static mrb_value
mrb_grn_table_array_reference(mrb_state *mrb, mrb_value self)
{
grn_ctx *ctx = (grn_ctx *)mrb->ud;
grn_obj *table;
grn_id key_domain_id;
mrb_value mrb_key;
grn_id record_id;
grn_mrb_value_to_raw_data_buffer buffer;
void *key;
unsigned int key_size;
mrb_get_args(mrb, "o", &mrb_key);
table = DATA_PTR(self);
if (table->header.type == GRN_DB) {
key_domain_id = GRN_DB_SHORT_TEXT;
} else {
key_domain_id = table->header.domain;
}
grn_mrb_value_to_raw_data_buffer_init(mrb, &buffer);
grn_mrb_value_to_raw_data(mrb, "key", mrb_key, key_domain_id,
&buffer, &key, &key_size);
record_id = grn_table_get(ctx, table, key, key_size);
grn_mrb_value_to_raw_data_buffer_fin(mrb, &buffer);
if (record_id == GRN_ID_NIL) {
return mrb_nil_value();
} else {
return mrb_fixnum_value(record_id);
}
}
int
table_get(void)
{
int i;
grn_obj buf;
grn_obj *table = grn_ctx_get(&ctx, "<t1>", 4);
if (!table) { return -1; }
GRN_TEXT_INIT(&buf, 0);
for (i = 0; i < nloops; i++) {
int key = GENKEY(i);
GRN_BULK_REWIND(&buf);
grn_text_itoh(&ctx, &buf, key, key_size);
{
grn_id rid = grn_table_get(&ctx, table, GRN_BULK_HEAD(&buf), key_size);
if (!rid) {
fprintf(stderr, "table_lookup failed");
} else {
grn_obj obj, *p;
GRN_TEXT_INIT(&obj, 0);
p = grn_obj_get_value(&ctx, table, rid, &obj);
if (!p) {
fprintf(stderr, "grn_obj_get_value failed\n");
} else {
if (memcmp(GRN_BULK_HEAD(p), GRN_BULK_HEAD(&buf), value_size)) {
fprintf(stderr, "value unmatch\n");
}
grn_obj_close(&ctx, p);
}
}
}
}
grn_obj_close(&ctx, &buf);
return 0;
}
static GList *
retrieve_record_ids(const gchar *term)
{
grn_id term_id;
grn_ii_cursor *cursor;
grn_ii_posting *posting;
term_id = grn_table_get(context, lexicon, term, strlen(term));
if (term_id == GRN_ID_NIL)
return NULL;
record_ids_free();
cursor = grn_ii_cursor_open(context, inverted_index, term_id,
GRN_ID_NIL, GRN_ID_MAX, 5, 0);
if (!cursor)
return NULL;
while ((posting = grn_ii_cursor_next(context, cursor))) {
record_ids = g_list_append(record_ids, g_strdup_printf("%d", posting->rid));
}
grn_ii_cursor_close(context, cursor);
return record_ids;
}
void
test_read_write(gconstpointer *data)
{
gint i;
int added;
grn_ctx *context;
grn_obj *table;
const gchar *path;
const gchar *value_string;
gint process_number = 0;
const gchar *process_number_string;
const gchar table_name[] = "performance-read-write";
grn_obj value;
grn_obj *retrieved_value;
grn_id id;
grn_rc rc;
i = GPOINTER_TO_INT(data);
process_number_string = g_getenv(GRN_TEST_ENV_PROCESS_NUMBER);
if (process_number_string)
process_number = atoi(process_number_string);
rc = grn_ctx_init(&contexts[i], GRN_CTX_USE_QL);
grn_test_assert(rc, cut_set_message("context: %d (%d)", i, process_number));
context = &contexts[i];
path = g_getenv(GRN_TEST_ENV_TABLE_PATH);
cut_assert_not_null(path);
tables[i] = grn_table_open(context, table_name, strlen(table_name),
path);
cut_assert_not_null(tables[i],
cut_message("table: %d (%d)", i, process_number));
table = tables[i];
grn_test_assert_nil(grn_table_get(context, table, &i, sizeof(grn_id)),
cut_message("lookup - fail: (%d:%d)", i, process_number));
value_string = cut_take_printf("value: (%d:%d)", i, process_number);
id = grn_table_add(context, table, &i, sizeof(grn_id), &added);
grn_test_assert_not_nil(id);
cut_assert_equal_int(1, added);
GRN_TEXT_INIT(&value, GRN_OBJ_DO_SHALLOW_COPY);
GRN_TEXT_SET_REF(&value, value_string, strlen(value_string));
grn_obj_set_value(context, table, id, &value, GRN_OBJ_SET);
retrieved_value = grn_obj_get_value(context, table, id, NULL);
grn_test_assert_not_nil(
id,
cut_message("lookup - success: (%d:%d)", i, process_number));
GRN_TEXT_PUTC(context, retrieved_value, '\0');
cut_assert_equal_string(value_string, GRN_BULK_HEAD(retrieved_value));
tables[i] = NULL;
grn_test_assert(grn_obj_close(context, table));
// contexts[i] = NULL;
grn_test_assert(grn_ctx_fin(context));
}
// <cmd> {<len> <column id>} <types> <#kvs> [{<len> <key>} {<len> <value>}]...
