void
test_read_write(gconstpointer *data)
{
gint i;
const gchar *key;
grn_ctx *context;
grn_pat *trie;
const gchar *path;
const gchar *value_string;
gint process_number = 0;
const gchar *process_number_string;
void *value;
int added;
grn_id id = GRN_ID_NIL;
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);
key = cut_take_printf("key: %d (%d:%d)", i, process_number, N_THREADS);
rc = grn_ctx_init(contexts[i], GRN_CTX_USE_QL);
cut_set_message("context: %d (%d)", i, process_number);
grn_test_assert(rc);
context = contexts[i];
path = g_getenv(GRN_TEST_ENV_PATRICIA_TRIE_PATH);
cut_assert_not_null(path);
tries[i] = grn_pat_open(context, path);
cut_assert_not_null(tries[i], "patricia trie: %d (%d)", i, process_number);
trie = tries[i];
cut_set_message("lookup - fail: %s (%d:%d)", key, i, process_number);
grn_test_assert_nil(grn_pat_get(context, trie, key, strlen(key), &value));
value_string = cut_take_printf("value: [%s] (%d:%d)", key, i, process_number);
rc = grn_io_lock(context, trie->io, -1);
if (rc != GRN_SUCCESS)
grn_test_assert(rc);
added = 0;
id = grn_pat_add(context, trie, key, strlen(key), &value, &added);
grn_io_unlock(trie->io);
grn_test_assert_not_nil(id);
cut_assert_equal_uint(1, added);
strcpy(value, value_string);
value = NULL;
id = grn_pat_get(context, trie, key, strlen(key), &value);
cut_set_message("lookup - success: %s (%d:%d)", key, i, process_number);
grn_test_assert_not_nil(id);
cut_assert_equal_string(value_string, value);
tries[i] = NULL;
grn_test_assert(grn_pat_close(context, trie));
contexts[i] = NULL;
grn_test_assert(grn_ctx_fin(context));
}
int
pat_get(const char *path)
{
int i;
grn_obj buf;
grn_pat *pat = grn_pat_open(&ctx, path);
if (!pat) { 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);
{
void *value;
grn_id rid = grn_pat_get(&ctx, pat, GRN_BULK_HEAD(&buf), key_size, &value);
if (!rid) {
fprintf(stderr, "table_lookup failed");
} else {
if (memcmp(value, GRN_BULK_HEAD(&buf), value_size)) {
fprintf(stderr, "value unmatch %d: %d != %d\n", i, *((int *)value), key);
}
}
}
}
grn_obj_close(&ctx, &buf);
grn_pat_close(&ctx, pat);
return 0;
}
grn_id
grn_token_next(grn_ctx *ctx, grn_token *token)
{
int status;
grn_id tid = GRN_ID_NIL;
grn_obj *table = token->table;
grn_obj *tokenizer = token->tokenizer;
while (token->status != GRN_TOKEN_DONE) {
if (tokenizer) {
grn_obj *curr_, *stat_;
((grn_proc *)tokenizer)->funcs[PROC_NEXT](ctx, 1, &table, &token->pctx.user_data);
stat_ = grn_ctx_pop(ctx);
curr_ = grn_ctx_pop(ctx);
token->curr = (const unsigned char *)GRN_TEXT_VALUE(curr_);
token->curr_size = GRN_TEXT_LEN(curr_);
status = GRN_UINT32_VALUE(stat_);
token->status = ((status & GRN_TOKENIZER_TOKEN_LAST) ||
(token->mode == GRN_TOKEN_GET &&
(status & GRN_TOKENIZER_TOKEN_REACH_END)))
? GRN_TOKEN_DONE : GRN_TOKEN_DOING;
token->force_prefix = 0;
if (token->curr_size == 0) {
char tokenizer_name[GRN_TABLE_MAX_KEY_SIZE];
