Scott Graham | d19529d | 2014-11-03 15:04:31 -0800 | [diff] [blame] | 1 | /* |
| 2 | ** 2006 July 10 |
| 3 | ** |
| 4 | ** The author disclaims copyright to this source code. |
| 5 | ** |
| 6 | ************************************************************************* |
| 7 | ** Defines the interface to tokenizers used by fulltext-search. There |
| 8 | ** are three basic components: |
| 9 | ** |
| 10 | ** sqlite3_tokenizer_module is a singleton defining the tokenizer |
| 11 | ** interface functions. This is essentially the class structure for |
| 12 | ** tokenizers. |
| 13 | ** |
| 14 | ** sqlite3_tokenizer is used to define a particular tokenizer, perhaps |
| 15 | ** including customization information defined at creation time. |
| 16 | ** |
| 17 | ** sqlite3_tokenizer_cursor is generated by a tokenizer to generate |
| 18 | ** tokens from a particular input. |
| 19 | */ |
| 20 | #ifndef _FTS3_TOKENIZER_H_ |
| 21 | #define _FTS3_TOKENIZER_H_ |
| 22 | |
| 23 | /* TODO(shess) Only used for SQLITE_OK and SQLITE_DONE at this time. |
| 24 | ** If tokenizers are to be allowed to call sqlite3_*() functions, then |
| 25 | ** we will need a way to register the API consistently. |
| 26 | */ |
| 27 | #include "sqlite3.h" |
| 28 | |
| 29 | /* |
| 30 | ** Structures used by the tokenizer interface. When a new tokenizer |
| 31 | ** implementation is registered, the caller provides a pointer to |
| 32 | ** an sqlite3_tokenizer_module containing pointers to the callback |
| 33 | ** functions that make up an implementation. |
| 34 | ** |
| 35 | ** When an fts3 table is created, it passes any arguments passed to |
| 36 | ** the tokenizer clause of the CREATE VIRTUAL TABLE statement to the |
| 37 | ** sqlite3_tokenizer_module.xCreate() function of the requested tokenizer |
| 38 | ** implementation. The xCreate() function in turn returns an |
| 39 | ** sqlite3_tokenizer structure representing the specific tokenizer to |
| 40 | ** be used for the fts3 table (customized by the tokenizer clause arguments). |
| 41 | ** |
| 42 | ** To tokenize an input buffer, the sqlite3_tokenizer_module.xOpen() |
| 43 | ** method is called. It returns an sqlite3_tokenizer_cursor object |
| 44 | ** that may be used to tokenize a specific input buffer based on |
| 45 | ** the tokenization rules supplied by a specific sqlite3_tokenizer |
| 46 | ** object. |
| 47 | */ |
| 48 | typedef struct sqlite3_tokenizer_module sqlite3_tokenizer_module; |
| 49 | typedef struct sqlite3_tokenizer sqlite3_tokenizer; |
| 50 | typedef struct sqlite3_tokenizer_cursor sqlite3_tokenizer_cursor; |
| 51 | |
| 52 | struct sqlite3_tokenizer_module { |
| 53 | |
| 54 | /* |
James Robinson | fa68617 | 2015-02-23 18:32:32 -0800 | [diff] [blame^] | 55 | ** Structure version. Should always be set to 0 or 1. |
Scott Graham | d19529d | 2014-11-03 15:04:31 -0800 | [diff] [blame] | 56 | */ |
| 57 | int iVersion; |
| 58 | |
| 59 | /* |
| 60 | ** Create a new tokenizer. The values in the argv[] array are the |
| 61 | ** arguments passed to the "tokenizer" clause of the CREATE VIRTUAL |
| 62 | ** TABLE statement that created the fts3 table. For example, if |
| 63 | ** the following SQL is executed: |
| 64 | ** |
| 65 | ** CREATE .. USING fts3( ... , tokenizer <tokenizer-name> arg1 arg2) |
| 66 | ** |
| 67 | ** then argc is set to 2, and the argv[] array contains pointers |
| 68 | ** to the strings "arg1" and "arg2". |
| 69 | ** |
| 70 | ** This method should return either SQLITE_OK (0), or an SQLite error |
| 71 | ** code. If SQLITE_OK is returned, then *ppTokenizer should be set |
| 72 | ** to point at the newly created tokenizer structure. The generic |
James Robinson | fa68617 | 2015-02-23 18:32:32 -0800 | [diff] [blame^] | 73 | ** sqlite3_tokenizer.pModule variable should not be initialized by |
Scott Graham | d19529d | 2014-11-03 15:04:31 -0800 | [diff] [blame] | 74 | ** this callback. The caller will do so. |
| 75 | */ |
| 76 | int (*xCreate)( |
| 77 | int argc, /* Size of argv array */ |
| 78 | const char *const*argv, /* Tokenizer argument strings */ |
