void* EGGAPI eggSySRecorder_get_dict(char* infotype, char* analyName )
{
if(POINTER_IS_INVALID(g_sysRecorder_handle))
{
return EGG_NULL;
}
if(POINTER_IS_INVALID(infotype))
{
return EGG_NULL;
}
if(POINTER_IS_INVALID(analyName))
{
return EGG_NULL;
}
pthread_mutex_lock( &g_sysRecorder_handle->mutex);
HEGGTOPCOLLECTOR hTopCollector = eggTopCollector_new(0);
HEGGQUERY hQuery1 = eggQuery_new_string(EGG_SYS_TYPEINFO, infotype, strlen(infotype), "");
HEGGQUERY hQuery2 = eggQuery_new_string(EGG_SYS_ANALYNAME, analyName, strlen(analyName), "");
hQuery1 = eggQuery_and(hQuery1, hQuery2);
eggIndexSearcher_search_with_query(g_sysRecorder_handle->hIndexSearcher, hTopCollector, hQuery1);
eggQuery_delete(hQuery1);
pthread_mutex_unlock( &g_sysRecorder_handle->mutex);
return hTopCollector;
}
HEGGQUERY
add_query (HEGGQUERY eggQuery, char* field, char* key)
{
HEGGQUERY new_query = NULL;
if(!strcmp(field, "keywords"))
{
new_query = eggQuery_new_string("title", key,strlen(key),ANALYZER_CWSLEX);
HEGGQUERY new_query_two = NULL;
if(!strcmp(G->op, "search"))
{
new_query_two = eggQuery_new_string("content", key,strlen(key),ANALYZER_CWSLEX);
}
else
{
new_query_two = eggQuery_new_string("content_index", key,strlen(key),ANALYZER_CWSLEX);
}
new_query = eggQuery_or(new_query,new_query_two);
}
if(!strcmp(field, "category"))
{
new_query = eggQuery_new_string("category", key, strlen(key), NULL);
/*
if(!strcmp(G->op, "search"))
{
// char* key_with_space = addSpace(key,30);
new_query = eggQuery_new_string("category", key, strlen(key), NULL);
// free(key_with_space);
}
else
{
new_query = eggQuery_new_string("category", key, strlen(key), NULL);
}
*/
}
if(new_query != NULL)
{
if(eggQuery == NULL)
{
eggQuery = new_query;
}
else
{
eggQuery = eggQuery_and(eggQuery,new_query);
}
}
return eggQuery;
} /* ----- end of function get_queryadd_query ----- */
static HEGGQUERY nextQuery(char *fieldName, char *keybuf)
{
HEGGQUERY h1 = NULL;
if (keybuf[0] == '"')
{
char *key;
char *p = keybuf+1;
while (p[0])
{
if (p[0] == '"')
{
break;
}
else if (p[0] == '\\')
{
p++;
if (p[0] == '\0')
{
fprintf(stderr, "input string invalid \\\n");
exit(-1);
}
p++;
}
else
{
p++;
}
}
p[0] = '\0';
key = keybuf+1;
h1 = eggQuery_new_string(fieldName, key, p-key, NULL);
}
else if (keybuf[0] == 'i')
{
int i;
i = atoi(keybuf+1);
h1 = eggQuery_new_int32(fieldName, i);
}
else if (keybuf[0] == 'l')
{
int64_t l;
l = strtoll(keybuf+1, NULL, 0);
h1 = eggQuery_new_int64(fieldName, l);
}
else if (keybuf[0] == 'f')
{
double f;
f = strtod(keybuf+1, NULL);
h1 = eggQuery_new_double(fieldName, f);
}
else
{
fprintf(stderr, "please \"string\" or i32 or l64 or f3.3333 \n");
exit(-1);
}
return h1;
}
int main()
{
HEGGHANDLE hHandle = eggPath_open("file:///tmp/");
HEGGINDEXREADER hIndexReader = eggIndexReader_open(hHandle);
HEGGINDEXSEARCHER hIndexSearcher = eggIndexSearcher_new(hIndexReader);
HEGGQUERY hq = eggQuery_new_string("content", "is good", strlen("is good"), ANALYZER_CWSLEX);
if(hq != EGG_NULL)
{
printf("query init OK! \n");
}
//填0取所有结果,非0按填的值取个数
HEGGTOPCOLLECTOR hTopCollector = eggTopCollector_new(0);
eggTopCollector_set_sorttype(hTopCollector, EGG_TOPSORT_WEIGHT);
printf("eggTopCollector sortType is EGG_TOPSORT_WEIGHT \n");
EBOOL ret = eggIndexSearcher_search_with_query(hIndexSearcher, hTopCollector, hq);
if (ret == EGG_TRUE)
{
//对最后结果进行排序
//EGG_TOPSORT_WEIGHT: 按document的weight排序
//EGG_TOPSORT_SCORE： 按查询关键字的相关度排序（打分排序）
//EGG_TOPSORT_NOT： 不排序
HEGGSCOREDOC lp_score_doc = eggTopCollector_top_docs(hTopCollector);
count_t cnt = eggTopCollector_total_hits(hTopCollector);
index_t i = 0;
printf("have hit %u documents\n", cnt);
while (i != cnt)
{
HEGGDOCUMENT lp_eggDocument = EGG_NULL;
eggIndexReader_get_document(hIndexReader,
lp_score_doc[i].idDoc, &lp_eggDocument);
HEGGFIELD lp_field = eggDocument_get_field(lp_eggDocument, "content");
unsigned len = 0;
char *val = eggField_get_value(lp_field, &len);
printf("id : [%llu], content : [%s], weight : [%d]\n", EGGDID_DOCID(&lp_score_doc[i].idDoc), val, eggDocument_get_weight(lp_eggDocument));
lp_field = 0;
