PostgreSQL Source Code  git master
pg_wchar.h
Go to the documentation of this file.
1 /*-------------------------------------------------------------------------
2  *
3  * pg_wchar.h
4  * multibyte-character support
5  *
6  * Portions Copyright (c) 1996-2022, 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  * kana -> 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 a string that's required in any encoding to convert
320  * at least one character to any other encoding. In other words, if you feed
321  * MAX_CONVERSION_INPUT_LENGTH bytes to any encoding conversion function, it
322  * is guaranteed to be able to convert something without needing more input
323  * (assuming the input is valid).
324  *
325  * Currently, the maximum case is the conversion UTF8 -> SJIS JIS X0201 half
326  * width kana, where a pair of UTF-8 characters is converted into a single
327  * SHIFT_JIS_2004 character (the reverse of the worst case for
328  * MAX_CONVERSION_GROWTH). It needs 6 bytes of input. In theory, a
329  * user-defined conversion function might have more complicated cases, although
330  * for the reverse mapping you would probably also need to bump up
331  * MAX_CONVERSION_GROWTH. But there is no need to be stingy here, so make it
332  * generous.
333  */
334 #define MAX_CONVERSION_INPUT_LENGTH 16
335 
336 /*
337  * Maximum byte length of the string equivalent to any one Unicode code point,
338  * in any backend encoding. The current value assumes that a 4-byte UTF-8
339  * character might expand by MAX_CONVERSION_GROWTH, which is a huge
340  * overestimate. But in current usage we don't allocate large multiples of
341  * this, so there's little point in being stingy.
342  */
343 #define MAX_UNICODE_EQUIVALENT_STRING 16
344 
345 /*
346  * Table for mapping an encoding number to official encoding name and
347  * possibly other subsidiary data. Be careful to check encoding number
348  * before accessing a table entry!
349  *
350  * if (PG_VALID_ENCODING(encoding))
351  * pg_enc2name_tbl[ encoding ];
352  */
353 typedef struct pg_enc2name
354 {
355  const char *name;
357 #ifdef WIN32
358  unsigned codepage; /* codepage for WIN32 */
359 #endif
360 } pg_enc2name;
361 
363 
364 /*
365  * Encoding names for gettext
366  */
367 typedef struct pg_enc2gettext
368 {
370  const char *name;
372 
374 
375 /*
376  * pg_wchar stuff
377  */
378 typedef int (*mb2wchar_with_len_converter) (const unsigned char *from,
379  pg_wchar *to,
380  int len);
381 
382 typedef int (*wchar2mb_with_len_converter) (const pg_wchar *from,
383  unsigned char *to,
384  int len);
385 
386 typedef int (*mblen_converter) (const unsigned char *mbstr);
387 
388 typedef int (*mbdisplaylen_converter) (const unsigned char *mbstr);
389 
390 typedef bool (*mbcharacter_incrementer) (unsigned char *mbstr, int len);
391 
392 typedef int (*mbchar_verifier) (const unsigned char *mbstr, int len);
393 
394 typedef int (*mbstr_verifier) (const unsigned char *mbstr, int len);
395 
396 typedef struct
397 {
398  mb2wchar_with_len_converter mb2wchar_with_len; /* convert a multibyte
399  * string to a wchar */
400  wchar2mb_with_len_converter wchar2mb_with_len; /* convert a wchar string
401  * to a multibyte */
402  mblen_converter mblen; /* get byte length of a char */
403  mbdisplaylen_converter dsplen; /* get display width of a char */
404  mbchar_verifier mbverifychar; /* verify multibyte character */
405  mbstr_verifier mbverifystr; /* verify multibyte string */
406  int maxmblen; /* max bytes for a char in this encoding */
407 } pg_wchar_tbl;
408 
410 
411 /*
412  * Data structures for conversions between UTF-8 and other encodings
413  * (UtfToLocal() and LocalToUtf()). In these data structures, characters of
414  * either encoding are represented by uint32 words; hence we can only support
415  * characters up to 4 bytes long. For example, the byte sequence 0xC2 0x89
416  * would be represented by 0x0000C289, and 0xE8 0xA2 0xB4 by 0x00E8A2B4.
