1 /*-------------------------------------------------------------------------
2 *
3 * pg_wchar.h
4 * multibyte-character support
5 *
6 * Portions Copyright (c) 1996-2020, PostgreSQL Global Development Group
7 * Portions Copyright (c) 1994, Regents of the University of California
8 *
9 * src/include/mb/pg_wchar.h
10 *
11 * NOTES
12 * This is used both by the backend and by frontends, but should not be
13 * included by libpq client programs. In particular, a libpq client
14 * should not assume that the encoding IDs used by the version of libpq
15 * it's linked to match up with the IDs declared here.
16 *
17 *-------------------------------------------------------------------------
18 */
19 #ifndef PG_WCHAR_H
20 #define PG_WCHAR_H
21
22 /*
23 * The pg_wchar type
24 */
25 typedef unsigned int pg_wchar;
26
27 /*
28 * Maximum byte length of multibyte characters in any backend encoding
29 */
30 #define MAX_MULTIBYTE_CHAR_LEN 4
31
32 /*
33 * various definitions for EUC
34 */
35 #define SS2 0x8e /* single shift 2 (JIS0201) */
36 #define SS3 0x8f /* single shift 3 (JIS0212) */
37
38 /*
39 * SJIS validation macros
40 */
41 #define ISSJISHEAD(c) (((c) >= 0x81 && (c) <= 0x9f) || ((c) >= 0xe0 && (c) <= 0xfc))
42 #define ISSJISTAIL(c) (((c) >= 0x40 && (c) <= 0x7e) || ((c) >= 0x80 && (c) <= 0xfc))
43
44 /*----------------------------------------------------
45 * MULE Internal Encoding (MIC)
46 *
47 * This encoding follows the design used within XEmacs; it is meant to
48 * subsume many externally-defined character sets. Each character includes
49 * identification of the character set it belongs to, so the encoding is
50 * general but somewhat bulky.
51 *
52 * Currently PostgreSQL supports 5 types of MULE character sets:
53 *
54 * 1) 1-byte ASCII characters. Each byte is below 0x80.
55 *
56 * 2) "Official" single byte charsets such as ISO-8859-1 (Latin1).
57 * Each MULE character consists of 2 bytes: LC1 + C1, where LC1 is
58 * an identifier for the charset (in the range 0x81 to 0x8d) and C1
59 * is the character code (in the range 0xa0 to 0xff).
60 *
61 * 3) "Private" single byte charsets such as SISHENG. Each MULE
62 * character consists of 3 bytes: LCPRV1 + LC12 + C1, where LCPRV1
63 * is a private-charset flag, LC12 is an identifier for the charset,
64 * and C1 is the character code (in the range 0xa0 to 0xff).
65 * LCPRV1 is either 0x9a (if LC12 is in the range 0xa0 to 0xdf)
66 * or 0x9b (if LC12 is in the range 0xe0 to 0xef).
67 *
68 * 4) "Official" multibyte charsets such as JIS X0208. Each MULE
69 * character consists of 3 bytes: LC2 + C1 + C2, where LC2 is
70 * an identifier for the charset (in the range 0x90 to 0x99) and C1
71 * and C2 form the character code (each in the range 0xa0 to 0xff).
72 *
73 * 5) "Private" multibyte charsets such as CNS 11643-1992 Plane 3.
74 * Each MULE character consists of 4 bytes: LCPRV2 + LC22 + C1 + C2,
75 * where LCPRV2 is a private-charset flag, LC22 is an identifier for
76 * the charset, and C1 and C2 form the character code (each in the range
77 * 0xa0 to 0xff). LCPRV2 is either 0x9c (if LC22 is in the range 0xf0
78 * to 0xf4) or 0x9d (if LC22 is in the range 0xf5 to 0xfe).
79 *
80 * "Official" encodings are those that have been assigned code numbers by
81 * the XEmacs project; "private" encodings have Postgres-specific charset
82 * identifiers.
83 *
84 * See the "XEmacs Internals Manual", available at http://www.xemacs.org,
85 * for more details. Note that for historical reasons, Postgres'
86 * private-charset flag values do not match what XEmacs says they should be,
87 * so this isn't really exactly MULE (not that private charsets would be
88 * interoperable anyway).
