/*
* Tokenize a string using the table as lexicon.
*
* @overload tokenize(string, options={})
* @param [String] string The string to be tokenized.
* @param [::Hash] options
* @option options [Bool] :add (true) Adds a new token to the table if true.
* Returned tokens include the new token. Otherwise, a new token is
* just ignored.
* @return [::Array<Groonga::Record>] Tokenized tokens.
*/
static VALUE
rb_grn_table_key_support_tokenize (int argc, VALUE *argv, VALUE self)
{
VALUE rb_string, rb_add_p;
VALUE rb_options;
VALUE rb_tokens = Qnil;
grn_ctx *context;
grn_obj *table;
char *string;
int string_size;
grn_bool add_p;
grn_obj tokens;
rb_scan_args(argc, argv, "11", &rb_string, &rb_options);
rb_grn_scan_options(rb_options,
"add", &rb_add_p,
NULL);
if (NIL_P(rb_add_p)) {
rb_add_p = Qtrue;
}
rb_grn_table_key_support_deconstruct(SELF(self), &table, &context,
NULL, NULL, NULL,
NULL, NULL, NULL,
NULL);
string = StringValueCStr(rb_string);
string_size = RSTRING_LEN(rb_string);
add_p = RVAL2CBOOL(rb_add_p);
GRN_RECORD_INIT(&tokens, GRN_OBJ_VECTOR, grn_obj_id(context, table));
grn_table_tokenize(context, table, string, string_size, &tokens, add_p);
if (context->rc == GRN_SUCCESS) {
rb_tokens = GRNUVECTOR2RVAL(context, &tokens, table, self);
}
grn_obj_unlink(context, &tokens);
rb_grn_context_check(context, self);
return rb_tokens;
}
static grn_obj *
command_tag_synonym(grn_ctx *ctx, GNUC_UNUSED int nargs, GNUC_UNUSED grn_obj **args,
GNUC_UNUSED grn_user_data *user_data)
{
GNUC_UNUSED grn_obj *flags = grn_ctx_pop(ctx);
grn_obj *newvalue = grn_ctx_pop(ctx);
grn_obj *oldvalue = grn_ctx_pop(ctx);
GNUC_UNUSED grn_obj *id = grn_ctx_pop(ctx);
grn_obj buf;
grn_obj record;
grn_obj *domain;
grn_obj *table;
grn_obj *column;
int i,n;
if (GRN_BULK_VSIZE(newvalue) == 0 || GRN_INT32_VALUE(flags) == 0) {
return NULL;
}
table = grn_ctx_at(ctx, oldvalue->header.domain);
if (table && !is_table(table)) {
GRN_PLUGIN_LOG(ctx, GRN_LOG_WARNING,
"[tag-synonym] "
"hooked column must be reference type");
return NULL;
}
column = grn_obj_column(ctx,
table,
SYNONYM_COLUMN_NAME,
SYNONYM_COLUMN_NAME_LEN);
if (!column) {
GRN_PLUGIN_LOG(ctx, GRN_LOG_WARNING,
"[tag-synonym] "
"couldn't open synonym column");
return NULL;
}
GRN_TEXT_INIT(&buf, 0);
domain = grn_ctx_at(ctx, newvalue->header.domain);
if (domain && is_string(domain)) {
GRN_RECORD_INIT(&record, GRN_OBJ_VECTOR, oldvalue->header.domain);
grn_table_tokenize(ctx, table, GRN_TEXT_VALUE(newvalue), GRN_TEXT_LEN(newvalue), &record, GRN_TRUE);
} else if (newvalue->header.type == GRN_UVECTOR) {
record = *newvalue;
}
if (is_string(domain) || newvalue->header.type == GRN_UVECTOR) {
grn_obj value;
GRN_RECORD_INIT(newvalue, GRN_OBJ_VECTOR, oldvalue->header.domain);
GRN_UINT32_INIT(&value, 0);
n = grn_vector_size(ctx, &record);
for (i = 0; i < n; i++) {
grn_id tid;
tid = grn_uvector_get_element(ctx, &record, i, NULL);
GRN_BULK_REWIND(&value);
grn_obj_get_value(ctx, column, tid, &value);
if (GRN_UINT32_VALUE(&value)) {
GRN_PLUGIN_LOG(ctx, GRN_LOG_INFO,
"[tag-synonym] "
"changed: tid %d -> %d", tid, GRN_UINT32_VALUE(&value));
tid = GRN_UINT32_VALUE(&value);
}
grn_uvector_add_element(ctx, newvalue, tid, 0);
}
grn_obj_unlink(ctx, &value);
} else {
grn_id tid;
grn_obj value;
tid = GRN_RECORD_VALUE(newvalue);
GRN_UINT32_INIT(&value, 0);
grn_obj_get_value(ctx, column, tid, &value);
if (GRN_UINT32_VALUE(&value)) {
GRN_PLUGIN_LOG(ctx, GRN_LOG_INFO,
"[tag-synonym] "
"changed: tid %d -> %d", tid, GRN_UINT32_VALUE(&value));
tid = GRN_UINT32_VALUE(&value);
GRN_BULK_REWIND(newvalue);
GRN_RECORD_SET(ctx, newvalue, tid);
}
grn_obj_unlink(ctx, &value);
}
grn_obj_unlink(ctx, &buf);
return NULL;
}