417  *
418  * There are three possible ways a character can be mapped:
419  *
420  * 1. Using a radix tree, from source to destination code.
421  * 2. Using a sorted array of source -> destination code pairs. This
422  * method is used for "combining" characters. There are so few of
423  * them that building a radix tree would be wasteful.
424  * 3. Using a conversion function.
425  */
426 
427 /*
428  * Radix tree for character conversion.
429  *
430  * Logically, this is actually four different radix trees, for 1-byte,
431  * 2-byte, 3-byte and 4-byte inputs. The 1-byte tree is a simple lookup
432  * table from source to target code. The 2-byte tree consists of two levels:
433  * one lookup table for the first byte, where the value in the lookup table
434  * points to a lookup table for the second byte. And so on.
435  *
436  * Physically, all the trees are stored in one big array, in 'chars16' or
437  * 'chars32', depending on the maximum value that needs to be represented. For
438  * each level in each tree, we also store lower and upper bound of allowed
439  * values - values outside those bounds are considered invalid, and are left
440  * out of the tables.
441  *
442  * In the intermediate levels of the trees, the values stored are offsets
443  * into the chars[16|32] array.
444  *
445  * In the beginning of the chars[16|32] array, there is always a number of
446  * zeros, so that you safely follow an index from an intermediate table
447  * without explicitly checking for a zero. Following a zero any number of
448  * times will always bring you to the dummy, all-zeros table in the
449  * beginning. This helps to shave some cycles when looking up values.
450  */
451 typedef struct
452 {
453  /*
454  * Array containing all the values. Only one of chars16 or chars32 is
455  * used, depending on how wide the values we need to represent are.
456  */
457  const uint16 *chars16;
458  const uint32 *chars32;
459 
460  /* Radix tree for 1-byte inputs */
461  uint32 b1root; /* offset of table in the chars[16|32] array */
462  uint8 b1_lower; /* min allowed value for a single byte input */
463  uint8 b1_upper; /* max allowed value for a single byte input */
464 
465  /* Radix tree for 2-byte inputs */
466  uint32 b2root; /* offset of 1st byte's table */
467  uint8 b2_1_lower; /* min/max allowed value for 1st input byte */
468  uint8 b2_1_upper;
469  uint8 b2_2_lower; /* min/max allowed value for 2nd input byte */
470  uint8 b2_2_upper;
471 
472  /* Radix tree for 3-byte inputs */
473  uint32 b3root; /* offset of 1st byte's table */
474  uint8 b3_1_lower; /* min/max allowed value for 1st input byte */
475  uint8 b3_1_upper;
476  uint8 b3_2_lower; /* min/max allowed value for 2nd input byte */
477  uint8 b3_2_upper;
478  uint8 b3_3_lower; /* min/max allowed value for 3rd input byte */
479  uint8 b3_3_upper;
480 
481  /* Radix tree for 4-byte inputs */
482  uint32 b4root; /* offset of 1st byte's table */
483  uint8 b4_1_lower; /* min/max allowed value for 1st input byte */
484  uint8 b4_1_upper;
485  uint8 b4_2_lower; /* min/max allowed value for 2nd input byte */
486  uint8 b4_2_upper;
487  uint8 b4_3_lower; /* min/max allowed value for 3rd input byte */
488  uint8 b4_3_upper;
489  uint8 b4_4_lower; /* min/max allowed value for 4th input byte */
490  uint8 b4_4_upper;
491 
493 
494 /*
495  * UTF-8 to local code conversion map (for combined characters)
496  */
497 typedef struct
498 {
499  uint32 utf1; /* UTF-8 code 1 */
500  uint32 utf2; /* UTF-8 code 2 */
501  uint32 code; /* local code */
503 
504 /*
505  * local code to UTF-8 conversion map (for combined characters)
506  */
507 typedef struct
508 {
509  uint32 code; /* local code */
510  uint32 utf1; /* UTF-8 code 1 */
511  uint32 utf2; /* UTF-8 code 2 */
513 
514 /*
515  * callback function for algorithmic encoding conversions (in either direction)
516  *
517  * if function returns zero, it does not know how to convert the code
518  */
519 typedef uint32 (*utf_local_conversion_func) (uint32 code);
520 
521 /*
522  * Support macro for encoding conversion functions to validate their
523  * arguments. (This could be made more compact if we included fmgr.h
524  * here, but we don't want to do that because this header file is also
525  * used by frontends.)
