tidy-ext 0.1.7

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (65) hide show
  1. data/.gitignore +4 -0
  2. data/LICENSE +50 -0
  3. data/README +12 -0
  4. data/Rakefile +60 -0
  5. data/VERSION +1 -0
  6. data/ext/tidy/access.c +3310 -0
  7. data/ext/tidy/access.h +279 -0
  8. data/ext/tidy/alloc.c +107 -0
  9. data/ext/tidy/attrask.c +209 -0
  10. data/ext/tidy/attrdict.c +2398 -0
  11. data/ext/tidy/attrdict.h +122 -0
  12. data/ext/tidy/attrget.c +213 -0
  13. data/ext/tidy/attrs.c +1911 -0
  14. data/ext/tidy/attrs.h +374 -0
  15. data/ext/tidy/buffio.c +232 -0
  16. data/ext/tidy/buffio.h +118 -0
  17. data/ext/tidy/charsets.c +1032 -0
  18. data/ext/tidy/charsets.h +14 -0
  19. data/ext/tidy/clean.c +2674 -0
  20. data/ext/tidy/clean.h +87 -0
  21. data/ext/tidy/config.c +1746 -0
  22. data/ext/tidy/config.h +153 -0
  23. data/ext/tidy/entities.c +419 -0
  24. data/ext/tidy/entities.h +24 -0
  25. data/ext/tidy/extconf.rb +5 -0
  26. data/ext/tidy/fileio.c +106 -0
  27. data/ext/tidy/fileio.h +46 -0
  28. data/ext/tidy/forward.h +69 -0
  29. data/ext/tidy/iconvtc.c +105 -0
  30. data/ext/tidy/iconvtc.h +15 -0
  31. data/ext/tidy/istack.c +373 -0
  32. data/ext/tidy/lexer.c +3825 -0
  33. data/ext/tidy/lexer.h +617 -0
  34. data/ext/tidy/localize.c +1882 -0
  35. data/ext/tidy/mappedio.c +329 -0
  36. data/ext/tidy/mappedio.h +16 -0
  37. data/ext/tidy/message.h +207 -0
  38. data/ext/tidy/parser.c +4408 -0
  39. data/ext/tidy/parser.h +76 -0
  40. data/ext/tidy/platform.h +636 -0
  41. data/ext/tidy/pprint.c +2276 -0
  42. data/ext/tidy/pprint.h +93 -0
  43. data/ext/tidy/ruby-tidy.c +195 -0
  44. data/ext/tidy/streamio.c +1407 -0
  45. data/ext/tidy/streamio.h +222 -0
  46. data/ext/tidy/tagask.c +286 -0
  47. data/ext/tidy/tags.c +955 -0
  48. data/ext/tidy/tags.h +235 -0
  49. data/ext/tidy/tidy-int.h +129 -0
  50. data/ext/tidy/tidy.h +1097 -0
  51. data/ext/tidy/tidyenum.h +622 -0
  52. data/ext/tidy/tidylib.c +1751 -0
  53. data/ext/tidy/tmbstr.c +306 -0
  54. data/ext/tidy/tmbstr.h +92 -0
  55. data/ext/tidy/utf8.c +539 -0
  56. data/ext/tidy/utf8.h +52 -0
  57. data/ext/tidy/version.h +14 -0
  58. data/ext/tidy/win32tc.c +795 -0
  59. data/ext/tidy/win32tc.h +19 -0
  60. data/spec/spec_helper.rb +5 -0
  61. data/spec/tidy/compat_spec.rb +44 -0
  62. data/spec/tidy/remote_uri_spec.rb +14 -0
  63. data/spec/tidy/test1.html +5 -0
  64. data/spec/tidy/tidy_spec.rb +34 -0
  65. metadata +125 -0
data/ext/tidy/tags.c ADDED
@@ -0,0 +1,955 @@
1
+ /* tags.c -- recognize HTML tags
2
+
3
+ (c) 1998-2008 (W3C) MIT, ERCIM, Keio University
4
+ See tidy.h for the copyright notice.
5
+
6
+ CVS Info :
7
+
8
+ $Author: hoehrmann $
9
+ $Date: 2008/08/09 11:55:27 $
10
+ $Revision: 1.71 $
11
+
12
+ The HTML tags are stored as 8 bit ASCII strings.
13
+
14
+ */
15
+
16
+ #include "tidy-int.h"
17
+ #include "message.h"
18
+ #include "tmbstr.h"
19
+
20
+ /* Attribute checking methods */
21
+ static CheckAttribs CheckIMG;
22
+ static CheckAttribs CheckLINK;
23
+ static CheckAttribs CheckAREA;
24
+ static CheckAttribs CheckTABLE;
25
+ static CheckAttribs CheckCaption;
26
+ static CheckAttribs CheckSCRIPT;
27
+ static CheckAttribs CheckSTYLE;
28
+ static CheckAttribs CheckHTML;
29
+ static CheckAttribs CheckFORM;
30
+ static CheckAttribs CheckMETA;
31
+
32
+ #define VERS_ELEM_A (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
33
+ #define VERS_ELEM_ABBR (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
34
+ #define VERS_ELEM_ACRONYM (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
35
+ #define VERS_ELEM_ADDRESS (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
36
+ #define VERS_ELEM_APPLET (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|xxxx|xxxx|xxxx|xxxx)
37
+ #define VERS_ELEM_AREA (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
38
+ #define VERS_ELEM_B (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
39
+ #define VERS_ELEM_BASE (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
40
+ #define VERS_ELEM_BASEFONT (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|xxxx|xxxx|xxxx|xxxx)
41
+ #define VERS_ELEM_BDO (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
42
+ #define VERS_ELEM_BIG (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
43
+ #define VERS_ELEM_BLOCKQUOTE (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
44
+ #define VERS_ELEM_BODY (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
45
+ #define VERS_ELEM_BR (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
46
+ #define VERS_ELEM_BUTTON (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
47
+ #define VERS_ELEM_CAPTION (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
48
+ #define VERS_ELEM_CENTER (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|xxxx|xxxx|xxxx|xxxx)
49
+ #define VERS_ELEM_CITE (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
50
+ #define VERS_ELEM_CODE (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
51
+ #define VERS_ELEM_COL (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
52
+ #define VERS_ELEM_COLGROUP (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
53
+ #define VERS_ELEM_DD (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
54
+ #define VERS_ELEM_DEL (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
55
+ #define VERS_ELEM_DFN (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
56
+ #define VERS_ELEM_DIR (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|xxxx|xxxx|xxxx|xxxx)
57
+ #define VERS_ELEM_DIV (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
58
+ #define VERS_ELEM_DL (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