89 *
90 * Note that XEmacs's implementation is different from what emacs does.
91 * We follow emacs's implementation, rather than XEmacs's.
92 *----------------------------------------------------
93 */
94
95 /*
96 * Charset identifiers (also called "leading bytes" in the MULE documentation)
97 */
98
99 /*
100 * Charset IDs for official single byte encodings (0x81-0x8e)
101 */
102 #define LC_ISO8859_1 0x81 /* ISO8859 Latin 1 */
103 #define LC_ISO8859_2 0x82 /* ISO8859 Latin 2 */
104 #define LC_ISO8859_3 0x83 /* ISO8859 Latin 3 */
105 #define LC_ISO8859_4 0x84 /* ISO8859 Latin 4 */
106 #define LC_TIS620 0x85 /* Thai (not supported yet) */
107 #define LC_ISO8859_7 0x86 /* Greek (not supported yet) */
108 #define LC_ISO8859_6 0x87 /* Arabic (not supported yet) */
109 #define LC_ISO8859_8 0x88 /* Hebrew (not supported yet) */
110 #define LC_JISX0201K 0x89 /* Japanese 1 byte kana */
111 #define LC_JISX0201R 0x8a /* Japanese 1 byte Roman */
112 /* Note that 0x8b seems to be unused as of Emacs 20.7.
113 * However, there might be a chance that 0x8b could be used
114 * in later versions of Emacs.
115 */
116 #define LC_KOI8_R 0x8b /* Cyrillic KOI8-R */
117 #define LC_ISO8859_5 0x8c /* ISO8859 Cyrillic */
118 #define LC_ISO8859_9 0x8d /* ISO8859 Latin 5 (not supported yet) */
119 #define LC_ISO8859_15 0x8e /* ISO8859 Latin 15 (not supported yet) */
120 /* #define CONTROL_1 0x8f control characters (unused) */
121
122 /* Is a leading byte for "official" single byte encodings? */
123 #define IS_LC1(c) ((unsigned char)(c) >= 0x81 && (unsigned char)(c) <= 0x8d)
124
125 /*
126 * Charset IDs for official multibyte encodings (0x90-0x99)
127 * 0x9a-0x9d are free. 0x9e and 0x9f are reserved.
128 */
129 #define LC_JISX0208_1978 0x90 /* Japanese Kanji, old JIS (not supported) */
130 #define LC_GB2312_80 0x91 /* Chinese */
131 #define LC_JISX0208 0x92 /* Japanese Kanji (JIS X 0208) */
132 #define LC_KS5601 0x93 /* Korean */
133 #define LC_JISX0212 0x94 /* Japanese Kanji (JIS X 0212) */
134 #define LC_CNS11643_1 0x95 /* CNS 11643-1992 Plane 1 */
135 #define LC_CNS11643_2 0x96 /* CNS 11643-1992 Plane 2 */
136 #define LC_JISX0213_1 0x97 /* Japanese Kanji (JIS X 0213 Plane 1)
137 * (not supported) */
138 #define LC_BIG5_1 0x98 /* Plane 1 Chinese traditional (not
139 * supported) */
140 #define LC_BIG5_2 0x99 /* Plane 1 Chinese traditional (not
141 * supported) */
142
143 /* Is a leading byte for "official" multibyte encodings? */
144 #define IS_LC2(c) ((unsigned char)(c) >= 0x90 && (unsigned char)(c) <= 0x99)
145
146 /*
147 * Postgres-specific prefix bytes for "private" single byte encodings
148 * (According to the MULE docs, we should be using 0x9e for this)
149 */
150 #define LCPRV1_A 0x9a
151 #define LCPRV1_B 0x9b
152 #define IS_LCPRV1(c) ((unsigned char)(c) == LCPRV1_A || (unsigned char)(c) == LCPRV1_B)
153 #define IS_LCPRV1_A_RANGE(c) \
154 ((unsigned char)(c) >= 0xa0 && (unsigned char)(c) <= 0xdf)
155 #define IS_LCPRV1_B_RANGE(c) \
156 ((unsigned char)(c) >= 0xe0 && (unsigned char)(c) <= 0xef)
157
158 /*
