2
** The author disclaims copyright to this source code.
4
*************************************************************************
5
** Implementation of the "simple" full-text-search tokenizer.
9
** The code in this file is only compiled if:
11
** * The FTS1 module is being built as an extension
12
** (in which case SQLITE_CORE is not defined), or
14
** * The FTS1 module is being built into the core of
15
** SQLite (in which case SQLITE_ENABLE_FTS1 is defined).
17
#if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS1)
26
#include "fts1_tokenizer.h"
28
typedef struct simple_tokenizer {
29
sqlite3_tokenizer base;
30
char delim[128]; /* flag ASCII delimiters */
33
typedef struct simple_tokenizer_cursor {
34
sqlite3_tokenizer_cursor base;
35
const char *pInput; /* input we are tokenizing */
36
int nBytes; /* size of the input */
37
int iOffset; /* current position in pInput */
38
int iToken; /* index of next token to be returned */
39
char *pToken; /* storage for current token */
40
int nTokenAllocated; /* space allocated to zToken buffer */
41
} simple_tokenizer_cursor;
44
/* Forward declaration */
45
static const sqlite3_tokenizer_module simpleTokenizerModule;
47
static int isDelim(simple_tokenizer *t, unsigned char c){
48
return c<0x80 && t->delim[c];
52
** Create a new tokenizer instance.
54
static int simpleCreate(
55
int argc, const char * const *argv,
56
sqlite3_tokenizer **ppTokenizer
60
t = (simple_tokenizer *) calloc(sizeof(*t), 1);
61
if( t==NULL ) return SQLITE_NOMEM;
63
/* TODO(shess) Delimiters need to remain the same from run to run,
64
** else we need to reindex. One solution would be a meta-table to
65
** track such information in the database, then we'd only want this
66
** information on the initial create.
69
int i, n = strlen(argv[1]);
71
unsigned char ch = argv[1][i];
72
/* We explicitly don't support UTF-8 delimiters for now. */
80
/* Mark non-alphanumeric ASCII characters as delimiters */
82
for(i=1; i<0x80; i++){
83
t->delim[i] = !isalnum(i);
87
*ppTokenizer = &t->base;
92
** Destroy a tokenizer
94
static int simpleDestroy(sqlite3_tokenizer *pTokenizer){
100
** Prepare to begin tokenizing a particular string. The input
101
** string to be tokenized is pInput[0..nBytes-1]. A cursor
102
** used to incrementally tokenize this string is returned in
105
static int simpleOpen(
106
sqlite3_tokenizer *pTokenizer, /* The tokenizer */
107
const char *pInput, int nBytes, /* String to be tokenized */
108
sqlite3_tokenizer_cursor **ppCursor /* OUT: Tokenization cursor */
110
simple_tokenizer_cursor *c;
112
c = (simple_tokenizer_cursor *) malloc(sizeof(*c));
113
if( c==NULL ) return SQLITE_NOMEM;
118
}else if( nBytes<0 ){
119
c->nBytes = (int)strlen(pInput);
123
c->iOffset = 0; /* start tokenizing at the beginning */
125
c->pToken = NULL; /* no space allocated, yet. */
126
c->nTokenAllocated = 0;
128
*ppCursor = &c->base;
133
** Close a tokenization cursor previously opened by a call to
134
** simpleOpen() above.
136
static int simpleClose(sqlite3_tokenizer_cursor *pCursor){
137
simple_tokenizer_cursor *c = (simple_tokenizer_cursor *) pCursor;
144
** Extract the next token from a tokenization cursor. The cursor must
145
** have been opened by a prior call to simpleOpen().
147
static int simpleNext(
148
sqlite3_tokenizer_cursor *pCursor, /* Cursor returned by simpleOpen */
149
const char **ppToken, /* OUT: *ppToken is the token text */
150
int *pnBytes, /* OUT: Number of bytes in token */
151
int *piStartOffset, /* OUT: Starting offset of token */
152
int *piEndOffset, /* OUT: Ending offset of token */
153
int *piPosition /* OUT: Position integer of token */
155
simple_tokenizer_cursor *c = (simple_tokenizer_cursor *) pCursor;
156
simple_tokenizer *t = (simple_tokenizer *) pCursor->pTokenizer;
157
unsigned char *p = (unsigned char *)c->pInput;
159
while( c->iOffset<c->nBytes ){
162
/* Scan past delimiter characters */
163
while( c->iOffset<c->nBytes && isDelim(t, p[c->iOffset]) ){
167
/* Count non-delimiter characters. */
168
iStartOffset = c->iOffset;
169
while( c->iOffset<c->nBytes && !isDelim(t, p[c->iOffset]) ){
173
if( c->iOffset>iStartOffset ){
174
int i, n = c->iOffset-iStartOffset;
175
if( n>c->nTokenAllocated ){
176
c->nTokenAllocated = n+20;
177
c->pToken = realloc(c->pToken, c->nTokenAllocated);
178
if( c->pToken==NULL ) return SQLITE_NOMEM;
181
/* TODO(shess) This needs expansion to handle UTF-8
182
** case-insensitivity.
184
unsigned char ch = p[iStartOffset+i];
185
c->pToken[i] = ch<0x80 ? tolower(ch) : ch;
187
*ppToken = c->pToken;
189
*piStartOffset = iStartOffset;
190
*piEndOffset = c->iOffset;
191
*piPosition = c->iToken++;
200
** The set of routines that implement the simple tokenizer
202
static const sqlite3_tokenizer_module simpleTokenizerModule = {
212
** Allocate a new simple tokenizer. Return a pointer to the new
213
** tokenizer in *ppModule
215
void sqlite3Fts1SimpleTokenizerModule(
216
sqlite3_tokenizer_module const**ppModule
218
*ppModule = &simpleTokenizerModule;
221
#endif /* !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS1) */