static grn_obj *
command_echo(grn_ctx *ctx, GNUC_UNUSED int nargs, GNUC_UNUSED grn_obj **args,
grn_user_data *user_data)
{
grn_obj *var;
char *input = NULL;
unsigned int input_len = 0;
var = grn_plugin_proc_get_var(ctx, user_data, "input", -1);
GRN_PLUGIN_LOG(ctx, GRN_LOG_NOTICE, "[echo] input = %.*s",
(int)GRN_TEXT_LEN(var), GRN_TEXT_VALUE(var));
if(GRN_TEXT_LEN(var) != 0) {
input = GRN_TEXT_VALUE(var);
input_len = GRN_TEXT_LEN(var);
}
grn_ctx_output_array_open(ctx, "RESULT", 2);
grn_ctx_output_cstr(ctx, input);
grn_ctx_output_int64(ctx, input_len);
grn_ctx_output_array_close(ctx);
return NULL;
}
static void
parse_synonyms_file_line(grn_ctx *ctx, const char *line, size_t line_length,
grn_obj *key, grn_obj *value)
{
size_t i = 0;
if (is_comment_mark(line[i])) {
return;
}
while (i < line_length) {
char character = line[i];
i++;
if (character == '\t') {
break;
}
GRN_TEXT_PUTC(ctx, key, character);
}
if (i == line_length) {
return;
}
GRN_TEXT_PUTS(ctx, value, "((");
while (i < line_length) {
char character = line[i];
i++;
if (character == '\t') {
GRN_TEXT_PUTS(ctx, value, ") OR (");
} else {
GRN_TEXT_PUTC(ctx, value, character);
}
}
GRN_TEXT_PUTS(ctx, value, "))");
{
grn_id id;
void *value_location = NULL;
id = grn_hash_add(ctx, synonyms, GRN_TEXT_VALUE(key), GRN_TEXT_LEN(key),
&value_location, NULL);
if (id == GRN_ID_NIL) {
GRN_PLUGIN_LOG(ctx, GRN_LOG_WARNING,
"[plugin][query-expander][tsv] "
"failed to register key: <%.*s>",
(int)GRN_TEXT_LEN(key), GRN_TEXT_VALUE(key));
return;
}
if (GRN_TEXT_LEN(value) <= MAX_SYNONYM_BYTES - 1) {
GRN_TEXT_PUTC(ctx, value, '\0');
} else {
grn_bulk_truncate(ctx, value, MAX_SYNONYM_BYTES - 1);
GRN_TEXT_PUTC(ctx, value, '\0');
}
grn_memcpy(value_location, GRN_TEXT_VALUE(value), GRN_TEXT_LEN(value));
}
}
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;
}
static grn_obj *
command_tag_synonym_add(grn_ctx *ctx, GNUC_UNUSED int nargs, GNUC_UNUSED grn_obj **args,
grn_user_data *user_data)
{
grn_obj *var, *proc, *table, *column;
unsigned int nhooks = 0;
char *table_name = NULL;
unsigned int table_len = 0;
char *column_name = NULL;
unsigned int column_len = 0;
var = grn_plugin_proc_get_var(ctx, user_data, "table", -1);
if (GRN_TEXT_LEN(var) != 0) {
table_name = GRN_TEXT_VALUE(var);
table_len = GRN_TEXT_LEN(var);
}
var = grn_plugin_proc_get_var(ctx, user_data, "column", -1);
if (GRN_TEXT_LEN(var) != 0) {
column_name = GRN_TEXT_VALUE(var);
column_len = GRN_TEXT_LEN(var);
}
table = grn_ctx_get(ctx, table_name, table_len);
column = grn_obj_column(ctx, table, column_name, column_len);
{
grn_obj *range;
grn_obj *col;
grn_id range_id;
range_id = grn_obj_get_range(ctx, column);
range = grn_ctx_at(ctx, range_id);
if (!range) {
GRN_PLUGIN_LOG(ctx, GRN_LOG_ERROR,
"[tag-synonym] "
"hooked column must be reference type");
return NULL;
}
col = grn_obj_column(ctx,
range,
SYNONYM_COLUMN_NAME,
SYNONYM_COLUMN_NAME_LEN);
if (!col) {
GRN_PLUGIN_LOG(ctx, GRN_LOG_ERROR,
"[tag-synonym] "
"couldn't open synonym column");
return NULL;
}
}
proc = grn_ctx_get(ctx, "tag_synonym", -1);
{
grn_obj data;
default_set_value_hook_data hook_data = { grn_obj_id(ctx, proc), 0 };
GRN_TEXT_INIT(&data, GRN_OBJ_DO_SHALLOW_COPY);
GRN_TEXT_SET_REF(&data, &hook_data, sizeof(hook_data));
grn_obj_add_hook(ctx, column, GRN_HOOK_SET, 0, proc, &data);
grn_obj_unlink(ctx, &data);
}
//grn_ctx_output_array_open(ctx, "RESULT", 1);
nhooks = grn_obj_get_nhooks(ctx, column, GRN_HOOK_SET);
grn_ctx_output_int32(ctx, nhooks);
//grn_ctx_output_array_close(ctx);
return NULL;
}
/*
This function is called for a full text search query or a document to be
indexed. This means that both short/long strings are given.