526  */
527 #define CHECK_ENCODING_CONVERSION_ARGS(srcencoding,destencoding) \
528  check_encoding_conversion_args(PG_GETARG_INT32(0), \
529  PG_GETARG_INT32(1), \
530  PG_GETARG_INT32(4), \
531  (srcencoding), \
532  (destencoding))
533 
534 
535 /*
536  * Some handy functions for Unicode-specific tests.
537  */
538 static inline bool
540 {
541  return (c > 0 && c <= 0x10FFFF);
542 }
543 
544 static inline bool
546 {
547  return (c >= 0xD800 && c <= 0xDBFF);
548 }
549 
550 static inline bool
552 {
553  return (c >= 0xDC00 && c <= 0xDFFF);
554 }
555 
556 static inline pg_wchar
558 {
559  return ((first & 0x3FF) << 10) + 0x10000 + (second & 0x3FF);
560 }
561 
562 
563 /*
564  * These functions are considered part of libpq's exported API and
565  * are also declared in libpq-fe.h.
566  */
567 extern int pg_char_to_encoding(const char *name);
568 extern const char *pg_encoding_to_char(int encoding);
569 extern int pg_valid_server_encoding_id(int encoding);
570 
571 /*
572  * These functions are available to frontend code that links with libpgcommon
573  * (in addition to the ones just above). The constant tables declared
574  * earlier in this file are also available from libpgcommon.
575  */
576 extern int pg_encoding_mblen(int encoding, const char *mbstr);
577 extern int pg_encoding_mblen_bounded(int encoding, const char *mbstr);
578 extern int pg_encoding_dsplen(int encoding, const char *mbstr);
579 extern int pg_encoding_verifymbchar(int encoding, const char *mbstr, int len);
580 extern int pg_encoding_verifymbstr(int encoding, const char *mbstr, int len);
581 extern int pg_encoding_max_length(int encoding);
582 extern int pg_valid_client_encoding(const char *name);
583 extern int pg_valid_server_encoding(const char *name);
584 extern bool is_encoding_supported_by_icu(int encoding);
585 extern const char *get_encoding_name_for_icu(int encoding);
586 
587 extern unsigned char *unicode_to_utf8(pg_wchar c, unsigned char *utf8string);
588 extern pg_wchar utf8_to_unicode(const unsigned char *c);
589 extern bool pg_utf8_islegal(const unsigned char *source, int length);
590 extern int pg_utf_mblen(const unsigned char *s);
591 extern int pg_mule_mblen(const unsigned char *s);
592 
593 /*
594  * The remaining functions are backend-only.
