| 1 | /* -*- c-basic-offset: 2 -*- */ |
| 2 | /* |
| 3 | Copyright(C) 2009-2017 Brazil |
| 4 | |
| 5 | This library is free software; you can redistribute it and/or |
| 6 | modify it under the terms of the GNU Lesser General Public |
| 7 | License version 2.1 as published by the Free Software Foundation. |
| 8 | |
| 9 | This library is distributed in the hope that it will be useful, |
| 10 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 11 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 12 | Lesser General Public License for more details. |
| 13 | |
| 14 | You should have received a copy of the GNU Lesser General Public |
| 15 | License along with this library; if not, write to the Free Software |
| 16 | Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 17 | */ |
| 18 | |
| 19 | #include "grn_load.h" |
| 20 | #include "grn_ctx_impl.h" |
| 21 | #include "grn_db.h" |
| 22 | #include "grn_util.h" |
| 23 | |
| 24 | static void |
| 25 | grn_loader_save_error(grn_ctx *ctx, grn_loader *loader) |
| 26 | { |
| 27 | loader->rc = ctx->rc; |
| 28 | grn_strcpy(loader->errbuf, GRN_CTX_MSGSIZE, ctx->errbuf); |
| 29 | } |
| 30 | |
| 31 | static grn_obj * |
| 32 | values_add(grn_ctx *ctx, grn_loader *loader) |
| 33 | { |
| 34 | grn_obj *res; |
| 35 | uint32_t curr_size = loader->values_size * sizeof(grn_obj); |
| 36 | if (curr_size < GRN_TEXT_LEN(&loader->values)) { |
| 37 | res = (grn_obj *)(GRN_TEXT_VALUE(&loader->values) + curr_size); |
| 38 | res->header.domain = GRN_DB_TEXT; |
| 39 | GRN_BULK_REWIND(res); |
| 40 | } else { |
| 41 | if (grn_bulk_space(ctx, &loader->values, sizeof(grn_obj))) { return NULL; } |
| 42 | res = (grn_obj *)(GRN_TEXT_VALUE(&loader->values) + curr_size); |
| 43 | GRN_TEXT_INIT(res, 0); |
| 44 | } |
| 45 | loader->values_size++; |
| 46 | loader->last = res; |
| 47 | return res; |
| 48 | } |
| 49 | |
| 50 | static grn_obj * |
| 51 | values_next(grn_ctx *ctx, grn_obj *value) |
| 52 | { |
| 53 | if (value->header.domain == GRN_JSON_LOAD_OPEN_BRACKET || |
| 54 | value->header.domain == GRN_JSON_LOAD_OPEN_BRACE) { |
| 55 | value += GRN_UINT32_VALUE(value); |
| 56 | } |
| 57 | return value + 1; |
| 58 | } |
| 59 | |
| 60 | static int |
| 61 | values_len(grn_ctx *ctx, grn_obj *head, grn_obj *tail) |
| 62 | { |
| 63 | int len; |
| 64 | for (len = 0; head < tail; head = values_next(ctx, head), len++) ; |
| 65 | return len; |
| 66 | } |
| 67 | |
| 68 | static grn_id |
| 69 | loader_add(grn_ctx *ctx, grn_obj *key) |
| 70 | { |
| 71 | int added = 0; |
| 72 | grn_loader *loader = &ctx->impl->loader; |
| 73 | grn_id id = grn_table_add_by_key(ctx, loader->table, key, &added); |
| 74 | if (id == GRN_ID_NIL) { |
| 75 | grn_loader_save_error(ctx, loader); |
| 76 | return id; |
| 77 | } |
| 78 | if (!added && loader->ifexists) { |
| 79 | grn_obj *v = grn_expr_get_var_by_offset(ctx, loader->ifexists, 0); |
| 80 | grn_obj *result; |
| 81 | GRN_RECORD_SET(ctx, v, id); |
| 82 | result = grn_expr_exec(ctx, loader->ifexists, 0); |
| 83 | if (!grn_obj_is_true(ctx, result)) { |
| 84 | id = 0; |
| 85 | } |
| 86 | } |
| 87 | return id; |
| 88 | } |
| 89 | |
| 90 | static void |
| 91 | add_weight_vector(grn_ctx *ctx, |
| 92 | grn_obj *column, |
| 93 | grn_obj *value, |
| 94 | grn_obj *vector) |
| 95 | { |
| 96 | unsigned int i, n; |
| 97 | grn_obj weight_buffer; |
| 98 | |
| 99 | n = GRN_UINT32_VALUE(value); |
| 100 | GRN_UINT32_INIT(&weight_buffer, 0); |
| 101 | for (i = 0; i < n; i += 2) { |
| 102 | grn_rc rc; |
| 103 | grn_obj *key, *weight; |
| 104 | |
| 105 | key = value + 1 + i; |
| 106 | weight = key + 1; |
| 107 | |
| 108 | GRN_BULK_REWIND(&weight_buffer); |
| 109 | rc = grn_obj_cast(ctx, weight, &weight_buffer, GRN_TRUE); |
| 110 | if (rc != GRN_SUCCESS) { |
| 111 | grn_obj *range; |
| 112 | range = grn_ctx_at(ctx, weight_buffer.header.domain); |
| 113 | ERR_CAST(column, range, weight); |
| 114 | grn_obj_unlink(ctx, range); |
| 115 | break; |
| 116 | } |
| 117 | grn_vector_add_element(ctx, |
| 118 | vector, |
| 119 | GRN_BULK_HEAD(key), |
| 120 | GRN_BULK_VSIZE(key), |
| 121 | GRN_UINT32_VALUE(&weight_buffer), |
| 122 | key->header.domain); |
| 123 | } |
| 124 | GRN_OBJ_FIN(ctx, &weight_buffer); |
| 125 | } |
| 126 | |
| 127 | static void |
| 128 | set_vector(grn_ctx *ctx, grn_obj *column, grn_id id, grn_obj *vector) |
| 129 | { |
| 130 | int n = GRN_UINT32_VALUE(vector); |
| 131 | grn_obj buf, *v = vector + 1; |
| 132 | grn_id range_id; |
| 133 | grn_obj *range; |
| 134 | |
| 135 | range_id = DB_OBJ(column)->range; |
| 136 | range = grn_ctx_at(ctx, range_id); |
| 137 | if (grn_obj_is_table(ctx, range)) { |
| 138 | GRN_RECORD_INIT(&buf, GRN_OBJ_VECTOR, range_id); |
| 139 | while (n--) { |
| 140 | grn_bool cast_failed = GRN_FALSE; |
| 141 | grn_obj record, *element = v; |
| 142 | if (range_id != element->header.domain) { |
| 143 | GRN_RECORD_INIT(&record, 0, range_id); |
| 144 | if (grn_obj_cast(ctx, element, &record, GRN_TRUE)) { |
| 145 | cast_failed = GRN_TRUE; |
| 146 | ERR_CAST(column, range, element); |
| 147 | } |
| 148 | element = &record; |
| 149 | } |
| 150 | if (!cast_failed) { |
| 151 | GRN_UINT32_PUT(ctx, &buf, GRN_RECORD_VALUE(element)); |
| 152 | } |
| 153 | if (element == &record) { GRN_OBJ_FIN(ctx, element); } |
| 154 | v = values_next(ctx, v); |
| 155 | } |
| 156 | } else { |
| 157 | if (((struct _grn_type *)range)->obj.header.flags & GRN_OBJ_KEY_VAR_SIZE) { |
| 158 | GRN_TEXT_INIT(&buf, GRN_OBJ_VECTOR); |
| 159 | while (n--) { |
| 160 | switch (v->header.domain) { |
| 161 | case GRN_DB_TEXT : |
| 162 | { |
| 163 | grn_bool cast_failed = GRN_FALSE; |
| 164 | grn_obj casted_element, *element = v; |