159 * Postgres-specific prefix bytes for "private" multibyte encodings
160 * (According to the MULE docs, we should be using 0x9f for this)
161 */
162 #define LCPRV2_A 0x9c
163 #define LCPRV2_B 0x9d
164 #define IS_LCPRV2(c) ((unsigned char)(c) == LCPRV2_A || (unsigned char)(c) == LCPRV2_B)
165 #define IS_LCPRV2_A_RANGE(c) \
166 ((unsigned char)(c) >= 0xf0 && (unsigned char)(c) <= 0xf4)
167 #define IS_LCPRV2_B_RANGE(c) \
168 ((unsigned char)(c) >= 0xf5 && (unsigned char)(c) <= 0xfe)
169
170 /*
171 * Charset IDs for private single byte encodings (0xa0-0xef)
172 */
173 #define LC_SISHENG 0xa0 /* Chinese SiSheng characters for
174 * PinYin/ZhuYin (not supported) */
175 #define LC_IPA 0xa1 /* IPA (International Phonetic
176 * Association) (not supported) */
177 #define LC_VISCII_LOWER 0xa2 /* Vietnamese VISCII1.1 lower-case (not
178 * supported) */
179 #define LC_VISCII_UPPER 0xa3 /* Vietnamese VISCII1.1 upper-case (not
180 * supported) */
181 #define LC_ARABIC_DIGIT 0xa4 /* Arabic digit (not supported) */
182 #define LC_ARABIC_1_COLUMN 0xa5 /* Arabic 1-column (not supported) */
183 #define LC_ASCII_RIGHT_TO_LEFT 0xa6 /* ASCII (left half of ISO8859-1) with
184 * right-to-left direction (not
185 * supported) */
186 #define LC_LAO 0xa7 /* Lao characters (ISO10646 0E80..0EDF)
187 * (not supported) */
188 #define LC_ARABIC_2_COLUMN 0xa8 /* Arabic 1-column (not supported) */
189
190 /*
191 * Charset IDs for private multibyte encodings (0xf0-0xff)
192 */
193 #define LC_INDIAN_1_COLUMN 0xf0 /* Indian charset for 1-column width
194 * glyphs (not supported) */
195 #define LC_TIBETAN_1_COLUMN 0xf1 /* Tibetan 1-column width glyphs (not
196 * supported) */
197 #define LC_UNICODE_SUBSET_2 0xf2 /* Unicode characters of the range
198 * U+2500..U+33FF. (not supported) */
199 #define LC_UNICODE_SUBSET_3 0xf3 /* Unicode characters of the range
200 * U+E000..U+FFFF. (not supported) */
201 #define LC_UNICODE_SUBSET 0xf4 /* Unicode characters of the range
202 * U+0100..U+24FF. (not supported) */
203 #define LC_ETHIOPIC 0xf5 /* Ethiopic characters (not supported) */
204 #define LC_CNS11643_3 0xf6 /* CNS 11643-1992 Plane 3 */
205 #define LC_CNS11643_4 0xf7 /* CNS 11643-1992 Plane 4 */
206 #define LC_CNS11643_5 0xf8 /* CNS 11643-1992 Plane 5 */
207 #define LC_CNS11643_6 0xf9 /* CNS 11643-1992 Plane 6 */
208 #define LC_CNS11643_7 0xfa /* CNS 11643-1992 Plane 7 */
209 #define LC_INDIAN_2_COLUMN 0xfb /* Indian charset for 2-column width
210 * glyphs (not supported) */
211 #define LC_TIBETAN 0xfc /* Tibetan (not supported) */
212 /* #define FREE 0xfd free (unused) */
213 /* #define FREE 0xfe free (unused) */
214 /* #define FREE 0xff free (unused) */
215
216 /*----------------------------------------------------
217 * end of MULE stuff
218 *----------------------------------------------------
219 */
220
221 /*
222 * PostgreSQL encoding identifiers
223 *
224 * WARNING: the order of this enum must be same as order of entries
225 * in the pg_enc2name_tbl[] array (in src/common/encnames.c), and
226 * in the pg_wchar_table[] array (in src/common/wchar.c)!
