summaryrefslogtreecommitdiff
path: root/ext/fts3/fts3_porter.c
blob: db175acdbbea8e1c6d29c40aeafc874b0edf0eb2 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
/*
** 2006 September 30
**
** The author disclaims copyright to this source code.  In place of
** a legal notice, here is a blessing:
**
**    May you do good and not evil.
**    May you find forgiveness for yourself and forgive others.
**    May you share freely, never taking more than you give.
**
*************************************************************************
** Implementation of the full-text-search tokenizer that implements
** a Porter stemmer.
*/

/*
** The code in this file is only compiled if:
**
**     * The FTS3 module is being built as an extension
**       (in which case SQLITE_CORE is not defined), or
**
**     * The FTS3 module is being built into the core of
**       SQLite (in which case SQLITE_ENABLE_FTS3 is defined).
*/
#include "fts3Int.h"
#if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3)

#include <assert.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "fts3_tokenizer.h"

/*
** Class derived from sqlite3_tokenizer
*/
typedef struct porter_tokenizer {
  sqlite3_tokenizer base;      /* Base class */
} porter_tokenizer;

/*
** Class derived from sqlite3_tokenizer_cursor
*/
typedef struct porter_tokenizer_cursor {
  sqlite3_tokenizer_cursor base;
  const char *zInput;          /* input we are tokenizing */
  int nInput;                  /* size of the input */
  int iOffset;                 /* current position in zInput */
  int iToken;                  /* index of next token to be returned */
  char *zToken;                /* storage for current token */
  int nAllocated;              /* space allocated to zToken buffer */
} porter_tokenizer_cursor;


/*
** Create a new tokenizer instance.
*/
static int porterCreate(
  int argc, const char * const *argv,
  sqlite3_tokenizer **ppTokenizer
){
  porter_tokenizer *t;

  UNUSED_PARAMETER(argc);
  UNUSED_PARAMETER(argv);

  t = (porter_tokenizer *) sqlite3_malloc(sizeof(*t));
  if( t==NULL ) return SQLITE_NOMEM;
  memset(t, 0, sizeof(*t));
  *ppTokenizer = &t->base;
  return SQLITE_OK;
}

/*
** Destroy a tokenizer
*/
static int porterDestroy(sqlite3_tokenizer *pTokenizer){
  sqlite3_free(pTokenizer);
  return SQLITE_OK;
}

/*
** Prepare to begin tokenizing a particular string.  The input
** string to be tokenized is zInput[0..nInput-1].  A cursor
** used to incrementally tokenize this string is returned in 
** *ppCursor.
*/
static int porterOpen(
  sqlite3_tokenizer *pTokenizer,         /* The tokenizer */
  const char *zInput, int nInput,        /* String to be tokenized */
  sqlite3_tokenizer_cursor **ppCursor    /* OUT: Tokenization cursor */
){
  porter_tokenizer_cursor *c;

  UNUSED_PARAMETER(pTokenizer);

  c = (porter_tokenizer_cursor *) sqlite3_malloc(sizeof(*c));
  if( c==NULL ) return SQLITE_NOMEM;

  c->zInput = zInput;
  if( zInput==0 ){
    c->nInput = 0;
  }else if( nInput<0 ){
    c->nInput = (int)strlen(zInput);
  }else{
    c->nInput = nInput;
  }
  c->iOffset = 0;                 /* start tokenizing at the beginning */
  c->iToken = 0;
  c->zToken = NULL;               /* no space allocated, yet. */
  c->nAllocated = 0;

  *ppCursor = &c->base;
  return SQLITE_OK;
}

/*
** Close a tokenization cursor previously opened by a call to
** porterOpen() above.
*/
static int porterClose(sqlite3_tokenizer_cursor *pCursor){
  porter_tokenizer_cursor *c = (porter_tokenizer_cursor *) pCursor;
  sqlite3_free(c->zToken);
  sqlite3_free(c);
  return SQLITE_OK;
}
/*
** Vowel or consonant
*/
static const char cType[] = {
   0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0,
   1, 1, 1, 2, 1
};

