sqlite3_android.cpp revision 4b2aeb8073512909db336dd297563f6d07df7537 
1/*
2 * Copyright (C) 2007 The Android Open Source Project
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 *      http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
17#define LOG_TAG "sqlite3_android"
18
19#include <ctype.h>
20#include <stdlib.h>
21#include <string.h>
22#include <unistd.h>
23
24#include <unicode/ucol.h>
25#include <unicode/uiter.h>
26#include <unicode/ustring.h>
27#include <unicode/utypes.h>
28#include <cutils/log.h>
29
30#include "sqlite3_android.h"
31#include "PhoneNumberUtils.h"
32#include "PhonebookIndex.h"
33#include "PhoneticStringUtils.h"
34
35#define ENABLE_ANDROID_LOG 0
36#define SMALL_BUFFER_SIZE 10
37
38static int collate16(void *p, int n1, const void *v1, int n2, const void *v2)
39{
40    UCollator *coll = (UCollator *) p;
41    UCollationResult result = ucol_strcoll(coll, (const UChar *) v1, n1,
42                                                 (const UChar *) v2, n2);
43
44    if (result == UCOL_LESS) {
45        return -1;
46    } else if (result == UCOL_GREATER) {
47        return 1;
48    } else {
49        return 0;
50    }
51}
52
53static int collate8(void *p, int n1, const void *v1, int n2, const void *v2)
54{
55    UCollator *coll = (UCollator *) p;
56    UCharIterator i1, i2;
57    UErrorCode status = U_ZERO_ERROR;
58
59    uiter_setUTF8(&i1, (const char *) v1, n1);
60    uiter_setUTF8(&i2, (const char *) v2, n2);
61
62    UCollationResult result = ucol_strcollIter(coll, &i1, &i2, &status);
63
64    if (U_FAILURE(status)) {
65//        LOGE("Collation iterator error: %d\n", status);
66    }
67
68    if (result == UCOL_LESS) {
69        return -1;
70    } else if (result == UCOL_GREATER) {
71        return 1;
72    } else {
73        return 0;
74    }
75}
76
77/**
78 * Obtains the first UNICODE letter from the supplied string, normalizes and returns it.
79 */
80static void get_phonebook_index(
81    sqlite3_context * context, int argc, sqlite3_value ** argv)
82{
83    if (argc != 2) {
84      sqlite3_result_null(context);
85      return;
86    }
87
88    char const * src = (char const *)sqlite3_value_text(argv[0]);
89    char const * locale = (char const *)sqlite3_value_text(argv[1]);
90    if (src == NULL || src[0] == 0 || locale == NULL) {
91      sqlite3_result_null(context);
92      return;
93    }
94
95    UCharIterator iter;
96    uiter_setUTF8(&iter, src, -1);
97
98    UBool isError = FALSE;
99    UChar index[SMALL_BUFFER_SIZE];
100    uint32_t len = android::GetPhonebookIndex(&iter, locale, index, sizeof(index), &isError);
101    if (isError) {
102      sqlite3_result_null(context);
103      return;
104    }
105
106    uint32_t outlen = 0;
107    uint8_t out[SMALL_BUFFER_SIZE];
108    for (uint32_t i = 0; i < len; i++) {
109      U8_APPEND(out, outlen, sizeof(out), index[i], isError);
110      if (isError) {
111        sqlite3_result_null(context);
112        return;
113      }
114    }
115
116    if (outlen == 0) {
117      sqlite3_result_null(context);
118      return;
119    }
120
121    sqlite3_result_text(context, (const char*)out, outlen, SQLITE_TRANSIENT);
122}
123
124static void get_phonetically_sortable_string(
125    sqlite3_context * context, int argc, sqlite3_value ** argv)
126{
127    if (argc != 1) {
128      sqlite3_result_null(context);
129      return;
130    }
131    char const * src = (char const *)sqlite3_value_text(argv[0]);
132    char * ret;
133    size_t len;
134
135    if (!android::GetPhoneticallySortableString(src, &ret, &len)) {
136        // Put this text at the end of a list.