59
+ #define VERS_ELEM_DT (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
60
+ #define VERS_ELEM_EM (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
61
+ #define VERS_ELEM_FIELDSET (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
62
+ #define VERS_ELEM_FONT (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|xxxx|xxxx|xxxx|xxxx)
63
+ #define VERS_ELEM_FORM (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
64
+ #define VERS_ELEM_FRAME (xxxx|xxxx|xxxx|xxxx|xxxx|H40F|H41F|X10F|xxxx|xxxx|xxxx|xxxx|xxxx)
65
+ #define VERS_ELEM_FRAMESET (xxxx|xxxx|xxxx|xxxx|xxxx|H40F|H41F|X10F|xxxx|xxxx|xxxx|xxxx|xxxx)
66
+ #define VERS_ELEM_H1 (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
67
+ #define VERS_ELEM_H2 (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
68
+ #define VERS_ELEM_H3 (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
69
+ #define VERS_ELEM_H4 (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
70
+ #define VERS_ELEM_H5 (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
71
+ #define VERS_ELEM_H6 (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
72
+ #define VERS_ELEM_HEAD (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
73
+ #define VERS_ELEM_HR (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
74
+ #define VERS_ELEM_HTML (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
75
+ #define VERS_ELEM_I (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
76
+ #define VERS_ELEM_IFRAME (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|xxxx|xxxx|xxxx|xxxx)
77
+ #define VERS_ELEM_IMG (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
78
+ #define VERS_ELEM_INPUT (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
79
+ #define VERS_ELEM_INS (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
80
+ #define VERS_ELEM_ISINDEX (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|xxxx|xxxx|xxxx|xxxx)
81
+ #define VERS_ELEM_KBD (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
82
+ #define VERS_ELEM_LABEL (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
83
+ #define VERS_ELEM_LEGEND (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
84
+ #define VERS_ELEM_LI (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
85
+ #define VERS_ELEM_LINK (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
86
+ #define VERS_ELEM_LISTING (HT20|HT32|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx)
87
+ #define VERS_ELEM_MAP (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
88
+ #define VERS_ELEM_MENU (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|xxxx|xxxx|xxxx|xxxx)
89
+ #define VERS_ELEM_META (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
90
+ #define VERS_ELEM_NEXTID (HT20|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx)
91
+ #define VERS_ELEM_NOFRAMES (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|xxxx|xxxx|xxxx|xxxx)
92
+ #define VERS_ELEM_NOSCRIPT (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
93
+ #define VERS_ELEM_OBJECT (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
94
+ #define VERS_ELEM_OL (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
95
+ #define VERS_ELEM_OPTGROUP (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
96
+ #define VERS_ELEM_OPTION (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
97
+ #define VERS_ELEM_P (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
98
+ #define VERS_ELEM_PARAM (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
99
+ #define VERS_ELEM_PLAINTEXT (HT20|HT32|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx)
100
+ #define VERS_ELEM_PRE (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
101
+ #define VERS_ELEM_Q (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
102
+ #define VERS_ELEM_RB (xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|XH11|xxxx)
103
+ #define VERS_ELEM_RBC (xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|XH11|xxxx)
104
+ #define VERS_ELEM_RP (xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|XH11|xxxx)
105
+ #define VERS_ELEM_RT (xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|XH11|xxxx)
106
+ #define VERS_ELEM_RTC (xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|XH11|xxxx)
107
+ #define VERS_ELEM_RUBY (xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|XH11|xxxx)
108
+ #define VERS_ELEM_S (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|xxxx|xxxx|xxxx|xxxx)
109
+ #define VERS_ELEM_SAMP (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
110
+ #define VERS_ELEM_SCRIPT (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
111
+ #define VERS_ELEM_SELECT (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
112
+ #define VERS_ELEM_SMALL (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
113
+ #define VERS_ELEM_SPAN (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
114
+ #define VERS_ELEM_STRIKE (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|xxxx|xxxx|xxxx|xxxx)
115
+ #define VERS_ELEM_STRONG (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
116
+ #define VERS_ELEM_STYLE (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
117
+ #define VERS_ELEM_SUB (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
118
+ #define VERS_ELEM_SUP (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
119
+ #define VERS_ELEM_TABLE (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
120
+ #define VERS_ELEM_TBODY (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
121
+ #define VERS_ELEM_TD (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
122
+ #define VERS_ELEM_TEXTAREA (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
