whatpm/t/tokenizer-test-2.dat

#data escaped
<!DOCTYPE HTML><html><head><title></title></head><body><p>\u0000</p></body>
#errors
NULL
#document escaped
| <!DOCTYPE HTML>
| <html>
|   <head>
|     <title>
|   <body>
|     <p>
|       "\uFFFD"

#data escaped
\u0000
#errors
NULL
#document-fragment p escaped
| "\uFFFD"

#data escaped
<!DOCTYPE HTML><p>\u000D
#errors
#document escaped
| <!DOCTYPE HTML>
| <html>
|   <head>
|   <body>
|     <p>
|       "\u000A"

#data escaped
<!DOCTYPE HTML><p>\u000D</p>
#errors
#document escaped
| <!DOCTYPE HTML>
| <html>
|   <head>
|   <body>
|     <p>
|       "\u000A"

#data escaped
<!DOCTYPE HTML><p>\u000D\u000D
#errors
#document escaped
| <!DOCTYPE HTML>
| <html>
|   <head>
|   <body>
|     <p>
|       "\u000A\u000A"

#data escaped
<!DOCTYPE HTML><p>\u000D\u0000
#errors
NULL
#document escaped
| <!DOCTYPE HTML>
| <html>
|   <head>
|   <body>
|     <p>
|       "\u000A\uFFFD"

#data escaped
\u000D
#errors
#document-fragment p escaped
| "\u000A"

#data escaped
\u000D\u000D
#errors
#document-fragment p escaped
| "\u000A\u000A"

#data escaped
\u000Da
#errors
#document-fragment p escaped
| "\u000Aa"

#data escaped
<!DOCTYPE HTML><p>\u000D\u000A</p>
#errors
#document escaped
| <!DOCTYPE HTML>
| <html>
|   <head>
|   <body>
|     <p>
|       "\u000A"

#data escaped
\u000D\u000A
#errors
#document-fragment p escaped
| "\u000A"

#data escaped
<!DOCTYPE HTML><p>\u000A</p>
#errors
#document escaped
| <!DOCTYPE HTML>
| <html>
|   <head>
|   <body>
|     <p>
|       "\u000A"

#data escaped
<!DOCTYPE HTML><p>\u000D\u0000</p>
#errors
NULL
#document escaped
| <!DOCTYPE HTML>
| <html>
|   <head>
|   <body>
|     <p>
|       "\u000A\uFFFD"

#data escaped
<!DOCTYPE HTML><p>\u000D\u000D</p>
#errors
#document escaped
| <!DOCTYPE HTML>
| <html>
|   <head>
|   <body>
|     <p>
|       "\u000A\u000A"

#data escaped
\u000A
#errors
#document-fragment p escaped
| "\u000A"

#data escaped
\u000D\u0000
#errors
NULL
#document-fragment p escaped
| "\u000A\uFFFD"

#data escaped
\u000D\u000D
#errors
#document-fragment p escaped
| "\u000A\u000A"

#data
<script><!-- </script> --></script>
#errors
#document-fragment p

#data
<style><!-- </style> --></style>
#errors
#document-fragment p
| <style>
|   "<!-- </style> -->"

#data
<script><!----x></script>foo
#errors
no DOCTYPE
no --></script> before EOF
#document
| <html>
|   <head>
|     <script>
|       "<!----x></script>foo"
|   <body>

#data escaped
\u0001
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\u0001"

#data escaped
\u0001
#errors
control character
#document-fragment p escaped
| "\u0001"

#data escaped
\u0002
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\u0002"

#data escaped
\u0002
#errors
control character
#document-fragment p escaped
| "\u0002"

#data escaped
\u000b
#errors
no DOCTYPE
#document escaped
| <html>
|   <head>
|   <body>
|     "\u000b"

#data escaped
\u000b
#errors
#document-fragment p escaped
| "\u000b"

#data escaped
\u000e
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\u000e"

#data escaped
\u000e
#errors
control character
#document-fragment p escaped
| "\u000e"

#data escaped
\u001f
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\u001f"

#data escaped
\u001f
#errors
control character
#document-fragment p escaped
| "\u001f"

#data escaped
\u007f
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\u007f"

#data escaped
\u007f
#errors
control character
#document-fragment p escaped
| "\u007f"