void put_or_set(server_t *s, grn_ctx *ctx, int set)
{
uint32_t len;
HOG_RECV(s, &len, sizeof(len), return);
len = ntohl(len);
char *buf = hog_alloc(NULL, len);
HOG_RECV(s, buf, len, goto cleanup);
grn_obj *col, *table;
col = grn_ctx_get(ctx, buf, len);
if(grn_obj_is_table(ctx, col)) table = col;
else table = grn_column_table(ctx, col);
// get key and value types
char types[2];
HOG_RECV(s, types, 2, goto cleanup);
// receive keys and values
uint32_t nkeys;
HOG_RECV(s, &nkeys, sizeof(nkeys), goto cleanup);
nkeys = ntohl(nkeys);
grn_obj value;
GRN_OBJ_INIT(&value, GRN_BULK, 0, types[1]);
for(uint32_t i = 0; i < nkeys; ++i){
HOG_RECV(s, &len, sizeof(len), goto value_fin);
len = ntohl(len);
buf = hog_alloc(buf, len);
HOG_RECV(s, buf, len, goto value_fin);
ntoh_buf(buf, len, types[0]);
grn_id id;
if(set){
id = grn_table_get(ctx, table, buf, len);
}else{
id = grn_table_add(ctx, table, buf, len, NULL);
}
HOG_RECV(s, &len, sizeof(len), goto value_fin);
len = ntohl(len);
buf = hog_alloc(buf, len);
HOG_RECV(s, buf, len, goto value_fin);
if(id == GRN_ID_NIL) continue;
ntoh_buf(buf, len, types[1]);
GRN_BULK_REWIND(&value);
grn_bulk_write(ctx, &value, buf, len);
grn_obj_set_value(ctx, col, id, &value, GRN_OBJ_SET);
}
submit_one(s->socket);
value_fin:
GRN_OBJ_FIN(ctx, &value);
cleanup:
free(buf);
}
/*
* テーブルに主キーが _key_ のレコードがあるならtrueを返す。
*
* @overload has_key?(key)
*/
static VALUE
rb_grn_table_key_support_has_key (VALUE self, VALUE rb_key)
{
grn_ctx *context;
grn_obj *table, *key, *domain;
grn_id id, domain_id;
rb_grn_table_key_support_deconstruct(SELF(self), &table, &context,
&key, &domain_id, &domain,
NULL, NULL, NULL,
NULL);
GRN_BULK_REWIND(key);
RVAL2GRNKEY(rb_key, context, key, domain_id, domain, self);
id = grn_table_get(context, table, GRN_BULK_HEAD(key), GRN_BULK_VSIZE(key));
return id == GRN_ID_NIL ? Qfalse : Qtrue;
}
int
column_get(void)
{
int i, s = 0;
grn_obj buf;
grn_obj *table = grn_ctx_get(&ctx, "<t1>", 4);
grn_obj *column = grn_ctx_get(&ctx, "<t1>.c1", 7);
if (!table || !column) { return -1; }
GRN_TEXT_INIT(&buf, 0);
for (i = 0; i < nloops; i++) {
int key = GENKEY(i);
GRN_BULK_REWIND(&buf);
grn_text_itoh(&ctx, &buf, key, key_size);
{
grn_id rid = grn_table_get(&ctx, table, GRN_BULK_HEAD(&buf), key_size);
if (!rid) {
fprintf(stderr, "table_lookup failed\n");
} else {
grn_obj obj, *p;
unsigned int v = key % value_size;
GRN_TEXT_INIT(&obj, 0);
p = grn_obj_get_value(&ctx, column, rid, &obj);
if (!p) {
fprintf(stderr, "grn_obj_get_value failed\n");
} else {
if (GRN_BULK_VSIZE(p) != v) {
fprintf(stderr, "value_size unmatch %d (%ld:%u)\n", i, GRN_BULK_VSIZE(p), v);
} else {
if (v && GRN_BULK_HEAD(p)[v-1] != GRN_BULK_HEAD(&buf)[0]) {
fprintf(stderr, "value unmatch\n");
} else {
s++;
}
}
grn_obj_close(&ctx, p);
}
}
}
}
grn_obj_close(&ctx, &buf);
if (i != s) { printf("successed: %d\n", s); }
return 0;
}
static grn_obj *
