OSDN Git Service

Initial commit
[wordring-tm/wordring-tm.git] / third_party / tidy-html5-master / src / win32tc.c
1 /* win32tc.c -- Interface to Win32 transcoding routines
2
3   (c) 1998-2008 (W3C) MIT, ERCIM, Keio University
4   See tidy.h for the copyright notice.
5
6 */
7
8 /* keep these here to keep file non-empty */
9 #include "tidy.h"
10 #include "forward.h"
11 #include "streamio.h"
12 #include "tmbstr.h"
13 #include "utf8.h"
14
15 #ifdef TIDY_WIN32_MLANG_SUPPORT
16
17 #define VC_EXTRALEAN
18 #define CINTERFACE
19 #define COBJMACROS
20
21 #include <windows.h>
22 #include <mlang.h>
23
24 #undef COBJMACROS
25 #undef CINTERFACE
26 #undef VC_EXTRALEAN
27
28 /* maximum number of bytes for a single character */
29 #define TC_INBUFSIZE  16
30
31 /* maximum number of characters per byte sequence */
32 #define TC_OUTBUFSIZE 16
33
34 #define CreateMLangObject(p) \
35   CoCreateInstance( \
36         &CLSID_CMLangConvertCharset, \
37         NULL, \
38         CLSCTX_ALL, \
39         &IID_IMLangConvertCharset, \
40         (VOID **)&p);
41
42
43 /* Character Set to Microsoft Windows Codepage Identifier map,     */
44 /* from <rotor/sscli/clr/src/classlibnative/nls/encodingdata.cpp>. */
45
46 /* note: the 'safe' field indicates whether this encoding can be   */
47 /* read/written character-by-character; this does not apply to     */
48 /* various stateful encodings such as ISO-2022 or UTF-7, these     */
49 /* must be read/written as a complete stream. It is possible that  */
50 /* some 'unsafe' encodings are marked as 'save'.                   */
51
52 /* todo: cleanup; Tidy should use only a single mapping table to   */
53 /* circumvent unsupported aliases in other transcoding libraries,  */
54 /* enable reverse lookup of encoding names and ease maintenance.   */
55
56 static struct _nameWinCPMap
57 {
58     tmbstr name;
59     uint wincp;
60     Bool safe;
61 } const NameWinCPMap[] = {
62   { "cp037",                                            37, yes },
63   { "csibm037",                                         37, yes },
64   { "ebcdic-cp-ca",                                     37, yes },
65   { "ebcdic-cp-nl",                                     37, yes },
66   { "ebcdic-cp-us",                                     37, yes },
67   { "ebcdic-cp-wt",                                     37, yes },
68   { "ibm037",                                           37, yes },
69   { "cp437",                                           437, yes },
70   { "cspc8codepage437",                                437, yes },
71   { "ibm437",                                          437, yes },
72   { "cp500",                                           500, yes },
73   { "csibm500",                                        500, yes },
74   { "ebcdic-cp-be",                                    500, yes },
75   { "ebcdic-cp-ch",                                    500, yes },
76   { "ibm500",                                          500, yes },
77   { "asmo-708",                                        708, yes },
78   { "dos-720",                                         720, yes },
79   { "ibm737",                                          737, yes },
80   { "ibm775",                                          775, yes },
81   { "cp850",                                           850, yes },
82   { "ibm850",                                          850, yes },
83   { "cp852",                                           852, yes },
84   { "ibm852",                                          852, yes },
85   { "cp855",                                           855, yes },
86   { "ibm855",                                          855, yes },
87   { "cp857",                                           857, yes },
88   { "ibm857",                                          857, yes },
89   { "ccsid00858",                                      858, yes },
90   { "cp00858",                                         858, yes },
91   { "cp858",                                           858, yes },
92   { "ibm00858",                                        858, yes },
93   { "pc-multilingual-850+euro",                        858, yes },
94   { "cp860",                                           860, yes },
95   { "ibm860",                                          860, yes },
96   { "cp861",                                           861, yes },
97   { "ibm861",                                          861, yes },
98   { "cp862",                                           862, yes },
99   { "dos-862",                                         862, yes },
100   { "ibm862",                                          862, yes },
101   { "cp863",                                           863, yes },
102   { "ibm863",                                          863, yes },
103   { "cp864",                                           864, yes },
104   { "ibm864",                                          864, yes },