#data escaped
\u0080
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\u0080"

#data escaped
\u0080
#errors
control character
#document-fragment p escaped
| "\u0080"

#data escaped
\u0085
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\u0085"

#data escaped
\u0085
#errors
control character
#document-fragment p escaped
| "\u0085"

#data escaped
\u009f
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\u009f"

#data escaped
\u009f
#errors
control character
#document-fragment p escaped
| "\u009f"

#data escaped
\ud800
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\ud800"

#data escaped
\ud800
#errors
control character
#document-fragment p escaped
| "\ud800"

#data escaped
\udfff
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\udfff"

#data escaped
\udfff
#errors
control character
#document-fragment p escaped
| "\udfff"

#data escaped
\ufdd0
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\ufdd0"

#data escaped
\ufdd0
#errors
control character
#document-fragment p escaped
| "\ufdd0"

#data escaped
\ufddf
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\ufddf"

#data escaped
\ufddf
#errors
control character
#document-fragment p escaped
| "\ufddf"

#data escaped
\ufffe
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\ufffe"

#data escaped
\ufffe
#errors
control character
#document-fragment p escaped
| "\ufffe"

#data escaped
\uffff
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\uffff"

#data escaped
\uffff
#errors
control character
#document-fragment p escaped
| "\uffff"

#data escaped
\U0010ffff
#errors
no DOCTYPE
control character
#document escaped
| <html>
|   <head>
|   <body>
|     "\U0010ffff"

#data escaped
\U0010ffff
#errors
control character
#document-fragment p escaped
| "\U0010ffff"

#data
<!DOCTYPE HTML SYSTEM "x"
#errors
not HTML5
#document
| <!DOCTYPE HTML SYSTEM "x">
| <html>
|   <head>
|   <body>

#data
<!DOCTYPE HTML SYSTEM "x"y>
#errors
string after SYSTEM literal
not HTML5
#document
| <!DOCTYPE HTML SYSTEM "x">
| <html>
|   <head>
|   <body>