227 *
228 * If you add some encoding don't forget to check
229 * PG_ENCODING_BE_LAST macro.
230 *
231 * PG_SQL_ASCII is default encoding and must be = 0.
232 *
233 * XXX We must avoid renumbering any backend encoding until libpq's major
234 * version number is increased beyond 5; it turns out that the backend
235 * encoding IDs are effectively part of libpq's ABI as far as 8.2 initdb and
236 * psql are concerned.
237 */
238 typedef enum pg_enc
239 {
240 PG_SQL_ASCII = 0, /* SQL/ASCII */
241 PG_EUC_JP, /* EUC for Japanese */
242 PG_EUC_CN, /* EUC for Chinese */
243 PG_EUC_KR, /* EUC for Korean */
244 PG_EUC_TW, /* EUC for Taiwan */
245 PG_EUC_JIS_2004, /* EUC-JIS-2004 */
246 PG_UTF8, /* Unicode UTF8 */
247 PG_MULE_INTERNAL, /* Mule internal code */
248 PG_LATIN1, /* ISO-8859-1 Latin 1 */
249 PG_LATIN2, /* ISO-8859-2 Latin 2 */
250 PG_LATIN3, /* ISO-8859-3 Latin 3 */
251 PG_LATIN4, /* ISO-8859-4 Latin 4 */
252 PG_LATIN5, /* ISO-8859-9 Latin 5 */
253 PG_LATIN6, /* ISO-8859-10 Latin6 */
254 PG_LATIN7, /* ISO-8859-13 Latin7 */
255 PG_LATIN8, /* ISO-8859-14 Latin8 */
256 PG_LATIN9, /* ISO-8859-15 Latin9 */
257 PG_LATIN10, /* ISO-8859-16 Latin10 */
258 PG_WIN1256, /* windows-1256 */
259 PG_WIN1258, /* Windows-1258 */
260 PG_WIN866, /* (MS-DOS CP866) */
261 PG_WIN874, /* windows-874 */
262 PG_KOI8R, /* KOI8-R */
263 PG_WIN1251, /* windows-1251 */
264 PG_WIN1252, /* windows-1252 */
265 PG_ISO_8859_5, /* ISO-8859-5 */
266 PG_ISO_8859_6, /* ISO-8859-6 */
267 PG_ISO_8859_7, /* ISO-8859-7 */
268 PG_ISO_8859_8, /* ISO-8859-8 */
269 PG_WIN1250, /* windows-1250 */
270 PG_WIN1253, /* windows-1253 */
271 PG_WIN1254, /* windows-1254 */
272 PG_WIN1255, /* windows-1255 */
273 PG_WIN1257, /* windows-1257 */
274 PG_KOI8U, /* KOI8-U */
275 /* PG_ENCODING_BE_LAST points to the above entry */
276
277 /* followings are for client encoding only */
278 PG_SJIS, /* Shift JIS (Windows-932) */
279 PG_BIG5, /* Big5 (Windows-950) */
280 PG_GBK, /* GBK (Windows-936) */
281 PG_UHC, /* UHC (Windows-949) */
282 PG_GB18030, /* GB18030 */
283 PG_JOHAB, /* EUC for Korean JOHAB */
284 PG_SHIFT_JIS_2004, /* Shift-JIS-2004 */
285 _PG_LAST_ENCODING_ /* mark only */
286
287 } pg_enc;
288
289 #define PG_ENCODING_BE_LAST PG_KOI8U
290
291 /*
292 * Please use these tests before access to pg_enc2name_tbl[]
293 * or to other places...