105   { "cp865",                                           865, yes },
106   { "ibm865",                                          865, yes },
107   { "cp866",                                           866, yes },
108   { "ibm866",                                          866, yes },
109   { "cp869",                                           869, yes },
110   { "ibm869",                                          869, yes },
111   { "cp870",                                           870, yes },
112   { "csibm870",                                        870, yes },
113   { "ebcdic-cp-roece",                                 870, yes },
114   { "ebcdic-cp-yu",                                    870, yes },
115   { "ibm870",                                          870, yes },
116   { "dos-874",                                         874, yes },
117   { "iso-8859-11",                                     874, yes },
118   { "tis-620",                                         874, yes },
119   { "windows-874",                                     874, yes },
120   { "cp875",                                           875, yes },
121   { "csshiftjis",                                      932, yes },
122   { "cswindows31j",                                    932, yes },
123   { "ms_kanji",                                        932, yes },
124   { "shift-jis",                                       932, yes },
125   { "shift_jis",                                       932, yes },
126   { "sjis",                                            932, yes },
127   { "x-ms-cp932",                                      932, yes },
128   { "x-sjis",                                          932, yes },
129   { "chinese",                                         936, yes },
130   { "cn-gb",                                           936, yes },
131   { "csgb2312",                                        936, yes },
132   { "csgb231280",                                      936, yes },
133   { "csiso58gb231280",                                 936, yes },
134   { "gb2312",                                          936, yes },
135   { "gb2312-80",                                       936, yes },
136   { "gb231280",                                        936, yes },
137   { "gb_2312-80",                                      936, yes },
138   { "gbk",                                             936, yes },
139   { "iso-ir-58",                                       936, yes },
140   { "csksc56011987",                                   949, yes },
141   { "iso-ir-149",                                      949, yes },
142   { "korean",                                          949, yes },
143   { "ks-c-5601",                                       949, yes },
144   { "ks-c5601",                                        949, yes },
145   { "ks_c_5601",                                       949, yes },
146   { "ks_c_5601-1987",                                  949, yes },
147   { "ks_c_5601-1989",                                  949, yes },
148   { "ks_c_5601_1987",                                  949, yes },
149   { "ksc5601",                                         949, yes },
150   { "ksc_5601",                                        949, yes },
151   { "big5",                                            950, yes },
152   { "big5-hkscs",                                      950, yes },
153   { "cn-big5",                                         950, yes },
154   { "csbig5",                                          950, yes },
155   { "x-x-big5",                                        950, yes },
156   { "cp1026",                                         1026, yes },
157   { "csibm1026",                                      1026, yes },
158   { "ibm1026",                                        1026, yes },
159   { "ibm01047",                                       1047, yes },
160   { "ccsid01140",                                     1140, yes },
161   { "cp01140",                                        1140, yes },
162   { "ebcdic-us-37+euro",                              1140, yes },
163   { "ibm01140",                                       1140, yes },
164   { "ccsid01141",                                     1141, yes },
165   { "cp01141",                                        1141, yes },
166   { "ebcdic-de-273+euro",                             1141, yes },
167   { "ibm01141",                                       1141, yes },
168   { "ccsid01142",                                     1142, yes },
169   { "cp01142",                                        1142, yes },
170   { "ebcdic-dk-277+euro",                             1142, yes },
171   { "ebcdic-no-277+euro",                             1142, yes },
172   { "ibm01142",                                       1142, yes },
173   { "ccsid01143",                                     1143, yes },