The return value of this function is ignored. When an error occurs in this
function, `ctx->rc' is overwritten with an error code (not GRN_SUCCESS).
*/
static grn_obj *
mecab_init(grn_ctx *ctx, int nargs, grn_obj **args, grn_user_data *user_data)
{
char *buf, *p;
const char *s;
grn_mecab_tokenizer *tokenizer;
unsigned int bufsize;
grn_tokenizer_query *query;
grn_obj *normalized_query;
const char *normalized_string;
unsigned int normalized_string_length;
query = grn_tokenizer_query_open(ctx, nargs, args);
if (!query) {
return NULL;
}
if (!sole_mecab) {
grn_plugin_mutex_lock(ctx, sole_mecab_mutex);
if (!sole_mecab) {
sole_mecab = mecab_new2("-Owakati");
if (!sole_mecab) {
GRN_PLUGIN_ERROR(ctx, GRN_TOKENIZER_ERROR,
"[tokenizer][mecab] "
"mecab_new2() failed on mecab_init(): %s",
mecab_strerror(NULL));
} else {
sole_mecab_encoding = get_mecab_encoding(sole_mecab);
}
}
grn_plugin_mutex_unlock(ctx, sole_mecab_mutex);
}
if (!sole_mecab) {
grn_tokenizer_query_close(ctx, query);
return NULL;
}
if (query->encoding != sole_mecab_encoding) {
grn_tokenizer_query_close(ctx, query);
GRN_PLUGIN_ERROR(ctx, GRN_TOKENIZER_ERROR,
"[tokenizer][mecab] "
"MeCab dictionary charset (%s) does not match "
"the table encoding: <%s>",
grn_enctostr(sole_mecab_encoding),
grn_enctostr(query->encoding));
return NULL;
}
if (!(tokenizer = GRN_PLUGIN_MALLOC(ctx, sizeof(grn_mecab_tokenizer)))) {
grn_tokenizer_query_close(ctx, query);
GRN_PLUGIN_ERROR(ctx, GRN_NO_MEMORY_AVAILABLE,
"[tokenizer][mecab] "
"memory allocation to grn_mecab_tokenizer failed");
return NULL;
}
tokenizer->mecab = sole_mecab;
tokenizer->query = query;
normalized_query = query->normalized_query;
grn_string_get_normalized(ctx,
normalized_query,
&normalized_string,
&normalized_string_length,
NULL);
tokenizer->have_tokenized_delimiter =
grn_tokenizer_have_tokenized_delimiter(ctx,
normalized_string,
normalized_string_length,
query->encoding);
if (tokenizer->have_tokenized_delimiter) {
tokenizer->buf = NULL;
tokenizer->next = normalized_string;
tokenizer->end = tokenizer->next + normalized_string_length;
} else {
grn_plugin_mutex_lock(ctx, sole_mecab_mutex);
s = mecab_sparse_tostr2(tokenizer->mecab,
normalized_string,
normalized_string_length);
if (!s) {
GRN_PLUGIN_ERROR(ctx, GRN_TOKENIZER_ERROR,
"[tokenizer][mecab] "
"mecab_sparse_tostr() failed len=%d err=%s",
normalized_string_length,
mecab_strerror(tokenizer->mecab));
} else {
bufsize = strlen(s) + 1;
if (!(buf = GRN_PLUGIN_MALLOC(ctx, bufsize))) {
GRN_PLUGIN_LOG(ctx, GRN_LOG_ALERT,
"[tokenizer][mecab] "
"buffer allocation on mecab_init failed !");
} else {
memcpy(buf, s, bufsize);
}
}
grn_plugin_mutex_unlock(ctx, sole_mecab_mutex);
if (!s || !buf) {
grn_tokenizer_query_close(ctx, tokenizer->query);
GRN_PLUGIN_FREE(ctx, tokenizer);
return NULL;
}
/* A certain version of mecab returns trailing lf or spaces. */
for (p = buf + bufsize - 2;
buf <= p && isspace(*(unsigned char *)p);
p--) { *p = '\0'; }
tokenizer->buf = buf;
tokenizer->next = buf;
tokenizer->end = p + 1;
}
user_data->ptr = tokenizer;
grn_tokenizer_token_init(ctx, &(tokenizer->token));
return NULL;
}