294 */
295 #define PG_VALID_BE_ENCODING(_enc) \
296 ((_enc) >= 0 && (_enc) <= PG_ENCODING_BE_LAST)
297
298 #define PG_ENCODING_IS_CLIENT_ONLY(_enc) \
299 ((_enc) > PG_ENCODING_BE_LAST && (_enc) < _PG_LAST_ENCODING_)
300
301 #define PG_VALID_ENCODING(_enc) \
302 ((_enc) >= 0 && (_enc) < _PG_LAST_ENCODING_)
303
304 /* On FE are possible all encodings */
305 #define PG_VALID_FE_ENCODING(_enc) PG_VALID_ENCODING(_enc)
306
307 /*
308 * When converting strings between different encodings, we assume that space
309 * for converted result is 4-to-1 growth in the worst case. The rate for
310 * currently supported encoding pairs are within 3 (SJIS JIS X0201 half width
311 * kanna -> UTF8 is the worst case). So "4" should be enough for the moment.
312 *
313 * Note that this is not the same as the maximum character width in any
314 * particular encoding.
315 */
316 #define MAX_CONVERSION_GROWTH 4
317
318 /*
319 * Maximum byte length of the string equivalent to any one Unicode code point,
320 * in any backend encoding. The current value assumes that a 4-byte UTF-8
321 * character might expand by MAX_CONVERSION_GROWTH, which is a huge
322 * overestimate. But in current usage we don't allocate large multiples of
323 * this, so there's little point in being stingy.
324 */
325 #define MAX_UNICODE_EQUIVALENT_STRING 16
326
327 /*
328 * Table for mapping an encoding number to official encoding name and
329 * possibly other subsidiary data. Be careful to check encoding number
330 * before accessing a table entry!
331 *
332 * if (PG_VALID_ENCODING(encoding))
333 * pg_enc2name_tbl[ encoding ];
334 */
335 typedef struct pg_enc2name
336 {
337 const char *name;
338 pg_enc encoding;
339 #ifdef WIN32
340 unsigned codepage; /* codepage for WIN32 */
341 #endif
342 } pg_enc2name;
343
344 extern const pg_enc2name pg_enc2name_tbl[];
345
346 /*
347 * Encoding names for gettext
348 */
349 typedef struct pg_enc2gettext
350 {
351 pg_enc encoding;
352 const char *name;
353 } pg_enc2gettext;
354
355 extern const pg_enc2gettext pg_enc2gettext_tbl[];
356
357 /*
358 * pg_wchar stuff
359 */
360 typedef int (*mb2wchar_with_len_converter) (const unsigned char *from,
361 pg_wchar *to,
362 int len);
363
364 typedef int (*wchar2mb_with_len_converter) (const pg_wchar *from,
365 unsigned char *to,
366 int len);
367
368 typedef int (*mblen_converter) (const unsigned char *mbstr);
369
370 typedef int (*mbdisplaylen_converter) (const unsigned char *mbstr);
371
372 typedef bool (*mbcharacter_incrementer) (unsigned char *mbstr, int len);
373
374 typedef int (*mbverifier) (const unsigned char *mbstr, int len);
375
376 typedef struct
377 {
378 mb2wchar_with_len_converter mb2wchar_with_len; /* convert a multibyte
379 * string to a wchar */
380 wchar2mb_with_len_converter wchar2mb_with_len; /* convert a wchar string
381 * to a multibyte */
382 mblen_converter mblen; /* get byte length of a char */
383 mbdisplaylen_converter dsplen; /* get display width of a char */
384 mbverifier mbverify; /* verify multibyte sequence */
385 int maxmblen; /* max bytes for a char in this encoding */
386 } pg_wchar_tbl;
387
388 extern const pg_wchar_tbl pg_wchar_table[];
389
390 /*
391 * Data structures for conversions between UTF-8 and other encodings
392 * (UtfToLocal() and LocalToUtf()). In these data structures, characters of
393 * either encoding are represented by uint32 words; hence we can only support
394 * characters up to 4 bytes long. For example, the byte sequence 0xC2 0x89
395 * would be represented by 0x0000C289, and 0xE8 0xA2 0xB4 by 0x00E8A2B4.
396 *
397 * There are three possible ways a character can be mapped:
398 *
399 * 1. Using a radix tree, from source to destination code.
400 * 2. Using a sorted array of source -> destination code pairs. This
401 * method is used for "combining" characters. There are so few of
402 * them that building a radix tree would be wasteful.
403 * 3. Using a conversion function.