/*
** isConsonant() and isVowel() determine if their first character in
** the string they point to is a consonant or a vowel, according
** to Porter ruls.  
**
** A consonate is any letter other than 'a', 'e', 'i', 'o', or 'u'.
** 'Y' is a consonant unless it follows another consonant,
** in which case it is a vowel.
**
** In these routine, the letters are in reverse order.  So the 'y' rule
** is that 'y' is a consonant unless it is followed by another
** consonent.
*/
static int isVowel(const char*);
static int isConsonant(const char *z){
  int j;
  char x = *z;
  if( x==0 ) return 0;
  assert( x>='a' && x<='z' );
  j = cType[x-'a'];
  if( j<2 ) return j;
  return z[1]==0 || isVowel(z + 1);
}
static int isVowel(const char *z){
  int j;
  char x = *z;
  if( x==0 ) return 0;
  assert( x>='a' && x<='z' );
  j = cType[x-'a'];
  if( j<2 ) return 1-j;
  return isConsonant(z + 1);
}

/*
** Let any sequence of one or more vowels be represented by V and let
** C be sequence of one or more consonants.  Then every word can be
** represented as:
**
**           [C] (VC){m} [V]
**
** In prose:  A word is an optional consonant followed by zero or
** vowel-consonant pairs followed by an optional vowel.  "m" is the
** number of vowel consonant pairs.  This routine computes the value
** of m for the first i bytes of a word.
**
** Return true if the m-value for z is 1 or more.  In other words,
** return true if z contains at least one vowel that is followed
** by a consonant.
**
** In this routine z[] is in reverse order.  So we are really looking
** for an instance of of a consonant followed by a vowel.
*/
static int m_gt_0(const char *z){
  while( isVowel(z) ){ z++; }
  if( *z==0 ) return 0;
  while( isConsonant(z) ){ z++; }
  return *z!=0;
}

/* Like mgt0 above except we are looking for a value of m which is
** exactly 1
*/
static int m_eq_1(const char *z){
  while( isVowel(z) ){ z++; }
  if( *z==0 ) return 0;
  while( isConsonant(z) ){ z++; }
  if( *z==0 ) return 0;
  while( isVowel(z) ){ z++; }
  if( *z==0 ) return 1;
  while( isConsonant(z) ){ z++; }
  return *z==0;
}

/* Like mgt0 above except we are looking for a value of m>1 instead
** or m>0
*/
static int m_gt_1(const char *z){
  while( isVowel(z) ){ z++; }
  if( *z==0 ) return 0;
  while( isConsonant(z) ){ z++; }
  if( *z==0 ) return 0;
  while( isVowel(z) ){ z++; }
  if( *z==0 ) return 0;
  while( isConsonant(z) ){ z++; }
  return *z!=0;
}

/*
** Return TRUE if there is a vowel anywhere within z[0..n-1]
*/
static int hasVowel(const char *z){
  while( isConsonant(z) ){ z++; }
  return *z!=0;
}

/*
** Return TRUE if the word ends in a double consonant.
**
** The text is reversed here. So we are really looking at
** the first two characters of z[].
*/
static int doubleConsonant(const char *z){
  return isConsonant(z) && z[0]==z[1];
}

/*
** Return TRUE if the word ends with three letters which
** are consonant-vowel-consonent and where the final consonant
** is not 'w', 'x', or 'y'.
**
** The word is reversed here.  So we are really checking the
** first three letters and the first one cannot be in [wxy].
*/
static int star_oh(const char *z){
  return
    isConsonant(z) &&
    z[0]!='w' && z[0]!='x' && z[0]!='y' &&
    isVowel(z+1) &&
    isConsonant(z+2);
}

/*
** If the word ends with zFrom and xCond() is true for the stem
** of the word that preceeds the zFrom ending, then change the 
** ending to zTo.
**
** The input word *pz and zFrom are both in reverse order.  zTo
** is in normal order. 
**
** Return TRUE if zFrom matches.  Return FALSE if zFrom does not
** match.  Not that TRUE is returned even if xCond() fails and
** no substitution occurs.
*/
static int stem(
  char **pz,             /* The word being stemmed (Reversed) */
  const char *zFrom,     /* If the ending matches this... (Reversed) */
  const char *zTo,       /* ... change the ending to this (not reversed) */
  int (*xCond)(const char*)   /* Condition that must be true */
){
  char *z = *pz;
  while( *zFrom && *zFrom==*z ){ z++; zFrom++; }
  if( *zFrom!=0 ) return 0;
  if( xCond && !xCond(z) ) return 1;
  while( *zTo ){
    *(--z) = *(zTo++);
  }
  *pz = z;
  return 1;
}