yangram_next(grn_ctx *ctx, GNUC_UNUSED int nargs, GNUC_UNUSED grn_obj **args,
grn_user_data *user_data)
{
grn_yangram_tokenizer *tokenizer = user_data->ptr;
const unsigned char *string_end = tokenizer->end;
const unsigned char *token_top = tokenizer->next;
const unsigned char *token_next = token_top;
const unsigned char *token_tail = token_top;
int token_size = 0;
grn_bool is_token_grouped = GRN_FALSE;
const unsigned char *token_ctypes = NULL;
unsigned int ctypes_skip_size;
int char_length = 0;
grn_tokenizer_status status = 0;
grn_bool is_token_hit = GRN_FALSE;
grn_obj *lexicon = args[0];
if (tokenizer->phrase_table) {
if (tokenizer->nhits > 0 &&
token_top - (const unsigned char *)tokenizer->scan_start >
tokenizer->hits[tokenizer->current_hit].offset) {
tokenizer->current_hit++;
}
if (tokenizer->current_hit >= tokenizer->nhits) {
tokenizer->scan_start = tokenizer->scan_rest;
unsigned int scan_rest_length = tokenizer->end - (const unsigned char *)tokenizer->scan_rest;
if (scan_rest_length > 0) {
tokenizer->nhits = grn_pat_scan(ctx, (grn_pat *)tokenizer->phrase_table,
tokenizer->scan_rest,
scan_rest_length,
tokenizer->hits, MAX_N_HITS, &(tokenizer->scan_rest));
tokenizer->current_hit = 0;
}
}
if (tokenizer->nhits > 0 &&
tokenizer->current_hit < tokenizer->nhits &&
token_top - (const unsigned char *)tokenizer->scan_start ==
tokenizer->hits[tokenizer->current_hit].offset) {
is_token_hit = GRN_TRUE;
}
}
if (tokenizer->ctypes) {
token_ctypes = tokenizer->ctypes + tokenizer->ctypes_next;
} else {
token_ctypes = NULL;
}
if (is_token_hit) {
token_size = forward_scan_hit_token_tail(ctx, tokenizer, &token_tail,
tokenizer->hits[tokenizer->current_hit].length);
token_next = token_tail;
tokenizer->current_hit++;
} else {
is_token_grouped = is_token_group(tokenizer, token_ctypes);
if (is_token_grouped) {
token_size = forward_grouped_token_tail(ctx, tokenizer, token_ctypes, &token_tail);
token_next = token_tail;
} else {
token_size = forward_ngram_token_tail(ctx, tokenizer, token_ctypes, &token_tail);
char_length = grn_plugin_charlen(ctx, (char *)token_next,
tokenizer->rest_length,
tokenizer->query->encoding);
token_next += char_length;
}
}
if (token_top == token_tail || token_next == string_end) {
ctypes_skip_size = 0;
} else {
if (is_token_grouped || is_token_hit) {
ctypes_skip_size = token_size;
} else {
ctypes_skip_size = 1;
}
}
if (tokenizer->use_vgram > 0 && !is_token_grouped) {
grn_bool maybe_vgram = GRN_FALSE;
grn_id id;
id = grn_table_get(ctx, tokenizer->vgram_table,
(const char *)token_top, token_tail - token_top);
if (id) {
maybe_vgram = GRN_TRUE;
}
if (tokenizer->use_vgram >= VGRAM_BOTH && !maybe_vgram) {
if (token_tail < string_end &&
!is_group_border(ctx, tokenizer, token_tail, token_ctypes, token_size)) {
grn_id id;
const unsigned char *token_next_tail;
char_length = grn_plugin_charlen(ctx, (char *)token_tail,