int tokenizer_name_length;
tokenizer_name_length =
grn_obj_name(ctx, token->tokenizer,
tokenizer_name, GRN_TABLE_MAX_KEY_SIZE);
GRN_LOG(ctx, GRN_WARN,
"[token_next] ignore an empty token: <%.*s>: <%.*s>",
tokenizer_name_length, tokenizer_name,
token->orig_blen, token->orig);
continue;
}
if (token->curr_size > GRN_TABLE_MAX_KEY_SIZE) {
GRN_LOG(ctx, GRN_WARN,
"[token_next] ignore too long token. "
"Token must be less than or equal to %d: <%d>(<%.*s>)",
GRN_TABLE_MAX_KEY_SIZE,
token->curr_size,
token->curr_size, token->curr);
continue;
}
if (status & GRN_TOKENIZER_TOKEN_UNMATURED) {
if (status & GRN_TOKENIZER_TOKEN_OVERLAP) {
if (token->mode == GRN_TOKEN_GET) { token->pos++; continue; }
} else {
if (status & GRN_TOKENIZER_TOKEN_LAST) { token->force_prefix = 1; }
}
}
} else {
token->status = GRN_TOKEN_DONE;
}
if (token->mode == GRN_TOKEN_ADD) {
switch (table->header.type) {
case GRN_TABLE_PAT_KEY :
if (grn_io_lock(ctx, ((grn_pat *)table)->io, grn_lock_timeout)) {
tid = GRN_ID_NIL;
} else {
tid = grn_pat_add(ctx, (grn_pat *)table, token->curr, token->curr_size,
NULL, NULL);
grn_io_unlock(((grn_pat *)table)->io);
}
break;
case GRN_TABLE_DAT_KEY :
if (grn_io_lock(ctx, ((grn_dat *)table)->io, grn_lock_timeout)) {
tid = GRN_ID_NIL;
} else {
tid = grn_dat_add(ctx, (grn_dat *)table, token->curr, token->curr_size,
NULL, NULL);
grn_io_unlock(((grn_dat *)table)->io);
}
break;
case GRN_TABLE_HASH_KEY :
if (grn_io_lock(ctx, ((grn_hash *)table)->io, grn_lock_timeout)) {
tid = GRN_ID_NIL;
} else {
tid = grn_hash_add(ctx, (grn_hash *)table, token->curr, token->curr_size,
NULL, NULL);
grn_io_unlock(((grn_hash *)table)->io);
}
break;
case GRN_TABLE_NO_KEY :
if (token->curr_size == sizeof(grn_id)) {
tid = *((grn_id *)token->curr);
} else {
tid = GRN_ID_NIL;
}
break;
}
} else {
switch (table->header.type) {
case GRN_TABLE_PAT_KEY :
tid = grn_pat_get(ctx, (grn_pat *)table, token->curr, token->curr_size, NULL);
break;
case GRN_TABLE_DAT_KEY :
tid = grn_dat_get(ctx, (grn_dat *)table, token->curr, token->curr_size, NULL);
break;
case GRN_TABLE_HASH_KEY :
tid = grn_hash_get(ctx, (grn_hash *)table, token->curr, token->curr_size, NULL);
break;
case GRN_TABLE_NO_KEY :
if (token->curr_size == sizeof(grn_id)) {
tid = *((grn_id *)token->curr);
} else {
tid = GRN_ID_NIL;
}
break;
}
}
if (tid == GRN_ID_NIL && token->status != GRN_TOKEN_DONE) {
token->status = GRN_TOKEN_NOT_FOUND;
}
token->pos++;
break;
}
return tid;
}
grn_id
grn_token_next(grn_ctx *ctx, grn_token *token)
{
int status;
grn_id tid = GRN_ID_NIL;
grn_obj *table = token->table;
grn_obj *tokenizer = token->tokenizer;
while (token->status != grn_token_done) {
if (tokenizer) {
grn_obj *curr_, *stat_;
((grn_proc *)tokenizer)->funcs[PROC_NEXT](ctx, 1, &table, &token->pctx.user_data);
stat_ = grn_ctx_pop(ctx);
curr_ = grn_ctx_pop(ctx);
token->curr = GRN_TEXT_VALUE(curr_);