| 165 | if (range_id != element->header.domain) { |
| 166 | GRN_OBJ_INIT(&casted_element, GRN_BULK, 0, range_id); |
| 167 | if (grn_obj_cast(ctx, element, &casted_element, GRN_TRUE)) { |
| 168 | cast_failed = GRN_TRUE; |
| 169 | ERR_CAST(column, range, element); |
| 170 | } |
| 171 | element = &casted_element; |
| 172 | } |
| 173 | if (!cast_failed) { |
| 174 | grn_vector_add_element(ctx, &buf, |
| 175 | GRN_TEXT_VALUE(element), |
| 176 | GRN_TEXT_LEN(element), |
| 177 | 0, |
| 178 | element->header.domain); |
| 179 | } |
| 180 | if (element == &casted_element) { GRN_OBJ_FIN(ctx, element); } |
| 181 | break; |
| 182 | } |
| 183 | case GRN_JSON_LOAD_OPEN_BRACE : |
| 184 | add_weight_vector(ctx, column, v, &buf); |
| 185 | n -= GRN_UINT32_VALUE(v); |
| 186 | break; |
| 187 | default : |
| 188 | ERR(GRN_INVALID_ARGUMENT, "array must contain string or object" ); |
| 189 | break; |
| 190 | } |
| 191 | v = values_next(ctx, v); |
| 192 | } |
| 193 | } else { |
| 194 | grn_id value_size = ((grn_db_obj *)range)->range; |
| 195 | GRN_VALUE_FIX_SIZE_INIT(&buf, GRN_OBJ_VECTOR, range_id); |
| 196 | while (n--) { |
| 197 | grn_bool cast_failed = GRN_FALSE; |
| 198 | grn_obj casted_element, *element = v; |
| 199 | if (range_id != element->header.domain) { |
| 200 | GRN_OBJ_INIT(&casted_element, GRN_BULK, 0, range_id); |
| 201 | if (grn_obj_cast(ctx, element, &casted_element, GRN_TRUE)) { |
| 202 | cast_failed = GRN_TRUE; |
| 203 | ERR_CAST(column, range, element); |
| 204 | } |
| 205 | element = &casted_element; |
| 206 | } |
| 207 | if (!cast_failed) { |
| 208 | grn_bulk_write(ctx, &buf, GRN_TEXT_VALUE(element), value_size); |
| 209 | } |
| 210 | if (element == &casted_element) { GRN_OBJ_FIN(ctx, element); } |
| 211 | v = values_next(ctx, v); |
| 212 | } |
| 213 | } |
| 214 | } |
| 215 | grn_obj_set_value(ctx, column, id, &buf, GRN_OBJ_SET); |
| 216 | GRN_OBJ_FIN(ctx, &buf); |
| 217 | } |
| 218 | |
| 219 | static void |
| 220 | set_weight_vector(grn_ctx *ctx, grn_obj *column, grn_id id, grn_obj *value) |
| 221 | { |
| 222 | if (!grn_obj_is_weight_vector_column(ctx, column)) { |
| 223 | char column_name[GRN_TABLE_MAX_KEY_SIZE]; |
| 224 | int column_name_size; |
| 225 | column_name_size = grn_obj_name(ctx, column, column_name, |
| 226 | GRN_TABLE_MAX_KEY_SIZE); |
| 227 | ERR(GRN_INVALID_ARGUMENT, |
| 228 | "<%.*s>: columns except weight vector column don't support object value" , |
| 229 | column_name_size, column_name); |
| 230 | return; |
| 231 | } |
| 232 | |
| 233 | { |
| 234 | grn_obj vector; |
| 235 | |
| 236 | GRN_TEXT_INIT(&vector, GRN_OBJ_VECTOR); |
| 237 | add_weight_vector(ctx, column, value, &vector); |
| 238 | grn_obj_set_value(ctx, column, id, &vector, GRN_OBJ_SET); |
| 239 | GRN_OBJ_FIN(ctx, &vector); |
| 240 | } |
| 241 | } |
| 242 | |
| 243 | static inline int |
| 244 | name_equal(const char *p, unsigned int size, const char *name) |
| 245 | { |
| 246 | if (strlen(name) != size) { return 0; } |
| 247 | if (*p != GRN_DB_PSEUDO_COLUMN_PREFIX) { return 0; } |
| 248 | return !memcmp(p + 1, name + 1, size - 1); |
| 249 | } |
| 250 | |
| 251 | static void |
| 252 | report_set_column_value_failure(grn_ctx *ctx, |
| 253 | grn_obj *key, |
| 254 | const char *column_name, |
| 255 | unsigned int column_name_size, |
| 256 | grn_obj *column_value) |
| 257 | { |
| 258 | grn_obj key_inspected, column_value_inspected; |
| 259 | |
| 260 | GRN_TEXT_INIT(&key_inspected, 0); |
| 261 | GRN_TEXT_INIT(&column_value_inspected, 0); |
| 262 | grn_inspect_limited(ctx, &key_inspected, key); |
| 263 | grn_inspect_limited(ctx, &column_value_inspected, column_value); |
| 264 | GRN_LOG(ctx, GRN_LOG_ERROR, |
| 265 | "[table][load] failed to set column value: %s: " |
| 266 | "key: <%.*s>, column: <%.*s>, value: <%.*s>" , |
| 267 | ctx->errbuf, |
| 268 | (int)GRN_TEXT_LEN(&key_inspected), |
| 269 | GRN_TEXT_VALUE(&key_inspected), |
| 270 | column_name_size, |
| 271 | column_name, |
| 272 | (int)GRN_TEXT_LEN(&column_value_inspected), |
| 273 | GRN_TEXT_VALUE(&column_value_inspected)); |
| 274 | GRN_OBJ_FIN(ctx, &key_inspected); |
| 275 | GRN_OBJ_FIN(ctx, &column_value_inspected); |
| 276 | } |
| 277 | |
| 278 | static grn_id |
| 279 | parse_id_value(grn_ctx *ctx, grn_obj *value) |
| 280 | { |
| 281 | switch (value->header.type) { |
| 282 | case GRN_DB_UINT32 : |
| 283 | return GRN_UINT32_VALUE(value); |
| 284 | case GRN_DB_INT32 : |
| 285 | return GRN_INT32_VALUE(value); |
| 286 | default : |
| 287 | { |
| 288 | grn_id id = GRN_ID_NIL; |
| 289 | grn_obj casted_value; |
| 290 | GRN_UINT32_INIT(&casted_value, 0); |
| 291 | if (grn_obj_cast(ctx, value, &casted_value, GRN_FALSE) != GRN_SUCCESS) { |
| 292 | grn_obj inspected; |
| 293 | GRN_TEXT_INIT(&inspected, 0); |
| 294 | grn_inspect(ctx, &inspected, value); |
| 295 | ERR(GRN_INVALID_ARGUMENT, |
| 296 | "<%s>: failed to cast to <UInt32>: <%.*s>" , |
| 297 | GRN_COLUMN_NAME_ID, |
| 298 | (int)GRN_TEXT_LEN(&inspected), |
| 299 | GRN_TEXT_VALUE(&inspected)); |
| 300 | GRN_OBJ_FIN(ctx, &inspected); |
| 301 | } else { |
| 302 | id = GRN_UINT32_VALUE(&casted_value); |
| 303 | } |
| 304 | GRN_OBJ_FIN(ctx, &casted_value); |
| 305 | return id; |
| 306 | } |
| 307 | } |
| 308 | } |
| 309 | |
| 310 | static void |
| 311 | bracket_close(grn_ctx *ctx, grn_loader *loader) |
| 312 | { |
| 313 | grn_id id = GRN_ID_NIL; |
| 314 | grn_obj *value, *value_end, *id_value = NULL, *key_value = NULL; |
| 315 | grn_obj *col, **cols; /* Columns except _id and _key. */ |
| 316 | uint32_t i, begin; |
| 317 | uint32_t ncols; /* Number of columns except _id and _key. */ |