/*
** This is the fallback stemmer used when the porter stemmer is
** inappropriate.  The input word is copied into the output with
** US-ASCII case folding.  If the input word is too long (more
** than 20 bytes if it contains no digits or more than 6 bytes if
** it contains digits) then word is truncated to 20 or 6 bytes
** by taking 10 or 3 bytes from the beginning and end.
*/
static void copy_stemmer(const char *zIn, int nIn, char *zOut, int *pnOut){
  int i, mx, j;
  int hasDigit = 0;
  for(i=0; i<nIn; i++){
    char c = zIn[i];
    if( c>='A' && c<='Z' ){
      zOut[i] = c - 'A' + 'a';
    }else{
      if( c>='0' && c<='9' ) hasDigit = 1;
      zOut[i] = c;
    }
  }
  mx = hasDigit ? 3 : 10;
  if( nIn>mx*2 ){
    for(j=mx, i=nIn-mx; i<nIn; i++, j++){
      zOut[j] = zOut[i];
    }
    i = j;
  }
  zOut[i] = 0;
  *pnOut = i;
}


/*
** Stem the input word zIn[0..nIn-1].  Store the output in zOut.
** zOut is at least big enough to hold nIn bytes.  Write the actual
** size of the output word (exclusive of the '\0' terminator) into *pnOut.
**
** Any upper-case characters in the US-ASCII character set ([A-Z])
** are converted to lower case.  Upper-case UTF characters are
** unchanged.
**
** Words that are longer than about 20 bytes are stemmed by retaining
** a few bytes from the beginning and the end of the word.  If the
** word contains digits, 3 bytes are taken from the beginning and
** 3 bytes from the end.  For long words without digits, 10 bytes
** are taken from each end.  US-ASCII case folding still applies.
** 
** If the input word contains not digits but does characters not 
** in [a-zA-Z] then no stemming is attempted and this routine just 
** copies the input into the input into the output with US-ASCII
** case folding.
**
** Stemming never increases the length of the word.  So there is
** no chance of overflowing the zOut buffer.
*/
static void porter_stemmer(const char *zIn, int nIn, char *zOut, int *pnOut){
  int i, j;
  char zReverse[28];
  char *z, *z2;
  if( nIn<3 || nIn>=(int)sizeof(zReverse)-7 ){
    /* The word is too big or too small for the porter stemmer.
    ** Fallback to the copy stemmer */
    copy_stemmer(zIn, nIn, zOut, pnOut);
    return;
  }
  for(i=0, j=sizeof(zReverse)-6; i<nIn; i++, j--){
    char c = zIn[i];
    if( c>='A' && c<='Z' ){
      zReverse[j] = c + 'a' - 'A';
    }else if( c>='a' && c<='z' ){
      zReverse[j] = c;
    }else{
      /* The use of a character not in [a-zA-Z] means that we fallback
      ** to the copy stemmer */
      copy_stemmer(zIn, nIn, zOut, pnOut);
      return;
    }
  }
  memset(&zReverse[sizeof(zReverse)-5], 0, 5);
  z = &zReverse[j+1];


  /* Step 1a */
  if( z[0]=='s' ){
    if(
     !stem(&z, "sess", "ss", 0) &&
     !stem(&z, "sei", "i", 0)  &&
     !stem(&z, "ss", "ss", 0)
    ){
      z++;
    }
  }

  /* Step 1b */  
  z2 = z;
  if( stem(&z, "dee", "ee", m_gt_0) ){
    /* Do nothing.  The work was all in the test */
  }else if( 
     (stem(&z, "gni", "", hasVowel) || stem(&z, "de", "", hasVowel))
      && z!=z2
  ){
     if( stem(&z, "ta", "ate", 0) ||
         stem(&z, "lb", "ble", 0) ||
         stem(&z, "zi", "ize", 0) ){
       /* Do nothing.  The work was all in the test */
     }else if( doubleConsonant(z) && (*z!='l' && *z!='s' && *z!='z') ){
       z++;
     }else if( m_eq_1(z) && star_oh(z) ){
       *(--z) = 'e';
     }
  }