token->curr_size = GRN_TEXT_LEN(curr_);
status = GRN_UINT32_VALUE(stat_);
token->status = ((status & GRN_TOKEN_LAST) ||
(!token->add && (status & GRN_TOKEN_REACH_END)))
? grn_token_done : grn_token_doing;
token->force_prefix = 0;
if (status & GRN_TOKEN_UNMATURED) {
if (status & GRN_TOKEN_OVERLAP) {
if (!token->add) { token->pos++; continue; }
} else {
if (status & GRN_TOKEN_LAST) { token->force_prefix = 1; }
}
}
} else {
token->curr = token->orig;
token->curr_size = token->orig_blen;
token->status = grn_token_done;
}
if (token->add) {
switch (table->header.type) {
case GRN_TABLE_PAT_KEY :
if (grn_io_lock(ctx, ((grn_pat *)table)->io, 10000000)) {
tid = GRN_ID_NIL;
} else {
tid = grn_pat_add(ctx, (grn_pat *)table, token->curr, token->curr_size,
NULL, NULL);
grn_io_unlock(((grn_pat *)table)->io);
}
break;
case GRN_TABLE_HASH_KEY :
if (grn_io_lock(ctx, ((grn_hash *)table)->io, 10000000)) {
tid = GRN_ID_NIL;
} else {
tid = grn_hash_add(ctx, (grn_hash *)table, token->curr, token->curr_size,
NULL, NULL);
grn_io_unlock(((grn_hash *)table)->io);
}
break;
case GRN_TABLE_NO_KEY :
if (token->curr_size == sizeof(grn_id)) {
tid = *((grn_id *)token->curr);
} else {
tid = GRN_ID_NIL;
}
break;
}
} else {
switch (table->header.type) {
case GRN_TABLE_PAT_KEY :
tid = grn_pat_get(ctx, (grn_pat *)table, token->curr, token->curr_size, NULL);
break;
case GRN_TABLE_HASH_KEY :
tid = grn_hash_get(ctx, (grn_hash *)table, token->curr, token->curr_size, NULL);
break;
case GRN_TABLE_NO_KEY :
if (token->curr_size == sizeof(grn_id)) {
tid = *((grn_id *)token->curr);
} else {
tid = GRN_ID_NIL;
}
break;
}
}
if (tid == GRN_ID_NIL && token->status != grn_token_done) {
token->status = grn_token_not_found;
}
token->pos++;
break;
}
return tid;
}
grn_id
grn_token_cursor_next(grn_ctx *ctx, grn_token_cursor *token_cursor)
{
int status;
grn_id tid = GRN_ID_NIL;
grn_obj *table = token_cursor->table;
grn_obj *tokenizer = token_cursor->tokenizer;
while (token_cursor->status != GRN_TOKEN_CURSOR_DONE) {
if (tokenizer) {
grn_obj *curr_, *stat_;
((grn_proc *)tokenizer)->funcs[PROC_NEXT](ctx, 1, &table, &token_cursor->pctx.user_data);
stat_ = grn_ctx_pop(ctx);
curr_ = grn_ctx_pop(ctx);
status = grn_token_cursor_next_apply_token_filters(ctx, token_cursor,
curr_, stat_);
token_cursor->status =
((status & GRN_TOKEN_LAST) ||
(token_cursor->mode == GRN_TOKENIZE_GET &&
(status & GRN_TOKEN_REACH_END)))
? GRN_TOKEN_CURSOR_DONE : GRN_TOKEN_CURSOR_DOING;
token_cursor->force_prefix = GRN_FALSE;
#define SKIP_FLAGS \
(GRN_TOKEN_SKIP | GRN_TOKEN_SKIP_WITH_POSITION)
if (status & SKIP_FLAGS) {
if (status & GRN_TOKEN_SKIP) {
token_cursor->pos++;
}
if (token_cursor->status == GRN_TOKEN_CURSOR_DONE && tid == GRN_ID_NIL) {
token_cursor->status = GRN_TOKEN_CURSOR_DONE_SKIP;
break;
} else {
continue;
}
}
#undef SKIP_FLAGS
if (status & GRN_TOKEN_FORCE_PREFIX) {
token_cursor->force_prefix = GRN_TRUE;
}