595  */
596 extern int pg_mb2wchar(const char *from, pg_wchar *to);
597 extern int pg_mb2wchar_with_len(const char *from, pg_wchar *to, int len);
599  const char *from, pg_wchar *to, int len);
600 extern int pg_wchar2mb(const pg_wchar *from, char *to);
601 extern int pg_wchar2mb_with_len(const pg_wchar *from, char *to, int len);
603  const pg_wchar *from, char *to, int len);
604 extern int pg_char_and_wchar_strcmp(const char *s1, const pg_wchar *s2);
605 extern int pg_wchar_strncmp(const pg_wchar *s1, const pg_wchar *s2, size_t n);
606 extern int pg_char_and_wchar_strncmp(const char *s1, const pg_wchar *s2, size_t n);
607 extern size_t pg_wchar_strlen(const pg_wchar *wstr);
608 extern int pg_mblen(const char *mbstr);
609 extern int pg_dsplen(const char *mbstr);
610 extern int pg_mbstrlen(const char *mbstr);
611 extern int pg_mbstrlen_with_len(const char *mbstr, int len);
612 extern int pg_mbcliplen(const char *mbstr, int len, int limit);
613 extern int pg_encoding_mbcliplen(int encoding, const char *mbstr,
614  int len, int limit);
615 extern int pg_mbcharcliplen(const char *mbstr, int len, int limit);
616 extern int pg_database_encoding_max_length(void);
618 
619 extern int PrepareClientEncoding(int encoding);
620 extern int SetClientEncoding(int encoding);
621 extern void InitializeClientEncoding(void);
622 extern int pg_get_client_encoding(void);
623 extern const char *pg_get_client_encoding_name(void);
624 
625 extern void SetDatabaseEncoding(int encoding);
626 extern int GetDatabaseEncoding(void);
627 extern const char *GetDatabaseEncodingName(void);
628 extern void SetMessageEncoding(int encoding);
629 extern int GetMessageEncoding(void);
630 
631 #ifdef ENABLE_NLS
632 extern int pg_bind_textdomain_codeset(const char *domainname);
633 #endif
634 
635 extern unsigned char *pg_do_encoding_conversion(unsigned char *src, int len,
636  int src_encoding,
637  int dest_encoding);
638 extern int pg_do_encoding_conversion_buf(Oid proc,
639  int src_encoding,
640  int dest_encoding,
641  unsigned char *src, int srclen,
642  unsigned char *dst, int dstlen,
643  bool noError);
644 
645 extern char *pg_client_to_server(const char *s, int len);
646 extern char *pg_server_to_client(const char *s, int len);
647 extern char *pg_any_to_server(const char *s, int len, int encoding);
648 extern char *pg_server_to_any(const char *s, int len, int encoding);
649 
650 extern void pg_unicode_to_server(pg_wchar c, unsigned char *s);
651 
652 extern unsigned short BIG5toCNS(unsigned short big5, unsigned char *lc);
653 extern unsigned short CNStoBIG5(unsigned short cns, unsigned char lc);
654 
655 extern int UtfToLocal(const unsigned char *utf, int len,
656  unsigned char *iso,
657  const pg_mb_radix_tree *map,
658  const pg_utf_to_local_combined *cmap, int cmapsize,
659  utf_local_conversion_func conv_func,
660  int encoding, bool noError);
661 extern int LocalToUtf(const unsigned char *iso, int len,
662  unsigned char *utf,
663  const pg_mb_radix_tree *map,
664  const pg_local_to_utf_combined *cmap, int cmapsize,
665  utf_local_conversion_func conv_func,
666  int encoding, bool noError);
667 
668 extern bool pg_verifymbstr(const char *mbstr, int len, bool noError);
669 extern bool pg_verify_mbstr(int encoding, const char *mbstr, int len,
670  bool noError);
671 extern int pg_verify_mbstr_len(int encoding, const char *mbstr, int len,
672  bool noError);
673 
674 extern void check_encoding_conversion_args(int src_encoding,
675  int dest_encoding,
676  int len,
677  int expected_src_encoding,
678  int expected_dest_encoding);
679 
680 extern void report_invalid_encoding(int encoding, const char *mbstr, int len) pg_attribute_noreturn();
681 extern void report_untranslatable_char(int src_encoding, int dest_encoding,
682  const char *mbstr, int len) pg_attribute_noreturn();
683 
684 extern int local2local(const unsigned char *l, unsigned char *p, int len,
685  int src_encoding, int dest_encoding,
686  const unsigned char *tab, bool noError);
687 extern int latin2mic(const unsigned char *l, unsigned char *p, int len,
688  int lc, int encoding, bool noError);
689 extern int mic2latin(const unsigned char *mic, unsigned char *p, int len,
690  int lc, int encoding, bool noError);
691 extern int latin2mic_with_table(const unsigned char *l, unsigned char *p,
692  int len, int lc, int encoding,
693  const unsigned char *tab, bool noError);
694 extern int mic2latin_with_table(const unsigned char *mic, unsigned char *p,
695  int len, int lc, int encoding,
696  const unsigned char *tab, bool noError);
697 
698 #ifdef WIN32
699 extern WCHAR *pgwin32_message_to_UTF16(const char *str, int len, int *utf16len);
700 #endif
701 
702 
703 /*
704  * Verify a chunk of bytes for valid ASCII.
