/* {{{ php_robbe_globals_construct */
static void php_friso_globals_construct(zend_friso_globals *friso_globals)
{
friso_globals->friso = friso_new();
friso_globals->config = friso_new_config();
friso_init_from_ifile(friso_globals->friso,
friso_globals->config, INI_STR("friso.ini_file"));
}
int init_friso(fstring path) {
friso = friso_new();
config = friso_new_config();
if ( friso_init_from_ifile(friso, config, path) != 1 ) {
printf("fail to initialize friso and config.");
return -1;
}
//set the task.
task = friso_new_task();
return 0;
}
int main(int argc, char **argv)
{
lex_entry_t e;
int lex = __LEX_CJK_WORDS__;
char _line[__LENGTH__];
clock_t s_time, e_time;
friso_t friso;
s_time = clock();
friso = friso_new();
friso->dic = friso_dic_new();
//__CJK_WORDS__
friso_dic_load( friso, __LEX_CJK_WORDS__, "../dict/lex-main.lex", __LENGTH__ );
friso_dic_load( friso, __LEX_CJK_WORDS__, "../dict/lex-admin.lex", __LENGTH__ );
friso_dic_load( friso, __LEX_CJK_WORDS__, "../dict/lex-chars.lex", __LENGTH__ );
friso_dic_load( friso, __LEX_CJK_WORDS__, "../dict/lex-cn-mz.lex", __LENGTH__ );
friso_dic_load( friso, __LEX_CJK_WORDS__, "../dict/lex-cn-place.lex", __LENGTH__ );
friso_dic_load( friso, __LEX_CJK_WORDS__, "../dict/lex-company.lex", __LENGTH__ );
friso_dic_load( friso, __LEX_CJK_WORDS__, "../dict/lex-festival.lex", __LENGTH__ );
friso_dic_load( friso, __LEX_CJK_WORDS__, "../dict/lex-flname.lex", __LENGTH__ );
friso_dic_load( friso, __LEX_CJK_WORDS__, "../dict/lex-food.lex", __LENGTH__ );
friso_dic_load( friso, __LEX_CJK_WORDS__, "../dict/lex-lang.lex", __LENGTH__ );
friso_dic_load( friso, __LEX_CJK_WORDS__, "../dict/lex-nation.lex", __LENGTH__ );
friso_dic_load( friso, __LEX_CJK_WORDS__, "../dict/lex-net.lex", __LENGTH__ );
friso_dic_load( friso, __LEX_CJK_WORDS__, "../dict/lex-org.lex", __LENGTH__ );
//__CJK_UNITS__
friso_dic_load( friso, __LEX_CJK_UNITS__, "../dict/lex-units.lex", __LENGTH__ );
//__MIX_WORDS__
friso_dic_load( friso, __LEX_ECM_WORDS__, "../dict/lex-mixed.lex", __LENGTH__ );
//__CN_LNAME__
friso_dic_load( friso, __LEX_CN_LNAME__, "../dict/lex-lname.lex", __LENGTH__ );
//__CN_SNAME__
friso_dic_load( friso, __LEX_CN_SNAME__, "../dict/lex-sname.lex", __LENGTH__ );
//__CN_DNAME1__
friso_dic_load( friso, __LEX_CN_DNAME1__, "../dict/lex-dname-1.lex", __LENGTH__ );
//__CN_DNAME2__
friso_dic_load( friso, __LEX_CN_DNAME2__, "../dict/lex-dname-2.lex", __LENGTH__ );
//__CN_LNA__
friso_dic_load( friso, __LEX_CN_LNA__, "../dict/lex-lna.lex", __LENGTH__ );
e_time = clock();
printf("Done, cost: %f sec, size=%d\n", ( double ) ( e_time - s_time ) / CLOCKS_PER_SEC, \
friso_all_dic_size( friso->dic ) );
while ( 1 ) {
printf("friso-%d>> ", lex);
scanf("%s", _line);
if ( strcmp( _line, "quit" ) == 0 ) {
break;
} else if ( strcmp( _line, "help" ) == 0 ) {
___PRINT_HELP_INFO___
} else if ( strcmp( _line, "#set" ) == 0 ) {
printf("lex_t>> ");
scanf("%d", &lex);
} else {
s_time = clock();
e = friso_dic_get( friso->dic, lex, _line );
e_time = clock();
if ( e != NULL ) {
printf("word=%s, syn=%s, fre=%d, cost:%fsec\n",
e->word, e->syn==NULL? "NULL" : (char *)e->syn->items[0], e->fre,
(double) ( e_time - s_time ) / CLOCKS_PER_SEC );
} else {
printf("%s was not found.\n", _line);
}
}
}
//friso_dic_free( friso->dic );
friso_free(friso);
return 0;
}
//creat a new friso with initialize item from a configuration file.