tokenizer->rest_length,
tokenizer->query->encoding);
token_next_tail = token_tail + char_length;
id = grn_table_get(ctx, tokenizer->vgram_table,
(const char *)token_next, token_next_tail - token_next);
if (id) {
maybe_vgram = GRN_TRUE;
}
} else if (token_tail == string_end &&
tokenizer->query->tokenize_mode == GRN_TOKENIZE_GET) {
maybe_vgram = GRN_TRUE;
}
}
if (maybe_vgram) {
if (token_tail < string_end &&
!is_group_border(ctx, tokenizer, token_tail, token_ctypes, token_size)) {
char_length = grn_plugin_charlen(ctx, (char *)token_tail,
tokenizer->rest_length,
tokenizer->query->encoding);
token_size++;
token_tail += char_length;
if (tokenizer->use_vgram == VGRAM_QUAD) {
if (token_tail < string_end &&
!is_group_border(ctx, tokenizer, token_tail, token_ctypes, token_size)) {
id = grn_table_get(ctx, tokenizer->vgram_table,
(const char *)token_top, token_tail - token_top);
if (id) {
char_length = grn_plugin_charlen(ctx, (char *)token_tail,
tokenizer->rest_length,
tokenizer->query->encoding);
token_size++;
token_tail += char_length;
}
} else {
if (tokenizer->query->tokenize_mode == GRN_TOKENIZE_GET) {
grn_id tid;
tid = grn_table_get(ctx, lexicon,
(const char *)token_top, token_tail - token_top);
if (tid == GRN_ID_NIL) {
int added;
grn_table_add(ctx, lexicon,
(const char *)token_top, token_tail - token_top, &added);
}
status |= GRN_TOKEN_FORCE_PREFIX;
}
}
}
} else {
if (tokenizer->query->tokenize_mode == GRN_TOKENIZE_GET) {
grn_id tid;
tid = grn_table_get(ctx, lexicon,
(const char *)token_top, token_tail - token_top);
if (tid == GRN_ID_NIL) {
int added;
grn_table_add(ctx, lexicon,
(const char *)token_top, token_tail - token_top, &added);
}
status |= GRN_TOKEN_FORCE_PREFIX;
}
}
}
}
if (token_top == token_tail || token_next == string_end) {
status |= GRN_TOKEN_LAST;
}
if (token_tail == string_end) {
status |= GRN_TOKEN_REACH_END;
}
if (!is_token_grouped && !is_token_hit && token_size < tokenizer->ngram_unit) {
status |= GRN_TOKEN_UNMATURED;
}
if (tokenizer->pushed_token_tail &&
token_top < tokenizer->pushed_token_tail) {
status |= GRN_TOKEN_OVERLAP;
if (tokenizer->skip_overlap &&
!grn_ii_overlap_token_skip_enable &&
!(status & GRN_TOKEN_REACH_END) &&
!(status & GRN_TOKEN_SKIP_WITH_POSITION) &&
tokenizer->query->tokenize_mode == GRN_TOKENIZE_GET) {
if (token_tail <= tokenizer->pushed_token_tail) {
status |= GRN_TOKEN_SKIP;
} else {
if (!is_group_border(ctx, tokenizer, token_tail, token_ctypes, token_size)) {
status |= GRN_TOKEN_SKIP;
}
}
}
}
if (!(status & GRN_TOKEN_SKIP) &&
!(status & GRN_TOKEN_SKIP_WITH_POSITION)) {
tokenizer->pushed_token_tail = token_tail;
}
tokenizer->next = token_next;
tokenizer->rest_length = string_end - token_next;
tokenizer->ctypes_next = tokenizer->ctypes_next + ctypes_skip_size;
grn_tokenizer_token_push(ctx,
&(tokenizer->token),
(const char *)token_top,
token_tail - token_top,
status);
return NULL;
}