123
+ #define VERS_ELEM_TFOOT (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
124
+ #define VERS_ELEM_TH (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
125
+ #define VERS_ELEM_THEAD (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
126
+ #define VERS_ELEM_TITLE (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
127
+ #define VERS_ELEM_TR (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
128
+ #define VERS_ELEM_TT (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx)
129
+ #define VERS_ELEM_U (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|xxxx|xxxx|xxxx|xxxx)
130
+ #define VERS_ELEM_UL (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
131
+ #define VERS_ELEM_VAR (HT20|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10)
132
+ #define VERS_ELEM_XMP (HT20|HT32|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx)
133
+
134
+ static const Dict tag_defs[] =
135
+ {
136
+ { TidyTag_UNKNOWN, "unknown!", VERS_UNKNOWN, NULL, (0), NULL, NULL },
137
+
138
+ /* W3C defined elements */
139
+ { TidyTag_A, "a", VERS_ELEM_A, &TY_(W3CAttrsFor_A)[0], (CM_INLINE), TY_(ParseInline), NULL },
140
+ { TidyTag_ABBR, "abbr", VERS_ELEM_ABBR, &TY_(W3CAttrsFor_ABBR)[0], (CM_INLINE), TY_(ParseInline), NULL },
141
+ { TidyTag_ACRONYM, "acronym", VERS_ELEM_ACRONYM, &TY_(W3CAttrsFor_ACRONYM)[0], (CM_INLINE), TY_(ParseInline), NULL },
142
+ { TidyTag_ADDRESS, "address", VERS_ELEM_ADDRESS, &TY_(W3CAttrsFor_ADDRESS)[0], (CM_BLOCK), TY_(ParseInline), NULL },
143
+ { TidyTag_APPLET, "applet", VERS_ELEM_APPLET, &TY_(W3CAttrsFor_APPLET)[0], (CM_OBJECT|CM_IMG|CM_INLINE|CM_PARAM), TY_(ParseBlock), NULL },
144
+ { TidyTag_AREA, "area", VERS_ELEM_AREA, &TY_(W3CAttrsFor_AREA)[0], (CM_BLOCK|CM_EMPTY), TY_(ParseEmpty), CheckAREA },
145
+ { TidyTag_B, "b", VERS_ELEM_B, &TY_(W3CAttrsFor_B)[0], (CM_INLINE), TY_(ParseInline), NULL },
146
+ { TidyTag_BASE, "base", VERS_ELEM_BASE, &TY_(W3CAttrsFor_BASE)[0], (CM_HEAD|CM_EMPTY), TY_(ParseEmpty), NULL },
147
+ { TidyTag_BASEFONT, "basefont", VERS_ELEM_BASEFONT, &TY_(W3CAttrsFor_BASEFONT)[0], (CM_INLINE|CM_EMPTY), TY_(ParseEmpty), NULL },
148
+ { TidyTag_BDO, "bdo", VERS_ELEM_BDO, &TY_(W3CAttrsFor_BDO)[0], (CM_INLINE), TY_(ParseInline), NULL },
149
+ { TidyTag_BIG, "big", VERS_ELEM_BIG, &TY_(W3CAttrsFor_BIG)[0], (CM_INLINE), TY_(ParseInline), NULL },
150
+ { TidyTag_BLOCKQUOTE, "blockquote", VERS_ELEM_BLOCKQUOTE, &TY_(W3CAttrsFor_BLOCKQUOTE)[0], (CM_BLOCK), TY_(ParseBlock), NULL },
151
+ { TidyTag_BODY, "body", VERS_ELEM_BODY, &TY_(W3CAttrsFor_BODY)[0], (CM_HTML|CM_OPT|CM_OMITST), TY_(ParseBody), NULL },
152
+ { TidyTag_BR, "br", VERS_ELEM_BR, &TY_(W3CAttrsFor_BR)[0], (CM_INLINE|CM_EMPTY), TY_(ParseEmpty), NULL },
153
+ { TidyTag_BUTTON, "button", VERS_ELEM_BUTTON, &TY_(W3CAttrsFor_BUTTON)[0], (CM_INLINE), TY_(ParseBlock), NULL },
154
+ { TidyTag_CAPTION, "caption", VERS_ELEM_CAPTION, &TY_(W3CAttrsFor_CAPTION)[0], (CM_TABLE), TY_(ParseInline), CheckCaption },
155
+ { TidyTag_CENTER, "center", VERS_ELEM_CENTER, &TY_(W3CAttrsFor_CENTER)[0], (CM_BLOCK), TY_(ParseBlock), NULL },
156
+ { TidyTag_CITE, "cite", VERS_ELEM_CITE, &TY_(W3CAttrsFor_CITE)[0], (CM_INLINE), TY_(ParseInline), NULL },
157
+ { TidyTag_CODE, "code", VERS_ELEM_CODE, &TY_(W3CAttrsFor_CODE)[0], (CM_INLINE), TY_(ParseInline), NULL },
158
+ { TidyTag_COL, "col", VERS_ELEM_COL, &TY_(W3CAttrsFor_COL)[0], (CM_TABLE|CM_EMPTY), TY_(ParseEmpty), NULL },
159
+ { TidyTag_COLGROUP, "colgroup", VERS_ELEM_COLGROUP, &TY_(W3CAttrsFor_COLGROUP)[0], (CM_TABLE|CM_OPT), TY_(ParseColGroup), NULL },
160
+ { TidyTag_DD, "dd", VERS_ELEM_DD, &TY_(W3CAttrsFor_DD)[0], (CM_DEFLIST|CM_OPT|CM_NO_INDENT), TY_(ParseBlock), NULL },
161
+ { TidyTag_DEL, "del", VERS_ELEM_DEL, &TY_(W3CAttrsFor_DEL)[0], (CM_INLINE|CM_BLOCK|CM_MIXED), TY_(ParseInline), NULL },
162
+ { TidyTag_DFN, "dfn", VERS_ELEM_DFN, &TY_(W3CAttrsFor_DFN)[0], (CM_INLINE), TY_(ParseInline), NULL },
163
+ { TidyTag_DIR, "dir", VERS_ELEM_DIR, &TY_(W3CAttrsFor_DIR)[0], (CM_BLOCK|CM_OBSOLETE), TY_(ParseList), NULL },
164
+ { TidyTag_DIV, "div", VERS_ELEM_DIV, &TY_(W3CAttrsFor_DIV)[0], (CM_BLOCK), TY_(ParseBlock), NULL },
165
+ { TidyTag_DL, "dl", VERS_ELEM_DL, &TY_(W3CAttrsFor_DL)[0], (CM_BLOCK), TY_(ParseDefList), NULL },
166
+ { TidyTag_DT, "dt", VERS_ELEM_DT, &TY_(W3CAttrsFor_DT)[0], (CM_DEFLIST|CM_OPT|CM_NO_INDENT), TY_(ParseInline), NULL },
167
+ { TidyTag_EM, "em", VERS_ELEM_EM, &TY_(W3CAttrsFor_EM)[0], (CM_INLINE), TY_(ParseInline), NULL },
168
+ { TidyTag_FIELDSET, "fieldset", VERS_ELEM_FIELDSET, &TY_(W3CAttrsFor_FIELDSET)[0], (CM_BLOCK), TY_(ParseBlock), NULL },
169
+ { TidyTag_FONT, "font", VERS_ELEM_FONT, &TY_(W3CAttrsFor_FONT)[0], (CM_INLINE), TY_(ParseInline), NULL },
170
+ { TidyTag_FORM, "form", VERS_ELEM_FORM, &TY_(W3CAttrsFor_FORM)[0], (CM_BLOCK), TY_(ParseBlock), CheckFORM },
171
+ { TidyTag_FRAME, "frame", VERS_ELEM_FRAME, &TY_(W3CAttrsFor_FRAME)[0], (CM_FRAMES|CM_EMPTY), TY_(ParseEmpty), NULL },
172
+ { TidyTag_FRAMESET, "frameset", VERS_ELEM_FRAMESET, &TY_(W3CAttrsFor_FRAMESET)[0], (CM_HTML|CM_FRAMES), TY_(ParseFrameSet), NULL },
173
+ { TidyTag_H1, "h1", VERS_ELEM_H1, &TY_(W3CAttrsFor_H1)[0], (CM_BLOCK|CM_HEADING), TY_(ParseInline), NULL },
174
+ { TidyTag_H2, "h2", VERS_ELEM_H2, &TY_(W3CAttrsFor_H2)[0], (CM_BLOCK|CM_HEADING), TY_(ParseInline), NULL },
175
+ { TidyTag_H3, "h3", VERS_ELEM_H3, &TY_(W3CAttrsFor_H3)[0], (CM_BLOCK|CM_HEADING), TY_(ParseInline), NULL },
176
+ { TidyTag_H4, "h4", VERS_ELEM_H4, &TY_(W3CAttrsFor_H4)[0], (CM_BLOCK|CM_HEADING), TY_(ParseInline), NULL },
177
+ { TidyTag_H5, "h5", VERS_ELEM_H5, &TY_(W3CAttrsFor_H5)[0], (CM_BLOCK|CM_HEADING), TY_(ParseInline), NULL },