705  *
706  * Returns false if the input contains any zero bytes or bytes with the
707  * high-bit set. Input len must be a multiple of 8.
708  */
709 static inline bool
710 is_valid_ascii(const unsigned char *s, int len)
711 {
712  uint64 chunk,
713  highbit_cum = UINT64CONST(0),
714  zero_cum = UINT64CONST(0x8080808080808080);
715 
716  Assert(len % sizeof(chunk) == 0);
717 
718  while (len > 0)
719  {
720  memcpy(&chunk, s, sizeof(chunk));
721 
722  /*
723  * Capture any zero bytes in this chunk.
724  *
725  * First, add 0x7f to each byte. This sets the high bit in each byte,
726  * unless it was a zero. If any resulting high bits are zero, the
727  * corresponding high bits in the zero accumulator will be cleared.
728  *
729  * If none of the bytes in the chunk had the high bit set, the max
730  * value each byte can have after the addition is 0x7f + 0x7f = 0xfe,
731  * and we don't need to worry about carrying over to the next byte. If
732  * any input bytes did have the high bit set, it doesn't matter
733  * because we check for those separately.
734  */
735  zero_cum &= (chunk + UINT64CONST(0x7f7f7f7f7f7f7f7f));
736 
737  /* Capture any set bits in this chunk. */
738  highbit_cum |= chunk;
739 
740  s += sizeof(chunk);
741  len -= sizeof(chunk);
742  }
743 
744  /* Check if any high bits in the high bit accumulator got set. */
745  if (highbit_cum & UINT64CONST(0x8080808080808080))
746  return false;
747 
748  /* Check if any high bits in the zero accumulator got cleared. */
749  if (zero_cum != UINT64CONST(0x8080808080808080))
750  return false;
751 
752  return true;
753 }
754 
755 #endif /* PG_WCHAR_H */
unsigned short uint16
Definition: c.h:451
unsigned int uint32
Definition: c.h:452
#define PGDLLIMPORT
Definition: c.h:1342
#define pg_attribute_noreturn()
Definition: c.h:190
unsigned char bool
Definition: c.h:402
unsigned char uint8
Definition: c.h:450
const char * name
Definition: encode.c:561
Assert(fmt[strlen(fmt) - 1] !='\n')
unsigned int pg_wchar
Definition: mbprint.c:31
const void size_t len
int32 encoding
Definition: pg_database.h:41
static rewind_source * source
Definition: pg_rewind.c:81
const char * pg_get_client_encoding_name(void)
Definition: mbutils.c:345
int pg_encoding_mblen_bounded(int encoding, const char *mbstr)
Definition: wchar.c:2141
unsigned char * pg_do_encoding_conversion(unsigned char *src, int len, int src_encoding, int dest_encoding)
Definition: mbutils.c:356
int mic2latin_with_table(const unsigned char *mic, unsigned char *p, int len, int lc, int encoding, const unsigned char *tab, bool noError)
Definition: conv.c:257
bool pg_utf8_islegal(const unsigned char *source, int length)
Definition: wchar.c:2012
char * pg_any_to_server(const char *s, int len, int encoding)
Definition: mbutils.c:676
int GetDatabaseEncoding(void)
Definition: mbutils.c:1210
int pg_encoding_wchar2mb_with_len(int encoding, const pg_wchar *from, char *to, int len)
Definition: mbutils.c:958
