KDb

icu.cpp
1/*
2** 2007 May 6
3**
4** The author disclaims copyright to this source code. In place of
5** a legal notice, here is a blessing:
6**
7** May you do good and not evil.
8** May you find forgiveness for yourself and forgive others.
9** May you share freely, never taking more than you give.
10**
11*************************************************************************
12** $Id: icu.c,v 1.7 2007/12/13 21:54:11 drh Exp $
13**
14** This file implements an integration between the ICU library
15** ("International Components for Unicode", an open-source library
16** for handling unicode data) and SQLite. The integration uses
17** ICU to provide the following to SQLite:
18**
19** * An implementation of the SQL regexp() function (and hence REGEXP
20** operator) using the ICU uregex_XX() APIs.
21**
22** * Implementations of the SQL scalar upper() and lower() functions
23** for case mapping.
24**
25** * Integration of ICU and SQLite collation seqences.
26**
27** * An implementation of the LIKE operator that uses ICU to
28** provide case-independent matching.
29*/
30
31#include "sqliteicu.h"
32
33#if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_ICU)
34
35/* Include ICU headers */
36#include <unicode/utypes.h>
37#include <unicode/uregex.h>
38#include <unicode/ustring.h>
39#include <unicode/ucol.h>
40#include <unicode/uvernum.h>
41#if U_ICU_VERSION_MAJOR_NUM>=51
42#include <unicode/utf_old.h>
43#endif
44
45#include <assert.h>
46
47#ifndef SQLITE_CORE
48 #include "sqlite3ext.h"
49 SQLITE_EXTENSION_INIT1
50#else
51 #include "sqlite3.h"
52#endif
53
54/*
55** Maximum length (in bytes) of the pattern in a LIKE or GLOB
56** operator.
57*/
58#ifndef SQLITE_MAX_LIKE_PATTERN_LENGTH
59# define SQLITE_MAX_LIKE_PATTERN_LENGTH 50000
60#endif
61
62/*
63** Version of sqlite3_free() that is always a function, never a macro.
64*/
65static void xFree(void *p){
66 sqlite3_free(p);
67}
68
69/*
70** Compare two UTF-8 strings for equality where the first string is
71** a "LIKE" expression. Return true (1) if they are the same and
72** false (0) if they are different.
73*/
74static int icuLikeCompare(
75 const uint8_t *zPattern, /* LIKE pattern */
76 const uint8_t *zString, /* The UTF-8 string to compare against */
77 const UChar32 uEsc /* The escape character */
78){
79 static const int MATCH_ONE = (UChar32)'_';
80 static const int MATCH_ALL = (UChar32)'%';
81
82 int iPattern = 0; /* Current byte index in zPattern */
83 int iString = 0; /* Current byte index in zString */
84
85 int prevEscape = 0; /* True if the previous character was uEsc */
86
87 while( zPattern[iPattern]!=0 ){
88
89 /* Read (and consume) the next character from the input pattern. */
90 UChar32 uPattern;
91 U8_NEXT_UNSAFE(zPattern, iPattern, uPattern);
92 assert(uPattern!=0);
93
94 /* There are now 4 possibilities:
95 **
96 ** 1. uPattern is an unescaped match-all character "%",
97 ** 2. uPattern is an unescaped match-one character "_",
98 ** 3. uPattern is an unescaped escape character, or
99 ** 4. uPattern is to be handled as an ordinary character
100 */
101 if( !prevEscape && uPattern==MATCH_ALL ){
102 /* Case 1. */
103 uint8_t c;
104
105 /* Skip any MATCH_ALL or MATCH_ONE characters that follow a
106 ** MATCH_ALL. For each MATCH_ONE, skip one character in the
107 ** test string.
108 */
109 while( (c=zPattern[iPattern]) == MATCH_ALL || c == MATCH_ONE ){
110 if( c==MATCH_ONE ){
111 if( zString[iString]==0 ) return 0;
112 U8_FWD_1_UNSAFE(zString, iString);
113 }
114 iPattern++;
115 }
116
117 if( zPattern[iPattern]==0 ) return 1;
118
119 while( zString[iString] ){
120 if( icuLikeCompare(&zPattern[iPattern], &zString[iString], uEsc) ){
121 return 1;
122 }
123 U8_FWD_1_UNSAFE(zString, iString);
124 }
125 return 0;
126
127 }else if( !prevEscape && uPattern==MATCH_ONE ){
128 /* Case 2. */
129 if( zString[iString]==0 ) return 0;
130 U8_FWD_1_UNSAFE(zString, iString);
131
132 }else if( !prevEscape && uPattern==uEsc){
133 /* Case 3. */
134 prevEscape = 1;
135
136 }else{
137 /* Case 4. */
138 UChar32 uString;
139 U8_NEXT_UNSAFE(zString, iString, uString);
140 uString = u_foldCase(uString, U_FOLD_CASE_DEFAULT);
141 uPattern = u_foldCase(uPattern, U_FOLD_CASE_DEFAULT);
142 if( uString!=uPattern ){
143 return 0;
144 }
145 prevEscape = 0;
146 }
147 }
148
149 return zString[iString]==0;
150}
151
152/*
153** Implementation of the like() SQL function. This function implements
154** the build-in LIKE operator. The first argument to the function is the
155** pattern and the second argument is the string. So, the SQL statements:
156**
157** A LIKE B
158**
159** is implemented as like(B, A). If there is an escape character E,
160**
161** A LIKE B ESCAPE E
162**
163** is mapped to like(B, A, E).