178
+ { TidyTag_H6, "h6", VERS_ELEM_H6, &TY_(W3CAttrsFor_H6)[0], (CM_BLOCK|CM_HEADING), TY_(ParseInline), NULL },
179
+ { TidyTag_HEAD, "head", VERS_ELEM_HEAD, &TY_(W3CAttrsFor_HEAD)[0], (CM_HTML|CM_OPT|CM_OMITST), TY_(ParseHead), NULL },
180
+ { TidyTag_HR, "hr", VERS_ELEM_HR, &TY_(W3CAttrsFor_HR)[0], (CM_BLOCK|CM_EMPTY), TY_(ParseEmpty), NULL },
181
+ { TidyTag_HTML, "html", VERS_ELEM_HTML, &TY_(W3CAttrsFor_HTML)[0], (CM_HTML|CM_OPT|CM_OMITST), TY_(ParseHTML), CheckHTML },
182
+ { TidyTag_I, "i", VERS_ELEM_I, &TY_(W3CAttrsFor_I)[0], (CM_INLINE), TY_(ParseInline), NULL },
183
+ { TidyTag_IFRAME, "iframe", VERS_ELEM_IFRAME, &TY_(W3CAttrsFor_IFRAME)[0], (CM_INLINE), TY_(ParseBlock), NULL },
184
+ { TidyTag_IMG, "img", VERS_ELEM_IMG, &TY_(W3CAttrsFor_IMG)[0], (CM_INLINE|CM_IMG|CM_EMPTY), TY_(ParseEmpty), CheckIMG },
185
+ { TidyTag_INPUT, "input", VERS_ELEM_INPUT, &TY_(W3CAttrsFor_INPUT)[0], (CM_INLINE|CM_IMG|CM_EMPTY), TY_(ParseEmpty), NULL },
186
+ { TidyTag_INS, "ins", VERS_ELEM_INS, &TY_(W3CAttrsFor_INS)[0], (CM_INLINE|CM_BLOCK|CM_MIXED), TY_(ParseInline), NULL },
187
+ { TidyTag_ISINDEX, "isindex", VERS_ELEM_ISINDEX, &TY_(W3CAttrsFor_ISINDEX)[0], (CM_BLOCK|CM_EMPTY), TY_(ParseEmpty), NULL },
188
+ { TidyTag_KBD, "kbd", VERS_ELEM_KBD, &TY_(W3CAttrsFor_KBD)[0], (CM_INLINE), TY_(ParseInline), NULL },
189
+ { TidyTag_LABEL, "label", VERS_ELEM_LABEL, &TY_(W3CAttrsFor_LABEL)[0], (CM_INLINE), TY_(ParseInline), NULL },
190
+ { TidyTag_LEGEND, "legend", VERS_ELEM_LEGEND, &TY_(W3CAttrsFor_LEGEND)[0], (CM_INLINE), TY_(ParseInline), NULL },
191
+ { TidyTag_LI, "li", VERS_ELEM_LI, &TY_(W3CAttrsFor_LI)[0], (CM_LIST|CM_OPT|CM_NO_INDENT), TY_(ParseBlock), NULL },
192
+ { TidyTag_LINK, "link", VERS_ELEM_LINK, &TY_(W3CAttrsFor_LINK)[0], (CM_HEAD|CM_EMPTY), TY_(ParseEmpty), CheckLINK },
193
+ { TidyTag_LISTING, "listing", VERS_ELEM_LISTING, &TY_(W3CAttrsFor_LISTING)[0], (CM_BLOCK|CM_OBSOLETE), TY_(ParsePre), NULL },
194
+ { TidyTag_MAP, "map", VERS_ELEM_MAP, &TY_(W3CAttrsFor_MAP)[0], (CM_INLINE), TY_(ParseBlock), NULL },
195
+ { TidyTag_MENU, "menu", VERS_ELEM_MENU, &TY_(W3CAttrsFor_MENU)[0], (CM_BLOCK|CM_OBSOLETE), TY_(ParseList), NULL },
196
+ { TidyTag_META, "meta", VERS_ELEM_META, &TY_(W3CAttrsFor_META)[0], (CM_HEAD|CM_EMPTY), TY_(ParseEmpty), CheckMETA },
197
+ { TidyTag_NOFRAMES, "noframes", VERS_ELEM_NOFRAMES, &TY_(W3CAttrsFor_NOFRAMES)[0], (CM_BLOCK|CM_FRAMES), TY_(ParseNoFrames), NULL },
198
+ { TidyTag_NOSCRIPT, "noscript", VERS_ELEM_NOSCRIPT, &TY_(W3CAttrsFor_NOSCRIPT)[0], (CM_BLOCK|CM_INLINE|CM_MIXED), TY_(ParseBlock), NULL },
199
+ { TidyTag_OBJECT, "object", VERS_ELEM_OBJECT, &TY_(W3CAttrsFor_OBJECT)[0], (CM_OBJECT|CM_HEAD|CM_IMG|CM_INLINE|CM_PARAM), TY_(ParseBlock), NULL },
200
+ { TidyTag_OL, "ol", VERS_ELEM_OL, &TY_(W3CAttrsFor_OL)[0], (CM_BLOCK), TY_(ParseList), NULL },
201
+ { TidyTag_OPTGROUP, "optgroup", VERS_ELEM_OPTGROUP, &TY_(W3CAttrsFor_OPTGROUP)[0], (CM_FIELD|CM_OPT), TY_(ParseOptGroup), NULL },
202
+ { TidyTag_OPTION, "option", VERS_ELEM_OPTION, &TY_(W3CAttrsFor_OPTION)[0], (CM_FIELD|CM_OPT), TY_(ParseText), NULL },
203
+ { TidyTag_P, "p", VERS_ELEM_P, &TY_(W3CAttrsFor_P)[0], (CM_BLOCK|CM_OPT), TY_(ParseInline), NULL },
204
+ { TidyTag_PARAM, "param", VERS_ELEM_PARAM, &TY_(W3CAttrsFor_PARAM)[0], (CM_INLINE|CM_EMPTY), TY_(ParseEmpty), NULL },
205
+ { TidyTag_PLAINTEXT, "plaintext", VERS_ELEM_PLAINTEXT, &TY_(W3CAttrsFor_PLAINTEXT)[0], (CM_BLOCK|CM_OBSOLETE), TY_(ParsePre), NULL },
206
+ { TidyTag_PRE, "pre", VERS_ELEM_PRE, &TY_(W3CAttrsFor_PRE)[0], (CM_BLOCK), TY_(ParsePre), NULL },
207
+ { TidyTag_Q, "q", VERS_ELEM_Q, &TY_(W3CAttrsFor_Q)[0], (CM_INLINE), TY_(ParseInline), NULL },
208
+ { TidyTag_RB, "rb", VERS_ELEM_RB, &TY_(W3CAttrsFor_RB)[0], (CM_INLINE), TY_(ParseInline), NULL },
209
+ { TidyTag_RBC, "rbc", VERS_ELEM_RBC, &TY_(W3CAttrsFor_RBC)[0], (CM_INLINE), TY_(ParseInline), NULL },
210
+ { TidyTag_RP, "rp", VERS_ELEM_RP, &TY_(W3CAttrsFor_RP)[0], (CM_INLINE), TY_(ParseInline), NULL },
211
+ { TidyTag_RT, "rt", VERS_ELEM_RT, &TY_(W3CAttrsFor_RT)[0], (CM_INLINE), TY_(ParseInline), NULL },
212
+ { TidyTag_RTC, "rtc", VERS_ELEM_RTC, &TY_(W3CAttrsFor_RTC)[0], (CM_INLINE), TY_(ParseInline), NULL },
213
+ { TidyTag_RUBY, "ruby", VERS_ELEM_RUBY, &TY_(W3CAttrsFor_RUBY)[0], (CM_INLINE), TY_(ParseInline), NULL },
214
+ { TidyTag_S, "s", VERS_ELEM_S, &TY_(W3CAttrsFor_S)[0], (CM_INLINE), TY_(ParseInline), NULL },
215
+ { TidyTag_SAMP, "samp", VERS_ELEM_SAMP, &TY_(W3CAttrsFor_SAMP)[0], (CM_INLINE), TY_(ParseInline), NULL },
216
+ { TidyTag_SCRIPT, "script", VERS_ELEM_SCRIPT, &TY_(W3CAttrsFor_SCRIPT)[0], (CM_HEAD|CM_MIXED|CM_BLOCK|CM_INLINE), TY_(ParseScript), CheckSCRIPT },
217
+ { TidyTag_SELECT, "select", VERS_ELEM_SELECT, &TY_(W3CAttrsFor_SELECT)[0], (CM_INLINE|CM_FIELD), TY_(ParseSelect), NULL },
218
+ { TidyTag_SMALL, "small", VERS_ELEM_SMALL, &TY_(W3CAttrsFor_SMALL)[0], (CM_INLINE), TY_(ParseInline), NULL },
219
+ { TidyTag_SPAN, "span", VERS_ELEM_SPAN, &TY_(W3CAttrsFor_SPAN)[0], (CM_INLINE), TY_(ParseInline), NULL },
220
+ { TidyTag_STRIKE, "strike", VERS_ELEM_STRIKE, &TY_(W3CAttrsFor_STRIKE)[0], (CM_INLINE), TY_(ParseInline), NULL },
221
+ { TidyTag_STRONG, "strong", VERS_ELEM_STRONG, &TY_(W3CAttrsFor_STRONG)[0], (CM_INLINE), TY_(ParseInline), NULL },
222
+ { TidyTag_STYLE, "style", VERS_ELEM_STYLE, &TY_(W3CAttrsFor_STYLE)[0], (CM_HEAD), TY_(ParseScript), CheckSTYLE },
223
+ { TidyTag_SUB, "sub", VERS_ELEM_SUB, &TY_(W3CAttrsFor_SUB)[0], (CM_INLINE), TY_(ParseInline), NULL },
224
+ { TidyTag_SUP, "sup", VERS_ELEM_SUP, &TY_(W3CAttrsFor_SUP)[0], (CM_INLINE), TY_(ParseInline), NULL },