| 318 | uint32_t nvalues; /* Number of values in brackets. */ |
| 319 | uint32_t depth; |
| 320 | grn_bool is_record_load = GRN_FALSE; |
| 321 | |
| 322 | cols = (grn_obj **)GRN_BULK_HEAD(&loader->columns); |
| 323 | ncols = GRN_BULK_VSIZE(&loader->columns) / sizeof(grn_obj *); |
| 324 | GRN_UINT32_POP(&loader->level, begin); |
| 325 | value = (grn_obj *)GRN_TEXT_VALUE(&loader->values) + begin; |
| 326 | value_end = (grn_obj *)GRN_TEXT_VALUE(&loader->values) + loader->values_size; |
| 327 | GRN_ASSERT(value->header.domain == GRN_JSON_LOAD_OPEN_BRACKET); |
| 328 | GRN_UINT32_SET(ctx, value, loader->values_size - begin - 1); |
| 329 | value++; |
| 330 | depth = GRN_BULK_VSIZE(&loader->level); |
| 331 | if (depth > sizeof(uint32_t) * loader->emit_level) { |
| 332 | return; |
| 333 | } |
| 334 | if (depth == 0 || !loader->table || |
| 335 | loader->columns_status == GRN_LOADER_COLUMNS_BROKEN) { |
| 336 | goto exit; |
| 337 | } |
| 338 | nvalues = values_len(ctx, value, value_end); |
| 339 | |
| 340 | if (loader->columns_status == GRN_LOADER_COLUMNS_UNSET) { |
| 341 | /* |
| 342 | * Target columns and _id or _key are not specified yet and values are |
| 343 | * handled as column names and "_id" or "_key". |
| 344 | */ |
| 345 | for (i = 0; i < nvalues; i++) { |
| 346 | const char *col_name; |
| 347 | unsigned int col_name_size; |
| 348 | if (value->header.domain != GRN_DB_TEXT) { |
| 349 | grn_obj buffer; |
| 350 | GRN_TEXT_INIT(&buffer, 0); |
| 351 | grn_inspect(ctx, &buffer, value); |
| 352 | ERR(GRN_INVALID_ARGUMENT, |
| 353 | "column name must be string: <%.*s>" , |
| 354 | (int)GRN_TEXT_LEN(&buffer), GRN_TEXT_VALUE(&buffer)); |
| 355 | grn_loader_save_error(ctx, loader); |
| 356 | GRN_OBJ_FIN(ctx, &buffer); |
| 357 | loader->columns_status = GRN_LOADER_COLUMNS_BROKEN; |
| 358 | goto exit; |
| 359 | } |
| 360 | col_name = GRN_TEXT_VALUE(value); |
| 361 | col_name_size = GRN_TEXT_LEN(value); |
| 362 | col = grn_obj_column(ctx, loader->table, col_name, col_name_size); |
| 363 | if (!col) { |
| 364 | ERR(GRN_INVALID_ARGUMENT, "nonexistent column: <%.*s>" , |
| 365 | col_name_size, col_name); |
| 366 | grn_loader_save_error(ctx, loader); |
| 367 | loader->columns_status = GRN_LOADER_COLUMNS_BROKEN; |
| 368 | goto exit; |
| 369 | } |
| 370 | if (name_equal(col_name, col_name_size, GRN_COLUMN_NAME_ID)) { |
| 371 | grn_obj_unlink(ctx, col); |
| 372 | if (loader->id_offset != -1 || loader->key_offset != -1) { |
| 373 | /* _id and _key must not appear more than once. */ |
| 374 | if (loader->id_offset != -1) { |
| 375 | ERR(GRN_INVALID_ARGUMENT, |
| 376 | "duplicated id and key columns: <%s> at %d and <%s> at %d" , |
| 377 | GRN_COLUMN_NAME_ID, i, |
| 378 | GRN_COLUMN_NAME_ID, loader->id_offset); |
| 379 | } else { |
| 380 | ERR(GRN_INVALID_ARGUMENT, |
| 381 | "duplicated id and key columns: <%s> at %d and <%s> at %d" , |
| 382 | GRN_COLUMN_NAME_ID, i, |
| 383 | GRN_COLUMN_NAME_KEY, loader->key_offset); |
| 384 | } |
| 385 | grn_loader_save_error(ctx, loader); |
| 386 | loader->columns_status = GRN_LOADER_COLUMNS_BROKEN; |
| 387 | goto exit; |
| 388 | } |
| 389 | loader->id_offset = i; |
| 390 | } else if (name_equal(col_name, col_name_size, GRN_COLUMN_NAME_KEY)) { |
| 391 | grn_obj_unlink(ctx, col); |
| 392 | if (loader->id_offset != -1 || loader->key_offset != -1) { |
| 393 | /* _id and _key must not appear more than once. */ |
| 394 | if (loader->id_offset != -1) { |
| 395 | ERR(GRN_INVALID_ARGUMENT, |
| 396 | "duplicated id and key columns: <%s> at %d and <%s> at %d" , |
| 397 | GRN_COLUMN_NAME_KEY, i, |
| 398 | GRN_COLUMN_NAME_ID, loader->id_offset); |
| 399 | } else { |
| 400 | ERR(GRN_INVALID_ARGUMENT, |
| 401 | "duplicated id and key columns: <%s> at %d and <%s> at %d" , |
| 402 | GRN_COLUMN_NAME_KEY, i, |
| 403 | GRN_COLUMN_NAME_KEY, loader->key_offset); |
| 404 | } |
| 405 | grn_loader_save_error(ctx, loader); |
| 406 | loader->columns_status = GRN_LOADER_COLUMNS_BROKEN; |
| 407 | goto exit; |
| 408 | } |
| 409 | loader->key_offset = i; |
| 410 | } else { |
| 411 | GRN_PTR_PUT(ctx, &loader->columns, col); |
| 412 | } |
| 413 | value++; |
| 414 | } |
| 415 | switch (loader->table->header.type) { |
| 416 | case GRN_TABLE_HASH_KEY : |
| 417 | case GRN_TABLE_PAT_KEY : |
| 418 | case GRN_TABLE_DAT_KEY : |
| 419 | if (loader->id_offset == -1 && loader->key_offset == -1) { |
| 420 | ERR(GRN_INVALID_ARGUMENT, "missing id or key column" ); |
| 421 | grn_loader_save_error(ctx, loader); |
| 422 | loader->columns_status = GRN_LOADER_COLUMNS_BROKEN; |
| 423 | goto exit; |
| 424 | } |
| 425 | break; |
| 426 | } |
| 427 | loader->columns_status = GRN_LOADER_COLUMNS_SET; |
| 428 | goto exit; |
| 429 | } |
| 430 | |
| 431 | is_record_load = GRN_TRUE; |
| 432 | |
| 433 | /* Target columns and _id or _key are already specified. */ |
| 434 | if (!nvalues) { |
| 435 | /* |
| 436 | * Accept empty arrays because a dump command may output a load command |
| 437 | * which contains empty arrays for a table with deleted records. |
| 438 | */ |
| 439 | id = grn_table_add(ctx, loader->table, NULL, 0, NULL); |
| 440 | } else { |
| 441 | uint32_t expected_nvalues = ncols; |
| 442 | if (loader->id_offset != -1 || loader->key_offset != -1) { |
| 443 | expected_nvalues++; |
| 444 | } |
| 445 | if (nvalues != expected_nvalues) { |
| 446 | ERR(GRN_INVALID_ARGUMENT, |
| 447 | "unexpected #values: expected:%u, actual:%u" , |
| 448 | expected_nvalues, nvalues); |
| 449 | grn_loader_save_error(ctx, loader); |
| 450 | goto exit; |
| 451 | } |
| 452 | if (loader->id_offset != -1) { |
| 453 | id_value = value + loader->id_offset; |
| 454 | id = parse_id_value(ctx, id_value); |
