = true
|
148
|
+
@type = START_DOCUMENT
|
149
|
+
@unresolvedEntity = false
|
150
|
+
|
151
|
+
@name = nil
|
152
|
+
@namespace = nil
|
153
|
+
|
154
|
+
@attributeName = []
|
155
|
+
@attributeQName = []
|
156
|
+
@attributeNamespace = []
|
157
|
+
@attributePrefix = []
|
158
|
+
@attributeValue = []
|
159
|
+
|
160
|
+
@emptyElement = false
|
161
|
+
end
|
162
|
+
|
163
|
+
def nextEvent
|
164
|
+
begin
|
165
|
+
if (nil == @inputBuffer) and (nil == @input) then
|
166
|
+
raise "no input defined"
|
167
|
+
end
|
168
|
+
|
169
|
+
@whitespace = true
|
170
|
+
@unresolvedEntity = false
|
171
|
+
|
172
|
+
@text = nil
|
173
|
+
|
174
|
+
parseNextEvent
|
175
|
+
|
176
|
+
return @type
|
177
|
+
rescue RuntimeError => message
|
178
|
+
if nil != @inputBuffer then
|
179
|
+
message = sprintf("parse error: '%s' -- String input, line %d, column %d", message, @line, @column)
|
180
|
+
elsif nil != @input then
|
181
|
+
if @input.kind_of? File then
|
182
|
+
message = sprintf("parse error: '%s' -- file '%s', line %d, column %d", message, @input.path, @line, @column)
|
183
|
+
else
|
184
|
+
message = sprintf("parse error: '%s' -- unnamed IO stream, line %d, column %d", message, @line, @column)
|
185
|
+
end
|
186
|
+
else
|
187
|
+
message = sprintf("parse error: '%s' -- unknown source, line %d, column %d", message, @line, @column)
|
188
|
+
end
|
189
|
+
raise message
|
190
|
+
end
|
191
|
+
end
|
192
|
+
|
193
|
+
private
|
194
|
+
def initialize
|
195
|
+
self.processNamespace = true
|
196
|
+
self.reportNamespaceAttributes = false
|
197
|
+
|
198
|
+
self.input = nil
|
199
|
+
|
200
|
+
self.entityMap = {"amp"=>"&",
|
201
|
+
"apos"=>"'",
|
202
|
+
"gt"=>">",
|
203
|
+
"lt"=>"<",
|
204
|
+
"quot"=>"\""}
|
205
|
+
end
|
206
|
+
|
207
|
+
def expect(e)
|
208
|
+
c = read
|
209
|
+
if (nil == c) or (c != e) then
|
210
|
+
msg = sprintf("unexpectedChar:: expect '%s' got '%s'\n", (''<<e), (''<<c))