174   { "cp01143",                                        1143, yes },
175   { "ebcdic-fi-278+euro",                             1143, yes },
176   { "ebcdic-se-278+euro",                             1143, yes },
177   { "ibm01143",                                       1143, yes },
178   { "ccsid01144",                                     1144, yes },
179   { "cp01144",                                        1144, yes },
180   { "ebcdic-it-280+euro",                             1144, yes },
181   { "ibm01144",                                       1144, yes },
182   { "ccsid01145",                                     1145, yes },
183   { "cp01145",                                        1145, yes },
184   { "ebcdic-es-284+euro",                             1145, yes },
185   { "ibm01145",                                       1145, yes },
186   { "ccsid01146",                                     1146, yes },
187   { "cp01146",                                        1146, yes },
188   { "ebcdic-gb-285+euro",                             1146, yes },
189   { "ibm01146",                                       1146, yes },
190   { "ccsid01147",                                     1147, yes },
191   { "cp01147",                                        1147, yes },
192   { "ebcdic-fr-297+euro",                             1147, yes },
193   { "ibm01147",                                       1147, yes },
194   { "ccsid01148",                                     1148, yes },
195   { "cp01148",                                        1148, yes },
196   { "ebcdic-international-500+euro",                  1148, yes },
197   { "ibm01148",                                       1148, yes },
198   { "ccsid01149",                                     1149, yes },
199   { "cp01149",                                        1149, yes },
200   { "ebcdic-is-871+euro",                             1149, yes },
201   { "ibm01149",                                       1149, yes },
202   { "iso-10646-ucs-2",                                1200, yes },
203   { "ucs-2",                                          1200, yes },
204   { "unicode",                                        1200, yes },
205   { "utf-16",                                         1200, yes },
206   { "utf-16le",                                       1200, yes },
207   { "unicodefffe",                                    1201, yes },
208   { "utf-16be",                                       1201, yes },
209   { "windows-1250",                                   1250, yes },
210   { "x-cp1250",                                       1250, yes },
211   { "windows-1251",                                   1251, yes },
212   { "x-cp1251",                                       1251, yes },
213   { "windows-1252",                                   1252, yes },
214   { "x-ansi",                                         1252, yes },
215   { "windows-1253",                                   1253, yes },
216   { "windows-1254",                                   1254, yes },
217   { "windows-1255",                                   1255, yes },
218   { "cp1256",                                         1256, yes },
219   { "windows-1256",                                   1256, yes },
220   { "windows-1257",                                   1257, yes },
221   { "windows-1258",                                   1258, yes },
222   { "johab",                                          1361, yes },
223   { "macintosh",                                     10000, yes },
224   { "x-mac-japanese",                                10001, yes },
225   { "x-mac-chinesetrad",                             10002, yes },
226   { "x-mac-korean",                                  10003, yes },
227   { "x-mac-arabic",                                  10004, yes },
228   { "x-mac-hebrew",                                  10005, yes },
229   { "x-mac-greek",                                   10006, yes },
230   { "x-mac-cyrillic",                                10007, yes },
231   { "x-mac-chinesesimp",                             10008, yes },
232   { "x-mac-romanian",                                10010, yes },
233   { "x-mac-ukrainian",                               10017, yes },
234   { "x-mac-thai",                                    10021, yes },
235   { "x-mac-ce",                                      10029, yes },
236   { "x-mac-icelandic",                               10079, yes },
237   { "x-mac-turkish",                                 10081, yes },
238   { "x-mac-croatian",                                10082, yes },
239   { "x-chinese-cns",                                 20000, yes },
240   { "x-cp20001",                                     20001, yes },
241   { "x-chinese-eten",                                20002, yes },
242   { "x-cp20003",                                     20003, yes },