137        sqlite3_result_text(context, "\xF0\x9F\xBF\xBD", -1, SQLITE_STATIC);
138        // sqlite3_result_null(context);
139    } else {
140        sqlite3_result_text(context, ret, len, free);
141    }
142}
143
144static void get_normalized_string(
145    sqlite3_context * context, int argc, sqlite3_value ** argv)
146{
147    if (argc != 1) {
148      sqlite3_result_null(context);
149      return;
150    }
151    char const * src = (char const *)sqlite3_value_text(argv[0]);
152    char * ret;
153    size_t len;
154
155    if (!android::GetNormalizedString(src, &ret, &len)) {
156        // Probably broken string. Return 0 length string.
157        sqlite3_result_text(context, "", -1, SQLITE_STATIC);
158    } else {
159        sqlite3_result_text(context, ret, len, free);
160    }
161}
162
163static void phone_numbers_equal(sqlite3_context * context, int argc, sqlite3_value ** argv)
164{
165    if (argc != 2 && argc != 3) {
166        sqlite3_result_int(context, 0);
167        return;
168    }
169
170    char const * num1 = (char const *)sqlite3_value_text(argv[0]);
171    char const * num2 = (char const *)sqlite3_value_text(argv[1]);
172
173    bool use_strict = false;
174    if (argc == 3) {
175        use_strict = (sqlite3_value_int(argv[2]) != 0);
176    }
177
178    if (num1 == NULL || num2 == NULL) {
179        sqlite3_result_null(context);
180        return;
181    }
182
183    bool equal =
184        (use_strict ?
185         android::phone_number_compare_strict(num1, num2) :
186         android::phone_number_compare_loose(num1, num2));
187
188    if (equal) {
189        sqlite3_result_int(context, 1);
190    } else {
191        sqlite3_result_int(context, 0);
192    }
193}
194
195#if ENABLE_ANDROID_LOG
196static void android_log(sqlite3_context * context, int argc, sqlite3_value ** argv)
197{
198    char const * tag = "sqlite_trigger";
199    char const * msg = "";
200    int msgIndex = 0;
201
202    switch (argc) {
203        case 2:
204            tag = (char const *)sqlite3_value_text(argv[0]);
205            if (tag == NULL) {
206                tag = "sqlite_trigger";
207            }
208            msgIndex = 1;
209        case 1:
210            msg = (char const *)sqlite3_value_text(argv[msgIndex]);
211            if (msg == NULL) {
212                msg = "";
213            }
214            LOG(LOG_INFO, tag, msg);
215            sqlite3_result_int(context, 1);
216            return;
217
218        default:
219            sqlite3_result_int(context, 0);
220            return;
221    }
222}
223#endif
224
225static void delete_file(sqlite3_context * context, int argc, sqlite3_value ** argv)
226{
227    if (argc != 1) {
228        sqlite3_result_int(context, 0);
229        return;
230    }
231
232    char const * path = (char const *)sqlite3_value_text(argv[0]);
233    if (path == NULL) {
234        sqlite3_result_null(context);
235        return;
236    }
237
238    if (strncmp("/sdcard/", path, 8) != 0) {
239        sqlite3_result_null(context);
240        return;
241    }
242    if (strstr(path, "/../") != NULL) {
243        sqlite3_result_null(context);
244        return;
245    }
246
247    int err = unlink(path);
248    if (err != -1) {
249        // No error occured, return true
250        sqlite3_result_int(context, 1);
251    } else {
252        // An error occured, return false
253        sqlite3_result_int(context, 0);
254    }
255}
256
257static void tokenize_auxdata_delete(void * data)
258{
259    sqlite3_stmt * statement = (sqlite3_stmt *)data;
260    sqlite3_finalize(statement);
261}
262
263static void base16Encode(char* dest, const char* src, uint32_t size)
264{
265    static const char * BASE16_TABLE = "0123456789abcdef";
266    for (uint32_t i = 0; i < size; i++) {
267        char ch = *src++;
268        *dest++ = BASE16_TABLE[ (ch & 0xf0) >> 4 ];
269        *dest++ = BASE16_TABLE[ (ch & 0x0f)      ];
270    }
271}
272
273struct SqliteUserData {
274    sqlite3 * handle;
275    UCollator* collator;
276};
277
278/**
279 * This function is invoked as:
280 *
281 *  _TOKENIZE('<token_table>', <data_row_id>, <data>, <delimiter>,
282 *             <use_token_index>, <data_tag>)
283 *
284 * If <use_token_index> is omitted, it is treated as 0.