|
211
|
+
raise msg
|
212
|
+
end
|
213
|
+
return c
|
214
|
+
end
|
215
|
+
|
216
|
+
def read
|
217
|
+
# This is consumes the first thing in the peek buffer.
|
218
|
+
|
219
|
+
result = readFromInput
|
220
|
+
|
221
|
+
if result == ?\n then
|
222
|
+
# counting newlines for line count... not great for the mac
|
223
|
+
@line += 1
|
224
|
+
@column = 1
|
225
|
+
else
|
226
|
+
@column += 1
|
227
|
+
end
|
228
|
+
|
229
|
+
return result
|
230
|
+
end
|
231
|
+
|
232
|
+
def readFromInput
|
233
|
+
if (nil == @inputBuffer) or (@inputBufferLength <= @inputBufferPosition) then
|
234
|
+
getMoreInput
|
235
|
+
end
|
236
|
+
|
237
|
+
if nil != @inputBuffer then
|
238
|
+
c = @inputBuffer[@inputBufferPosition]
|
239
|
+
@inputBufferPosition += 1
|
240
|
+
return c
|
241
|
+
else
|
242
|
+
return nil
|
243
|
+
end
|
244
|
+
end
|
245
|
+
|
246
|
+
def getMoreInput
|
247
|
+
if nil == @input then
|
248
|
+
return nil
|
249
|
+
end
|
250
|
+
@inputBuffer = @nextInputBuffer
|
251
|
+
@inputBufferPosition = 0
|
252
|
+
if nil == @inputBuffer then
|
253
|
+
@inputBuffer = @input.gets
|
254
|
+
@inputBufferPosition = 0
|
255
|
+
if nil == @inputBuffer then
|
256
|
+
@inputBufferLength = 0
|
257
|
+
return nil
|
258
|
+
end
|
259
|
+
end
|
260
|
+
@inputBufferLength = @inputBuffer.length
|
261
|
+
@nextInputBuffer = @input.gets
|
262
|
+
end
|
263
|
+
|
264
|
+
def peekAt0
|
265
|
+
if nil == @inputBuffer then
|
266
|
+
getMoreInput
|
267
|
+
end
|
268
|
+
if @inputBufferPosition < @inputBufferLength then
|
269
|
+
return @inputBuffer[@inputBufferPosition]
|
270
|
+
else
|
271
|
+
if (nil != @nextInputBuffer) and (0 < @nextInputBuffer.length) then
|
272
|
+
return @nextInputBuffer[0]
|
273
|
+
else
|
274
|
+
return nil
|
275
|
+
end
|
276
|
+
end
|
277
|
+
end
|
278
|
+
|
279
|
+
def peekAt1
|
280
|
+
if nil == @inputBuffer then
|
281
|
+
getMoreInput
|
282
|
+
end
|
283
|
+
if (@inputBufferPosition + 1) < @inputBufferLength then
|
284
|
+
return @inputBuffer[@inputBufferPosition + 1]
|
285
|
+
else
|
286
|
+
if @inputBufferPosition < @inputBufferLength then
|
287
|
+
if (nil != @nextInputBuffer) and (0 < @nextInputBuffer.length) then
|
288
|
+
return @nextInputBuffer[0]
|
289
|
+
else
|
290
|
+
return nil
|
291
|
+
end
|
292
|
+
else
|
293
|
+
if (nil != @nextInputBuffer) and (1 < @nextInputBuffer.length) then
|
294
|
+
return @nextInputBuffer[1]
|
295
|
+
else
|
296
|
+
return nil
|
297
|
+
end
|
298
|
+
end
|
299
|
+
end
|
300
|
+
end
|
301
|
+
|
302
|
+
def parseNextEvent
|
303
|
+
@attributeName.clear
|
304
|
+
@attributeQName.clear
|
305
|
+
@attributeNamespace.clear
|
306
|
+
@attributePrefix.clear
|
307
|
+
@attributeValue.clear
|
308
|
+
|
309
|
+
if @emptyElement then
|
310
|
+
# the last event was an empty start element like <start/>
|
311
|
+
@type = END_ELEMENT
|
312
|
+
@emptyElement = false
|
313
|
+
return
|
314
|
+
end
|
315
|
+
|
316
|
+
@prefix = nil
|
317
|
+
@name = nil
|
318
|
+
@qname = nil
|
319
|
+
@namespace = nil
|
320
|
+
@type = peekType
|
321
|
+
|
322
|
+
case @type
|
323
|
+
when END_DOCUMENT
|
324
|
+
# nothing to do
|
325
|
+
when ENTITY_REF
|
326
|
+
parseEntity
|
327
|
+
@text = @textBuffer
|
328
|
+
@textBuffer = ''
|
329
|
+
when START_ELEMENT
|
330
|
+
parseStartElement
|
331
|
+
when END_ELEMENT
|
332
|
+
parseEndElement
|
333
|
+
when TEXT
|
334
|
+
parseText(?<, false)
|
335
|
+
@text = @textBuffer
|
336
|
+
@textBuffer = ''
|
337
|
+
if 0 == @elementName.length then
|
338
|
+
if(@whitespace) then
|
339
|
+
@type = IGNORABLE_WHITESPACE
|
340
|
+
end
|
341
|
+
end
|
342
|
+
when CDATA_SECTION
|
343
|
+
printf("\nCDATA... WOW I *am* used\n")