243   { "x-cp20004",                                     20004, yes },
244   { "x-cp20005",                                     20005, yes },
245   { "irv",                                           20105, yes },
246   { "x-ia5",                                         20105, yes },
247   { "din_66003",                                     20106, yes },
248   { "german",                                        20106, yes },
249   { "x-ia5-german",                                  20106, yes },
250   { "sen_850200_b",                                  20107, yes },
251   { "swedish",                                       20107, yes },
252   { "x-ia5-swedish",                                 20107, yes },
253   { "norwegian",                                     20108, yes },
254   { "ns_4551-1",                                     20108, yes },
255   { "x-ia5-norwegian",                               20108, yes },
256   { "ansi_x3.4-1968",                                20127, yes },
257   { "ansi_x3.4-1986",                                20127, yes },
258   { "ascii",                                         20127, yes },
259   { "cp367",                                         20127, yes },
260   { "csascii",                                       20127, yes },
261   { "ibm367",                                        20127, yes },
262   { "iso-ir-6",                                      20127, yes },
263   { "iso646-us",                                     20127, yes },
264   { "iso_646.irv:1991",                              20127, yes },
265   { "us",                                            20127, yes },
266   { "us-ascii",                                      20127, yes },
267   { "x-cp20261",                                     20261, yes },
268   { "x-cp20269",                                     20269, yes },
269   { "cp273",                                         20273, yes },
270   { "csibm273",                                      20273, yes },
271   { "ibm273",                                        20273, yes },
272   { "csibm277",                                      20277, yes },
273   { "ebcdic-cp-dk",                                  20277, yes },
274   { "ebcdic-cp-no",                                  20277, yes },
275   { "ibm277",                                        20277, yes },
276   { "cp278",                                         20278, yes },
277   { "csibm278",                                      20278, yes },
278   { "ebcdic-cp-fi",                                  20278, yes },
279   { "ebcdic-cp-se",                                  20278, yes },
280   { "ibm278",                                        20278, yes },
281   { "cp280",                                         20280, yes },
282   { "csibm280",                                      20280, yes },
283   { "ebcdic-cp-it",                                  20280, yes },
284   { "ibm280",                                        20280, yes },
285   { "cp284",                                         20284, yes },
286   { "csibm284",                                      20284, yes },
287   { "ebcdic-cp-es",                                  20284, yes },
288   { "ibm284",                                        20284, yes },
289   { "cp285",                                         20285, yes },
290   { "csibm285",                                      20285, yes },
291   { "ebcdic-cp-gb",                                  20285, yes },
292   { "ibm285",                                        20285, yes },
293   { "cp290",                                         20290, yes },
294   { "csibm290",                                      20290, yes },
295   { "ebcdic-jp-kana",                                20290, yes },
296   { "ibm290",                                        20290, yes },
297   { "cp297",                                         20297, yes },
298   { "csibm297",                                      20297, yes },
299   { "ebcdic-cp-fr",                                  20297, yes },
300   { "ibm297",                                        20297, yes },
301   { "cp420",                                         20420, yes },
302   { "csibm420",                                      20420, yes },
303   { "ebcdic-cp-ar1",                                 20420, yes },
304   { "ibm420",                                        20420, yes },
305   { "cp423",                                         20423, yes },
306   { "csibm423",                                      20423, yes },
307   { "ebcdic-cp-gr",                                  20423, yes },
308   { "ibm423",                                        20423, yes },
309   { "cp424",                                         20424, yes },
310   { "csibm424",                                      20424, yes },
311   { "ebcdic-cp-he",                                  20424, yes },