285 * If <data_tag> is omitted, it is treated as NULL.
286 *
287 * It will split <data> on each instance of <delimiter> and insert each token
288 * into <token_table>. The following columns in <token_table> are used:
289 * token TEXT, source INTEGER, token_index INTEGER, tag (any type)
290 * The token_index column is not required if <use_token_index> is 0.
291 * The tag column is not required if <data_tag> is NULL.
292 *
293 * One row is inserted for each token in <data>.
294 * In each inserted row, 'source' is <data_row_id>.
295 * In the first inserted row, 'token' is the hex collation key of
296 * the entire <data> string, and 'token_index' is 0.
297 * In each row I (where 1 <= I < N, and N is the number of tokens in <data>)
298 * 'token' will be set to the hex collation key of the I:th token (0-based).
299 * If <use_token_index> != 0, 'token_index' is set to I.
300 * If <data_tag> is not NULL, 'tag' is set to <data_tag>.
301 *
302 * In other words, there will be one row for the entire string,
303 * and one row for each token except the first one.
304 *
305 * The function returns the number of tokens generated.
306 */
307static void tokenize(sqlite3_context * context, int argc, sqlite3_value ** argv)
308{
309    //LOGD("enter tokenize");
310    int err;
311    int useTokenIndex = 0;
312    int useDataTag = 0;
313
314    if (!(argc >= 4 || argc <= 6)) {
315        LOGE("Tokenize requires 4 to 6 arguments");
316        sqlite3_result_null(context);
317        return;
318    }
319
320    if (argc > 4) {
321        useTokenIndex = sqlite3_value_int(argv[4]);
322    }
323
324    if (argc > 5) {
325        useDataTag = (sqlite3_value_type(argv[5]) != SQLITE_NULL);
326    }
327
328    sqlite3 * handle = sqlite3_context_db_handle(context);
329    UCollator* collator = (UCollator*)sqlite3_user_data(context);
330    char const * tokenTable = (char const *)sqlite3_value_text(argv[0]);
331    if (tokenTable == NULL) {
332        LOGE("tokenTable null");
333        sqlite3_result_null(context);
334        return;
335    }
336
337    // Get or create the prepared statement for the insertions
338    sqlite3_stmt * statement = (sqlite3_stmt *)sqlite3_get_auxdata(context, 0);
339    if (!statement) {
340        char const * tokenIndexCol = useTokenIndex ? ", token_index" : "";
341        char const * tokenIndexParam = useTokenIndex ? ", ?" : "";
342        char const * dataTagCol = useDataTag ? ", tag" : "";
343        char const * dataTagParam = useDataTag ? ", ?" : "";
344        char * sql = sqlite3_mprintf("INSERT INTO %s (token, source%s%s) VALUES (?, ?%s%s);",
345                tokenTable, tokenIndexCol, dataTagCol, tokenIndexParam, dataTagParam);
346        err = sqlite3_prepare_v2(handle, sql, -1, &statement, NULL);
347        sqlite3_free(sql);
348        if (err) {
349            LOGE("prepare failed");
350            sqlite3_result_null(context);
351            return;
352        }
353        // This binds the statement to the table it was compiled against, which is argv[0].
354        // If this function is ever called with a different table the finalizer will be called
355        // and sqlite3_get_auxdata() will return null above, forcing a recompile for the new table.