|
344
|
+
# how is this ever called?
|
345
|
+
@type = parseLookFurther
|
346
|
+
else
|
347
|
+
@type = parseLookFurther
|
348
|
+
end
|
349
|
+
|
350
|
+
end
|
351
|
+
|
352
|
+
def parseLookFurther
|
353
|
+
# this could be a comment, processing instruction, or CDATA section
|
354
|
+
expect(?<)
|
355
|
+
|
356
|
+
demand = nil
|
357
|
+
delimiter = nil # **first** (not last) character in delimting string
|
358
|
+
|
359
|
+
@text = @textBuffer
|
360
|
+
@textBuffer = ''
|
361
|
+
|
362
|
+
c = read
|
363
|
+
if ?? == c then
|
364
|
+
result = PROCESSING_INSTRUCTION
|
365
|
+
demand = nil
|
366
|
+
delimiter = ??
|
367
|
+
elsif ?! == c then
|
368
|
+
cc = peekAt0
|
369
|
+
if ?- == cc then
|
370
|
+
result = COMMENT
|
371
|
+
demand = '--'
|
372
|
+
delimiter = ?-
|
373
|
+
elsif ?[ == cc then
|
374
|
+
result = CDATA_SECTION
|
375
|
+
demand = '[CDATA['
|
376
|
+
delimiter = ?]
|
377
|
+
else
|
378
|
+
result = DOCTYPE
|
379
|
+
demand = 'DOCTYPE'
|
380
|
+
delimiter = nil
|
381
|
+
end
|
382
|
+
else
|
383
|
+
# this should never happen because we'll get an illegal name execption
|
384
|
+
# first
|
385
|
+
raise "illegal <{c}"
|
386
|
+
end
|
387
|
+
|
388
|
+
if nil != demand then
|
389
|
+
demand.each_byte do
|
390
|
+
| d |
|
391
|
+
expect d
|
392
|
+
end
|
393
|
+
end
|
394
|
+
|
395
|
+
if DOCTYPE == result then
|
396
|
+
parseDoctype
|
397
|
+
else
|
398
|
+
while true do
|
399
|
+
c = read
|
400
|
+
if nil == c then
|
401
|
+
raise "unexpectedEOF"
|
402
|
+
end
|
403
|
+
|
404
|
+
|
405
|
+
if ((?? == delimiter) or (c == delimiter)) and
|
406
|
+
(peekAt0 == delimiter) and
|
407
|
+
(peekAt1 == ?>) then
|
408
|
+
if ?? == delimiter then
|
409
|
+
@text << c
|
410
|
+
end
|
411
|
+
break
|
412
|
+
end
|
413
|
+
@text << c
|
414
|
+
end
|
415
|
+
|
416
|
+
read
|
417
|
+
read
|
418
|
+
end
|
419
|
+
|
420
|
+
return result
|
421
|
+
end
|
422
|
+
|
423
|
+
def parseDoctype
|
424
|
+
depth = 1
|
425
|
+
quoted = false
|
426
|
+
|
427
|
+
@text = ''
|
428
|
+
|
429
|
+
while true do
|
430
|
+
c = read
|
431
|
+
|
432
|
+
case c
|
433
|
+
when ?', ?" # for the sake of vim '
|
434
|
+
quoted = !quoted
|
435
|
+
when ?<
|
436
|
+
if not quoted then
|
437
|
+
depth += 1
|
438
|
+
end
|
439
|
+
when ?>
|
440
|
+
if not quoted then
|
441
|
+
depth -= 1
|
442
|
+
if 0 == depth then
|
443
|
+
return
|
444
|
+
end
|
445
|
+
end
|
446
|
+
when nil
|
447
|
+
raise "unexpectedEOF"
|
448
|
+
end
|
449
|
+
@text << c
|
450
|
+
end
|
451
|
+
end
|
452
|
+
|
453
|
+
def peekType
|
454
|
+
c = peekAt0
|
455
|
+
case c
|
456
|
+
when nil, 0
|
457
|
+
return END_DOCUMENT
|
458
|
+
when ?&
|
459
|
+
return ENTITY_REF
|
460
|
+
when ?<
|
461
|
+
case peekAt1
|
462
|
+
when ?/
|
463
|
+
return END_ELEMENT
|
464
|
+
when ?[
|
465
|
+
return CDATA_SECTION
|
466
|
+
when ??, ?!