__EXTERN_API__ friso_t friso_new_from_ifile( string __ifile ) {
FILE *__stream;
char __chars__[256], __key__[128], *__line__, __lexi__[128];
uint_t i, t, __hit__ = 0, __length__;
friso_t e = friso_new();
if ( ( __stream = fopen( __ifile, "rb" ) ) != NULL ) {
//initialize the entry with the value from the ifile.
while ( ( __line__ = file_get_line( __chars__, __stream ) ) != NULL ) {
//comments filter.
if ( __line__[0] == '#' ) continue;
if ( __line__[0] == '\t' ) continue;
if ( __line__[0] == ' ' || __line__[0] == '\0' ) continue;
__length__ = strlen( __line__ );
for ( i = 0; i < __length__; i++ ) {
if ( __line__[i] == ' ' || __line__[i] == '\t' || __line__[i] == '=' ) break;
__key__[i] = __line__[i];
}
__key__[i] = '\0';
//position the euqals char '='.
if ( __line__[i] == ' ' || __line__[i] == '\t' ) {
for ( i++ ; i < __length__; i++ )
if ( __line__[i] == '=' ) break;
}
//clear the left whitespace of the value.
for ( i++; i < __length__
&& ( __line__[i] == ' ' || __line__[i] == '\t' ); i++ );
for ( t = 0; i < __length__; i++, t++ ) {
if ( __line__[i] == ' ' || __line__[i] == '\t' ) break;
__line__[t] = __line__[i];
}
__line__[t] = '\0';
//printf("key=%s, value=%s\n", __key__, __line__ );
if ( strcmp( __key__, "friso.lex_dir" ) == 0 ) {
/*
* here copy the value of the lex_dir.
* cause we need the value of friso.max_len to finish all
* the work when we call function friso_dic_load_from_ifile to
* initiliaze the friso dictionary.
*/
if ( __hit__ == 0 ) {
__hit__ = t;
for ( t = 0; t < __hit__; t++ ) {
__lexi__[t] = __line__[t];
}
__lexi__[t] = '\0';
}
} else if ( strcmp( __key__, "friso.max_len" ) == 0 ) {
e->max_len = atoi( __line__ );
} else if ( strcmp( __key__, "friso.r_name" ) == 0 ) {
e->r_name = atoi( __line__ );
} else if ( strcmp( __key__, "friso.mix_len" ) == 0 ) {
e->mix_len = atoi( __line__ );
} else if ( strcmp( __key__, "friso.lna_len" ) == 0 ) {
e->lna_len = atoi( __line__ );
} else if ( strcmp( __key__, "friso.nthreshold" ) == 0 ) {
e->nthreshold = atoi( __line__ );
} else if ( strcmp( __key__, "friso.mode" ) == 0 ) {
e->mode = ( friso_mode_t ) atoi( __line__ );
}
}
/*
* intialize the friso dictionary here.
* use the setting from the ifile parse above.
* we copied the value in the __lexi__.
*/
if ( __hit__ != 0 ) {
e->dic = friso_dic_new();
friso_dic_load_from_ifile( e->dic, __lexi__, e->max_len * 3 );
}
fclose( __stream );
}
return e;
}
int main(int argc, char **argv)
{
clock_t s_time, e_time;
char line[__INPUT_LENGTH__] = {0};
int i;
fstring __path__ = NULL, mode = NULL;
friso_t friso;
friso_config_t config;
friso_task_t task;
//get the lexicon directory
for ( i = 0; i < argc; i++ ) {
if ( strcasecmp( "-init", argv[i] ) == 0 ) {
__path__ = argv[i+1];
}
}
if ( __path__ == NULL ) {
println("Usage: friso -init lexicon path");
exit(0);
}
s_time = clock();
//initialize
friso = friso_new();
config = friso_new_config();
/*friso_dic_t dic = friso_dic_new();
friso_dic_load_from_ifile( dic, __path__, __LENGTH__ );
friso_set_dic( friso, dic );
friso_set_mode( friso, __FRISO_COMPLEX_MODE__ );*/
if ( friso_init_from_ifile(friso, config, __path__) != 1 ) {
printf("fail to initialize friso and config.");
goto err;
}
switch ( config->mode )
{
case __FRISO_SIMPLE_MODE__:
mode = "Simple";
break;
case __FRISO_COMPLEX_MODE__:
mode = "Complex";
break;
case __FRISO_DETECT_MODE__:
mode = "Detect";
break;
}
//friso_set_mode( config, __FRISO_DETECT_MODE__ );
//printf("clr_stw=%d\n", friso->clr_stw);
//printf("match c++?%d\n", friso_dic_match( friso->dic, __LEX_ENPUN_WORDS__, "c++" ));
//printf("match(研究)?%d\n", friso_dic_match( friso->dic, __LEX_CJK_WORDS__, "研究"));
e_time = clock();
printf("Initialized in %fsec\n", (double) ( e_time - s_time ) / CLOCKS_PER_SEC );
printf("Mode: %s\n", mode);
printf("+-Version: %s (%s)\n", friso_version(), friso->charset == FRISO_UTF8 ? "UTF-8" : "GBK" );
___ABOUT___;
//set the task.
task = friso_new_task();
while ( 1 )
{
print("friso>> ");
getLine( stdin, line );
//exit the programe
if ( strcasecmp( line, "quit" ) == 0 ) {
___EXIT_INFO___
}
//for ( i = 0; i < 1000000; i++ ) {
//set the task text.
friso_set_text( task, line );
println("分词结果:");
s_time = clock();
while ( ( config->next_token( friso, config, task ) ) != NULL )
{
//printf("%s[%d, %d, %d] ", task->token->word,
// task->token->offset, task->token->length, task->token->rlen );
printf("%s ", task->token->word );
}
//}
e_time = clock();
printf("\nDone, cost < %fsec\n", ( (double)(e_time - s_time) ) / CLOCKS_PER_SEC );
}
friso_free_task( task );
//error block.
err:
friso_free_config(config);
friso_free(friso);
return 0;
}