1	wakaba	1.1	#data escaped
2	wakaba	1.2	<!DOCTYPE HTML><html><head><title></title></head><body><p>\u0000</p></body>
3	wakaba	1.1	#errors
4			NULL
5			#document escaped
6	wakaba	1.2	\| <!DOCTYPE HTML>
7	wakaba	1.1	\| <html>
8			\| <head>
9			\| <title>
10			\| <body>
11			\| <p>
12			\| "\uFFFD"
13
14	wakaba	1.3	#data escaped
15			\u0000
16			#errors
17			NULL
18			#document-fragment p escaped
19			\| "\uFFFD"
20
21			#data escaped
22			<!DOCTYPE HTML><p>\u000D
23			#errors
24			#document escaped
25			\| <!DOCTYPE HTML>
26			\| <html>
27			\| <head>
28			\| <body>
29			\| <p>
30			\| "\u000A"
31
32			#data escaped
33			<!DOCTYPE HTML><p>\u000D</p>
34			#errors
35			#document escaped
36			\| <!DOCTYPE HTML>
37			\| <html>
38			\| <head>
39			\| <body>
40			\| <p>
41			\| "\u000A"
42
43			#data escaped
44	wakaba	1.5	<!DOCTYPE HTML><p>\u000D\u000D
45	wakaba	1.4	#errors
46			#document escaped
47			\| <!DOCTYPE HTML>
48			\| <html>
49			\| <head>
50			\| <body>
51			\| <p>
52			\| "\u000A\u000A"
53
54			#data escaped
55	wakaba	1.6	<!DOCTYPE HTML><p>\u000D\u0000
56			#errors
57			NULL
58			#document escaped
59			\| <!DOCTYPE HTML>
60			\| <html>
61			\| <head>
62			\| <body>
63			\| <p>
64			\| "\u000A\uFFFD"
65
66			#data escaped
67	wakaba	1.3	\u000D
68			#errors
69			#document-fragment p escaped
70			\| "\u000A"
71	wakaba	1.4
72			#data escaped
73			\u000D\u000D
74			#errors
75			#document-fragment p escaped
76			\| "\u000A\u000A"
77	wakaba	1.3
78			#data escaped
79			\u000Da
80			#errors
81			#document-fragment p escaped
82			\| "\u000Aa"
83
84			#data escaped
85			<!DOCTYPE HTML><p>\u000D\u000A</p>
86			#errors
87			#document escaped
88			\| <!DOCTYPE HTML>
89			\| <html>
90			\| <head>
91			\| <body>
92			\| <p>
93			\| "\u000A"
94
95			#data escaped
96			\u000D\u000A
97			#errors
98			#document-fragment p escaped
99			\| "\u000A"
100
101			#data escaped
102			<!DOCTYPE HTML><p>\u000A</p>
103			#errors
104			#document escaped
105			\| <!DOCTYPE HTML>
106			\| <html>
107			\| <head>
108			\| <body>
109			\| <p>
110			\| "\u000A"
111
112			#data escaped
113	wakaba	1.6	<!DOCTYPE HTML><p>\u000D\u0000</p>
114			#errors
115			NULL
116			#document escaped
117			\| <!DOCTYPE HTML>
118			\| <html>
119			\| <head>
120			\| <body>
121			\| <p>
122			\| "\u000A\uFFFD"
123
124			#data escaped
125			<!DOCTYPE HTML><p>\u000D\u000D</p>
126			#errors
127			#document escaped
128			\| <!DOCTYPE HTML>
129			\| <html>
130			\| <head>
131			\| <body>
132			\| <p>
133			\| "\u000A\u000A"
134
135			#data escaped
136	wakaba	1.3	\u000A
137			#errors
138			#document-fragment p escaped
139			\| "\u000A"
140	wakaba	1.6
141			#data escaped
142			\u000D\u0000
143			#errors
144			NULL
145			#document-fragment p escaped
146			\| "\u000A\uFFFD"
147
148			#data escaped
149			\u000D\u000D
150			#errors
151			#document-fragment p escaped
152			\| "\u000A\u000A"
153	wakaba	1.3
154			#data
155			<script><!-- </script> --></script>
156			#errors
157			#document-fragment p
158
159			#data
160			<style><!-- </style> --></style>
161			#errors
162			#document-fragment p
163			\| <style>
164			\| "<!-- </style> -->"
165
166	wakaba	1.7	#data
167			<script><!----x></script>foo
168			#errors
169			no DOCTYPE
170			no --></script> before EOF
171			#document
172			\| <html>
173			\| <head>
174			\| <script>
175			\| "<!----x></script>foo"
176			\| <body>
177	wakaba	1.8
178			#data escaped
179			\u0001
180			#errors
181			no DOCTYPE
182			control character
183			#document escaped
184			\| <html>
185			\| <head>
186			\| <body>
187			\| "\u0001"
188
189			#data escaped
190			\u0001
191			#errors
192			control character
193			#document-fragment p escaped
194			\| "\u0001"
195
196			#data escaped
197			\u0002
198			#errors
199			no DOCTYPE
200			control character
201			#document escaped
202			\| <html>
203			\| <head>
204			\| <body>
205			\| "\u0002"
206
207			#data escaped
208			\u0002
209			#errors
210			control character
211			#document-fragment p escaped
212			\| "\u0002"
213
214			#data escaped
215			\u000b
216			#errors
217			no DOCTYPE
218			#document escaped
219			\| <html>
220			\| <head>
221			\| <body>
222	wakaba	1.10	\| "\u000b"
223	wakaba	1.8
224			#data escaped
225			\u000b
226			#errors
227			#document-fragment p escaped
228			\| "\u000b"
229
230			#data escaped
231			\u000e
232			#errors
233			no DOCTYPE
234			control character