312   { "ibm424",                                        20424, yes },
313   { "x-ebcdic-koreanextended",                       20833, yes },
314   { "csibmthai",                                     20838, yes },
315   { "ibm-thai",                                      20838, yes },
316   { "cskoi8r",                                       20866, yes },
317   { "koi",                                           20866, yes },
318   { "koi8",                                          20866, yes },
319   { "koi8-r",                                        20866, yes },
320   { "koi8r",                                         20866, yes },
321   { "cp871",                                         20871, yes },
322   { "csibm871",                                      20871, yes },
323   { "ebcdic-cp-is",                                  20871, yes },
324   { "ibm871",                                        20871, yes },
325   { "cp880",                                         20880, yes },
326   { "csibm880",                                      20880, yes },
327   { "ebcdic-cyrillic",                               20880, yes },
328   { "ibm880",                                        20880, yes },
329   { "cp905",                                         20905, yes },
330   { "csibm905",                                      20905, yes },
331   { "ebcdic-cp-tr",                                  20905, yes },
332   { "ibm905",                                        20905, yes },
333   { "ccsid00924",                                    20924, yes },
334   { "cp00924",                                       20924, yes },
335   { "ebcdic-latin9--euro",                           20924, yes },
336   { "ibm00924",                                      20924, yes },
337   { "x-cp20936",                                     20936, yes },
338   { "x-cp20949",                                     20949, yes },
339   { "cp1025",                                        21025, yes },
340   { "x-cp21027",                                     21027, yes },
341   { "koi8-ru",                                       21866, yes },
342   { "koi8-u",                                        21866, yes },
343   { "cp819",                                         28591, yes },
344   { "csisolatin1",                                   28591, yes },
345   { "ibm819",                                        28591, yes },
346   { "iso-8859-1",                                    28591, yes },
347   { "iso-ir-100",                                    28591, yes },
348   { "iso8859-1",                                     28591, yes },
349   { "iso_8859-1",                                    28591, yes },
350   { "iso_8859-1:1987",                               28591, yes },
351   { "l1",                                            28591, yes },
352   { "latin1",                                        28591, yes },
353   { "csisolatin2",                                   28592, yes },
354   { "iso-8859-2",                                    28592, yes },
355   { "iso-ir-101",                                    28592, yes },
356   { "iso8859-2",                                     28592, yes },
357   { "iso_8859-2",                                    28592, yes },
358   { "iso_8859-2:1987",                               28592, yes },
359   { "l2",                                            28592, yes },
360   { "latin2",                                        28592, yes },
361   { "csisolatin3",                                   28593, yes },
362   { "iso-8859-3",                                    28593, yes },
363   { "iso-ir-109",                                    28593, yes },
364   { "iso_8859-3",                                    28593, yes },
365   { "iso_8859-3:1988",                               28593, yes },
366   { "l3",                                            28593, yes },
367   { "latin3",                                        28593, yes },
368   { "csisolatin4",                                   28594, yes },
369   { "iso-8859-4",                                    28594, yes },
370   { "iso-ir-110",                                    28594, yes },
371   { "iso_8859-4",                                    28594, yes },
372   { "iso_8859-4:1988",                               28594, yes },
373   { "l4",                                            28594, yes },
374   { "latin4",                                        28594, yes },
375   { "csisolatincyrillic",                            28595, yes },
376   { "cyrillic",                                      28595, yes },
377   { "iso-8859-5",                                    28595, yes },
378   { "iso-ir-144",                                    28595, yes },
379   { "iso_8859-5",                                    28595, yes },
380   { "iso_8859-5:1988",                               28595, yes },