404 */
405
406 /*
407 * Radix tree for character conversion.
408 *
409 * Logically, this is actually four different radix trees, for 1-byte,
410 * 2-byte, 3-byte and 4-byte inputs. The 1-byte tree is a simple lookup
411 * table from source to target code. The 2-byte tree consists of two levels:
412 * one lookup table for the first byte, where the value in the lookup table
413 * points to a lookup table for the second byte. And so on.
414 *
415 * Physically, all the trees are stored in one big array, in 'chars16' or
416 * 'chars32', depending on the maximum value that needs to be represented. For
417 * each level in each tree, we also store lower and upper bound of allowed
418 * values - values outside those bounds are considered invalid, and are left
419 * out of the tables.
420 *
421 * In the intermediate levels of the trees, the values stored are offsets
422 * into the chars[16|32] array.
423 *
424 * In the beginning of the chars[16|32] array, there is always a number of
425 * zeros, so that you safely follow an index from an intermediate table
426 * without explicitly checking for a zero. Following a zero any number of
427 * times will always bring you to the dummy, all-zeros table in the
428 * beginning. This helps to shave some cycles when looking up values.
429 */
430 typedef struct
431 {
432 /*
433 * Array containing all the values. Only one of chars16 or chars32 is
434 * used, depending on how wide the values we need to represent are.
435 */
436 const uint16 *chars16;
437 const uint32 *chars32;
438
439 /* Radix tree for 1-byte inputs */
440 uint32 b1root; /* offset of table in the chars[16|32] array */
441 uint8 b1_lower; /* min allowed value for a single byte input */
442 uint8 b1_upper; /* max allowed value for a single byte input */
443
444 /* Radix tree for 2-byte inputs */
445 uint32 b2root; /* offset of 1st byte's table */
446 uint8 b2_1_lower; /* min/max allowed value for 1st input byte */
447 uint8 b2_1_upper;
448 uint8 b2_2_lower; /* min/max allowed value for 2nd input byte */
449 uint8 b2_2_upper;
450
451 /* Radix tree for 3-byte inputs */
452 uint32 b3root; /* offset of 1st byte's table */
453 uint8 b3_1_lower; /* min/max allowed value for 1st input byte */
454 uint8 b3_1_upper;
455 uint8 b3_2_lower; /* min/max allowed value for 2nd input byte */
456 uint8 b3_2_upper;
457 uint8 b3_3_lower; /* min/max allowed value for 3rd input byte */
458 uint8 b3_3_upper;
459
460 /* Radix tree for 4-byte inputs */
461 uint32 b4root; /* offset of 1st byte's table */
462 uint8 b4_1_lower; /* min/max allowed value for 1st input byte */
463 uint8 b4_1_upper;
464 uint8 b4_2_lower; /* min/max allowed value for 2nd input byte */
465 uint8 b4_2_upper;
466 uint8 b4_3_lower; /* min/max allowed value for 3rd input byte */
467 uint8 b4_3_upper;
468 uint8 b4_4_lower; /* min/max allowed value for 4th input byte */
469 uint8 b4_4_upper;
470
471 } pg_mb_radix_tree;
472
473 /*
474 * UTF-8 to local code conversion map (for combined characters)
475 */
476 typedef struct
477 {
478 uint32 utf1; /* UTF-8 code 1 */
479 uint32 utf2; /* UTF-8 code 2 */
480 uint32 code; /* local code */
481 } pg_utf_to_local_combined;
482
483 /*
484 * local code to UTF-8 conversion map (for combined characters)
485 */
486 typedef struct
487 {
488 uint32 code; /* local code */
489 uint32 utf1; /* UTF-8 code 1 */
490 uint32 utf2; /* UTF-8 code 2 */
491 } pg_local_to_utf_combined;
492
493 /*
494 * callback function for algorithmic encoding conversions (in either direction)
495 *
496 * if function returns zero, it does not know how to convert the code
497 */
498 typedef uint32 (*utf_local_conversion_func) (uint32 code);
499
500 /*
501 * Support macro for encoding conversion functions to validate their
502 * arguments. (This could be made more compact if we included fmgr.h
503 * here, but we don't want to do that because this header file is also
504 * used by frontends.)