235			#document escaped
236			\| <html>
237			\| <head>
238			\| <body>
239			\| "\u000e"
240
241			#data escaped
242			\u000e
243			#errors
244			control character
245			#document-fragment p escaped
246			\| "\u000e"
247
248			#data escaped
249			\u001f
250			#errors
251			no DOCTYPE
252			control character
253			#document escaped
254			\| <html>
255			\| <head>
256			\| <body>
257			\| "\u001f"
258
259			#data escaped
260			\u001f
261			#errors
262			control character
263			#document-fragment p escaped
264			\| "\u001f"
265
266			#data escaped
267			\u007f
268			#errors
269			no DOCTYPE
270			control character
271			#document escaped
272			\| <html>
273			\| <head>
274			\| <body>
275			\| "\u007f"
276
277			#data escaped
278			\u007f
279			#errors
280			control character
281			#document-fragment p escaped
282			\| "\u007f"
283
284			#data escaped
285			\u0080
286			#errors
287			no DOCTYPE
288			control character
289			#document escaped
290			\| <html>
291			\| <head>
292			\| <body>
293			\| "\u0080"
294
295			#data escaped
296			\u0080
297			#errors
298			control character
299			#document-fragment p escaped
300			\| "\u0080"
301
302			#data escaped
303			\u0085
304			#errors
305			no DOCTYPE
306			control character
307			#document escaped
308			\| <html>
309			\| <head>
310			\| <body>
311			\| "\u0085"
312
313			#data escaped
314			\u0085
315			#errors
316			control character
317			#document-fragment p escaped
318			\| "\u0085"
319
320			#data escaped
321			\u009f
322			#errors
323			no DOCTYPE
324			control character
325			#document escaped
326			\| <html>
327			\| <head>
328			\| <body>
329			\| "\u009f"
330
331			#data escaped
332			\u009f
333			#errors
334			control character
335			#document-fragment p escaped
336			\| "\u009f"
337
338			#data escaped
339			\ud800
340			#errors
341			no DOCTYPE
342			control character
343			#document escaped
344			\| <html>
345			\| <head>
346			\| <body>
347			\| "\ud800"
348
349			#data escaped
350			\ud800
351			#errors
352			control character
353			#document-fragment p escaped
354			\| "\ud800"
355
356			#data escaped
357			\udfff
358			#errors
359			no DOCTYPE
360			control character
361			#document escaped
362			\| <html>
363			\| <head>
364			\| <body>
365			\| "\udfff"
366
367			#data escaped
368			\udfff
369			#errors
370			control character
371			#document-fragment p escaped
372			\| "\udfff"
373
374			#data escaped
375			\ufdd0
376			#errors
377			no DOCTYPE
378			control character
379			#document escaped
380			\| <html>
381			\| <head>
382			\| <body>
383			\| "\ufdd0"
384
385			#data escaped
386			\ufdd0
387			#errors
388			control character
389			#document-fragment p escaped
390			\| "\ufdd0"
391
392			#data escaped
393			\ufddf
394			#errors
395			no DOCTYPE
396			control character
397			#document escaped
398			\| <html>
399			\| <head>
400			\| <body>
401			\| "\ufddf"
402
403			#data escaped
404			\ufddf
405			#errors
406			control character
407			#document-fragment p escaped
408			\| "\ufddf"
409
410			#data escaped
411			\ufffe
412			#errors
413			no DOCTYPE
414			control character
415			#document escaped
416			\| <html>
417			\| <head>
418			\| <body>
419			\| "\ufffe"
420
421			#data escaped
422			\ufffe
423			#errors
424			control character
425			#document-fragment p escaped
426			\| "\ufffe"
427
428			#data escaped
429			\uffff
430			#errors
431			no DOCTYPE
432			control character
433			#document escaped
434			\| <html>
435			\| <head>
436			\| <body>
437			\| "\uffff"
438
439			#data escaped
440			\uffff
441			#errors
442			control character
443			#document-fragment p escaped
444			\| "\uffff"
445
446			#data escaped
447			\U0010ffff
448			#errors
449			no DOCTYPE
450			control character
451			#document escaped
452			\| <html>
453			\| <head>
454			\| <body>
455			\| "\U0010ffff"
456
457			#data escaped
458			\U0010ffff
459			#errors
460			control character
461			#document-fragment p escaped
462			\| "\U0010ffff"
463
464	wakaba	1.9	#data
465			<!DOCTYPE HTML SYSTEM "x"
466			#errors
467			not HTML5
468			#document
469			\| <!DOCTYPE HTML SYSTEM "x">
470			\| <html>
471			\| <head>
472			\| <body>
473
474			#data
475			<!DOCTYPE HTML SYSTEM "x"y>
476			#errors
477			string after SYSTEM literal
478			not HTML5
479			#document
480			\| <!DOCTYPE HTML SYSTEM "x">
481			\| <html>
482			\| <head>
483			\| <body>
484