356        sqlite3_set_auxdata(context, 0, statement, tokenize_auxdata_delete);
357    } else {
358        // Reset the cached statement so that binding the row ID will work properly
359        sqlite3_reset(statement);
360    }
361
362    // Bind the row ID of the source row
363    int64_t rowID = sqlite3_value_int64(argv[1]);
364    err = sqlite3_bind_int64(statement, 2, rowID);
365    if (err != SQLITE_OK) {
366        LOGE("bind failed");
367        sqlite3_result_null(context);
368        return;
369    }
370
371    // Bind <data_tag> to the tag column
372    if (useDataTag) {
373        int dataTagParamIndex = useTokenIndex ? 4 : 3;
374        err = sqlite3_bind_value(statement, dataTagParamIndex, argv[5]);
375        if (err != SQLITE_OK) {
376            LOGE("bind failed");
377            sqlite3_result_null(context);
378            return;
379        }
380    }
381
382    // Get the raw bytes for the string to tokenize
383    // the string will be modified by following code
384    // however, sqlite did not reuse the string, so it is safe to not dup it
385    UChar * origData = (UChar *)sqlite3_value_text16(argv[2]);
386    if (origData == NULL) {
387        sqlite3_result_null(context);
388        return;
389    }
390
391    // Get the raw bytes for the delimiter
392    const UChar * delim = (const UChar *)sqlite3_value_text16(argv[3]);
393    if (delim == NULL) {
394        LOGE("can't get delimiter");
395        sqlite3_result_null(context);
396        return;
397    }
398
399    UChar * token = NULL;
400    UChar *state;
401    int numTokens = 0;
402
403    do {
404        if (numTokens == 0) {
405            token = origData;
406        }
407
408        // Reset the program so we can use it to perform the insert
409        sqlite3_reset(statement);
410        UErrorCode status = U_ZERO_ERROR;
411        char keybuf[1024];
412        uint32_t result = ucol_getSortKey(collator, token, -1, (uint8_t*)keybuf, sizeof(keybuf)-1);
413        if (result > sizeof(keybuf)) {
414            // TODO allocate memory for this super big string
415            LOGE("ucol_getSortKey needs bigger buffer %d", result);
416            break;
417        }
418        uint32_t keysize = result-1;
419        uint32_t base16Size = keysize*2;
420        char *base16buf = (char*)malloc(base16Size);
421        base16Encode(base16buf, keybuf, keysize);
422        err = sqlite3_bind_text(statement, 1, base16buf, base16Size, SQLITE_STATIC);
423
424        if (err != SQLITE_OK) {
425            LOGE(" sqlite3_bind_text16 error %d", err);
426            free(base16buf);
427            break;
428        }
429
430        if (useTokenIndex) {
431            err = sqlite3_bind_int(statement, 3, numTokens);
432            if (err != SQLITE_OK) {
433                LOGE(" sqlite3_bind_int error %d", err);
434                free(base16buf);
435                break;
436            }
437        }
438
439        err = sqlite3_step(statement);
440        free(base16buf);
441
442        if (err != SQLITE_DONE) {
443            LOGE(" sqlite3_step error %d", err);
444            break;
445        }
446        numTokens++;
447        if (numTokens == 1) {
448            // first call
449            u_strtok_r(origData, delim, &state);
450        }
451    } while ((token = u_strtok_r(NULL, delim, &state)) != NULL);
452    sqlite3_result_int(context, numTokens);
453}
454
455static void localized_collator_dtor(UCollator* collator)
456{
457    ucol_close(collator);
458}
459
460#define LOCALIZED_COLLATOR_NAME "LOCALIZED"
461
462extern "C" int register_localized_collators(sqlite3* handle, const char* systemLocale, int utf16Storage)
463{
464    int err;
465    UErrorCode status = U_ZERO_ERROR;
466    void* icudata;
467
468    UCollator* collator = ucol_open(systemLocale, &status);
469    if (U_FAILURE(status)) {
470        return -1;
471    }
472
473    ucol_setAttribute(collator, UCOL_STRENGTH, UCOL_PRIMARY, &status);
474    if (U_FAILURE(status)) {
475        return -1;
476    }
477
478    status = U_ZERO_ERROR;
479    char buf[1024];
480    int n = ucol_getShortDefinitionString(collator, NULL, buf, 1024, &status);
481
482    if (utf16Storage) {
