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>

#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
223			#data escaped
224			\u000b
225			#errors
226			#document-fragment p escaped
227			\| "\u000b"
228
229			#data escaped
230			\u000e
231			#errors
232			no DOCTYPE
233			control character
234			#document escaped
235			\| <html>
236			\| <head>
237			\| <body>
238			\| "\u000e"
239
240			#data escaped
241			\u000e
242			#errors
243			control character
244			#document-fragment p escaped
245			\| "\u000e"
246
247			#data escaped
248			\u001f
249			#errors
250			no DOCTYPE
251			control character
252			#document escaped
253			\| <html>
254			\| <head>
255			\| <body>
256			\| "\u001f"
257
258			#data escaped
259			\u001f
260			#errors
261			control character
262			#document-fragment p escaped
263			\| "\u001f"
264
265			#data escaped
266			\u007f
267			#errors
268			no DOCTYPE
269			control character
270			#document escaped
271			\| <html>
272			\| <head>
273			\| <body>
274			\| "\u007f"
275
276			#data escaped
277			\u007f
278			#errors
279			control character
280			#document-fragment p escaped
281			\| "\u007f"
282
283			#data escaped
284			\u0080
285			#errors
286			no DOCTYPE
287			control character
288			#document escaped
289			\| <html>
290			\| <head>
291			\| <body>
292			\| "\u0080"
293
294			#data escaped
295			\u0080
296			#errors
297			control character
298			#document-fragment p escaped
299			\| "\u0080"
300
301			#data escaped
302			\u0085
303			#errors
304			no DOCTYPE
305			control character
306			#document escaped
307			\| <html>
308			\| <head>
309			\| <body>
310			\| "\u0085"
311
312			#data escaped
313			\u0085
314			#errors
315			control character
316			#document-fragment p escaped
317			\| "\u0085"
318
319			#data escaped
320			\u009f
321			#errors
322			no DOCTYPE
323			control character
324			#document escaped
325			\| <html>
326			\| <head>
327			\| <body>
328			\| "\u009f"
329
330			#data escaped
331			\u009f
332			#errors
333			control character
334			#document-fragment p escaped
335			\| "\u009f"
336
337			#data escaped
338			\ud800
339			#errors
340			no DOCTYPE
341			control character
342			#document escaped
343			\| <html>
344			\| <head>
345			\| <body>
346			\| "\ud800"
347
348			#data escaped
349			\ud800
350			#errors
351			control character
352			#document-fragment p escaped
353			\| "\ud800"
354
355			#data escaped
356			\udfff
357			#errors
358			no DOCTYPE
359			control character
360			#document escaped
361			\| <html>
362			\| <head>
363			\| <body>
364			\| "\udfff"
365
366			#data escaped
367			\udfff
368			#errors
369			control character
370			#document-fragment p escaped
371			\| "\udfff"
372
373			#data escaped
374			\ufdd0
375			#errors
376			no DOCTYPE
377			control character
378			#document escaped
379			\| <html>
380			\| <head>
381			\| <body>
382			\| "\ufdd0"
383
384			#data escaped
385			\ufdd0
386			#errors
387			control character
388			#document-fragment p escaped
389			\| "\ufdd0"
390
391			#data escaped
392			\ufddf
393			#errors
394			no DOCTYPE
395			control character
396			#document escaped
397			\| <html>
398			\| <head>
399			\| <body>
400			\| "\ufddf"
401
402			#data escaped
403			\ufddf
404			#errors
405			control character
406			#document-fragment p escaped
407			\| "\ufddf"
408
409			#data escaped
410			\ufffe
411			#errors
412			no DOCTYPE
413			control character
414			#document escaped
415			\| <html>
416			\| <head>
417			\| <body>
418			\| "\ufffe"
419
420			#data escaped
421			\ufffe
422			#errors
423			control character
424			#document-fragment p escaped
425			\| "\ufffe"
426
427			#data escaped
428			\uffff
429			#errors
430			no DOCTYPE
431			control character
432			#document escaped
433			\| <html>
434			\| <head>
435			\| <body>
436			\| "\uffff"
437
438			#data escaped
439			\uffff
440			#errors
441			control character
442			#document-fragment p escaped
443			\| "\uffff"
444
445			#data escaped
446			\U0010ffff
447			#errors
448			no DOCTYPE
449			control character
450			#document escaped
451			\| <html>
452			\| <head>
453			\| <body>
454			\| "\U0010ffff"
455
456			#data escaped
457			\U0010ffff
458			#errors
459			control character
460			#document-fragment p escaped
461			\| "\U0010ffff"
462
463	wakaba	1.9	#data
464			<!DOCTYPE HTML SYSTEM "x"
465			#errors
466			not HTML5
467			#document
468			\| <!DOCTYPE HTML SYSTEM "x">
469			\| <html>
470			\| <head>
471			\| <body>
472
473			#data
474			<!DOCTYPE HTML SYSTEM "x"y>
475			#errors
476			string after SYSTEM literal
477			not HTML5
478			#document
479			\| <!DOCTYPE HTML SYSTEM "x">
480			\| <html>
481			\| <head>
482			\| <body>
483