225
+ { TidyTag_TABLE, "table", VERS_ELEM_TABLE, &TY_(W3CAttrsFor_TABLE)[0], (CM_BLOCK), TY_(ParseTableTag), CheckTABLE },
226
+ { TidyTag_TBODY, "tbody", VERS_ELEM_TBODY, &TY_(W3CAttrsFor_TBODY)[0], (CM_TABLE|CM_ROWGRP|CM_OPT), TY_(ParseRowGroup), NULL },
227
+ { TidyTag_TD, "td", VERS_ELEM_TD, &TY_(W3CAttrsFor_TD)[0], (CM_ROW|CM_OPT|CM_NO_INDENT), TY_(ParseBlock), NULL },
228
+ { TidyTag_TEXTAREA, "textarea", VERS_ELEM_TEXTAREA, &TY_(W3CAttrsFor_TEXTAREA)[0], (CM_INLINE|CM_FIELD), TY_(ParseText), NULL },
229
+ { TidyTag_TFOOT, "tfoot", VERS_ELEM_TFOOT, &TY_(W3CAttrsFor_TFOOT)[0], (CM_TABLE|CM_ROWGRP|CM_OPT), TY_(ParseRowGroup), NULL },
230
+ { TidyTag_TH, "th", VERS_ELEM_TH, &TY_(W3CAttrsFor_TH)[0], (CM_ROW|CM_OPT|CM_NO_INDENT), TY_(ParseBlock), NULL },
231
+ { TidyTag_THEAD, "thead", VERS_ELEM_THEAD, &TY_(W3CAttrsFor_THEAD)[0], (CM_TABLE|CM_ROWGRP|CM_OPT), TY_(ParseRowGroup), NULL },
232
+ { TidyTag_TITLE, "title", VERS_ELEM_TITLE, &TY_(W3CAttrsFor_TITLE)[0], (CM_HEAD), TY_(ParseTitle), NULL },
233
+ { TidyTag_TR, "tr", VERS_ELEM_TR, &TY_(W3CAttrsFor_TR)[0], (CM_TABLE|CM_OPT), TY_(ParseRow), NULL },
234
+ { TidyTag_TT, "tt", VERS_ELEM_TT, &TY_(W3CAttrsFor_TT)[0], (CM_INLINE), TY_(ParseInline), NULL },
235
+ { TidyTag_U, "u", VERS_ELEM_U, &TY_(W3CAttrsFor_U)[0], (CM_INLINE), TY_(ParseInline), NULL },
236
+ { TidyTag_UL, "ul", VERS_ELEM_UL, &TY_(W3CAttrsFor_UL)[0], (CM_BLOCK), TY_(ParseList), NULL },
237
+ { TidyTag_VAR, "var", VERS_ELEM_VAR, &TY_(W3CAttrsFor_VAR)[0], (CM_INLINE), TY_(ParseInline), NULL },
238
+ { TidyTag_XMP, "xmp", VERS_ELEM_XMP, &TY_(W3CAttrsFor_XMP)[0], (CM_BLOCK|CM_OBSOLETE), TY_(ParsePre), NULL },
239
+ { TidyTag_NEXTID, "nextid", VERS_ELEM_NEXTID, &TY_(W3CAttrsFor_NEXTID)[0], (CM_HEAD|CM_EMPTY), TY_(ParseEmpty), NULL },
240
+
241
+ /* proprietary elements */
242
+ { TidyTag_ALIGN, "align", VERS_NETSCAPE, NULL, (CM_BLOCK), TY_(ParseBlock), NULL },
243
+ { TidyTag_BGSOUND, "bgsound", VERS_MICROSOFT, NULL, (CM_HEAD|CM_EMPTY), TY_(ParseEmpty), NULL },
244
+ { TidyTag_BLINK, "blink", VERS_PROPRIETARY, NULL, (CM_INLINE), TY_(ParseInline), NULL },
245
+ { TidyTag_COMMENT, "comment", VERS_MICROSOFT, NULL, (CM_INLINE), TY_(ParseInline), NULL },
246
+ { TidyTag_EMBED, "embed", VERS_NETSCAPE, NULL, (CM_INLINE|CM_IMG|CM_EMPTY), TY_(ParseEmpty), NULL },
247
+ { TidyTag_ILAYER, "ilayer", VERS_NETSCAPE, NULL, (CM_INLINE), TY_(ParseInline), NULL },
248
+ { TidyTag_KEYGEN, "keygen", VERS_NETSCAPE, NULL, (CM_INLINE|CM_EMPTY), TY_(ParseEmpty), NULL },
249
+ { TidyTag_LAYER, "layer", VERS_NETSCAPE, NULL, (CM_BLOCK), TY_(ParseBlock), NULL },
250
+ { TidyTag_MARQUEE, "marquee", VERS_MICROSOFT, NULL, (CM_INLINE|CM_OPT), TY_(ParseInline), NULL },
251
+ { TidyTag_MULTICOL, "multicol", VERS_NETSCAPE, NULL, (CM_BLOCK), TY_(ParseBlock), NULL },
252
+ { TidyTag_NOBR, "nobr", VERS_PROPRIETARY, NULL, (CM_INLINE), TY_(ParseInline), NULL },
253
+ { TidyTag_NOEMBED, "noembed", VERS_NETSCAPE, NULL, (CM_INLINE), TY_(ParseInline), NULL },
254
+ { TidyTag_NOLAYER, "nolayer", VERS_NETSCAPE, NULL, (CM_BLOCK|CM_INLINE|CM_MIXED), TY_(ParseBlock), NULL },
255
+ { TidyTag_NOSAVE, "nosave", VERS_NETSCAPE, NULL, (CM_BLOCK), TY_(ParseBlock), NULL },
256
+ { TidyTag_SERVER, "server", VERS_NETSCAPE, NULL, (CM_HEAD|CM_MIXED|CM_BLOCK|CM_INLINE), TY_(ParseScript), NULL },
257
+ { TidyTag_SERVLET, "servlet", VERS_SUN, NULL, (CM_OBJECT|CM_IMG|CM_INLINE|CM_PARAM), TY_(ParseBlock), NULL },
258
+ { TidyTag_SPACER, "spacer", VERS_NETSCAPE, NULL, (CM_INLINE|CM_EMPTY), TY_(ParseEmpty), NULL },
259
+ { TidyTag_WBR, "wbr", VERS_PROPRIETARY, NULL, (CM_INLINE|CM_EMPTY), TY_(ParseEmpty), NULL },
260
+
261
+ /* this must be the final entry */
262
+ { (TidyTagId)0, NULL, 0, NULL, (0), NULL, NULL }
263
+ };
264
+
265
+ #if ELEMENT_HASH_LOOKUP
266
+ static uint tagsHash(ctmbstr s)
267
+ {
268
+ uint hashval;
269
+
270
+ for (hashval = 0; *s != '\0'; s++)
271
+ hashval = *s + 31*hashval;
272
+
273
+ return hashval % ELEMENT_HASH_SIZE;
274
+ }
275
+
276
+ static const Dict *tagsInstall(TidyDocImpl* doc, TidyTagImpl* tags, const Dict* old)
277
+ {
278
+ DictHash *np;
279
+ uint hashval;
280
+
281
+ if (old)
282
+ {
283
+ np = (DictHash *)TidyDocAlloc(doc, sizeof(*np));
284
+ np->tag = old;
285
+
286
+ hashval = tagsHash(old->name);
287
+ np->next = tags->hashtab[hashval];
288
+ tags->hashtab[hashval] = np;
289
+ }
290
+
291
+ return old;
292
+ }
293
+
294
+ static void tagsRemoveFromHash( TidyDocImpl* doc, TidyTagImpl* tags, ctmbstr s )
295
+ {
296
+ uint h = tagsHash(s);
297
+ DictHash *p, *prev = NULL;
298
+ for (p = tags->hashtab[h]; p && p->tag; p = p->next)
299
+ {
300
+ if (TY_(tmbstrcmp)(s, p->tag->name) == 0)
301
+ {
302
+ DictHash* next = p->next;
303
+ if ( prev )
304
+ prev->next = next;
305
+ else
306
+ tags->hashtab[h] = next;
307
+ TidyDocFree(doc, p);
308
+ return;
309
+ }
310
+ prev = p;
311
+ }
312
+ }
313
+
314
+ static void tagsEmptyHash( TidyDocImpl* doc, TidyTagImpl* tags )
315
+ {
316
+ uint i;
317
+ DictHash *prev, *next;
318
+
319
+ for (i = 0; i < ELEMENT_HASH_SIZE; ++i)
320
+ {
321
+ prev = NULL;
322
+ next = tags->hashtab[i];
323
+
324
+ while(next)
325
+ {
326
+ prev = next->next;
327
+ TidyDocFree(doc, next);
328
+ next = prev;
329
+ }
330
+
331
+ tags->hashtab[i] = NULL;
332
+ }
333
+ }
334
+ #endif /* ELEMENT_HASH_LOOKUP */
335
+
336
+ static const Dict* tagsLookup( TidyDocImpl* doc, TidyTagImpl* tags, ctmbstr s )
337
+ {
338
+ const Dict *np;
339
+ #if ELEMENT_HASH_LOOKUP
340
+ const DictHash* p;
341
+ #endif
342
+
343
+ if (!s)
344
+ return NULL;
345
+
346
+ #if ELEMENT_HASH_LOOKUP
347
+ /* this breaks if declared elements get changed between two */