164*/
165static void icuLikeFunc(
166 sqlite3_context *context,
167 int argc,
168 sqlite3_value **argv
169){
170 const unsigned char *zA = sqlite3_value_text(argv[0]);
171 const unsigned char *zB = sqlite3_value_text(argv[1]);
172 UChar32 uEsc = 0;
173
174 /* Limit the length of the LIKE or GLOB pattern to avoid problems
175 ** of deep recursion and N*N behavior in patternCompare().
176 */
177 if( sqlite3_value_bytes(argv[0])>SQLITE_MAX_LIKE_PATTERN_LENGTH ){
178 sqlite3_result_error(context, "LIKE or GLOB pattern too complex", -1);
179 return;
180 }
181
182
183 if( argc==3 ){
184 /* The escape character string must consist of a single UTF-8 character.
185 ** Otherwise, return an error.
186 */
187 int nE= sqlite3_value_bytes(argv[2]);
188 const unsigned char *zE = sqlite3_value_text(argv[2]);
189 int i = 0;
190 if( zE==nullptr ) return;
191 U8_NEXT(zE, i, nE, uEsc);
192 if( i!=nE){
193 sqlite3_result_error(context,
194 "ESCAPE expression must be a single character", -1);
195 return;
196 }
197 }
198
199 if( zA && zB ){
200 sqlite3_result_int(context, icuLikeCompare(zA, zB, uEsc));
201 }
202}
203
204/*
205** This function is called when an ICU function called from within
206** the implementation of an SQL scalar function returns an error.
207**
208** The scalar function context passed as the first argument is
209** loaded with an error message based on the following two args.
210*/
211static void icuFunctionError(
212 sqlite3_context *pCtx, /* SQLite scalar function context */
213 const char *zName, /* Name of ICU function that failed */
214 UErrorCode e /* Error code returned by ICU function */
215){
216 char zBuf[128];
217 sqlite3_snprintf(128, zBuf, "ICU error: %s(): %s", zName, u_errorName(e));
218 zBuf[127] = '\0';
219 sqlite3_result_error(pCtx, zBuf, -1);
220}
221
222/*
223** Function to delete compiled regexp objects. Registered as
224** a destructor function with sqlite3_set_auxdata().
225*/
226static void icuRegexpDelete(void *p){
227 URegularExpression *pExpr = (URegularExpression *)p;
228 uregex_close(pExpr);
229}
230
231/*
232** Implementation of SQLite REGEXP operator. This scalar function takes
233** two arguments. The first is a regular expression pattern to compile
234** the second is a string to match against that pattern. If either
235** argument is an SQL NULL, then NULL is returned. Otherwise, the result
236** is 1 if the string matches the pattern, or 0 otherwise.
237**
238** SQLite maps the regexp() function to the regexp() operator such
239** that the following two are equivalent:
240**
241** zString REGEXP zPattern
242** regexp(zPattern, zString)
243**
244** Uses the following ICU regexp APIs:
245**
246** uregex_open()
247** uregex_matches()
248** uregex_close()
249*/
250static void icuRegexpFunc(sqlite3_context *p, int nArg, sqlite3_value **apArg){
251 UErrorCode status = U_ZERO_ERROR;
252 URegularExpression *pExpr;
253 UBool res;
254 const UChar *zString = static_cast<const UChar *>(sqlite3_value_text16(apArg[1]));
255
256 (void)nArg; /* Unused parameter */
257
258 /* If the left hand side of the regexp operator is NULL,
259 ** then the result is also NULL.