|
467
|
+
return LOOK_FURTHER
|
468
|
+
else
|
469
|
+
return START_ELEMENT
|
470
|
+
end
|
471
|
+
else
|
472
|
+
return TEXT
|
473
|
+
end
|
474
|
+
end
|
475
|
+
|
476
|
+
def parseEntity
|
477
|
+
@compactNewLine = false
|
478
|
+
|
479
|
+
expect(?&)
|
480
|
+
|
481
|
+
@name = ''
|
482
|
+
while true do
|
483
|
+
c = read
|
484
|
+
if ?; == c then
|
485
|
+
break
|
486
|
+
end
|
487
|
+
if nil == c then
|
488
|
+
raise "unexpectedEOF"
|
489
|
+
end
|
490
|
+
@name << c
|
491
|
+
end
|
492
|
+
|
493
|
+
if ?\# == @name[0] then
|
494
|
+
if ?x == @name[1] then
|
495
|
+
c = @name[2..@name.length].hex
|
496
|
+
else
|
497
|
+
c = @name[1..@name.length].to_i
|
498
|
+
end
|
499
|
+
@textBuffer << c
|
500
|
+
@whitespace &= (c <= ?\s)
|
501
|
+
else
|
502
|
+
value = entityMap[@name]
|
503
|
+
if nil != value then
|
504
|
+
@textBuffer << value
|
505
|
+
@whitespace = false
|
506
|
+
else
|
507
|
+
@unresolvedEntity = true
|
508
|
+
end
|
509
|
+
end
|
510
|
+
end
|
511
|
+
|
512
|
+
def parseStartElement
|
513
|
+
#read the "<" that got us here
|
514
|
+
expect(?<)
|
515
|
+
|
516
|
+
@qname = readName
|
517
|
+
@textBuffer = ''
|
518
|
+
|
519
|
+
while true do
|
520
|
+
skipWhitespace
|
521
|
+
c = peekAt0
|
522
|
+
if nil == c then
|
523
|
+
raise "unexpectedEOF"
|
524
|
+
end
|
525
|
+
if ?/ == c then
|
526
|
+
@emptyElement = true
|
527
|
+
|
528
|
+
read
|
529
|
+
skipWhitespace
|
530
|
+
expect(?>)
|
531
|
+
|
532
|
+
break
|
533
|
+
end
|
534
|
+
if ?> == c then
|
535
|
+
@emptyElement = false
|
536
|
+
read
|
537
|
+
break
|
538
|
+
end
|
539
|
+
|
540
|
+
aName = readName
|
541
|
+
@textBuffer = ''
|
542
|
+
if (nil == aName) or (0 == aName.length) then
|
543
|
+
raise "nameExpected"
|
544
|
+
end
|
545
|
+
|
546
|
+
skipWhitespace
|
547
|
+
expect(?=)
|
548
|
+
skipWhitespace
|
549
|
+
|
550
|
+
delimiter = read
|
551
|
+
if (?' != delimiter) and (?" != delimiter) then # for vim: '
|
552
|
+
raise "invalidDelimiter"
|
553
|
+
end
|
554
|
+
|
555
|
+
value = parseText(delimiter, true)
|
556
|
+
@textBuffer = ''
|
557
|
+
|
558
|
+
# skip the end delimiter
|
559
|
+
read
|
560
|
+
|
561
|
+
if processNamespace then
|
562
|
+