| 455 | if (grn_table_at(ctx, loader->table, id) == GRN_ID_NIL) { |
| 456 | id = grn_table_add(ctx, loader->table, NULL, 0, NULL); |
| 457 | } |
| 458 | } else if (loader->key_offset != -1) { |
| 459 | key_value = value + loader->key_offset; |
| 460 | id = loader_add(ctx, key_value); |
| 461 | } else { |
| 462 | id = grn_table_add(ctx, loader->table, NULL, 0, NULL); |
| 463 | } |
| 464 | } |
| 465 | if (id == GRN_ID_NIL) { |
| 466 | /* Target record is not available. */ |
| 467 | goto exit; |
| 468 | } |
| 469 | |
| 470 | for (i = 0; i < nvalues; i++, value = values_next(ctx, value)) { |
| 471 | if (i == loader->id_offset || i == loader->key_offset) { |
| 472 | /* Skip _id and _key, because it's already used to get id. */ |
| 473 | continue; |
| 474 | } |
| 475 | col = *cols; |
| 476 | if (value->header.domain == GRN_JSON_LOAD_OPEN_BRACKET) { |
| 477 | set_vector(ctx, col, id, value); |
| 478 | } else if (value->header.domain == GRN_JSON_LOAD_OPEN_BRACE) { |
| 479 | set_weight_vector(ctx, col, id, value); |
| 480 | } else { |
| 481 | grn_obj_set_value(ctx, col, id, value, GRN_OBJ_SET); |
| 482 | } |
| 483 | if (ctx->rc != GRN_SUCCESS) { |
| 484 | char column_name[GRN_TABLE_MAX_KEY_SIZE]; |
| 485 | unsigned int column_name_size; |
| 486 | grn_loader_save_error(ctx, loader); |
| 487 | column_name_size = grn_obj_name(ctx, col, column_name, |
| 488 | GRN_TABLE_MAX_KEY_SIZE); |
| 489 | report_set_column_value_failure(ctx, key_value, |
| 490 | column_name, column_name_size, |
| 491 | value); |
| 492 | ERRCLR(ctx); |
| 493 | } |
| 494 | cols++; |
| 495 | } |
| 496 | if (loader->each) { |
| 497 | grn_obj *v = grn_expr_get_var_by_offset(ctx, loader->each, 0); |
| 498 | GRN_RECORD_SET(ctx, v, id); |
| 499 | grn_expr_exec(ctx, loader->each, 0); |
| 500 | } |
| 501 | loader->nrecords++; |
| 502 | exit: |
| 503 | if (is_record_load) { |
| 504 | if (loader->output_ids) { |
| 505 | GRN_UINT32_PUT(ctx, &(loader->ids), id); |
| 506 | } |
| 507 | if (loader->output_errors) { |
| 508 | GRN_INT32_PUT(ctx, &(loader->return_codes), ctx->rc); |
| 509 | grn_vector_add_element(ctx, |
| 510 | &(loader->error_messages), |
| 511 | ctx->errbuf, |
| 512 | strlen(ctx->errbuf), |
| 513 | 0, |
| 514 | GRN_DB_TEXT); |
| 515 | } |
| 516 | } |
| 517 | loader->values_size = begin; |
| 518 | ERRCLR(ctx); |
| 519 | } |
| 520 | |
| 521 | static void |
| 522 | brace_close(grn_ctx *ctx, grn_loader *loader) |
| 523 | { |
| 524 | grn_id id = GRN_ID_NIL; |
| 525 | grn_obj *value, *value_begin, *value_end; |
| 526 | grn_obj *id_value = NULL, *key_value = NULL; |
| 527 | uint32_t begin; |
| 528 | |
| 529 | GRN_UINT32_POP(&loader->level, begin); |
| 530 | value_begin = (grn_obj *)GRN_TEXT_VALUE(&loader->values) + begin; |
| 531 | value_end = (grn_obj *)GRN_TEXT_VALUE(&loader->values) + loader->values_size; |
| 532 | GRN_ASSERT(value->header.domain == GRN_JSON_LOAD_OPEN_BRACE); |
| 533 | GRN_UINT32_SET(ctx, value_begin, loader->values_size - begin - 1); |
| 534 | value_begin++; |
| 535 | if (GRN_BULK_VSIZE(&loader->level) > sizeof(uint32_t) * loader->emit_level) { |
| 536 | return; |
| 537 | } |
| 538 | if (!loader->table) { |
| 539 | goto exit; |
| 540 | } |
| 541 | |
| 542 | /* Scan values to find _id or _key. */ |
| 543 | for (value = value_begin; value + 1 < value_end; |
| 544 | value = values_next(ctx, value)) { |
| 545 | const char *name = GRN_TEXT_VALUE(value); |
| 546 | unsigned int name_size = GRN_TEXT_LEN(value); |
| 547 | if (value->header.domain != GRN_DB_TEXT) { |
| 548 | grn_obj buffer; |
| 549 | GRN_TEXT_INIT(&buffer, 0); |
| 550 | grn_inspect(ctx, &buffer, value); |
| 551 | GRN_LOG(ctx, GRN_LOG_ERROR, |
| 552 | "column name must be string: <%.*s>" , |
| 553 | (int)GRN_TEXT_LEN(&buffer), GRN_TEXT_VALUE(&buffer)); |
| 554 | GRN_OBJ_FIN(ctx, &buffer); |
| 555 | goto exit; |
| 556 | } |
| 557 | value++; |
| 558 | if (name_equal(name, name_size, GRN_COLUMN_NAME_ID)) { |
| 559 | if (id_value || key_value) { |
| 560 | if (loader->table->header.type == GRN_TABLE_NO_KEY) { |
| 561 | GRN_LOG(ctx, GRN_LOG_ERROR, "duplicated '_id' column" ); |
| 562 | goto exit; |
| 563 | } else { |
| 564 | GRN_LOG(ctx, GRN_LOG_ERROR, |
| 565 | "duplicated key columns: %s and %s" , |
| 566 | id_value ? GRN_COLUMN_NAME_ID : GRN_COLUMN_NAME_KEY, |
| 567 | GRN_COLUMN_NAME_ID); |
| 568 | goto exit; |
| 569 | } |
| 570 | } |
| 571 | id_value = value; |
| 572 | } else if (name_equal(name, name_size, GRN_COLUMN_NAME_KEY)) { |
| 573 | if (id_value || key_value) { |
| 574 | GRN_LOG(ctx, GRN_LOG_ERROR, |
| 575 | "duplicated key columns: %s and %s" , |
| 576 | id_value ? GRN_COLUMN_NAME_ID : GRN_COLUMN_NAME_KEY, |
| 577 | GRN_COLUMN_NAME_KEY); |
| 578 | goto exit; |
| 579 | } |
| 580 | key_value = value; |
| 581 | } |
| 582 | } |
| 583 | |
| 584 | switch (loader->table->header.type) { |
| 585 | case GRN_TABLE_HASH_KEY : |
| 586 | case GRN_TABLE_PAT_KEY : |
| 587 | case GRN_TABLE_DAT_KEY : |
| 588 | /* The target table requires _id or _key. */ |
| 589 | if (!id_value && !key_value) { |
| 590 | GRN_LOG(ctx, GRN_LOG_ERROR, "neither _key nor _id is assigned" ); |
| 591 | goto exit; |
| 592 | } |
| 593 | break; |
| 594 | default : |
| 595 | /* The target table does not have _key. */ |
| 596 | if (key_value) { |
| 597 | GRN_LOG(ctx, GRN_LOG_ERROR, "nonexistent key value" ); |
| 598 | goto exit; |
| 599 | } |
| 600 | break; |
| 601 | } |
| 602 | |
| 603 | if (id_value) { |
| 604 | id = parse_id_value(ctx, id_value); |
| 605 | if (grn_table_at(ctx, loader->table, id) == GRN_ID_NIL) { |
| 606 | if (ctx->rc == GRN_SUCCESS) { |
| 607 | id = grn_table_add(ctx, loader->table, NULL, 0, NULL); |