348
+ /* parser runs since Tidy would use the cached version rather */
349
+ /* than the new one. */
350
+ /* However, as FreeDeclaredTags() correctly cleans the hash */
351
+ /* this should not be true anymore. */
352
+ for (p = tags->hashtab[tagsHash(s)]; p && p->tag; p = p->next)
353
+ if (TY_(tmbstrcmp)(s, p->tag->name) == 0)
354
+ return p->tag;
355
+
356
+ for (np = tag_defs + 1; np < tag_defs + N_TIDY_TAGS; ++np)
357
+ if (TY_(tmbstrcmp)(s, np->name) == 0)
358
+ return tagsInstall(doc, tags, np);
359
+
360
+ for (np = tags->declared_tag_list; np; np = np->next)
361
+ if (TY_(tmbstrcmp)(s, np->name) == 0)
362
+ return tagsInstall(doc, tags, np);
363
+ #else
364
+
365
+ for (np = tag_defs + 1; np < tag_defs + N_TIDY_TAGS; ++np)
366
+ if (TY_(tmbstrcmp)(s, np->name) == 0)
367
+ return np;
368
+
369
+ for (np = tags->declared_tag_list; np; np = np->next)
370
+ if (TY_(tmbstrcmp)(s, np->name) == 0)
371
+ return np;
372
+
373
+ #endif /* ELEMENT_HASH_LOOKUP */
374
+
375
+ return NULL;
376
+ }
377
+
378
+ static Dict* NewDict( TidyDocImpl* doc, ctmbstr name )
379
+ {
380
+ Dict *np = (Dict*) TidyDocAlloc( doc, sizeof(Dict) );
381
+ np->id = TidyTag_UNKNOWN;
382
+ np->name = name ? TY_(tmbstrdup)( doc->allocator, name ) : NULL;
383
+ np->versions = VERS_UNKNOWN;
384
+ np->attrvers = NULL;
385
+ np->model = CM_UNKNOWN;
386
+ np->parser = 0;
387
+ np->chkattrs = 0;
388
+ np->next = NULL;
389
+ return np;
390
+ }
391
+
392
+ static void FreeDict( TidyDocImpl* doc, Dict *d )
393
+ {
394
+ if ( d )
395
+ TidyDocFree( doc, d->name );
396
+ TidyDocFree( doc, d );
397
+ }
398
+
399
+ static void declare( TidyDocImpl* doc, TidyTagImpl* tags,
400
+ ctmbstr name, uint versions, uint model,
401
+ Parser *parser, CheckAttribs *chkattrs )
402
+ {
403
+ if ( name )
404
+ {
405
+ Dict* np = (Dict*) tagsLookup( doc, tags, name );
406
+ if ( np == NULL )
407
+ {
408
+ np = NewDict( doc, name );
409
+ np->next = tags->declared_tag_list;
410
+ tags->declared_tag_list = np;
411
+ }
412
+
413
+ /* Make sure we are not over-writing predefined tags */
414
+ if ( np->id == TidyTag_UNKNOWN )
415
+ {
416
+ np->versions = versions;
417
+ np->model |= model;
418
+ np->parser = parser;
419
+ np->chkattrs = chkattrs;
420
+ np->attrvers = NULL;
421
+ }
422
+ }
423
+ }
424
+
425
+ /* public interface for finding tag by name */
426
+ Bool TY_(FindTag)( TidyDocImpl* doc, Node *node )
427
+ {
428
+ const Dict *np = NULL;
429
+ if ( cfgBool(doc, TidyXmlTags) )
430
+ {
431
+ node->tag = doc->tags.xml_tags;
432
+ return yes;
433
+ }
434
+
435
+ if ( node->element && (np = tagsLookup(doc, &doc->tags, node->element)) )
436
+ {
437
+ node->tag = np;
438
+ return yes;
439
+ }
440
+
441
+ return no;
442
+ }
443
+
444
+ const Dict* TY_(LookupTagDef)( TidyTagId tid )
445
+ {
446
+ const Dict *np;
447
+
448
+ for (np = tag_defs + 1; np < tag_defs + N_TIDY_TAGS; ++np )
449
+ if (np->id == tid)
450
+ return np;
451
+
452
+ return NULL;
453
+ }
454
+
455
+ Parser* TY_(FindParser)( TidyDocImpl* doc, Node *node )
456
+ {
457
+ const Dict* np = tagsLookup( doc, &doc->tags, node->element );
458
+ if ( np )
459
+ return np->parser;
460
+ return NULL;
461
+ }
462
+
463
+ void TY_(DefineTag)( TidyDocImpl* doc, UserTagType tagType, ctmbstr name )
464
+ {
465
+ Parser* parser = 0;
466
+ uint cm = CM_UNKNOWN;
467
+ uint vers = VERS_PROPRIETARY;
468
+
469
+ switch (tagType)
470
+ {
471
+ case tagtype_empty:
472
+ cm = CM_EMPTY|CM_NO_INDENT|CM_NEW;
473
+ parser = TY_(ParseBlock);
474
+ break;
475
+
476
+ case tagtype_inline:
477
+ cm = CM_INLINE|CM_NO_INDENT|CM_NEW;
478
+ parser = TY_(ParseInline);
479
+ break;
480
+
481
+ case tagtype_block:
482
+ cm = CM_BLOCK|CM_NO_INDENT|CM_NEW;
483
+ parser = TY_(ParseBlock);
484
+ break;
485
+
486
+ case tagtype_pre:
487
+ cm = CM_BLOCK|CM_NO_INDENT|CM_NEW;
488
+ parser = TY_(ParsePre);
489
+ break;
490
+
491
+ case tagtype_null:
492
+ break;
493
+ }
494
+ if ( cm && parser )
495
+ declare( doc, &doc->tags, name, vers, cm, parser, 0 );
496
+ }
497
+
498
+ TidyIterator TY_(GetDeclaredTagList)( TidyDocImpl* doc )
499
+ {
500
+ return (TidyIterator) doc->tags.declared_tag_list;
501
+ }
502
+
503
+ ctmbstr TY_(GetNextDeclaredTag)( TidyDocImpl* ARG_UNUSED(doc),
504
+ UserTagType tagType, TidyIterator* iter )
505
+ {
506
+ ctmbstr name = NULL;
507
+ Dict* curr;
508
+ for ( curr = (Dict*) *iter; name == NULL && curr != NULL; curr = curr->next )
509
+ {
510
+ switch ( tagType )
511
+ {
512
+ case tagtype_empty:
513
+ if ( (curr->model & CM_EMPTY) != 0 )
514
+ name = curr->name;
515
+ break;
516
+
517
+ case tagtype_inline:
518
+ if ( (curr->model & CM_INLINE) != 0 )
519
+ name = curr->name;
520
+ break;
521
+
522
+ case tagtype_block:
523
+ if ( (curr->model & CM_BLOCK) != 0 &&
524
+ curr->parser == TY_(ParseBlock) )
525
+ name = curr->name;
526
+ break;
527
+
528
+ case tagtype_pre:
529
+ if ( (curr->model & CM_BLOCK) != 0 &&
530
+ curr->parser == TY_(ParsePre) )
531
+ name = curr->name;
532
+ break;
533
+
534
+ case tagtype_null:
535
+ break;
536
+ }
537
+ }
538
+ *iter = (TidyIterator) curr;
539
+ return name;
540
+ }
541
+
542
+ void TY_(InitTags)( TidyDocImpl* doc )
543
+ {
544
+ Dict* xml;
545
+ TidyTagImpl* tags = &doc->tags;
546
+
547
+ TidyClearMemory( tags, sizeof(TidyTagImpl) );
548
+
549
+ /* create dummy entry for all xml tags */
550
+ xml = NewDict( doc, NULL );
551
+ xml->versions = VERS_XML;
552
+ xml->model = CM_BLOCK;
553
+ xml->parser = 0;
554
+ xml->chkattrs = 0;
555
+ xml->attrvers = NULL;
556
+ tags->xml_tags = xml;
557
+ }
558
+
559
+ /* By default, zap all of them. But allow
560
+ ** an single type to be specified.