381   { "arabic",                                        28596, yes },
382   { "csisolatinarabic",                              28596, yes },
383   { "ecma-114",                                      28596, yes },
384   { "iso-8859-6",                                    28596, yes },
385   { "iso-ir-127",                                    28596, yes },
386   { "iso_8859-6",                                    28596, yes },
387   { "iso_8859-6:1987",                               28596, yes },
388   { "csisolatingreek",                               28597, yes },
389   { "ecma-118",                                      28597, yes },
390   { "elot_928",                                      28597, yes },
391   { "greek",                                         28597, yes },
392   { "greek8",                                        28597, yes },
393   { "iso-8859-7",                                    28597, yes },
394   { "iso-ir-126",                                    28597, yes },
395   { "iso_8859-7",                                    28597, yes },
396   { "iso_8859-7:1987",                               28597, yes },
397   { "csisolatinhebrew",                              28598, yes },
398   { "hebrew",                                        28598, yes },
399   { "iso-8859-8",                                    28598, yes },
400   { "iso-ir-138",                                    28598, yes },
401   { "iso_8859-8",                                    28598, yes },
402   { "iso_8859-8:1988",                               28598, yes },
403   { "logical",                                       28598, yes },
404   { "visual",                                        28598, yes },
405   { "csisolatin5",                                   28599, yes },
406   { "iso-8859-9",                                    28599, yes },
407   { "iso-ir-148",                                    28599, yes },
408   { "iso_8859-9",                                    28599, yes },
409   { "iso_8859-9:1989",                               28599, yes },
410   { "l5",                                            28599, yes },
411   { "latin5",                                        28599, yes },
412   { "iso-8859-13",                                   28603, yes },
413   { "csisolatin9",                                   28605, yes },
414   { "iso-8859-15",                                   28605, yes },
415   { "iso_8859-15",                                   28605, yes },
416   { "l9",                                            28605, yes },
417   { "latin9",                                        28605, yes },
418   { "x-europa",                                      29001, yes },
419   { "iso-8859-8-i",                                  38598, yes },
420   { "iso-2022-jp",                                   50220,  no },
421   { "csiso2022jp",                                   50221,  no },
422   { "csiso2022kr",                                   50225,  no },
423   { "iso-2022-kr",                                   50225,  no },
424   { "iso-2022-kr-7",                                 50225,  no },
425   { "iso-2022-kr-7bit",                              50225,  no },
426   { "cp50227",                                       50227,  no },
427   { "x-cp50227",                                     50227,  no },
428   { "cp930",                                         50930, yes },
429   { "x-ebcdic-japaneseanduscanada",                  50931, yes },
430   { "cp933",                                         50933, yes },
431   { "cp935",                                         50935, yes },
432   { "cp937",                                         50937, yes },
433   { "cp939",                                         50939, yes },
434   { "cseucpkdfmtjapanese",                           51932, yes },
435   { "euc-jp",                                        51932, yes },
436   { "extended_unix_code_packed_format_for_japanese", 51932, yes },
437   { "iso-2022-jpeuc",                                51932, yes },
438   { "x-euc",                                         51932, yes },
439   { "x-euc-jp",                                      51932, yes },
440   { "euc-cn",                                        51936, yes },
441   { "x-euc-cn",                                      51936, yes },
442   { "cseuckr",                                       51949, yes },
443   { "euc-kr",                                        51949, yes },
444   { "iso-2022-kr-8",                                 51949, yes },
445   { "iso-2022-kr-8bit",                              51949, yes },
446   { "hz-gb-2312",                                    52936,  no },
447   { "gb18030",                                       54936, yes },
448   { "x-iscii-de",                                    57002, yes },
449   { "x-iscii-be",                                    57003, yes },