@attributeQName.push aName
|
563
|
+
else
|
564
|
+
@attributeName.push aName
|
565
|
+
@attributeQName.push aName
|
566
|
+
@attributeNamespace.push nil
|
567
|
+
@attributePrefix.push nil
|
568
|
+
end
|
569
|
+
@attributeValue.push value
|
570
|
+
|
571
|
+
end
|
572
|
+
|
573
|
+
if processNamespace then
|
574
|
+
handleNamespaces
|
575
|
+
else
|
576
|
+
@name = @qname
|
577
|
+
@namespace = nil
|
578
|
+
@prefix = nil
|
579
|
+
end
|
580
|
+
|
581
|
+
if not @emptyElement then
|
582
|
+
@elementName.push @name
|
583
|
+
@elementQName.push @qname
|
584
|
+
@elementNamespace.push @namespace
|
585
|
+
@elementPrefix.push @prefix
|
586
|
+
end
|
587
|
+
|
588
|
+
#read
|
589
|
+
end
|
590
|
+
|
591
|
+
def parseEndElement
|
592
|
+
if 0 == @elementName.length then
|
593
|
+
raise "elementStackEmpty"
|
594
|
+
end
|
595
|
+
|
596
|
+
# read the "</" that we've only had a peek at
|
597
|
+
expect(?<)
|
598
|
+
expect(?/)
|
599
|
+
|
600
|
+
@qname = readName
|
601
|
+
startQName = @elementQName.pop
|
602
|
+
if @qname != startQName then
|
603
|
+
raise sprintf("unexpectedEndElement wanted '%s' found '%s'", startQName, @qname)
|
604
|
+
end
|
605
|
+
skipWhitespace
|
606
|
+
expect(?>)
|
607
|
+
|
608
|
+
@name = @elementName.pop
|
609
|
+
@prefix = @elementPrefix.pop
|
610
|
+
@namespace = @elementNamespace.pop
|
611
|
+
|
612
|
+
@elementNamespacePrefixStack.pop
|
613
|
+
@elementNamespaceValueStack.pop
|
614
|
+
@elementNamespaceDefaultStack.pop
|
615
|
+
end
|
616
|
+
|
617
|
+
def readName
|
618
|
+
c = peekAt0
|
619
|
+
if !nameStartChar(c) then
|
620
|
+
raise "nameExpected"
|
621
|
+
end
|
622
|
+
while true do
|
623
|
+
appendToTextBuffer(read)
|
624
|
+
c = peekAt0
|
625
|
+
if !nameChar(c) then
|
626
|
+
break
|
627
|
+
end
|
628
|
+
end
|
629
|
+
return @textBuffer
|
630
|
+
end
|
631
|
+
|
632
|
+
# is this method correct?? verify FIX ME
|
633
|
+
def nameStartChar(c)
|
634
|
+
if ((c < ?A) or (?Z < c)) and ((c < ?a) or (?z < c)) then
|
635
|
+
if (c != ?_) and (c != ?:) then
|
636
|
+
return false
|
637
|
+
end
|
638
|
+
end
|
639
|
+
return true
|
640
|
+
end
|
641
|
+
|
642
|
+
# is this method correct?? verify FIX ME
|
643
|
+
def nameChar(c)