| 608 | } |
| 609 | } |
| 610 | } else if (key_value) { |
| 611 | id = loader_add(ctx, key_value); |
| 612 | } else { |
| 613 | id = grn_table_add(ctx, loader->table, NULL, 0, NULL); |
| 614 | } |
| 615 | if (id == GRN_ID_NIL) { |
| 616 | /* Target record is not available. */ |
| 617 | goto exit; |
| 618 | } |
| 619 | |
| 620 | for (value = value_begin; value + 1 < value_end; |
| 621 | value = values_next(ctx, value)) { |
| 622 | grn_obj *col; |
| 623 | const char *name = GRN_TEXT_VALUE(value); |
| 624 | unsigned int name_size = GRN_TEXT_LEN(value); |
| 625 | value++; |
| 626 | if (value == id_value || value == key_value) { |
| 627 | /* Skip _id and _key, because it's already used to get id. */ |
| 628 | continue; |
| 629 | } |
| 630 | col = grn_obj_column(ctx, loader->table, name, name_size); |
| 631 | if (!col) { |
| 632 | GRN_LOG(ctx, GRN_LOG_ERROR, "invalid column('%.*s')" , |
| 633 | (int)name_size, name); |
| 634 | /* Automatic column creation is disabled. */ |
| 635 | /* |
| 636 | if (value->header.domain == GRN_JSON_LOAD_OPEN_BRACKET) { |
| 637 | grn_obj *v = value + 1; |
| 638 | col = grn_column_create(ctx, loader->table, name, name_size, |
| 639 | NULL, GRN_OBJ_PERSISTENT|GRN_OBJ_COLUMN_VECTOR, |
| 640 | grn_ctx_at(ctx, v->header.domain)); |
| 641 | } else { |
| 642 | col = grn_column_create(ctx, loader->table, name, name_size, |
| 643 | NULL, GRN_OBJ_PERSISTENT, |
| 644 | grn_ctx_at(ctx, value->header.domain)); |
| 645 | } |
| 646 | */ |
| 647 | } else { |
| 648 | if (value->header.domain == GRN_JSON_LOAD_OPEN_BRACKET) { |
| 649 | set_vector(ctx, col, id, value); |
| 650 | } else if (value->header.domain == GRN_JSON_LOAD_OPEN_BRACE) { |
| 651 | set_weight_vector(ctx, col, id, value); |
| 652 | } else { |
| 653 | grn_obj_set_value(ctx, col, id, value, GRN_OBJ_SET); |
| 654 | } |
| 655 | if (ctx->rc != GRN_SUCCESS) { |
| 656 | grn_loader_save_error(ctx, loader); |
| 657 | report_set_column_value_failure(ctx, key_value, |
| 658 | name, name_size, value); |
| 659 | ERRCLR(ctx); |
| 660 | } |
| 661 | grn_obj_unlink(ctx, col); |
| 662 | } |
| 663 | } |
| 664 | if (loader->each) { |
| 665 | value = grn_expr_get_var_by_offset(ctx, loader->each, 0); |
| 666 | GRN_RECORD_SET(ctx, value, id); |
| 667 | grn_expr_exec(ctx, loader->each, 0); |
| 668 | } |
| 669 | loader->nrecords++; |
| 670 | exit: |
| 671 | if (loader->output_ids) { |
| 672 | GRN_UINT32_PUT(ctx, &(loader->ids), id); |
| 673 | } |
| 674 | if (loader->output_errors) { |
| 675 | GRN_INT32_PUT(ctx, &(loader->return_codes), ctx->rc); |
| 676 | grn_vector_add_element(ctx, |
| 677 | &(loader->error_messages), |
| 678 | ctx->errbuf, |
| 679 | strlen(ctx->errbuf), |
| 680 | 0, |
| 681 | GRN_DB_TEXT); |
| 682 | } |
| 683 | loader->values_size = begin; |
| 684 | ERRCLR(ctx); |
| 685 | } |
| 686 | |
| 687 | #define JSON_READ_OPEN_BRACKET() do {\ |
| 688 | GRN_UINT32_PUT(ctx, &loader->level, loader->values_size);\ |
| 689 | values_add(ctx, loader);\ |
| 690 | loader->last->header.domain = GRN_JSON_LOAD_OPEN_BRACKET;\ |
| 691 | loader->stat = GRN_LOADER_TOKEN;\ |
| 692 | str++;\ |
| 693 | } while (0) |
| 694 | |
| 695 | #define JSON_READ_OPEN_BRACE() do {\ |
| 696 | GRN_UINT32_PUT(ctx, &loader->level, loader->values_size);\ |
| 697 | values_add(ctx, loader);\ |
| 698 | loader->last->header.domain = GRN_JSON_LOAD_OPEN_BRACE;\ |
| 699 | loader->stat = GRN_LOADER_TOKEN;\ |
| 700 | str++;\ |
| 701 | } while (0) |
| 702 | |
| 703 | static void |
| 704 | json_read(grn_ctx *ctx, grn_loader *loader, const char *str, unsigned int str_len) |
| 705 | { |
| 706 | const char *const beg = str; |
| 707 | char c; |
| 708 | int len; |
| 709 | const char *se = str + str_len; |
| 710 | while (str < se) { |
| 711 | c = *str; |
| 712 | switch (loader->stat) { |
| 713 | case GRN_LOADER_BEGIN : |
| 714 | if ((len = grn_isspace(str, ctx->encoding))) { |
| 715 | str += len; |
| 716 | continue; |
| 717 | } |
| 718 | switch (c) { |
| 719 | case '[' : |
| 720 | JSON_READ_OPEN_BRACKET(); |
| 721 | break; |
| 722 | case '{' : |
| 723 | JSON_READ_OPEN_BRACE(); |
| 724 | break; |
| 725 | default : |
| 726 | ERR(GRN_INVALID_ARGUMENT, |
| 727 | "JSON must start with '[' or '{': <%.*s>" , str_len, beg); |
| 728 | loader->stat = GRN_LOADER_END; |
| 729 | break; |
| 730 | } |
| 731 | break; |
| 732 | case GRN_LOADER_TOKEN : |
| 733 | if ((len = grn_isspace(str, ctx->encoding))) { |
| 734 | str += len; |
| 735 | continue; |
| 736 | } |
| 737 | switch (c) { |
| 738 | case '"' : |
| 739 | loader->stat = GRN_LOADER_STRING; |
| 740 | values_add(ctx, loader); |
| 741 | str++; |
| 742 | break; |
| 743 | case '[' : |
| 744 | JSON_READ_OPEN_BRACKET(); |
| 745 | break; |
| 746 | case '{' : |
| 747 | JSON_READ_OPEN_BRACE(); |
| 748 | break; |
| 749 | case ':' : |
| 750 | str++; |
| 751 | break; |
| 752 | case ',' : |
| 753 | str++; |
| 754 | break; |
| 755 | case ']' : |
| 756 | bracket_close(ctx, loader); |
| 757 | loader->stat = GRN_BULK_VSIZE(&loader->level) ? GRN_LOADER_TOKEN : GRN_LOADER_END; |
| 758 | if (ctx->rc == GRN_CANCEL) { |
| 759 | loader->stat = GRN_LOADER_END; |
| 760 | } |
| 761 | str++; |
| 762 | break; |
| 763 | case '}' : |
| 764 | brace_close(ctx, loader); |
| 765 | loader->stat = GRN_BULK_VSIZE(&loader->level) ? GRN_LOADER_TOKEN : GRN_LOADER_END; |
| 766 | if (ctx->rc == GRN_CANCEL) { |
| 767 | loader->stat = GRN_LOADER_END; |
| 768 | } |
| 769 | str++; |
| 770 | break; |
| 771 | case '+' : case '-' : case '0' : case '1' : case '2' : case '3' : |
| 772 | case '4' : case '5' : case '6' : case '7' : case '8' : case '9' : |
| 773 | loader->stat = GRN_LOADER_NUMBER; |