int UtfToLocal(const unsigned char *utf, int len, unsigned char *iso, const pg_mb_radix_tree *map, const pg_utf_to_local_combined *cmap, int cmapsize, utf_local_conversion_func conv_func, int encoding, bool noError)
Definition: conv.c:507
pg_enc
Definition: pg_wchar.h:223
@ PG_WIN1254
Definition: pg_wchar.h:255
@ PG_LATIN4
Definition: pg_wchar.h:235
@ _PG_LAST_ENCODING_
Definition: pg_wchar.h:269
@ PG_LATIN9
Definition: pg_wchar.h:240
@ PG_JOHAB
Definition: pg_wchar.h:267
@ PG_GB18030
Definition: pg_wchar.h:266
@ PG_SQL_ASCII
Definition: pg_wchar.h:224
@ PG_KOI8R
Definition: pg_wchar.h:246
@ PG_ISO_8859_6
Definition: pg_wchar.h:250
@ PG_WIN1253
Definition: pg_wchar.h:254
@ PG_KOI8U
Definition: pg_wchar.h:258
@ PG_LATIN6
Definition: pg_wchar.h:237
@ PG_MULE_INTERNAL
Definition: pg_wchar.h:231
@ PG_LATIN5
Definition: pg_wchar.h:236
@ PG_EUC_CN
Definition: pg_wchar.h:226
@ PG_UHC
Definition: pg_wchar.h:265
@ PG_LATIN2
Definition: pg_wchar.h:233
@ PG_ISO_8859_5
Definition: pg_wchar.h:249
@ PG_LATIN10
Definition: pg_wchar.h:241
@ PG_WIN1250
Definition: pg_wchar.h:253
@ PG_ISO_8859_7
Definition: pg_wchar.h:251
@ PG_SJIS
Definition: pg_wchar.h:262
@ PG_LATIN8
Definition: pg_wchar.h:239
@ PG_EUC_JP
Definition: pg_wchar.h:225
@ PG_GBK
Definition: pg_wchar.h:264
@ PG_LATIN3
Definition: pg_wchar.h:234
@ PG_WIN1256
Definition: pg_wchar.h:242
@ PG_LATIN1
Definition: pg_wchar.h:232
@ PG_EUC_TW
Definition: pg_wchar.h:228
@ PG_WIN1258
Definition: pg_wchar.h:243
@ PG_SHIFT_JIS_2004
Definition: pg_wchar.h:268
@ PG_WIN1252
Definition: pg_wchar.h:248
@ PG_LATIN7
Definition: pg_wchar.h:238
@ PG_UTF8
Definition: pg_wchar.h:230
@ PG_WIN1255
Definition: pg_wchar.h:256
@ PG_WIN1257
Definition: pg_wchar.h:257
@ PG_WIN1251
Definition: pg_wchar.h:247
@ PG_EUC_KR
Definition: pg_wchar.h:227
@ PG_WIN866
Definition: pg_wchar.h:244
@ PG_ISO_8859_8
Definition: pg_wchar.h:252
@ PG_WIN874
Definition: pg_wchar.h:245
@ PG_EUC_JIS_2004
Definition: pg_wchar.h:229
@ PG_BIG5
Definition: pg_wchar.h:263
const char * get_encoding_name_for_icu(int encoding)
Definition: encnames.c:470
uint32(* utf_local_conversion_func)(uint32 code)
Definition: pg_wchar.h:503
int pg_do_encoding_conversion_buf(Oid proc, int src_encoding, int dest_encoding, unsigned char *src, int srclen, unsigned char *dst, int dstlen, bool noError)
Definition: mbutils.c:469
unsigned char * unicode_to_utf8(pg_wchar c, unsigned char *utf8string)
Definition: wchar.c:483
int(* wchar2mb_with_len_converter)(const pg_wchar *from, unsigned char *to, int len)
Definition: pg_wchar.h:366
int pg_verify_mbstr_len(int encoding, const char *mbstr, int len, bool noError)
Definition: mbutils.c:1546
void InitializeClientEncoding(void)
Definition: mbutils.c:281
int pg_dsplen(const char *mbstr)
Definition: mbutils.c:973
mbcharacter_incrementer pg_database_encoding_character_incrementer(void)
Definition: mbutils.c:1472
size_t pg_wchar_strlen(const pg_wchar *wstr)
Definition: wstrncmp.c:70
char * pg_client_to_server(const char *s, int len)
Definition: mbutils.c:660
int pg_wchar2mb_with_len(const pg_wchar *from, char *to, int len)
Definition: mbutils.c:951
int pg_utf_mblen(const unsigned char *s)