| 79 | sqlite3_tokenizer **ppTokenizer /* OUT: Created tokenizer */ |
| 80 | ); |
| 81 | |
| 82 | /* |
| 83 | ** Destroy an existing tokenizer. The fts3 module calls this method |
| 84 | ** exactly once for each successful call to xCreate(). |
| 85 | */ |
| 86 | int (*xDestroy)(sqlite3_tokenizer *pTokenizer); |
| 87 | |
| 88 | /* |
| 89 | ** Create a tokenizer cursor to tokenize an input buffer. The caller |
| 90 | ** is responsible for ensuring that the input buffer remains valid |
| 91 | ** until the cursor is closed (using the xClose() method). |
| 92 | */ |
| 93 | int (*xOpen)( |
| 94 | sqlite3_tokenizer *pTokenizer, /* Tokenizer object */ |
| 95 | const char *pInput, int nBytes, /* Input buffer */ |
| 96 | sqlite3_tokenizer_cursor **ppCursor /* OUT: Created tokenizer cursor */ |
| 97 | ); |
| 98 | |
| 99 | /* |
| 100 | ** Destroy an existing tokenizer cursor. The fts3 module calls this |
| 101 | ** method exactly once for each successful call to xOpen(). |
| 102 | */ |
| 103 | int (*xClose)(sqlite3_tokenizer_cursor *pCursor); |
| 104 | |
| 105 | /* |
| 106 | ** Retrieve the next token from the tokenizer cursor pCursor. This |
| 107 | ** method should either return SQLITE_OK and set the values of the |
| 108 | ** "OUT" variables identified below, or SQLITE_DONE to indicate that |
| 109 | ** the end of the buffer has been reached, or an SQLite error code. |
| 110 | ** |
| 111 | ** *ppToken should be set to point at a buffer containing the |
| 112 | ** normalized version of the token (i.e. after any case-folding and/or |
| 113 | ** stemming has been performed). *pnBytes should be set to the length |
| 114 | ** of this buffer in bytes. The input text that generated the token is |
| 115 | ** identified by the byte offsets returned in *piStartOffset and |
| 116 | ** *piEndOffset. *piStartOffset should be set to the index of the first |
| 117 | ** byte of the token in the input buffer. *piEndOffset should be set |
| 118 | ** to the index of the first byte just past the end of the token in |
| 119 | ** the input buffer. |
| 120 | ** |
| 121 | ** The buffer *ppToken is set to point at is managed by the tokenizer |
| 122 | ** implementation. It is only required to be valid until the next call |
| 123 | ** to xNext() or xClose(). |
| 124 | */ |
| 125 | /* TODO(shess) current implementation requires pInput to be |
| 126 | ** nul-terminated. This should either be fixed, or pInput/nBytes |
| 127 | ** should be converted to zInput. |
| 128 | */ |
| 129 | int (*xNext)( |
| 130 | sqlite3_tokenizer_cursor *pCursor, /* Tokenizer cursor */ |
| 131 | const char **ppToken, int *pnBytes, /* OUT: Normalized text for token */ |
| 132 | int *piStartOffset, /* OUT: Byte offset of token in input buffer */ |
| 133 | int *piEndOffset, /* OUT: Byte offset of end of token in input buffer */ |
| 134 | int *piPosition /* OUT: Number of tokens returned before this one */ |
| 135 | ); |
James Robinson | fa68617 | 2015-02-23 18:32:32 -0800 | [diff] [blame^] | 136 | |
| 137 | /*********************************************************************** |
| 138 | ** Methods below this point are only available if iVersion>=1. |
| 139 | */ |
| 140 | |
| 141 | /* |
| 142 | ** Configure the language id of a tokenizer cursor. |
| 143 | */ |
| 144 | int (*xLanguageid)(sqlite3_tokenizer_cursor *pCsr, int iLangid); |
Scott Graham | d19529d | 2014-11-03 15:04:31 -0800 | [diff] [blame] | 145 | }; |
| 146 | |
| 147 | struct sqlite3_tokenizer { |
| 148 | const sqlite3_tokenizer_module *pModule; /* The module for this tokenizer */ |
| 149 | /* Tokenizer implementations will typically add additional fields */ |
| 150 | }; |
| 151 | |
| 152 | struct sqlite3_tokenizer_cursor { |
| 153 | sqlite3_tokenizer *pTokenizer; /* Tokenizer for this cursor. */ |
| 154 | /* Tokenizer implementations will typically add additional fields */ |
| 155 | }; |
| 156 | |
| 157 | int fts3_global_term_cnt(int iTerm, int iCol); |
| 158 | int fts3_term_cnt(int iTerm, int iCol); |
| 159 | |
| 160 | |
| 161 | #endif /* _FTS3_TOKENIZER_H_ */ |