| 774 | values_add(ctx, loader); |
| 775 | break; |
| 776 | default : |
| 777 | if (('A' <= c && c <= 'Z') || ('a' <= c && c <= 'z') || ('_' == c)) { |
| 778 | loader->stat = GRN_LOADER_SYMBOL; |
| 779 | values_add(ctx, loader); |
| 780 | } else { |
| 781 | if ((len = grn_charlen(ctx, str, se))) { |
| 782 | GRN_LOG(ctx, GRN_LOG_ERROR, "ignored invalid char('%c') at" , c); |
| 783 | GRN_LOG(ctx, GRN_LOG_ERROR, "%.*s" , (int)(str - beg) + len, beg); |
| 784 | GRN_LOG(ctx, GRN_LOG_ERROR, "%*s" , (int)(str - beg) + 1, "^" ); |
| 785 | str += len; |
| 786 | } else { |
| 787 | GRN_LOG(ctx, GRN_LOG_ERROR, "ignored invalid char(\\x%.2x) after" , c); |
| 788 | GRN_LOG(ctx, GRN_LOG_ERROR, "%.*s" , (int)(str - beg), beg); |
| 789 | str = se; |
| 790 | } |
| 791 | } |
| 792 | break; |
| 793 | } |
| 794 | break; |
| 795 | case GRN_LOADER_SYMBOL : |
| 796 | if (('A' <= c && c <= 'Z') || ('a' <= c && c <= 'z') || |
| 797 | ('0' <= c && c <= '9') || ('_' == c)) { |
| 798 | GRN_TEXT_PUTC(ctx, loader->last, c); |
| 799 | str++; |
| 800 | } else { |
| 801 | char *v = GRN_TEXT_VALUE(loader->last); |
| 802 | switch (*v) { |
| 803 | case 'n' : |
| 804 | if (GRN_TEXT_LEN(loader->last) == 4 && !memcmp(v, "null" , 4)) { |
| 805 | loader->last->header.domain = GRN_DB_VOID; |
| 806 | GRN_BULK_REWIND(loader->last); |
| 807 | } |
| 808 | break; |
| 809 | case 't' : |
| 810 | if (GRN_TEXT_LEN(loader->last) == 4 && !memcmp(v, "true" , 4)) { |
| 811 | loader->last->header.domain = GRN_DB_BOOL; |
| 812 | GRN_BOOL_SET(ctx, loader->last, GRN_TRUE); |
| 813 | } |
| 814 | break; |
| 815 | case 'f' : |
| 816 | if (GRN_TEXT_LEN(loader->last) == 5 && !memcmp(v, "false" , 5)) { |
| 817 | loader->last->header.domain = GRN_DB_BOOL; |
| 818 | GRN_BOOL_SET(ctx, loader->last, GRN_FALSE); |
| 819 | } |
| 820 | break; |
| 821 | default : |
| 822 | break; |
| 823 | } |
| 824 | loader->stat = GRN_BULK_VSIZE(&loader->level) ? GRN_LOADER_TOKEN : GRN_LOADER_END; |
| 825 | } |
| 826 | break; |
| 827 | case GRN_LOADER_NUMBER : |
| 828 | switch (c) { |
| 829 | case '+' : case '-' : case '.' : case 'e' : case 'E' : |
| 830 | case '0' : case '1' : case '2' : case '3' : case '4' : |
| 831 | case '5' : case '6' : case '7' : case '8' : case '9' : |
| 832 | GRN_TEXT_PUTC(ctx, loader->last, c); |
| 833 | str++; |
| 834 | break; |
| 835 | default : |
| 836 | { |
| 837 | const char *cur, *str = GRN_BULK_HEAD(loader->last); |
| 838 | const char *str_end = GRN_BULK_CURR(loader->last); |
| 839 | int64_t i = grn_atoll(str, str_end, &cur); |
| 840 | if (cur == str_end) { |
| 841 | loader->last->header.domain = GRN_DB_INT64; |
| 842 | GRN_INT64_SET(ctx, loader->last, i); |
| 843 | } else if (cur != str) { |
| 844 | uint64_t i = grn_atoull(str, str_end, &cur); |
| 845 | if (cur == str_end) { |
| 846 | loader->last->header.domain = GRN_DB_UINT64; |
| 847 | GRN_UINT64_SET(ctx, loader->last, i); |
| 848 | } else if (cur != str) { |
| 849 | double d; |
| 850 | char *end; |
| 851 | grn_obj buf; |
| 852 | GRN_TEXT_INIT(&buf, 0); |
| 853 | GRN_TEXT_PUT(ctx, &buf, str, GRN_BULK_VSIZE(loader->last)); |
| 854 | GRN_TEXT_PUTC(ctx, &buf, '\0'); |
| 855 | errno = 0; |
| 856 | d = strtod(GRN_TEXT_VALUE(&buf), &end); |
| 857 | if (!errno && end + 1 == GRN_BULK_CURR(&buf)) { |
| 858 | loader->last->header.domain = GRN_DB_FLOAT; |
| 859 | GRN_FLOAT_SET(ctx, loader->last, d); |
| 860 | } |
| 861 | GRN_OBJ_FIN(ctx, &buf); |
| 862 | } |
| 863 | } |
| 864 | } |
| 865 | loader->stat = GRN_BULK_VSIZE(&loader->level) ? GRN_LOADER_TOKEN : GRN_LOADER_END; |
| 866 | break; |
| 867 | } |
| 868 | break; |
| 869 | case GRN_LOADER_STRING : |
| 870 | switch (c) { |
| 871 | case '\\' : |
| 872 | loader->stat = GRN_LOADER_STRING_ESC; |
| 873 | str++; |
| 874 | break; |
| 875 | case '"' : |
| 876 | str++; |
| 877 | loader->stat = GRN_BULK_VSIZE(&loader->level) ? GRN_LOADER_TOKEN : GRN_LOADER_END; |
| 878 | /* |
| 879 | *(GRN_BULK_CURR(loader->last)) = '\0'; |
| 880 | GRN_LOG(ctx, GRN_LOG_ALERT, "read str(%s)", GRN_TEXT_VALUE(loader->last)); |
| 881 | */ |
| 882 | break; |
| 883 | default : |
| 884 | if ((len = grn_charlen(ctx, str, se))) { |
| 885 | GRN_TEXT_PUT(ctx, loader->last, str, len); |
| 886 | str += len; |
| 887 | } else { |
| 888 | GRN_LOG(ctx, GRN_LOG_ERROR, "ignored invalid char(\\x%.2x) after" , c); |
| 889 | GRN_LOG(ctx, GRN_LOG_ERROR, "%.*s" , (int)(str - beg), beg); |
| 890 | str = se; |
| 891 | } |
| 892 | break; |
| 893 | } |
| 894 | break; |
| 895 | case GRN_LOADER_STRING_ESC : |
| 896 | switch (c) { |
| 897 | case 'b' : |
| 898 | GRN_TEXT_PUTC(ctx, loader->last, '\b'); |
| 899 | loader->stat = GRN_LOADER_STRING; |
| 900 | break; |
| 901 | case 'f' : |
| 902 | GRN_TEXT_PUTC(ctx, loader->last, '\f'); |
| 903 | loader->stat = GRN_LOADER_STRING; |
| 904 | break; |
| 905 | case 'n' : |
| 906 | GRN_TEXT_PUTC(ctx, loader->last, '\n'); |
| 907 | loader->stat = GRN_LOADER_STRING; |
| 908 | break; |
| 909 | case 'r' : |
| 910 | GRN_TEXT_PUTC(ctx, loader->last, '\r'); |
| 911 | loader->stat = GRN_LOADER_STRING; |
| 912 | break; |
| 913 | case 't' : |
| 914 | GRN_TEXT_PUTC(ctx, loader->last, '\t'); |
| 915 | loader->stat = GRN_LOADER_STRING; |
| 916 | break; |
| 917 | case 'u' : |
| 918 | loader->stat = GRN_LOADER_UNICODE0; |
| 919 | break; |
| 920 | default : |
| 921 | GRN_TEXT_PUTC(ctx, loader->last, c); |
| 922 | loader->stat = GRN_LOADER_STRING; |
| 923 | break; |
| 924 | } |
| 925 | str++; |
| 926 | break; |
| 927 | case GRN_LOADER_UNICODE0 : |
| 928 | switch (c) { |
| 929 | case '0' : case '1' : case '2' : case '3' : case '4' : |
| 930 | case '5' : case '6' : case '7' : case '8' : case '9' : |