483        err = sqlite3_create_collation_v2(handle, LOCALIZED_COLLATOR_NAME, SQLITE_UTF16, collator,
484                collate16, (void(*)(void*))localized_collator_dtor);
485    } else {
486        err = sqlite3_create_collation_v2(handle, LOCALIZED_COLLATOR_NAME, SQLITE_UTF8, collator,
487                collate8, (void(*)(void*))localized_collator_dtor);
488    }
489    if (err != SQLITE_OK) {
490        return err;
491    }
492
493    // Register the _TOKENIZE function
494    err = sqlite3_create_function(handle, "_TOKENIZE", 4, SQLITE_UTF16, collator, tokenize, NULL, NULL);
495    if (err != SQLITE_OK) {
496        return err;
497    }
498    err = sqlite3_create_function(handle, "_TOKENIZE", 5, SQLITE_UTF16, collator, tokenize, NULL, NULL);
499    if (err != SQLITE_OK) {
500        return err;
501    }
502    err = sqlite3_create_function(handle, "_TOKENIZE", 6, SQLITE_UTF16, collator, tokenize, NULL, NULL);
503    if (err != SQLITE_OK) {
504        return err;
505    }
506
507    return SQLITE_OK;
508}
509
510
511extern "C" int register_android_functions(sqlite3 * handle, int utf16Storage)
512{
513    int err;
514    UErrorCode status = U_ZERO_ERROR;
515
516    UCollator * collator = ucol_open(NULL, &status);
517    if (U_FAILURE(status)) {
518        return -1;
519    }
520
521    if (utf16Storage) {
522        // Note that text should be stored as UTF-16
523        err = sqlite3_exec(handle, "PRAGMA encoding = 'UTF-16'", 0, 0, 0);
524        if (err != SQLITE_OK) {
525            return err;
526        }
527
528        // Register the UNICODE collation
529        err = sqlite3_create_collation_v2(handle, "UNICODE", SQLITE_UTF16, collator, collate16,
530                (void(*)(void*))localized_collator_dtor);
531    } else {
532        err = sqlite3_create_collation_v2(handle, "UNICODE", SQLITE_UTF8, collator, collate8,
533                (void(*)(void*))localized_collator_dtor);
534    }
535
536    if (err != SQLITE_OK) {
537        return err;
538    }
539
540    // Register the PHONE_NUM_EQUALS function
541    err = sqlite3_create_function(
542        handle, "PHONE_NUMBERS_EQUAL", 2,
543        SQLITE_UTF8, NULL, phone_numbers_equal, NULL, NULL);
544    if (err != SQLITE_OK) {
545        return err;
546    }
547
548    // Register the PHONE_NUM_EQUALS function with an additional argument "use_strict"
549    err = sqlite3_create_function(
550        handle, "PHONE_NUMBERS_EQUAL", 3,
551        SQLITE_UTF8, NULL, phone_numbers_equal, NULL, NULL);
552    if (err != SQLITE_OK) {
553        return err;
554    }
555
556    // Register the _DELETE_FILE function
557    err = sqlite3_create_function(handle, "_DELETE_FILE", 1, SQLITE_UTF8, NULL, delete_file, NULL, NULL);
558    if (err != SQLITE_OK) {
559        return err;
560    }
561
562#if ENABLE_ANDROID_LOG
563    // Register the _LOG function
564    err = sqlite3_create_function(handle, "_LOG", 1, SQLITE_UTF8, NULL, android_log, NULL, NULL);
565    if (err != SQLITE_OK) {
566        return err;
567    }
568#endif
569
570    // Register the GET_PHONETICALLY_SORTABLE_STRING function
571    err = sqlite3_create_function(handle,
572                                  "GET_PHONETICALLY_SORTABLE_STRING",
573                                  1, SQLITE_UTF8, NULL,
574                                  get_phonetically_sortable_string,
575                                  NULL, NULL);
576    if (err != SQLITE_OK) {
577        return err;
578    }
579
580    // Register the GET_NORMALIZED_STRING function
581    err = sqlite3_create_function(handle,
582                                  "GET_NORMALIZED_STRING",
583                                  1, SQLITE_UTF8, NULL,
584                                  get_normalized_string,
585                                  NULL, NULL);
586    if (err != SQLITE_OK) {
587        return err;
588    }
589
590    // Register the GET_PHONEBOOK_INDEX function
591    err = sqlite3_create_function(handle,
592        "GET_PHONEBOOK_INDEX",
593        2, SQLITE_UTF8, NULL,
594        get_phonebook_index,
595        NULL, NULL);
596    if (err != SQLITE_OK) {
597        return err;
598    }
599
600    return SQLITE_OK;
601}
602