diff options
Diffstat (limited to 'include/gram_index_engine.h')
| -rw-r--r-- | include/gram_index_engine.h | 68 |
1 files changed, 68 insertions, 0 deletions
diff --git a/include/gram_index_engine.h b/include/gram_index_engine.h new file mode 100644 index 0000000..a69e924 --- /dev/null +++ b/include/gram_index_engine.h @@ -0,0 +1,68 @@ +#ifndef _GRAM_INDEX_ENGINE_ +#define _GRAM_INDEX_ENGINE_ + +#ifdef __cplusplus +extern "C" { +#endif + +#define GIE_INSERT_OPT 0 +#define GIE_DELETE_OPT 1 +#define GIE_INPUT_FORMAT_SFH 1 +#define GIE_INPUT_FORMAT_PLAIN 0 + + +typedef struct +{ + /* data */ +}GIE_handle_t; + + +typedef struct +{ + unsigned int id; + unsigned int sfh_length;//size of fuzzy_hash + short operation;//GIE_INSERT_OPT or GIE_DELETE_OPT.if operation is GIE_DELETE_OPT, only id is needed; + short cfds_lvl; + char * sfh; + void * tag; +}GIE_digest_t; + + +typedef struct +{ + unsigned int id; + short cfds_lvl; + void * tag; +}GIE_result_t; + + +typedef struct +{ + unsigned int gram_value; + //unsigned int htable_num; + unsigned int position_accuracy; + short format; //if format==GIE_INPUT_FORMAT_SFH, means the input string is a GIE_INPUT_FORMAT_SFH string + //else id format==PALIN, means the input string is common string + short ED_reexamine;//if ED_reexamine==1, calculate edit distance to verify the final result +}GIE_create_para_t; + + +GIE_handle_t * GIE_create(const GIE_create_para_t * para); + + +int GIE_update(GIE_handle_t * handle, GIE_digest_t ** digests, int size); + + +//return actual matched result count +//return 0 when matched nothing; +//return -1 when error occurs; +int GIE_query(GIE_handle_t * handle, const char * data, int data_len, GIE_result_t * results, int result_size); + +void GIE_destory(GIE_handle_t * handle); +int GIE_string_similiarity(const char *str1, int len1, const char *str2, int len2); +int GIE_sfh_similiarity(const char *sfh1, int len1, const char *sfh2, int len2); + +#ifdef __cplusplus +} +#endif +#endif |