505 */
506 #define CHECK_ENCODING_CONVERSION_ARGS(srcencoding,destencoding) \
507 check_encoding_conversion_args(PG_GETARG_INT32(0), \
508 PG_GETARG_INT32(1), \
509 PG_GETARG_INT32(4), \
510 (srcencoding), \
511 (destencoding))
512
513
514 /*
515 * Some handy functions for Unicode-specific tests.
516 */
517 static inline bool
is_valid_unicode_codepoint(pg_wchar c)518 is_valid_unicode_codepoint(pg_wchar c)
519 {
520 return (c > 0 && c <= 0x10FFFF);
521 }
522
523 static inline bool
is_utf16_surrogate_first(pg_wchar c)524 is_utf16_surrogate_first(pg_wchar c)
525 {
526 return (c >= 0xD800 && c <= 0xDBFF);
527 }
528
529 static inline bool
is_utf16_surrogate_second(pg_wchar c)530 is_utf16_surrogate_second(pg_wchar c)
531 {
532 return (c >= 0xDC00 && c <= 0xDFFF);
533 }
534
535 static inline pg_wchar
surrogate_pair_to_codepoint(pg_wchar first,pg_wchar second)536 surrogate_pair_to_codepoint(pg_wchar first, pg_wchar second)
537 {
538 return ((first & 0x3FF) << 10) + 0x10000 + (second & 0x3FF);
539 }
540
541
542 /*
543 * These functions are considered part of libpq's exported API and
544 * are also declared in libpq-fe.h.
545 */
546 extern int pg_char_to_encoding(const char *name);
547 extern const char *pg_encoding_to_char(int encoding);
548 extern int pg_valid_server_encoding_id(int encoding);
549
550 /*
551 * These functions are available to frontend code that links with libpgcommon
552 * (in addition to the ones just above). The constant tables declared
553 * earlier in this file are also available from libpgcommon.
554 */
555 extern int pg_encoding_mblen(int encoding, const char *mbstr);
556 extern int pg_encoding_mblen_bounded(int encoding, const char *mbstr);
557 extern int pg_encoding_dsplen(int encoding, const char *mbstr);
558 extern int pg_encoding_verifymb(int encoding, const char *mbstr, int len);
559 extern int pg_encoding_max_length(int encoding);
560 extern int pg_valid_client_encoding(const char *name);
561 extern int pg_valid_server_encoding(const char *name);
562 extern bool is_encoding_supported_by_icu(int encoding);
563 extern const char *get_encoding_name_for_icu(int encoding);
564
565 extern unsigned char *unicode_to_utf8(pg_wchar c, unsigned char *utf8string);
566 extern pg_wchar utf8_to_unicode(const unsigned char *c);
567 extern bool pg_utf8_islegal(const unsigned char *source, int length);
568 extern int pg_utf_mblen(const unsigned char *s);
569 extern int pg_mule_mblen(const unsigned char *s);
570
571 /*
572 * The remaining functions are backend-only.