  /* Step 1c */
  if( z[0]=='y' && hasVowel(z+1) ){
    z[0] = 'i';
  }

  /* Step 2 */
  switch( z[1] ){
   case 'a':
     if( !stem(&z, "lanoita", "ate", m_gt_0) ){
       stem(&z, "lanoit", "tion", m_gt_0);
     }
     break;
   case 'c':
     if( !stem(&z, "icne", "ence", m_gt_0) ){
       stem(&z, "icna", "ance", m_gt_0);
     }
     break;
   case 'e':
     stem(&z, "rezi", "ize", m_gt_0);
     break;
   case 'g':
     stem(&z, "igol", "log", m_gt_0);
     break;
   case 'l':
     if( !stem(&z, "ilb", "ble", m_gt_0) 
      && !stem(&z, "illa", "al", m_gt_0)
      && !stem(&z, "iltne", "ent", m_gt_0)
      && !stem(&z, "ile", "e", m_gt_0)
     ){
       stem(&z, "ilsuo", "ous", m_gt_0);
     }
     break;
   case 'o':
     if( !stem(&z, "noitazi", "ize", m_gt_0)
      && !stem(&z, "noita", "ate", m_gt_0)
     ){
       stem(&z, "rota", "ate", m_gt_0);
     }
     break;
   case 's':
     if( !stem(&z, "msila", "al", m_gt_0)
      && !stem(&z, "ssenevi", "ive", m_gt_0)
      && !stem(&z, "ssenluf", "ful", m_gt_0)
     ){
       stem(&z, "ssensuo", "ous", m_gt_0);
     }
     break;
   case 't':
     if( !stem(&z, "itila", "al", m_gt_0)
      && !stem(&z, "itivi", "ive", m_gt_0)
     ){
       stem(&z, "itilib", "ble", m_gt_0);
     }
     break;
  }

  /* Step 3 */
  switch( z[0] ){
   case 'e':
     if( !stem(&z, "etaci", "ic", m_gt_0)
      && !stem(&z, "evita", "", m_gt_0)
     ){
       stem(&z, "ezila", "al", m_gt_0);
     }
     break;
   case 'i':
     stem(&z, "itici", "ic", m_gt_0);
     break;
   case 'l':
     if( !stem(&z, "laci", "ic", m_gt_0) ){
       stem(&z, "luf", "", m_gt_0);
     }
     break;
   case 's':
     stem(&z, "ssen", "", m_gt_0);
     break;
  }

  /* Step 4 */
  switch( z[1] ){
   case 'a':
     if( z[0]=='l' && m_gt_1(z+2) ){
       z += 2;
     }
     break;
   case 'c':
     if( z[0]=='e' && z[2]=='n' && (z[3]=='a' || z[3]=='e')  && m_gt_1(z+4)  ){
       z += 4;
     }
     break;
   case 'e':
     if( z[0]=='r' && m_gt_1(z+2) ){
       z += 2;
     }
     break;
   case 'i':
     if( z[0]=='c' && m_gt_1(z+2) ){
       z += 2;
     }
     break;
   case 'l':
     if( z[0]=='e' && z[2]=='b' && (z[3]=='a' || z[3]=='i') && m_gt_1(z+4) ){
       z += 4;
     }
     break;
   case 'n':
     if( z[0]=='t' ){
       if( z[2]=='a' ){
         if( m_gt_1(z+3) ){
           z += 3;
         }
       }else if( z[2]=='e' ){
         if( !stem(&z, "tneme", "", m_gt_1)
          && !stem(&z, "tnem", "", m_gt_1)
         ){
           stem(&z, "tne", "", m_gt_1);
         }
       }
     }
     break;
   case 'o':
     if( z[0]=='u' ){
       if( m_gt_1(z+2) ){
         z += 2;
       }
     }else if( z[3]=='s' || z[3]=='t' ){
       stem(&z, "noi", "", m_gt_1);
     }
     break;
   case 's':
     if( z[0]=='m' && z[2]=='i' && m_gt_1(z+3) ){
       z += 3;
     }
     break;
   case 't':
     if( !stem(&z, "eta", "", m_gt_1) ){
       stem(&z, "iti", "", m_gt_1);
     }
     break;
   case 'u':
     if( z[0]=='s' && z[2]=='o' && m_gt_1(z+3) ){
       z += 3;
     }
     break;
   case 'v':
   case 'z':
     if( z[0]=='e' && z[2]=='i' && m_gt_1(z+3) ){
       z += 3;
     }
     break;
  }