561
+ */
562
+ void TY_(FreeDeclaredTags)( TidyDocImpl* doc, UserTagType tagType )
563
+ {
564
+ TidyTagImpl* tags = &doc->tags;
565
+ Dict *curr, *next = NULL, *prev = NULL;
566
+
567
+ for ( curr=tags->declared_tag_list; curr; curr = next )
568
+ {
569
+ Bool deleteIt = yes;
570
+ next = curr->next;
571
+ switch ( tagType )
572
+ {
573
+ case tagtype_empty:
574
+ deleteIt = ( curr->model & CM_EMPTY ) != 0;
575
+ break;
576
+
577
+ case tagtype_inline:
578
+ deleteIt = ( curr->model & CM_INLINE ) != 0;
579
+ break;
580
+
581
+ case tagtype_block:
582
+ deleteIt = ( (curr->model & CM_BLOCK) != 0 &&
583
+ curr->parser == TY_(ParseBlock) );
584
+ break;
585
+
586
+ case tagtype_pre:
587
+ deleteIt = ( (curr->model & CM_BLOCK) != 0 &&
588
+ curr->parser == TY_(ParsePre) );
589
+ break;
590
+
591
+ case tagtype_null:
592
+ break;
593
+ }
594
+
595
+ if ( deleteIt )
596
+ {
597
+ #if ELEMENT_HASH_LOOKUP
598
+ tagsRemoveFromHash( doc, &doc->tags, curr->name );
599
+ #endif
600
+ FreeDict( doc, curr );
601
+ if ( prev )
602
+ prev->next = next;
603
+ else
604
+ tags->declared_tag_list = next;
605
+ }
606
+ else
607
+ prev = curr;
608
+ }
609
+ }
610
+
611
+ void TY_(FreeTags)( TidyDocImpl* doc )
612
+ {
613
+ TidyTagImpl* tags = &doc->tags;
614
+
615
+ #if ELEMENT_HASH_LOOKUP
616
+ tagsEmptyHash( doc, tags );
617
+ #endif
618
+ TY_(FreeDeclaredTags)( doc, tagtype_null );
619
+ FreeDict( doc, tags->xml_tags );
620
+
621
+ /* get rid of dangling tag references */
622
+ TidyClearMemory( tags, sizeof(TidyTagImpl) );
623
+ }
624
+
625
+
626
+ /* default method for checking an element's attributes */
627
+ void TY_(CheckAttributes)( TidyDocImpl* doc, Node *node )
628
+ {
629
+ AttVal *next, *attval = node->attributes;
630
+ while (attval)
631
+ {
632
+ next = attval->next;
633
+ TY_(CheckAttribute)( doc, node, attval );
634
+ attval = next;
635
+ }
636
+ }
637
+
638
+ /* methods for checking attributes for specific elements */
639
+
640
+ void CheckIMG( TidyDocImpl* doc, Node *node )
641
+ {
642
+ Bool HasAlt = TY_(AttrGetById)(node, TidyAttr_ALT) != NULL;
643
+ Bool HasSrc = TY_(AttrGetById)(node, TidyAttr_SRC) != NULL;
644
+ Bool HasUseMap = TY_(AttrGetById)(node, TidyAttr_USEMAP) != NULL;
645
+ Bool HasIsMap = TY_(AttrGetById)(node, TidyAttr_ISMAP) != NULL;
646
+ Bool HasDataFld = TY_(AttrGetById)(node, TidyAttr_DATAFLD) != NULL;
647
+
648
+ TY_(CheckAttributes)(doc, node);
649
+
650
+ if ( !HasAlt )
651
+ {
652
+ if ( cfg(doc, TidyAccessibilityCheckLevel) == 0 )
653
+ {
654
+ doc->badAccess |= BA_MISSING_IMAGE_ALT;
655
+ TY_(ReportMissingAttr)( doc, node, "alt" );
656
+ }
657
+
658
+ if ( cfgStr(doc, TidyAltText) )
659
+ TY_(AddAttribute)( doc, node, "alt", cfgStr(doc, TidyAltText) );
660
+ }
661
+
662
+ if ( !HasSrc && !HasDataFld )
663
+ TY_(ReportMissingAttr)( doc, node, "src" );
664
+
665
+ if ( cfg(doc, TidyAccessibilityCheckLevel) == 0 )
666
+ {
667
+ if ( HasIsMap && !HasUseMap )
668
+ TY_(ReportAttrError)( doc, node, NULL, MISSING_IMAGEMAP);
669
+ }
670
+ }
671
+
672
+ void CheckCaption(TidyDocImpl* doc, Node *node)
673
+ {
674
+ AttVal *attval;
675
+
676
+ TY_(CheckAttributes)(doc, node);
677
+
678
+ attval = TY_(AttrGetById)(node, TidyAttr_ALIGN);
679
+
680
+ if (!AttrHasValue(attval))
681
+ return;
682
+
683
+ if (AttrValueIs(attval, "left") || AttrValueIs(attval, "right"))
684
+ TY_(ConstrainVersion)(doc, VERS_HTML40_LOOSE);
685
+ else if (AttrValueIs(attval, "top") || AttrValueIs(attval, "bottom"))
686
+ TY_(ConstrainVersion)(doc, ~(VERS_HTML20|VERS_HTML32));
687
+ else
688
+ TY_(ReportAttrError)(doc, node, attval, BAD_ATTRIBUTE_VALUE);
689
+ }
690
+
691
+ void CheckHTML( TidyDocImpl* doc, Node *node )
692
+ {
693
+ TY_(CheckAttributes)(doc, node);
694
+ }
695
+
696
+ void CheckAREA( TidyDocImpl* doc, Node *node )
697
+ {
698
+ Bool HasAlt = TY_(AttrGetById)(node, TidyAttr_ALT) != NULL;
699
+ Bool HasHref = TY_(AttrGetById)(node, TidyAttr_HREF) != NULL;
700
+ Bool HasNohref = TY_(AttrGetById)(node, TidyAttr_NOHREF) != NULL;
701
+
702
+ TY_(CheckAttributes)(doc, node);
703
+
704
+ if ( !HasAlt )
705
+ {
706
+ if ( cfg(doc, TidyAccessibilityCheckLevel) == 0 )
707
+ {
708
+ doc->badAccess |= BA_MISSING_LINK_ALT;
709
+ TY_(ReportMissingAttr)( doc, node, "alt" );
710
+ }
711
+ }
712
+
713
+ if ( !HasHref && !HasNohref )
714
+ TY_(ReportMissingAttr)( doc, node, "href" );
715
+ }
716
+
717
+ void CheckTABLE( TidyDocImpl* doc, Node *node )
718
+ {
719
+ AttVal* attval;
720
+ Bool HasSummary = TY_(AttrGetById)(node, TidyAttr_SUMMARY) != NULL;
721
+
722
+ TY_(CheckAttributes)(doc, node);
723
+
724
+ /* a missing summary attribute is bad accessibility, no matter
725
+ what HTML version is involved; a document without is valid */
726
+ if (cfg(doc, TidyAccessibilityCheckLevel) == 0)
727
+ {
728
+ if (!HasSummary)
729
+ {
730
+ doc->badAccess |= BA_MISSING_SUMMARY;
731
+ TY_(ReportMissingAttr)( doc, node, "summary");
732
+ }
733
+ }
734
+
735
+ /* convert <table border> to <table border="1"> */
736
+ if ( cfgBool(doc, TidyXmlOut) && (attval = TY_(AttrGetById)(node, TidyAttr_BORDER)) )
737
+ {
738
+ if (attval->value == NULL)
739
+ attval->value = TY_(tmbstrdup)(doc->allocator, "1");
740
+ }
741
+ }
742
+
743
+ /* add missing type attribute when appropriate */
744
+ void CheckSCRIPT( TidyDocImpl* doc, Node *node )
745
+ {
746