450   { "x-iscii-ta",                                    57004, yes },
451   { "x-iscii-te",                                    57005, yes },
452   { "x-iscii-as",                                    57006, yes },
453   { "x-iscii-or",                                    57007, yes },
454   { "x-iscii-ka",                                    57008, yes },
455   { "x-iscii-ma",                                    57009, yes },
456   { "x-iscii-gu",                                    57010, yes },
457   { "x-iscii-pa",                                    57011, yes },
458   { "csunicode11utf7",                               65000,  no },
459   { "unicode-1-1-utf-7",                             65000,  no },
460   { "unicode-2-0-utf-7",                             65000,  no },
461   { "utf-7",                                         65000,  no },
462   { "x-unicode-1-1-utf-7",                           65000,  no },
463   { "x-unicode-2-0-utf-7",                           65000,  no },
464   { "unicode-1-1-utf-8",                             65001, yes },
465   { "unicode-2-0-utf-8",                             65001, yes },
466   { "utf-8",                                         65001, yes },
467   { "x-unicode-1-1-utf-8",                           65001, yes },
468   { "x-unicode-2-0-utf-8",                           65001, yes },
469
470   /* final entry */
471   { NULL,                                                0,  no }
472 };
473
474 uint TY_(Win32MLangGetCPFromName)(TidyAllocator *allocator, ctmbstr encoding)
475 {
476     uint i;
477     tmbstr enc;
478
479     /* ensure name is in lower case */
480     enc = TY_(tmbstrdup)(allocator,encoding);
481     enc = TY_(tmbstrtolower)(enc);
482
483     for (i = 0; NameWinCPMap[i].name; ++i)
484     {
485         if (TY_(tmbstrcmp)(NameWinCPMap[i].name, enc) == 0)
486         {
487             IMLangConvertCharset * p = NULL;
488             uint wincp = NameWinCPMap[i].wincp;
489             HRESULT hr;
490
491             TidyFree(allocator, enc);
492
493             /* currently no support for unsafe encodings */
494             if (!NameWinCPMap[i].safe)
495                 return 0;
496
497             /* hack for config.c */
498             CoInitialize(NULL);
499             hr = CreateMLangObject(p);
500
501             if (hr != S_OK || !p)
502             {
503                 wincp = 0;
504             }
505             else
506             {
507                 hr = IMLangConvertCharset_Initialize(p, wincp, 1200, 0);
508
509                 if (hr != S_OK)
510                     wincp = 0;
511
512                 IMLangConvertCharset_Release(p);
513                 p = NULL;
514             }
515
516             CoUninitialize();
517
518             return wincp;
519         }
520     }
521
522     TidyFree(allocator, enc);
523     return 0;
524 }
525
526 Bool TY_(Win32MLangInitInputTranscoder)(StreamIn * in, uint wincp)
527 {
528     IMLangConvertCharset * p = NULL;
529     HRESULT hr;
530
531     assert( in != NULL );
532
533     CoInitialize(NULL);
534
535     if (wincp == 0)
536     {
537         /* no codepage found for this encoding */
538         return no;
539     }
540
541     hr = CreateMLangObject(p);
542
543     if (hr != S_OK || !p)
544     {
545         /* MLang not supported */
546         return no;
547     }
548
549     hr = IMLangConvertCharset_Initialize(p, wincp, 1200, 0);
550
551     if (hr != S_OK)
552     {
553         /* encoding not supported, insufficient memory, etc. */
554         return no;
555     }
556
557     in->mlang = p;
558
559     return yes;
560 }
561
562 void TY_(Win32MLangUninitInputTranscoder)(StreamIn * in)
563 {
564     IMLangConvertCharset * p;
565
566     assert( in != NULL );
567
568     p = (IMLangConvertCharset *)in->mlang;
569     if (p)
570     {
571         IMLangConvertCharset_Release(p);
572         p = NULL;
573         in->mlang = NULL;
574     }
575
576     CoUninitialize();
577 }
578
579 #if 0
580 Bool Win32MLangInitOutputTranscoder(TidyAllocator *allocator, StreamOut * out, tmbstr encoding)
581 {
582     IMLangConvertCharset * p = NULL;
583     HRESULT hr;
584     uint wincp;
585
586     assert( out != NULL );
587
588     CoInitialize(NULL);
589
590     wincp = TY_(Win32MLangGetCPFromName)(allocator, encoding);
591     if (wincp == 0)
592     {
593         /* no codepage found for this encoding */
594         return no;
595     }
596
597     hr = CreateMLangObject(p);
598
599     if (hr != S_OK || !p)
600     {
601         /* MLang not supported */
602         return no;
603     }
604
605     IMLangConvertCharset_Initialize(p, 1200, wincp, MLCONVCHARF_NOBESTFITCHARS);
606
607     if (hr != S_OK)
608     {
609         /* encoding not supported, insufficient memory, etc. */
610         return no;
611     }
612
613     out->mlang = p;
614
615     return yes;
616 }
617
618 void Win32MLangUninitOutputTranscoder(StreamOut * out)