  /* Step 5a */
  if( z[0]=='e' ){
    if( m_gt_1(z+1) ){
      z++;
    }else if( m_eq_1(z+1) && !star_oh(z+1) ){
      z++;
    }
  }

  /* Step 5b */
  if( m_gt_1(z) && z[0]=='l' && z[1]=='l' ){
    z++;
  }

  /* z[] is now the stemmed word in reverse order.  Flip it back
  ** around into forward order and return.
  */
  *pnOut = i = (int)strlen(z);
  zOut[i] = 0;
  while( *z ){
    zOut[--i] = *(z++);
  }
}

/*
** Characters that can be part of a token.  We assume any character
** whose value is greater than 0x80 (any UTF character) can be
** part of a token.  In other words, delimiters all must have
** values of 0x7f or lower.
*/
static const char porterIdChar[] = {
/* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xA xB xC xD xE xF */
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,  /* 3x */
    0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,  /* 4x */
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1,  /* 5x */
    0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,  /* 6x */
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0,  /* 7x */
};
#define isDelim(C) (((ch=C)&0x80)==0 && (ch<0x30 || !porterIdChar[ch-0x30]))

/*
** Extract the next token from a tokenization cursor.  The cursor must
** have been opened by a prior call to porterOpen().
*/
static int porterNext(
  sqlite3_tokenizer_cursor *pCursor,  /* Cursor returned by porterOpen */
  const char **pzToken,               /* OUT: *pzToken is the token text */
  int *pnBytes,                       /* OUT: Number of bytes in token */
  int *piStartOffset,                 /* OUT: Starting offset of token */
  int *piEndOffset,                   /* OUT: Ending offset of token */
  int *piPosition                     /* OUT: Position integer of token */
){
  porter_tokenizer_cursor *c = (porter_tokenizer_cursor *) pCursor;
  const char *z = c->zInput;

  while( c->iOffset<c->nInput ){
    int iStartOffset, ch;

    /* Scan past delimiter characters */
    while( c->iOffset<c->nInput && isDelim(z[c->iOffset]) ){
      c->iOffset++;
    }

    /* Count non-delimiter characters. */
    iStartOffset = c->iOffset;
    while( c->iOffset<c->nInput && !isDelim(z[c->iOffset]) ){
      c->iOffset++;
    }

    if( c->iOffset>iStartOffset ){
      int n = c->iOffset-iStartOffset;
      if( n>c->nAllocated ){
        char *pNew;
        c->nAllocated = n+20;
        pNew = sqlite3_realloc(c->zToken, c->nAllocated);
        if( !pNew ) return SQLITE_NOMEM;
        c->zToken = pNew;
      }
      porter_stemmer(&z[iStartOffset], n, c->zToken, pnBytes);
      *pzToken = c->zToken;
      *piStartOffset = iStartOffset;
      *piEndOffset = c->iOffset;
      *piPosition = c->iToken++;
      return SQLITE_OK;
    }
  }
  return SQLITE_DONE;
}

/*
** The set of routines that implement the porter-stemmer tokenizer
*/
static const sqlite3_tokenizer_module porterTokenizerModule = {
  0,
  porterCreate,
  porterDestroy,
  porterOpen,
  porterClose,
  porterNext,
  0
};

/*
** Allocate a new porter tokenizer.  Return a pointer to the new
** tokenizer in *ppModule
*/
void sqlite3Fts3PorterTokenizerModule(
  sqlite3_tokenizer_module const**ppModule
){
  *ppModule = &porterTokenizerModule;
}

#endif /* !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) */