260 */
261 if( !zString ){
262 return;
263 }
264
265 pExpr = static_cast<URegularExpression*>(sqlite3_get_auxdata(p, 0));
266 if( !pExpr ){
267 const UChar *zPattern = static_cast<const UChar *>(sqlite3_value_text16(apArg[0]));
268 if( !zPattern ){
269 return;
270 }
271 pExpr = uregex_open(zPattern, -1, 0, nullptr, &status);
272
273 if( U_SUCCESS(status) ){
274 sqlite3_set_auxdata(p, 0, pExpr, icuRegexpDelete);
275 }else{
276 assert(!pExpr);
277 icuFunctionError(p, "uregex_open", status);
278 return;
279 }
280 }
281
282 /* Configure the text that the regular expression operates on. */
283 uregex_setText(pExpr, zString, -1, &status);
284 if( !U_SUCCESS(status) ){
285 icuFunctionError(p, "uregex_setText", status);
286 return;
287 }
288
289 /* Attempt the match */
290 res = uregex_matches(pExpr, 0, &status);
291 if( !U_SUCCESS(status) ){
292 icuFunctionError(p, "uregex_matches", status);
293 return;
294 }
295
296 /* Set the text that the regular expression operates on to a NULL
297 ** pointer. This is not really necessary, but it is tidier than
298 ** leaving the regular expression object configured with an invalid
299 ** pointer after this function returns.
300 */
301 uregex_setText(pExpr, nullptr, 0, &status);
302
303 /* Return 1 or 0. */
304 sqlite3_result_int(p, res ? 1 : 0);
305}
306
307/*
308** Implementations of scalar functions for case mapping - upper() and
309** lower(). Function upper() converts its input to upper-case (ABC).
310** Function lower() converts to lower-case (abc).
311**
312** ICU provides two types of case mapping, "general" case mapping and
313** "language specific". Refer to ICU documentation for the differences
314** between the two.
315**
316** To utilise "general" case mapping, the upper() or lower() scalar
317** functions are invoked with one argument:
318**
319** upper('ABC') -> 'abc'
320** lower('abc') -> 'ABC'
321**
322** To access ICU "language specific" case mapping, upper() or lower()
323** should be invoked with two arguments. The second argument is the name
324** of the locale to use. Passing an empty string ("") or SQL NULL value
325** as the second argument is the same as invoking the 1 argument version
326** of upper() or lower().
327**
328** lower('I', 'en_us') -> 'i'
329** lower('I', 'tr_tr') -> 'ı' (small dotless i)
330**
331** https://www.icu-project.org/userguide/posix.html#case_mappings
332*/
333static void icuCaseFunc16(sqlite3_context *p, int nArg, sqlite3_value **apArg){
334 const UChar *zInput;
335 UChar *zOutput;
336 int nInput;
337 int nOutput;
338
339 UErrorCode status = U_ZERO_ERROR;
340 const unsigned char *zLocale = nullptr;
341
342 assert(nArg==1 || nArg==2);
343 if( nArg==2 ){
344 zLocale = static_cast<const unsigned char *>(sqlite3_value_text(apArg[1]));
345 }
346
347 zInput = static_cast<const UChar *>(sqlite3_value_text16(apArg[0]));
348 if( !zInput ){
349 return;
350 }
351 nInput = sqlite3_value_bytes16(apArg[0]);
352
353 nOutput = nInput * 2 + 2;
354 zOutput = static_cast<UChar *>(sqlite3_malloc(nOutput));
355 if( !zOutput ){
356 return;
357 }
358
359 if( sqlite3_user_data(p) ){
360 u_strToUpper(zOutput, nOutput/2, zInput, nInput/2, reinterpret_cast<const char*>(zLocale), &status);
361 }else{
362 u_strToLower(zOutput, nOutput/2, zInput, nInput/2, reinterpret_cast<const char*>(zLocale), &status);
363 }
364
365 if( !U_SUCCESS(status) ){
366 icuFunctionError(p, "u_strToLower()/u_strToUpper", status);
367 return;
368 }
369
370 sqlite3_result_text16(p, zOutput, -1, xFree);
371}
372
373/*
374** Collation sequence destructor function. The pCtx argument points to
375** a UCollator structure previously allocated using ucol_open().
376*/
377static void icuCollationDel(void *pCtx){
378 UCollator *p = (UCollator *)pCtx;
379 ucol_close(p);
380}
381
382/*
383** Collation sequence comparison function. The pCtx argument points to
384** a UCollator structure previously allocated using ucol_open().
385*/
386static int icuCollationColl(
387 void *pCtx,
388 int nLeft,
389 const void *zLeft,
390 int nRight,
391 const void *zRight
392){
393 UCollationResult res;
394 UCollator *p = (UCollator *)pCtx;
395 res = ucol_strcoll(p, (UChar *)zLeft, nLeft/2, (UChar *)zRight, nRight/2);
396 switch( res ){
397 case UCOL_LESS: return -1;
398 case UCOL_GREATER: return +1;
399 case UCOL_EQUAL: return 0;
400 }
401 assert(!"Unexpected return value from ucol_strcoll()");
402 return 0;
403}
404
405/*
406** Implementation of the scalar function icu_load_collation().