|
644
|
+
if nil == c then return false end
|
645
|
+
if ((?A <= c) and (c <= ?Z)) then return true end
|
646
|
+
if ((?a <= c) and (c <= ?z)) then return true end
|
647
|
+
if ((?0 <= c) and (c <= ?9)) then return true end
|
648
|
+
if (?_ == c) then return true end
|
649
|
+
if (?- == c) then return true end
|
650
|
+
if (?. == c) then return true end
|
651
|
+
if (?: == c) then return true end
|
652
|
+
|
653
|
+
return false
|
654
|
+
end
|
655
|
+
|
656
|
+
def handleNamespaces
|
657
|
+
# This is called by parseStartElement to deal with namespaces. Updates knows
|
658
|
+
# name spaces based on the attributes in this start element. Then sets up
|
659
|
+
# the namespaces for this element itself (i.e. process the qname).
|
660
|
+
|
661
|
+
i = 0
|
662
|
+
|
663
|
+
defaultNamespace = @elementNamespaceDefaultStack.last
|
664
|
+
|
665
|
+
qnames = @attributeQName
|
666
|
+
@attributeQName = []
|
667
|
+
values = @attributeValue
|
668
|
+
@attributeValue = []
|
669
|
+
|
670
|
+
prefixList = []
|
671
|
+
valueList = []
|
672
|
+
|
673
|
+
while i < qnames.length do
|
674
|
+
qname = qnames[i]
|
675
|
+
value = values[i]
|
676
|
+
i += 1
|
677
|
+
|
678
|
+
if 'xmlns' == qname then
|
679
|
+
prefix = 'xmlns'
|
680
|
+
name = nil
|
681
|
+
namespace = lookupNamespace prefix
|
682
|
+
defaultNamespace = value
|
683
|
+
else
|
684
|
+
pieces = qname.split(':', 2)
|
685
|
+
if 2 == pieces.length then
|
686
|
+
namespace = value
|
687
|
+
prefix = pieces[0]
|
688
|
+
name = pieces[1]
|
689
|
+
|
690
|
+
if 0 == prefix.length then
|
691
|
+
raise "illegalEmptyAtributePrefix"
|
692
|
+
end
|
693
|
+
if 0 == name.length then
|
694
|
+
raise "illegalEmptyAttributeName"
|
695
|
+
end
|
696
|
+
else
|
697
|
+
# this is a un-prefixed non-xmlns attribute
|
698
|
+
@attributeQName.push qname
|
699
|
+
@attributeName.push qname
|
700
|
+
@attributeNamespace.push nil
|
701
|
+
@attributePrefix.push nil
|
702
|
+
@attributeValue.push value
|
703
|
+
|
704
|
+
next
|
705
|
+
end
|
706
|
+
end
|
707
|
+
|
708
|
+
# only prefixed attributes beyond here
|
709
|
+
|
710
|
+
if nil == namespace then
|
711
|
+
raise "illegalEmptyNamespace"
|
712
|
+
end
|
713
|
+
|
714
|
+
if "xmlns" != prefix then
|
715
|
+
anyQualifiedAttributes = true
|
716
|
+
|
717
|
+
@attributeQName.push qname
|
718
|
+
@attributeName.push name
|
719
|
+
@attributeNamespace.push namespace
|
720
|
+
@attributePrefix.push prefix
|
721
|
+
@attributeValue.push value
|
722
|
+
else
|
723
|
+
if (nil != name) and ((nil == namespace) or (0 == namespace.length)) then
|
724
|
+
raise "illegalNamespace"
|
725
|
+
end
|
726
|
+
|
727
|
+
prefixList.push name
|
728
|
+
valueList.push value
|
729
|
+
|
730
|
+
if @reportNamespaceAttributes then
|
731
|
+
@attributeQName.push qname
|
732
|
+
@attributeName.push name
|
733
|
+
@attributeNamespace.push namespace
|
734
|
+
@attributePrefix.push prefix
|
735
|
+
@attributeValue.push value
|
736
|
+
|
737
|
+
#why???