| 931 | loader->unichar = (c - '0') * 0x1000; |
| 932 | break; |
| 933 | case 'a' : case 'b' : case 'c' : case 'd' : case 'e' : case 'f' : |
| 934 | loader->unichar = (c - 'a' + 10) * 0x1000; |
| 935 | break; |
| 936 | case 'A' : case 'B' : case 'C' : case 'D' : case 'E' : case 'F' : |
| 937 | loader->unichar = (c - 'A' + 10) * 0x1000; |
| 938 | break; |
| 939 | default : |
| 940 | ;// todo : error |
| 941 | } |
| 942 | loader->stat = GRN_LOADER_UNICODE1; |
| 943 | str++; |
| 944 | break; |
| 945 | case GRN_LOADER_UNICODE1 : |
| 946 | switch (c) { |
| 947 | case '0' : case '1' : case '2' : case '3' : case '4' : |
| 948 | case '5' : case '6' : case '7' : case '8' : case '9' : |
| 949 | loader->unichar += (c - '0') * 0x100; |
| 950 | break; |
| 951 | case 'a' : case 'b' : case 'c' : case 'd' : case 'e' : case 'f' : |
| 952 | loader->unichar += (c - 'a' + 10) * 0x100; |
| 953 | break; |
| 954 | case 'A' : case 'B' : case 'C' : case 'D' : case 'E' : case 'F' : |
| 955 | loader->unichar += (c - 'A' + 10) * 0x100; |
| 956 | break; |
| 957 | default : |
| 958 | ;// todo : error |
| 959 | } |
| 960 | loader->stat = GRN_LOADER_UNICODE2; |
| 961 | str++; |
| 962 | break; |
| 963 | case GRN_LOADER_UNICODE2 : |
| 964 | switch (c) { |
| 965 | case '0' : case '1' : case '2' : case '3' : case '4' : |
| 966 | case '5' : case '6' : case '7' : case '8' : case '9' : |
| 967 | loader->unichar += (c - '0') * 0x10; |
| 968 | break; |
| 969 | case 'a' : case 'b' : case 'c' : case 'd' : case 'e' : case 'f' : |
| 970 | loader->unichar += (c - 'a' + 10) * 0x10; |
| 971 | break; |
| 972 | case 'A' : case 'B' : case 'C' : case 'D' : case 'E' : case 'F' : |
| 973 | loader->unichar += (c - 'A' + 10) * 0x10; |
| 974 | break; |
| 975 | default : |
| 976 | ;// todo : error |
| 977 | } |
| 978 | loader->stat = GRN_LOADER_UNICODE3; |
| 979 | str++; |
| 980 | break; |
| 981 | case GRN_LOADER_UNICODE3 : |
| 982 | switch (c) { |
| 983 | case '0' : case '1' : case '2' : case '3' : case '4' : |
| 984 | case '5' : case '6' : case '7' : case '8' : case '9' : |
| 985 | loader->unichar += (c - '0'); |
| 986 | break; |
| 987 | case 'a' : case 'b' : case 'c' : case 'd' : case 'e' : case 'f' : |
| 988 | loader->unichar += (c - 'a' + 10); |
| 989 | break; |
| 990 | case 'A' : case 'B' : case 'C' : case 'D' : case 'E' : case 'F' : |
| 991 | loader->unichar += (c - 'A' + 10); |
| 992 | break; |
| 993 | default : |
| 994 | ;// todo : error |
| 995 | } |
| 996 | { |
| 997 | uint32_t u = loader->unichar; |
| 998 | if (u < 0x80) { |
| 999 | GRN_TEXT_PUTC(ctx, loader->last, u); |
| 1000 | } else { |
| 1001 | if (u < 0x800) { |
| 1002 | GRN_TEXT_PUTC(ctx, loader->last, ((u >> 6) & 0x1f) | 0xc0); |
| 1003 | } else { |
| 1004 | GRN_TEXT_PUTC(ctx, loader->last, (u >> 12) | 0xe0); |
| 1005 | GRN_TEXT_PUTC(ctx, loader->last, ((u >> 6) & 0x3f) | 0x80); |
| 1006 | } |
| 1007 | GRN_TEXT_PUTC(ctx, loader->last, (u & 0x3f) | 0x80); |
| 1008 | } |
| 1009 | } |
| 1010 | loader->stat = GRN_LOADER_STRING; |
| 1011 | str++; |
| 1012 | break; |
| 1013 | case GRN_LOADER_END : |
| 1014 | str = se; |
| 1015 | break; |
| 1016 | } |
| 1017 | } |
| 1018 | } |
| 1019 | |
| 1020 | #undef JSON_READ_OPEN_BRACKET |
| 1021 | #undef JSON_READ_OPEN_BRACE |
| 1022 | |
| 1023 | /* |
| 1024 | * grn_loader_parse_columns parses a columns parameter. |
| 1025 | * Columns except _id and _key are appended to loader->columns. |
| 1026 | * If it contains _id or _key, loader->id_offset or loader->key_offset is set. |
| 1027 | */ |
| 1028 | static grn_rc |
| 1029 | grn_loader_parse_columns(grn_ctx *ctx, grn_loader *loader, |
| 1030 | const char *str, unsigned int str_size) |
| 1031 | { |
| 1032 | const char *ptr = str, *ptr_end = ptr + str_size, *rest; |
| 1033 | const char *tokens[256], *token_end; |
| 1034 | while (ptr < ptr_end) { |
| 1035 | int i, n = grn_tokenize(ptr, ptr_end - ptr, tokens, 256, &rest); |
| 1036 | for (i = 0; i < n; i++) { |
| 1037 | grn_obj *column; |
| 1038 | token_end = tokens[i]; |
| 1039 | while (ptr < token_end && (' ' == *ptr || ',' == *ptr)) { |
| 1040 | ptr++; |
| 1041 | } |
| 1042 | column = grn_obj_column(ctx, loader->table, ptr, token_end - ptr); |
| 1043 | if (!column) { |
| 1044 | ERR(GRN_INVALID_ARGUMENT, "nonexistent column: <%.*s>" , |
| 1045 | (int)(token_end - ptr), ptr); |
| 1046 | return ctx->rc; |
| 1047 | } |
| 1048 | if (name_equal(ptr, token_end - ptr, GRN_COLUMN_NAME_ID)) { |
| 1049 | grn_obj_unlink(ctx, column); |
| 1050 | if (loader->id_offset != -1 || loader->key_offset != -1) { |
| 1051 | /* _id and _key must not appear more than once. */ |
| 1052 | if (loader->id_offset != -1) { |
| 1053 | ERR(GRN_INVALID_ARGUMENT, |
| 1054 | "duplicated id and key columns: <%s> at %d and <%s> at %d" , |
| 1055 | GRN_COLUMN_NAME_ID, i, |
| 1056 | GRN_COLUMN_NAME_ID, loader->id_offset); |
| 1057 | } else { |
| 1058 | ERR(GRN_INVALID_ARGUMENT, |
| 1059 | "duplicated id and key columns: <%s> at %d and <%s> at %d" , |
| 1060 | GRN_COLUMN_NAME_ID, i, |
| 1061 | GRN_COLUMN_NAME_KEY, loader->key_offset); |
| 1062 | } |
| 1063 | return ctx->rc; |
| 1064 | } |
| 1065 | loader->id_offset = i; |
| 1066 | } else if (name_equal(ptr, token_end - ptr, GRN_COLUMN_NAME_KEY)) { |
| 1067 | grn_obj_unlink(ctx, column); |
| 1068 | if (loader->id_offset != -1 || loader->key_offset != -1) { |
| 1069 | /* _id and _key must not appear more than once. */ |
| 1070 | if (loader->id_offset != -1) { |
| 1071 | ERR(GRN_INVALID_ARGUMENT, |
| 1072 | "duplicated id and key columns: <%s> at %d and <%s> at %d" , |
| 1073 | GRN_COLUMN_NAME_KEY, i, |
| 1074 | GRN_COLUMN_NAME_ID, loader->id_offset); |