+ AttVal *lang, *type;
747
+ char buf[16];
748
+
749
+ TY_(CheckAttributes)(doc, node);
750
+
751
+ lang = TY_(AttrGetById)(node, TidyAttr_LANGUAGE);
752
+ type = TY_(AttrGetById)(node, TidyAttr_TYPE);
753
+
754
+ if (!type)
755
+ {
756
+ /* check for javascript */
757
+ if (lang)
758
+ {
759
+ /* Test #696799. lang->value can be NULL. */
760
+ buf[0] = '\0';
761
+ TY_(tmbstrncpy)(buf, lang->value, sizeof(buf));
762
+ buf[10] = '\0';
763
+
764
+ if (TY_(tmbstrncasecmp)(buf, "javascript", 10) == 0 ||
765
+ TY_(tmbstrncasecmp)(buf, "jscript", 7) == 0)
766
+ {
767
+ TY_(AddAttribute)(doc, node, "type", "text/javascript");
768
+ }
769
+ else if (TY_(tmbstrcasecmp)(buf, "vbscript") == 0)
770
+ {
771
+ /* per Randy Waki 8/6/01 */
772
+ TY_(AddAttribute)(doc, node, "type", "text/vbscript");
773
+ }
774
+ }
775
+ else
776
+ {
777
+ TY_(AddAttribute)(doc, node, "type", "text/javascript");
778
+ }
779
+
780
+ type = TY_(AttrGetById)(node, TidyAttr_TYPE);
781
+
782
+ if (type != NULL)
783
+ {
784
+ TY_(ReportAttrError)(doc, node, type, INSERTING_ATTRIBUTE);
785
+ }
786
+ else
787
+ {
788
+ TY_(ReportMissingAttr)(doc, node, "type");
789
+ }
790
+ }
791
+ }
792
+
793
+
794
+ /* add missing type attribute when appropriate */
795
+ void CheckSTYLE( TidyDocImpl* doc, Node *node )
796
+ {
797
+ AttVal *type = TY_(AttrGetById)(node, TidyAttr_TYPE);
798
+
799
+ TY_(CheckAttributes)( doc, node );
800
+
801
+ if ( !type || !type->value || !TY_(tmbstrlen)(type->value) )
802
+ {
803
+ type = TY_(RepairAttrValue)(doc, node, "type", "text/css");
804
+ TY_(ReportAttrError)( doc, node, type, INSERTING_ATTRIBUTE );
805
+ }
806
+ }
807
+
808
+ /* add missing type attribute when appropriate */
809
+ void CheckLINK( TidyDocImpl* doc, Node *node )
810
+ {
811
+ AttVal *rel = TY_(AttrGetById)(node, TidyAttr_REL);
812
+
813
+ TY_(CheckAttributes)( doc, node );
814
+
815
+ /* todo: <link rel="alternate stylesheet"> */
816
+ if (AttrValueIs(rel, "stylesheet"))
817
+ {
818
+ AttVal *type = TY_(AttrGetById)(node, TidyAttr_TYPE);
819
+ if (!type)
820
+ {
821
+ TY_(AddAttribute)( doc, node, "type", "text/css" );
822
+ type = TY_(AttrGetById)(node, TidyAttr_TYPE);
823
+ TY_(ReportAttrError)( doc, node, type, INSERTING_ATTRIBUTE );
824
+ }
825
+ }
826
+ }
827
+
828
+ /* reports missing action attribute */
829
+ void CheckFORM( TidyDocImpl* doc, Node *node )
830
+ {
831
+ AttVal *action = TY_(AttrGetById)(node, TidyAttr_ACTION);
832
+
833
+ TY_(CheckAttributes)(doc, node);
834
+
835
+ if (!action)
836
+ TY_(ReportMissingAttr)(doc, node, "action");
837
+ }
838
+
839
+ /* reports missing content attribute */
840
+ void CheckMETA( TidyDocImpl* doc, Node *node )
841
+ {
842
+ AttVal *content = TY_(AttrGetById)(node, TidyAttr_CONTENT);
843
+
844
+ TY_(CheckAttributes)(doc, node);
845
+
846
+ if (!content)
847
+ TY_(ReportMissingAttr)( doc, node, "content" );
848
+ /* name or http-equiv attribute must also be set */
849
+ }
850
+
851
+
852
+ Bool TY_(nodeIsText)( Node* node )
853
+ {
854
+ return ( node && node->type == TextNode );
855
+ }
856
+
857
+ Bool TY_(nodeHasText)( TidyDocImpl* doc, Node* node )
858
+ {
859
+ if ( doc && node )
860
+ {
861
+ uint ix;
862
+ Lexer* lexer = doc->lexer;
863
+ for ( ix = node->start; ix < node->end; ++ix )
864
+ {
865
+ /* whitespace */
866
+ if ( !TY_(IsWhite)( lexer->lexbuf[ix] ) )
867
+ return yes;
868
+ }
869
+ }
870
+ return no;
871
+ }
872
+
873
+ Bool TY_(nodeIsElement)( Node* node )
874
+ {
875
+ return ( node &&
876
+ (node->type == StartTag || node->type == StartEndTag) );
877
+ }
878
+
879
+ #if 0
880
+ /* Compare & result to operand. If equal, then all bits
881
+ ** requested are set.
882
+ */
883
+ Bool nodeMatchCM( Node* node, uint contentModel )
884
+ {
885
+ return ( node && node->tag &&
886
+ (node->tag->model & contentModel) == contentModel );
887
+ }
888
+ #endif
889
+
890
+ /* True if any of the bits requested are set.
891
+ */
892
+ Bool TY_(nodeHasCM)( Node* node, uint contentModel )
893
+ {
894
+ return ( node && node->tag &&
895
+ (node->tag->model & contentModel) != 0 );
896
+ }
897
+
898
+ Bool TY_(nodeCMIsBlock)( Node* node )
899
+ {
900
+ return TY_(nodeHasCM)( node, CM_BLOCK );
901
+ }
902
+ Bool TY_(nodeCMIsInline)( Node* node )
903
+ {
904
+ return TY_(nodeHasCM)( node, CM_INLINE );
905
+ }
906
+ Bool TY_(nodeCMIsEmpty)( Node* node )
907
+ {
908
+ return TY_(nodeHasCM)( node, CM_EMPTY );
909
+ }
910
+
911
+ Bool TY_(nodeIsHeader)( Node* node )
912
+ {
913
+ TidyTagId tid = TagId( node );
914
+ return ( tid && (
915
+ tid == TidyTag_H1 ||
916
+ tid == TidyTag_H2 ||
917
+ tid == TidyTag_H3 ||
918
+ tid == TidyTag_H4 ||
919
+ tid == TidyTag_H5 ||
920
+ tid == TidyTag_H6 ));
921
+ }
922
+
923
+ uint TY_(nodeHeaderLevel)( Node* node )
924
+ {
925
+ TidyTagId tid = TagId( node );
926
+ switch ( tid )
927
+ {
928
+ case TidyTag_H1:
929
+ return 1;
930
+ case TidyTag_H2:
931
+ return 2;
932
+ case TidyTag_H3:
933
+ return 3;
934
+ case TidyTag_H4:
935
+ return 4;
936
+ case TidyTag_H5:
937
+ return 5;
938
+ case TidyTag_H6:
939
+ return 6;
940
+ default:
941
+ {
942
+ /* fall through */
943
+ }
944
+ }
945
+ return 0;
946
+ }
947
+
948
+ /*
949
+ * local variables:
950
+ * mode: c
951
+ * indent-tabs-mode: nil
952
+ * c-basic-offset: 4
953
+ * eval: (c-set-offset 'substatement-open 0)
954
+ * end:
955
+ */