619 {
620     IMLangConvertCharset * p;
621
622     assert( out != NULL );
623
624     p = (IMLangConvertCharset *)out->mlang;
625     if (p)
626     {
627         IMLangConvertCharset_Release(p);
628         p = NULL;
629         out->mlang = NULL;
630     }
631
632     CoUninitialize();
633 }
634 #endif
635
636 int TY_(Win32MLangGetChar)(byte firstByte, StreamIn * in, uint * bytesRead)
637 {
638     IMLangConvertCharset * p;
639     TidyInputSource * source;
640     CHAR inbuf[TC_INBUFSIZE] = { 0 };
641     WCHAR outbuf[TC_OUTBUFSIZE] = { 0 };
642     HRESULT hr = S_OK;
643     size_t inbufsize = 0;
644
645     assert( in != NULL );
646     assert( &in->source != NULL );
647     assert( bytesRead != NULL );
648     assert( in->mlang != NULL );
649
650     p = (IMLangConvertCharset *)in->mlang;
651     source = &in->source;
652
653     inbuf[inbufsize++] = (CHAR)firstByte;
654
655     while(inbufsize < TC_INBUFSIZE)
656     {
657         UINT outbufsize = TC_OUTBUFSIZE;
658         UINT readNow = inbufsize;
659         int nextByte = EndOfStream;
660
661         hr = IMLangConvertCharset_DoConversionToUnicode(p, inbuf, &readNow, outbuf, &outbufsize);
662
663         assert( hr == S_OK );
664         assert( outbufsize <= 2 );
665
666         if (outbufsize == 2)
667         {
668             /* U+10000-U+10FFFF are returned as a pair of surrogates */
669             tchar m = (tchar)outbuf[0];
670             tchar n = (tchar)outbuf[1];
671             assert( TY_(IsHighSurrogate)(n) && TY_(IsLowSurrogate)(m) );
672             *bytesRead = readNow;
673             return (int)TY_(CombineSurrogatePair)(n, m);
674         }
675
676         if (outbufsize == 1)
677         {
678             /* we found the character   */
679             /* set bytesRead and return */
680             *bytesRead = readNow;
681             return (int)outbuf[0];
682         }
683
684         /* we need more bytes */
685         nextByte = source->getByte(source->sourceData);
686
687         if (nextByte == EndOfStream)
688         {
689             /* todo: error message for broken stream? */
690
691             *bytesRead = readNow;
692             return EndOfStream;
693         }
694
695         inbuf[inbufsize++] = (CHAR)nextByte;
696     }
697
698     /* No full character found after reading TC_INBUFSIZE bytes, */
699     /* give up to read this stream, it's obviously unreadable.   */
700
701     /* todo: error message for broken stream? */
702     return EndOfStream;
703 }
704
705 Bool Win32MLangIsConvertible(tchar c, StreamOut * out)
706 {
707     IMLangConvertCharset * p;
708     UINT i = 1;
709     HRESULT hr;
710     WCHAR inbuf[2] = { 0 };
711     UINT inbufsize = 0;
712
713     assert( c != 0 );
714     assert( c <= 0x10FFFF );
715     assert( out != NULL );
716     assert( out->mlang != NULL );
717
718     if (c > 0xFFFF)
719     {
720         tchar high = 0;
721         tchar low = 0;
722
723         TY_(SplitSurrogatePair)(c, &low, &high);
724
725         inbuf[inbufsize++] = (WCHAR)low;
726         inbuf[inbufsize++] = (WCHAR)high;
727     }
728     else
729         inbuf[inbufsize++] = (WCHAR)c;
730
731     p = (IMLangConvertCharset *)out->mlang;
732     hr = IMLangConvertCharset_DoConversionFromUnicode(p, inbuf, &inbufsize, NULL, NULL);
733
734     return hr == S_OK ? yes : no;
735 }
736
737 void Win32MLangPutChar(tchar c, StreamOut * out, uint * bytesWritten)
738 {
739     IMLangConvertCharset * p;
740     TidyOutputSink * sink;
741     CHAR outbuf[TC_OUTBUFSIZE] = { 0 };
742     UINT outbufsize = TC_OUTBUFSIZE;
743     HRESULT hr = S_OK;
744     WCHAR inbuf[2] = { 0 };
745     UINT inbufsize = 0;
746     uint i;
747
748     assert( c != 0 );
749     assert( c <= 0x10FFFF );
750     assert( bytesWritten != NULL );
751     assert( out != NULL );
752     assert( &out->sink != NULL );
753     assert( out->mlang != NULL );
754
755     p = (IMLangConvertCharset *)out->mlang;
756     sink = &out->sink;
757
758     if (c > 0xFFFF)
759     {
760         tchar high = 0;
761         tchar low = 0;
762
763         TY_(SplitSurrogatePair)(c, &low, &high);
764
765         inbuf[inbufsize++] = (WCHAR)low;
766         inbuf[inbufsize++] = (WCHAR)high;
767     }
768     else
769         inbuf[inbufsize++] = (WCHAR)c;
770
771     hr = IMLangConvertCharset_DoConversionFromUnicode(p, inbuf, &inbufsize, outbuf, &outbufsize);
772     
773     assert( hr == S_OK );
774     assert( outbufsize > 0 );
775     assert( inbufsize == 1 || inbufsize == 2 );
776
777     for (i = 0; i < outbufsize; ++i)
778         sink->putByte(sink->sinkData, (byte)(outbuf[i]));
779
780     *bytesWritten = outbufsize;
781
782     return;
783 }
784
785 #endif /* TIDY_WIN32_MLANG_SUPPORT */
786
787 /*
788  * local variables:
789  * mode: c
790  * indent-tabs-mode: nil
791  * c-basic-offset: 4
792  * eval: (c-set-offset 'substatement-open 0)
793  * end:
794  */