|
738
|
+
# anyQualifiedAttributes = true
|
739
|
+
end
|
740
|
+
end
|
741
|
+
|
742
|
+
end
|
743
|
+
|
744
|
+
@elementNamespacePrefixStack.push prefixList
|
745
|
+
@elementNamespaceValueStack.push valueList
|
746
|
+
@elementNamespaceDefaultStack.push defaultNamespace
|
747
|
+
|
748
|
+
if anyQualifiedAttributes then
|
749
|
+
# run over the attributes and make sure we have them qualified
|
750
|
+
for i in 0..(@attributeName.length-1) do
|
751
|
+
prefix = @attributePrefix[i]
|
752
|
+
|
753
|
+
if nil != prefix then
|
754
|
+
@attributeNamespace[i] = lookupNamespace prefix
|
755
|
+
end
|
756
|
+
end
|
757
|
+
end
|
758
|
+
|
759
|
+
# handle namespaces for the element name
|
760
|
+
pieces = @qname.split(':', 2)
|
761
|
+
if 2 == pieces.length then
|
762
|
+
@name = pieces[1]
|
763
|
+
@prefix = pieces[0]
|
764
|
+
@namespace = lookupNamespace @prefix
|
765
|
+
else
|
766
|
+
@name = @qname
|
767
|
+
@namespace = defaultNamespace
|
768
|
+
@prefix = nil
|
769
|
+
end
|
770
|
+
end
|
771
|
+
|
772
|
+
def lookupNamespace(prefix)
|
773
|
+
if nil == prefix then
|
774
|
+
raise "illegalPrefix"
|
775
|
+
end
|
776
|
+
if'xml' == prefix then
|
777
|
+
return 'http://www.w3.org/XML/1998/namespace'
|
778
|
+
end
|
779
|
+
if'xmlns' == prefix then
|
780
|
+
return 'http://www.w3.org/2000/xmlns/'
|
781
|
+
end
|
782
|
+
|
783
|
+
i = @elementNamespacePrefixStack.length - 1
|
784
|
+
while 0 <= i do
|
785
|
+
j = @elementNamespacePrefixStack[i].index(prefix)
|
786
|
+
if nil != j then
|
787
|
+
return @elementNamespaceValueStack[i][j]
|
788
|
+
end
|
789
|
+
|
790
|
+
i -= 1
|
791
|
+
end
|
792
|
+
raise "unknownNamespacePrefix"
|
793
|
+
end
|
794
|
+
|
795
|
+
def parseText(delimiter, resolve)
|
796
|
+
c = peekAt0
|
797
|
+
while (nil != c) and (delimiter != c) do
|
798
|
+
if ?& == c then
|
799
|
+
if !resolve then
|
800
|
+
break
|
801
|
+
end
|
802
|
+
|
803
|
+
parseEntity
|
804
|
+
|
805
|
+
if @unresolvedEntity then
|
806
|
+
raise "unresolvedEntity"
|
807
|
+
end
|
808
|
+
else
|
809
|
+
appendToTextBuffer(read)
|
810
|
+
end
|
811
|
+
|
812
|
+
c = peekAt0
|
813
|
+
end
|
814
|
+
|
815
|
+
return @textBuffer
|
816
|
+
end
|
817
|
+
|
818
|
+
def appendToTextBuffer(c)
|
819
|
+
if ((?\r == c) or (?\n == c)) and startElement? then
|
820
|
+
if (?\n == c) and @compactNewLine then
|
821
|
+
@compactNewLine = false
|
822
|
+
return
|
823
|
+
end
|
824
|
+
@compactNewLine = (?\r == c)
|
825
|
+
c = startElement? ? ?\s : ?\n
|
826
|
+
else
|
827
|
+
@compactNewLine = false
|
828
|
+
end
|
829
|
+
|
830
|
+
@textBuffer << c
|
831
|
+
@whitespace &= (c <= ?\s)
|
832
|
+
return @textBuffer
|
833
|
+
end
|
834
|
+
|
835
|
+
def skipWhitespace
|
836
|
+
while true do
|
837
|
+
c = peekAt0
|
838
|
+
if (nil == c) or (?\s < c) then
|
839
|
+
return
|
840
|
+
end
|
841
|
+
read
|
842
|
+
end
|
843
|
+
end
|
844
|
+
|
845
|
+
end
|