Definition: wchar.c:549
int latin2mic_with_table(const unsigned char *l, unsigned char *p, int len, int lc, int encoding, const unsigned char *tab, bool noError)
Definition: conv.c:194
int pg_mbstrlen_with_len(const char *mbstr, int len)
Definition: mbutils.c:1000
int pg_mb2wchar(const char *from, pg_wchar *to)
Definition: mbutils.c:922
int pg_char_to_encoding(const char *name)
Definition: encnames.c:550
bool(* mbcharacter_incrementer)(unsigned char *mbstr, int len)
Definition: pg_wchar.h:374
static bool is_valid_ascii(const unsigned char *s, int len)
Definition: pg_wchar.h:694
int(* mbchar_verifier)(const unsigned char *mbstr, int len)
Definition: pg_wchar.h:376
int pg_mbcharcliplen(const char *mbstr, int len, int limit)
Definition: mbutils.c:1068
int mic2latin(const unsigned char *mic, unsigned char *p, int len, int lc, int encoding, bool noError)
Definition: conv.c:127
struct pg_enc2name pg_enc2name
int pg_valid_server_encoding_id(int encoding)
Definition: encnames.c:514
unsigned short CNStoBIG5(unsigned short cns, unsigned char lc)
Definition: big5.c:345
int pg_wchar2mb(const pg_wchar *from, char *to)
Definition: mbutils.c:944
void report_untranslatable_char(int src_encoding, int dest_encoding, const char *mbstr, int len) pg_attribute_noreturn()
Definition: mbutils.c:1679
int pg_mule_mblen(const unsigned char *s)
Definition: wchar.c:832
PGDLLIMPORT const pg_enc2name pg_enc2name_tbl[]
Definition: encnames.c:309
int pg_mbstrlen(const char *mbstr)
Definition: mbutils.c:980
bool pg_verify_mbstr(int encoding, const char *mbstr, int len, bool noError)
Definition: mbutils.c:1515
PGDLLIMPORT const pg_enc2gettext pg_enc2gettext_tbl[]
Definition: encnames.c:361
bool pg_verifymbstr(const char *mbstr, int len, bool noError)
Definition: mbutils.c:1505
int LocalToUtf(const unsigned char *iso, int len, unsigned char *utf, const pg_mb_radix_tree *map, const pg_local_to_utf_combined *cmap, int cmapsize, utf_local_conversion_func conv_func, int encoding, bool noError)
Definition: conv.c:717
int pg_mbcliplen(const char *mbstr, int len, int limit)
Definition: mbutils.c:1026
int GetMessageEncoding(void)
Definition: mbutils.c:1257
pg_wchar utf8_to_unicode(const unsigned char *c)
Definition: wchar.c:679
int SetClientEncoding(int encoding)
Definition: mbutils.c:208
void SetMessageEncoding(int encoding)
Definition: mbutils.c:1114
void pg_unicode_to_server(pg_wchar c, unsigned char *s)
Definition: mbutils.c:864
static bool is_valid_unicode_codepoint(pg_wchar c)
Definition: pg_wchar.h:523
unsigned int pg_wchar
Definition: pg_wchar.h:25
int pg_valid_server_encoding(const char *name)
Definition: encnames.c:500
struct pg_enc2gettext pg_enc2gettext
void check_encoding_conversion_args(int src_encoding, int dest_encoding, int len, int expected_src_encoding, int expected_dest_encoding)
Definition: mbutils.c:1618
int pg_database_encoding_max_length(void)
Definition: mbutils.c:1495
int PrepareClientEncoding(int encoding)
Definition: mbutils.c:110
int pg_valid_client_encoding(const char *name)