eggDocument_delete(lp_eggDocument);
i++;
}
}
eggTopCollector_delete(hTopCollector);
eggQuery_delete(hq);
eggIndexSearcher_delete(hIndexSearcher);
eggIndexReader_close(hIndexReader);
eggPath_close(hHandle);
return 0;
}
void query::add(const char *keywords, std::string key)
{
HEGGQUERY new_query;
new_query = eggQuery_new_string(keywords, key.c_str(), key.length(), ANALYZER_CWSLEX);
if(query_str == NULL)
{
query_str = new_query;
}
}
int main(int argc, char* argv[])
{
HEGGHANDLE hHandle = eggPath_open("file:///egg/");
HEGGINDEXREADER hIndexReader = eggIndexReader_open(hHandle);
HEGGINDEXSEARCHER hIndexSearcher = eggIndexSearcher_new(hIndexReader);
HEGGQUERY hq = eggQuery_new_string("content", argv[1], strlen(argv[1]), "");
if(hq != EGG_NULL)
{
printf("query init OK! \n");
}
HEGGTOPCOLLECTOR hTopCollector = eggTopCollector_new(0);
EBOOL ret = eggIndexSearcher_search_with_query(hIndexSearcher, hTopCollector, hq);
if (ret == EGG_TRUE)
{
HEGGSCOREDOC lp_score_doc = eggTopCollector_top_docs(hTopCollector);
count_t cnt = eggTopCollector_total_hits(hTopCollector);
index_t i = 0;
printf("have hit %u documents\n", cnt);
while (i != cnt)
{
HEGGDOCUMENT lp_eggDocument = EGG_NULL;
eggIndexReader_get_document(hIndexReader,
lp_score_doc[i].idDoc, &lp_eggDocument);
HEGGFIELD lp_field = eggDocument_get_field(lp_eggDocument, "content");
unsigned len = 0;
char *val = eggField_get_value(lp_field, &len);
printf("id : [%llu], content : [%s], \n", EGGDID_DOCID(&lp_score_doc[i].idDoc), val);
lp_field = 0;
eggDocument_delete(lp_eggDocument);
i++;
}
}
eggTopCollector_delete(hTopCollector);
eggQuery_delete(hq);
eggIndexSearcher_delete(hIndexSearcher);
eggIndexReader_close(hIndexReader);
eggPath_close(hHandle);
return 0;
}
int main()
{
//ImLexAnalyzer* p_la = (ImLexAnalyzer*)ImCnLexAnalyzer_new();
// ImLexAnalyzer* p_la = (ImLexAnalyzer*)ImCnLexAnalyzer_new();
HEGGDIRECTORY hDirectory = eggDirectory_open("/ape/ImRoBot5/index/bbstest");
HEGGINDEXREADER hIndexReader = eggIndexReader_open(hDirectory);
HEGGINDEXSEARCHER hIndexSearcher = eggIndexSearcher_new(hIndexReader);
HEGGQUERY h1, h2, h3;
h1 = eggQuery_new_string("title", "人", 3, ANALYZER_CWSLEX);
//h2 = eggQuery_new_string("content", "new", 3, p_la);
// h2 = eggQuery_new_string("body", "some description", 16, p_la);
// h3 = eggQuery_new_int32("price", 199);
// h2 = eggQuery_and(h3, h2);
//h1 = eggQuery_or(h2, h1);
// h3 = h2 = 0;
HEGGTOPCOLLECTOR hTopCollector = eggTopCollector_new(0);
int ret = eggIndexSearcher_search_with_query(hIndexSearcher, hTopCollector, h1);
if (ret == EGG_TRUE)
{
eggTopCollector_normalized(hTopCollector, EGG_TOPSORT_SCORE);
// eggTopCollector_normalized(hTopCollector, EGG_TOPSORT_NOT);
HEGGSCOREDOC lp_score_doc = eggTopCollector_top_docs(hTopCollector);
count_t cnt = eggTopCollector_total_hits(hTopCollector);
printf("have hit %u documents\n", cnt);
if (cnt > 0)
{
printf("last document: id[%llu]\n", lp_score_doc[cnt-1].idDoc);
HEGGDOCUMENT lp_eggDocument = EGG_NULL;
eggIndexReader_get_document(hIndexReader,
lp_score_doc[cnt-1].idDoc, &lp_eggDocument);
HEGGFIELD lp_field = eggDocument_get_field(lp_eggDocument,"title");
unsigned len = 0;
char *val = eggField_get_value(lp_field, &len);
printf("last document: body[%.*s]\n", len, val);
lp_field = 0;
eggDocument_delete(lp_eggDocument);
}
}
eggTopCollector_delete(hTopCollector);
eggQuery_delete(h1);
eggIndexSearcher_delete(hIndexSearcher);
eggIndexReader_close(hIndexReader);
eggDirectory_close(hDirectory);
ImLexAnalyzer_delete(p_la);
}
void* EGGAPI eggSySRecorder_get_records(char* infotype )
{
if(POINTER_IS_INVALID(g_sysRecorder_handle))
{
return EGG_NULL;
}
if(POINTER_IS_INVALID(infotype))
{
return EGG_NULL;
}
pthread_mutex_lock( &g_sysRecorder_handle->mutex);
HEGGTOPCOLLECTOR hTopCollector = eggTopCollector_new(0);
HEGGQUERY hQuery = eggQuery_new_string(EGG_SYS_TYPEINFO, infotype, strlen(infotype), "");
eggIndexSearcher_search_with_query(g_sysRecorder_handle->hIndexSearcher, hTopCollector, hQuery);
eggQuery_delete(hQuery);
pthread_mutex_unlock( &g_sysRecorder_handle->mutex);
}