407**
408** This scalar function is used to add ICU collation based collation
409** types to an SQLite database connection. It is intended to be called
410** as follows:
411**
412** SELECT icu_load_collation(<locale>, <collation-name>);
413**
414** Where <locale> is a string containing an ICU locale identifier (i.e.
415** "en_AU", "tr_TR" etc.) and <collation-name> is the name of the
416** collation sequence to create.
417*/
418static void icuLoadCollation(
419 sqlite3_context *p,
420 int nArg,
421 sqlite3_value **apArg
422){
423 (void)nArg; /* Unused parameter */
424
425 sqlite3 *db = (sqlite3 *)sqlite3_user_data(p);
426 UErrorCode status = U_ZERO_ERROR;
427 const char *zLocale; /* Locale identifier - (eg. "jp_JP") */
428 const char *zName; /* SQL Collation sequence name (eg. "japanese") */
429 UCollator *pUCollator; /* ICU library collation object */
430 int rc; /* Return code from sqlite3_create_collation_x() */
431
432 assert(nArg==2);
433 zLocale = (const char *)sqlite3_value_text(apArg[0]);
434 zName = (const char *)sqlite3_value_text(apArg[1]);
435
436 if( !zLocale || !zName ){
437 return;
438 }
439
440 pUCollator = ucol_open(zLocale, &status);
441 if( !U_SUCCESS(status) ){
442 icuFunctionError(p, "ucol_open", status);
443 return;
444 }
445 assert(p);
446
447 rc = sqlite3_create_collation_v2(db, zName, SQLITE_UTF16, (void *)pUCollator,
448 icuCollationColl, icuCollationDel
449 );
450 if( rc!=SQLITE_OK ){
451 ucol_close(pUCollator);
452 sqlite3_result_error(p, "Error registering collation function", -1);
453 }
454}
455
456/*
457** Register the ICU extension functions with database db.
458*/
459KDB_SQLITE_ICU_EXPORT int sqlite3IcuInit(sqlite3 *db){
460 struct IcuScalar {
461 const char *zName; /* Function name */
462 int nArg; /* Number of arguments */
463 int enc; /* Optimal text encoding */
464 void *pContext; /* sqlite3_user_data() context */
465 void (*xFunc)(sqlite3_context*,int,sqlite3_value**);
466 } scalars[] = {
467 {"regexp", 2, SQLITE_ANY, nullptr, icuRegexpFunc},
468
469 {"lower", 1, SQLITE_UTF16, nullptr, icuCaseFunc16},
470 {"lower", 2, SQLITE_UTF16, nullptr, icuCaseFunc16},
471 {"upper", 1, SQLITE_UTF16, (void*)1, icuCaseFunc16},
472 {"upper", 2, SQLITE_UTF16, (void*)1, icuCaseFunc16},
473
474 {"lower", 1, SQLITE_UTF8, nullptr, icuCaseFunc16},
475 {"lower", 2, SQLITE_UTF8, nullptr, icuCaseFunc16},
476 {"upper", 1, SQLITE_UTF8, (void*)1, icuCaseFunc16},
477 {"upper", 2, SQLITE_UTF8, (void*)1, icuCaseFunc16},
478
479 {"like", 2, SQLITE_UTF8, nullptr, icuLikeFunc},
480 {"like", 3, SQLITE_UTF8, nullptr, icuLikeFunc},
481
482 {"icu_load_collation", 2, SQLITE_UTF8, (void*)db, icuLoadCollation},
483 };
484
485 int rc = SQLITE_OK;
486 int i;
487
488 for(i=0; rc==SQLITE_OK && i<(int)(sizeof(scalars)/sizeof(scalars[0])); i++){
489 struct IcuScalar *p = &scalars[i];
490 rc = sqlite3_create_function(
491 db, p->zName, p->nArg, p->enc, p->pContext, p->xFunc, nullptr, nullptr
492 );
493 }
494
495 return rc;
496}
497
498#if !defined SQLITE_CORE || !SQLITE_CORE
499KDB_SQLITE_ICU_EXPORT int sqlite3_extension_init(
500 sqlite3 *db,
501 char **pzErrMsg,
502 const struct sqlite3_api_routines *pApi
503){
504 (void)pzErrMsg; /* Unused parameter */
505 SQLITE_EXTENSION_INIT2(pApi)
506 return sqlite3IcuInit(db);
507}
508#endif
509
510#endif
Q_SCRIPTABLE CaptureState status()
This file is part of the KDE documentation.
Documentation copyright © 1996-2024 The KDE developers.
Generated on Mon Nov 4 2024 16:38:30 by doxygen 1.12.0 written by Dimitri van Heesch, © 1997-2006

KDE's Doxygen guidelines are available online.