| 1075 | } else { |
| 1076 | ERR(GRN_INVALID_ARGUMENT, |
| 1077 | "duplicated id and key columns: <%s> at %d and <%s> at %d" , |
| 1078 | GRN_COLUMN_NAME_KEY, i, |
| 1079 | GRN_COLUMN_NAME_KEY, loader->key_offset); |
| 1080 | } |
| 1081 | return ctx->rc; |
| 1082 | } |
| 1083 | loader->key_offset = i; |
| 1084 | } else { |
| 1085 | GRN_PTR_PUT(ctx, &loader->columns, column); |
| 1086 | } |
| 1087 | ptr = token_end; |
| 1088 | } |
| 1089 | ptr = rest; |
| 1090 | } |
| 1091 | switch (loader->table->header.type) { |
| 1092 | case GRN_TABLE_HASH_KEY : |
| 1093 | case GRN_TABLE_PAT_KEY : |
| 1094 | case GRN_TABLE_DAT_KEY : |
| 1095 | if (loader->id_offset == -1 && loader->key_offset == -1) { |
| 1096 | ERR(GRN_INVALID_ARGUMENT, "missing id or key column" ); |
| 1097 | return ctx->rc; |
| 1098 | } |
| 1099 | break; |
| 1100 | } |
| 1101 | return ctx->rc; |
| 1102 | } |
| 1103 | |
| 1104 | static grn_com_addr *addr; |
| 1105 | |
| 1106 | void |
| 1107 | grn_load_internal(grn_ctx *ctx, grn_load_input *input) |
| 1108 | { |
| 1109 | grn_loader *loader = &ctx->impl->loader; |
| 1110 | |
| 1111 | loader->emit_level = input->emit_level; |
| 1112 | if (ctx->impl->edge) { |
| 1113 | grn_edge *edge = grn_edges_add_communicator(ctx, addr); |
| 1114 | grn_obj *msg = grn_msg_open(ctx, edge->com, &ctx->impl->edge->send_old); |
| 1115 | /* build msg */ |
| 1116 | grn_edge_dispatch(ctx, edge, msg); |
| 1117 | } |
| 1118 | if (input->table.length > 0) { |
| 1119 | grn_ctx_loader_clear(ctx); |
| 1120 | loader->input_type = input->type; |
| 1121 | if (grn_db_check_name(ctx, input->table.value, input->table.length)) { |
| 1122 | GRN_DB_CHECK_NAME_ERR("[table][load]" , |
| 1123 | input->table.value, |
| 1124 | (int)(input->table.length)); |
| 1125 | loader->stat = GRN_LOADER_END; |
| 1126 | return; |
| 1127 | } |
| 1128 | loader->table = grn_ctx_get(ctx, input->table.value, input->table.length); |
| 1129 | if (!loader->table) { |
| 1130 | ERR(GRN_INVALID_ARGUMENT, |
| 1131 | "nonexistent table: <%.*s>" , |
| 1132 | (int)(input->table.length), |
| 1133 | input->table.value); |
| 1134 | loader->stat = GRN_LOADER_END; |
| 1135 | return; |
| 1136 | } |
| 1137 | if (input->columns.length > 0) { |
| 1138 | grn_rc rc = grn_loader_parse_columns(ctx, |
| 1139 | loader, |
| 1140 | input->columns.value, |
| 1141 | input->columns.length); |
| 1142 | if (rc != GRN_SUCCESS) { |
| 1143 | loader->columns_status = GRN_LOADER_COLUMNS_BROKEN; |
| 1144 | loader->stat = GRN_LOADER_END; |
| 1145 | return; |
| 1146 | } |
| 1147 | loader->columns_status = GRN_LOADER_COLUMNS_SET; |
| 1148 | } |
| 1149 | if (input->if_exists.length > 0) { |
| 1150 | grn_obj *v; |
| 1151 | GRN_EXPR_CREATE_FOR_QUERY(ctx, loader->table, loader->ifexists, v); |
| 1152 | if (loader->ifexists && v) { |
| 1153 | grn_expr_parse(ctx, |
| 1154 | loader->ifexists, |
| 1155 | input->if_exists.value, |
| 1156 | input->if_exists.length, |
| 1157 | NULL, GRN_OP_EQUAL, GRN_OP_AND, |
| 1158 | GRN_EXPR_SYNTAX_SCRIPT|GRN_EXPR_ALLOW_UPDATE); |
| 1159 | } |
| 1160 | } |
| 1161 | if (input->each.length > 0) { |
| 1162 | grn_obj *v; |
| 1163 | GRN_EXPR_CREATE_FOR_QUERY(ctx, loader->table, loader->each, v); |
| 1164 | if (loader->each && v) { |
| 1165 | grn_expr_parse(ctx, loader->each, |
| 1166 | input->each.value, |
| 1167 | input->each.length, |
| 1168 | NULL, GRN_OP_EQUAL, GRN_OP_AND, |
| 1169 | GRN_EXPR_SYNTAX_SCRIPT|GRN_EXPR_ALLOW_UPDATE); |
| 1170 | } |
| 1171 | } |
| 1172 | loader->output_ids = input->output_ids; |
| 1173 | loader->output_errors = input->output_errors; |
| 1174 | } else { |
| 1175 | if (!loader->table) { |
| 1176 | ERR(GRN_INVALID_ARGUMENT, "mandatory \"table\" parameter is absent" ); |
| 1177 | loader->stat = GRN_LOADER_END; |
| 1178 | return; |
| 1179 | } |
| 1180 | } |
| 1181 | switch (loader->input_type) { |
| 1182 | case GRN_CONTENT_JSON : |
| 1183 | json_read(ctx, loader, input->values.value, input->values.length); |
| 1184 | break; |
| 1185 | case GRN_CONTENT_NONE : |
| 1186 | case GRN_CONTENT_TSV : |
| 1187 | case GRN_CONTENT_XML : |
| 1188 | case GRN_CONTENT_MSGPACK : |
| 1189 | case GRN_CONTENT_GROONGA_COMMAND_LIST : |
| 1190 | ERR(GRN_FUNCTION_NOT_IMPLEMENTED, "unsupported input_type" ); |
| 1191 | loader->stat = GRN_LOADER_END; |
| 1192 | // todo |
| 1193 | break; |
| 1194 | } |
| 1195 | } |
| 1196 | |
| 1197 | grn_rc |
| 1198 | grn_load(grn_ctx *ctx, grn_content_type input_type, |
| 1199 | const char *table, unsigned int table_len, |
| 1200 | const char *columns, unsigned int columns_len, |
| 1201 | const char *values, unsigned int values_len, |
| 1202 | const char *ifexists, unsigned int ifexists_len, |
| 1203 | const char *each, unsigned int each_len) |
| 1204 | { |
| 1205 | if (!ctx || !ctx->impl) { |
| 1206 | ERR(GRN_INVALID_ARGUMENT, "db not initialized" ); |
| 1207 | return ctx->rc; |
| 1208 | } |
| 1209 | GRN_API_ENTER; |
| 1210 | { |
| 1211 | grn_load_input input; |
| 1212 | input.type = input_type; |
| 1213 | input.table.value = table; |
| 1214 | input.table.length = table_len; |
| 1215 | input.columns.value = columns; |
| 1216 | input.columns.length = columns_len; |
| 1217 | input.values.value = values; |
| 1218 | input.values.length = values_len; |
| 1219 | input.if_exists.value = ifexists; |
| 1220 | input.if_exists.length = ifexists_len; |
| 1221 | input.each.value = each; |
| 1222 | input.each.length = each_len; |
| 1223 | input.output_ids = GRN_FALSE; |
| 1224 | input.output_errors = GRN_FALSE; |
| 1225 | input.emit_level = 1; |
| 1226 | grn_load_internal(ctx, &input); |
| 1227 | } |
| 1228 | GRN_API_RETURN(ctx->rc); |
| 1229 | } |
| 1230 | |