Definition: encnames.c:486
int pg_encoding_dsplen(int encoding, const char *mbstr)
Definition: wchar.c:2150
const char * GetDatabaseEncodingName(void)
Definition: mbutils.c:1216
void report_invalid_encoding(int encoding, const char *mbstr, int len) pg_attribute_noreturn()
Definition: mbutils.c:1647
int pg_char_and_wchar_strcmp(const char *s1, const pg_wchar *s2)
Definition: wstrcmp.c:41
char * pg_server_to_client(const char *s, int len)
Definition: mbutils.c:738
unsigned short BIG5toCNS(unsigned short big5, unsigned char *lc)
Definition: big5.c:292
int pg_encoding_verifymbstr(int encoding, const char *mbstr, int len)
Definition: wchar.c:2176
int pg_get_client_encoding(void)
Definition: mbutils.c:336
bool is_encoding_supported_by_icu(int encoding)
Definition: encnames.c:459
int(* mb2wchar_with_len_converter)(const unsigned char *from, pg_wchar *to, int len)
Definition: pg_wchar.h:362
int(* mbstr_verifier)(const unsigned char *mbstr, int len)
Definition: pg_wchar.h:378
int local2local(const unsigned char *l, unsigned char *p, int len, int src_encoding, int dest_encoding, const unsigned char *tab, bool noError)
Definition: conv.c:33
const char * pg_encoding_to_char(int encoding)
Definition: encnames.c:588
int pg_encoding_max_length(int encoding)
Definition: wchar.c:2187
int pg_encoding_mblen(int encoding, const char *mbstr)
Definition: wchar.c:2129
PGDLLIMPORT const pg_wchar_tbl pg_wchar_table[]
Definition: wchar.c:2075
static pg_wchar surrogate_pair_to_codepoint(pg_wchar first, pg_wchar second)
Definition: pg_wchar.h:541
int pg_wchar_strncmp(const pg_wchar *s1, const pg_wchar *s2, size_t n)
Definition: wstrncmp.c:40
static bool is_utf16_surrogate_first(pg_wchar c)
Definition: pg_wchar.h:529
void SetDatabaseEncoding(int encoding)
Definition: mbutils.c:1104
int latin2mic(const unsigned char *l, unsigned char *p, int len, int lc, int encoding, bool noError)
Definition: conv.c:89
int pg_encoding_mbcliplen(int encoding, const char *mbstr, int len, int limit)
Definition: mbutils.c:1036
int pg_char_and_wchar_strncmp(const char *s1, const pg_wchar *s2, size_t n)
Definition: wstrncmp.c:55
int pg_encoding_verifymbchar(int encoding, const char *mbstr, int len)
Definition: wchar.c:2163
int pg_encoding_mb2wchar_with_len(int encoding, const char *from, pg_wchar *to, int len)
Definition: mbutils.c:936
char * pg_server_to_any(const char *s, int len, int encoding)
Definition: mbutils.c:749
int(* mbdisplaylen_converter)(const unsigned char *mbstr)
Definition: pg_wchar.h:372
int pg_mb2wchar_with_len(const char *from, pg_wchar *to, int len)
Definition: mbutils.c:929
static bool is_utf16_surrogate_second(pg_wchar c)
Definition: pg_wchar.h:535
int pg_mblen(const char *mbstr)
Definition: mbutils.c:966
int(* mblen_converter)(const unsigned char *mbstr)
Definition: pg_wchar.h:370
unsigned int Oid
Definition: postgres_ext.h:31
char * c
char * s1
char * s2
const char * name
Definition: pg_wchar.h:354
pg_enc encoding
Definition: pg_wchar.h:353
pg_enc encoding
Definition: pg_wchar.h:340
const char * name
Definition: pg_wchar.h:339