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
| <script>
|   "<!-- </script> -->"

#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
U+000B
#document escaped
| <html>
|   <head>
|   <body>
|     "\u000b"

#data escaped
\u000b
#errors
U+000B
#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 "" "x">
| <html>
|   <head>
|   <body>

#data
<!DOCTYPE HTML SYSTEM "x"y>
#errors
string after SYSTEM literal
not HTML5
#document
| <!DOCTYPE HTML "" "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	wakaba	1.12	\| <script>
159			\| "<!-- </script> -->"
160	wakaba	1.3
161			#data
162			<style><!-- </style> --></style>
163			#errors
164			#document-fragment p
165			\| <style>
166			\| "<!-- </style> -->"
167
168	wakaba	1.7	#data
169			<script><!----x></script>foo
170			#errors
171			no DOCTYPE
172			no --></script> before EOF
173			#document
174			\| <html>
175			\| <head>
176			\| <script>
177			\| "<!----x></script>foo"
178			\| <body>
179	wakaba	1.8
180			#data escaped
181			\u0001
182			#errors
183			no DOCTYPE
184			control character
185			#document escaped
186			\| <html>
187			\| <head>
188			\| <body>
189			\| "\u0001"
190
191			#data escaped
192			\u0001
193			#errors
194			control character
195			#document-fragment p escaped
196			\| "\u0001"
197
198			#data escaped
199			\u0002
200			#errors
201			no DOCTYPE
202			control character
203			#document escaped
204			\| <html>
205			\| <head>
206			\| <body>
207			\| "\u0002"
208
209			#data escaped
210			\u0002
211			#errors
212			control character
213			#document-fragment p escaped
214			\| "\u0002"
215
216			#data escaped
217			\u000b
218			#errors
219			no DOCTYPE
220	wakaba	1.11	U+000B
221	wakaba	1.8	#document escaped
222			\| <html>
223			\| <head>
224			\| <body>
225	wakaba	1.10	\| "\u000b"
226	wakaba	1.8
227			#data escaped
228			\u000b
229			#errors
230	wakaba	1.11	U+000B
231	wakaba	1.8	#document-fragment p escaped
232			\| "\u000b"
233
234			#data escaped
235			\u000e
236			#errors
237			no DOCTYPE
238			control character
239			#document escaped
240			\| <html>
241			\| <head>
242			\| <body>
243			\| "\u000e"
244
245			#data escaped
246			\u000e
247			#errors
248			control character
249			#document-fragment p escaped
250			\| "\u000e"
251
252			#data escaped
253			\u001f
254			#errors
255			no DOCTYPE
256			control character
257			#document escaped
258			\| <html>
259			\| <head>
260			\| <body>
261			\| "\u001f"
262
263			#data escaped
264			\u001f
265			#errors
266			control character
267			#document-fragment p escaped
268			\| "\u001f"
269
270			#data escaped
271			\u007f
272			#errors
273			no DOCTYPE
274			control character
275			#document escaped
276			\| <html>
277			\| <head>
278			\| <body>
279			\| "\u007f"
280
281			#data escaped
282			\u007f
283			#errors
284			control character
285			#document-fragment p escaped
286			\| "\u007f"
287
288			#data escaped
289			\u0080
290			#errors
291			no DOCTYPE
292			control character
293			#document escaped
294			\| <html>
295			\| <head>
296			\| <body>
297			\| "\u0080"
298
299			#data escaped
300			\u0080
301			#errors
302			control character
303			#document-fragment p escaped
304			\| "\u0080"
305
306			#data escaped
307			\u0085
308			#errors
309			no DOCTYPE
310			control character
311			#document escaped
312			\| <html>
313			\| <head>
314			\| <body>
315			\| "\u0085"
316
317			#data escaped
318			\u0085
319			#errors
320			control character
321			#document-fragment p escaped
322			\| "\u0085"
323
324			#data escaped
325			\u009f
326			#errors
327			no DOCTYPE
328			control character
329			#document escaped
330			\| <html>
331			\| <head>
332			\| <body>
333			\| "\u009f"
334
335			#data escaped
336			\u009f
337			#errors
338			control character
339			#document-fragment p escaped
340			\| "\u009f"
341
342			#data escaped
343			\ud800
344			#errors
345			no DOCTYPE
346			control character
347			#document escaped
348			\| <html>
349			\| <head>
350			\| <body>
351			\| "\ud800"
352
353			#data escaped
354			\ud800
355			#errors
356			control character
357			#document-fragment p escaped
358			\| "\ud800"
359
360			#data escaped
361			\udfff
362			#errors
363			no DOCTYPE
364			control character
365			#document escaped
366			\| <html>
367			\| <head>
368			\| <body>
369			\| "\udfff"
370
371			#data escaped
372			\udfff
373			#errors
374			control character
375			#document-fragment p escaped
376			\| "\udfff"
377
378			#data escaped
379			\ufdd0
380			#errors
381			no DOCTYPE
382			control character
383			#document escaped
384			\| <html>
385			\| <head>
386			\| <body>
387			\| "\ufdd0"
388
389			#data escaped
390			\ufdd0
391			#errors
392			control character
393			#document-fragment p escaped
394			\| "\ufdd0"
395
396			#data escaped
397			\ufddf
398			#errors
399			no DOCTYPE
400			control character
401			#document escaped
402			\| <html>
403			\| <head>
404			\| <body>
405			\| "\ufddf"
406
407			#data escaped
408			\ufddf
409			#errors
410			control character
411			#document-fragment p escaped
412			\| "\ufddf"
413
414			#data escaped
415			\ufffe
416			#errors
417			no DOCTYPE
418			control character
419			#document escaped
420			\| <html>
421			\| <head>
422			\| <body>
423			\| "\ufffe"
424
425			#data escaped
426			\ufffe
427			#errors
428			control character
429			#document-fragment p escaped
430			\| "\ufffe"
431
432			#data escaped
433			\uffff
434			#errors
435			no DOCTYPE
436			control character
437			#document escaped
438			\| <html>
439			\| <head>
440			\| <body>
441			\| "\uffff"
442
443			#data escaped
444			\uffff
445			#errors
446			control character
447			#document-fragment p escaped
448			\| "\uffff"
449
450			#data escaped
451			\U0010ffff
452			#errors
453			no DOCTYPE
454			control character
455			#document escaped
456			\| <html>
457			\| <head>
458			\| <body>
459			\| "\U0010ffff"
460
461			#data escaped
462			\U0010ffff
463			#errors
464			control character
465			#document-fragment p escaped
466			\| "\U0010ffff"
467
468	wakaba	1.9	#data
469			<!DOCTYPE HTML SYSTEM "x"
470			#errors
471			not HTML5
472			#document
473	wakaba	1.13	\| <!DOCTYPE HTML "" "x">
474	wakaba	1.9	\| <html>
475			\| <head>
476			\| <body>
477
478			#data
479			<!DOCTYPE HTML SYSTEM "x"y>
480			#errors
481			string after SYSTEM literal
482			not HTML5
483			#document
484	wakaba	1.13	\| <!DOCTYPE HTML "" "x">
485	wakaba	1.9	\| <html>
486			\| <head>
487			\| <body>
488