if (token_cursor->curr_size == 0) {
if (token_cursor->status != GRN_TOKEN_CURSOR_DONE) {
char tokenizer_name[GRN_TABLE_MAX_KEY_SIZE];
int tokenizer_name_length;
tokenizer_name_length =
grn_obj_name(ctx, token_cursor->tokenizer,
tokenizer_name, GRN_TABLE_MAX_KEY_SIZE);
GRN_LOG(ctx, GRN_WARN,
"[token_next] ignore an empty token: <%.*s>: <%.*s>",
tokenizer_name_length, tokenizer_name,
token_cursor->orig_blen, token_cursor->orig);
}
continue;
}
if (token_cursor->curr_size > GRN_TABLE_MAX_KEY_SIZE) {
GRN_LOG(ctx, GRN_WARN,
"[token_next] ignore too long token. "
"Token must be less than or equal to %d: <%d>(<%.*s>)",
GRN_TABLE_MAX_KEY_SIZE,
token_cursor->curr_size,
token_cursor->curr_size, token_cursor->curr);
continue;
}
if (status & GRN_TOKEN_UNMATURED) {
if (status & GRN_TOKEN_OVERLAP) {
if (token_cursor->mode == GRN_TOKENIZE_GET) {
token_cursor->pos++;
continue;
}
} else {
if (status & GRN_TOKEN_REACH_END) {
token_cursor->force_prefix = GRN_TRUE;
}
}
}
} else {
token_cursor->status = GRN_TOKEN_CURSOR_DONE;
}
if (token_cursor->mode == GRN_TOKENIZE_ADD) {
switch (table->header.type) {
case GRN_TABLE_PAT_KEY :
if (grn_io_lock(ctx, ((grn_pat *)table)->io, grn_lock_timeout)) {
tid = GRN_ID_NIL;
} else {
tid = grn_pat_add(ctx, (grn_pat *)table, token_cursor->curr, token_cursor->curr_size,
NULL, NULL);
grn_io_unlock(((grn_pat *)table)->io);
}
break;
case GRN_TABLE_DAT_KEY :
if (grn_io_lock(ctx, ((grn_dat *)table)->io, grn_lock_timeout)) {
tid = GRN_ID_NIL;
} else {
tid = grn_dat_add(ctx, (grn_dat *)table, token_cursor->curr, token_cursor->curr_size,
NULL, NULL);
grn_io_unlock(((grn_dat *)table)->io);
}
break;
case GRN_TABLE_HASH_KEY :
if (grn_io_lock(ctx, ((grn_hash *)table)->io, grn_lock_timeout)) {
tid = GRN_ID_NIL;
} else {
tid = grn_hash_add(ctx, (grn_hash *)table, token_cursor->curr, token_cursor->curr_size,
NULL, NULL);
grn_io_unlock(((grn_hash *)table)->io);
}
break;
case GRN_TABLE_NO_KEY :
if (token_cursor->curr_size == sizeof(grn_id)) {
tid = *((grn_id *)token_cursor->curr);
} else {
tid = GRN_ID_NIL;
}
break;
}
} else if (token_cursor->mode != GRN_TOKENIZE_ONLY) {
switch (table->header.type) {
case GRN_TABLE_PAT_KEY :
tid = grn_pat_get(ctx, (grn_pat *)table, token_cursor->curr, token_cursor->curr_size, NULL);
break;
case GRN_TABLE_DAT_KEY :
tid = grn_dat_get(ctx, (grn_dat *)table, token_cursor->curr, token_cursor->curr_size, NULL);
break;
case GRN_TABLE_HASH_KEY :
tid = grn_hash_get(ctx, (grn_hash *)table, token_cursor->curr, token_cursor->curr_size, NULL);
break;
case GRN_TABLE_NO_KEY :
if (token_cursor->curr_size == sizeof(grn_id)) {
tid = *((grn_id *)token_cursor->curr);
} else {
tid = GRN_ID_NIL;
}
break;
}
}
if (token_cursor->mode != GRN_TOKENIZE_ONLY &&
tid == GRN_ID_NIL && token_cursor->status != GRN_TOKEN_CURSOR_DONE) {
token_cursor->status = GRN_TOKEN_CURSOR_NOT_FOUND;
}
token_cursor->pos++;
break;
}
return tid;
}