573 */
574 extern int pg_mb2wchar(const char *from, pg_wchar *to);
575 extern int pg_mb2wchar_with_len(const char *from, pg_wchar *to, int len);
576 extern int pg_encoding_mb2wchar_with_len(int encoding,
577 const char *from, pg_wchar *to, int len);
578 extern int pg_wchar2mb(const pg_wchar *from, char *to);
579 extern int pg_wchar2mb_with_len(const pg_wchar *from, char *to, int len);
580 extern int pg_encoding_wchar2mb_with_len(int encoding,
581 const pg_wchar *from, char *to, int len);
582 extern int pg_char_and_wchar_strcmp(const char *s1, const pg_wchar *s2);
583 extern int pg_wchar_strncmp(const pg_wchar *s1, const pg_wchar *s2, size_t n);
584 extern int pg_char_and_wchar_strncmp(const char *s1, const pg_wchar *s2, size_t n);
585 extern size_t pg_wchar_strlen(const pg_wchar *wstr);
586 extern int pg_mblen(const char *mbstr);
587 extern int pg_dsplen(const char *mbstr);
588 extern int pg_mbstrlen(const char *mbstr);
589 extern int pg_mbstrlen_with_len(const char *mbstr, int len);
590 extern int pg_mbcliplen(const char *mbstr, int len, int limit);
591 extern int pg_encoding_mbcliplen(int encoding, const char *mbstr,
592 int len, int limit);
593 extern int pg_mbcharcliplen(const char *mbstr, int len, int limit);
594 extern int pg_database_encoding_max_length(void);
595 extern mbcharacter_incrementer pg_database_encoding_character_incrementer(void);
596
597 extern int PrepareClientEncoding(int encoding);
598 extern int SetClientEncoding(int encoding);
599 extern void InitializeClientEncoding(void);
600 extern int pg_get_client_encoding(void);
601 extern const char *pg_get_client_encoding_name(void);
602
603 extern void SetDatabaseEncoding(int encoding);
604 extern int GetDatabaseEncoding(void);
605 extern const char *GetDatabaseEncodingName(void);
606 extern void SetMessageEncoding(int encoding);
607 extern int GetMessageEncoding(void);
608
609 #ifdef ENABLE_NLS
610 extern int pg_bind_textdomain_codeset(const char *domainname);
611 #endif
612
613 extern unsigned char *pg_do_encoding_conversion(unsigned char *src, int len,
614 int src_encoding,
615 int dest_encoding);
616
617 extern char *pg_client_to_server(const char *s, int len);
618 extern char *pg_server_to_client(const char *s, int len);
619 extern char *pg_any_to_server(const char *s, int len, int encoding);
620 extern char *pg_server_to_any(const char *s, int len, int encoding);
621
622 extern void pg_unicode_to_server(pg_wchar c, unsigned char *s);
623
624 extern unsigned short BIG5toCNS(unsigned short big5, unsigned char *lc);
625 extern unsigned short CNStoBIG5(unsigned short cns, unsigned char lc);
626
627 extern void UtfToLocal(const unsigned char *utf, int len,
628 unsigned char *iso,
629 const pg_mb_radix_tree *map,
630 const pg_utf_to_local_combined *cmap, int cmapsize,
631 utf_local_conversion_func conv_func,
632 int encoding);
633 extern void LocalToUtf(const unsigned char *iso, int len,
634 unsigned char *utf,
635 const pg_mb_radix_tree *map,
636 const pg_local_to_utf_combined *cmap, int cmapsize,
637 utf_local_conversion_func conv_func,
638 int encoding);
639
640 extern bool pg_verifymbstr(const char *mbstr, int len, bool noError);
641 extern bool pg_verify_mbstr(int encoding, const char *mbstr, int len,
642 bool noError);
643 extern int pg_verify_mbstr_len(int encoding, const char *mbstr, int len,
644 bool noError);
645
646 extern void check_encoding_conversion_args(int src_encoding,
647 int dest_encoding,
648 int len,
649 int expected_src_encoding,
650 int expected_dest_encoding);
651
652 extern void report_invalid_encoding(int encoding, const char *mbstr, int len) pg_attribute_noreturn();
653 extern void report_untranslatable_char(int src_encoding, int dest_encoding,
654 const char *mbstr, int len) pg_attribute_noreturn();
655
656 extern void local2local(const unsigned char *l, unsigned char *p, int len,
657 int src_encoding, int dest_encoding, const unsigned char *tab);
658 extern void latin2mic(const unsigned char *l, unsigned char *p, int len,
659 int lc, int encoding);
660 extern void mic2latin(const unsigned char *mic, unsigned char *p, int len,
661 int lc, int encoding);
662 extern void latin2mic_with_table(const unsigned char *l, unsigned char *p,
663 int len, int lc, int encoding,
664 const unsigned char *tab);
665 extern void mic2latin_with_table(const unsigned char *mic, unsigned char *p,
666 int len, int lc, int encoding,
667 const unsigned char *tab);
668
669 #ifdef WIN32
670 extern WCHAR *pgwin32_message_to_UTF16(const char *str, int len, int *utf16len);
671 #endif
672
673 #endif /* PG_WCHAR_H */
674