| 1 | #include <errno.h> |
| 2 | #include <cstdlib> |
| 3 | |
| 4 | #include <Poco/String.h> |
| 5 | |
| 6 | #include <IO/ReadHelpers.h> |
| 7 | #include <IO/ReadBufferFromMemory.h> |
| 8 | #include <Common/typeid_cast.h> |
| 9 | #include <Parsers/DumpASTNode.h> |
| 10 | |
| 11 | #include <Parsers/IAST.h> |
| 12 | #include <Parsers/ASTExpressionList.h> |
| 13 | #include <Parsers/ASTFunction.h> |
| 14 | #include <Parsers/ASTIdentifier.h> |
| 15 | #include <Parsers/ASTLiteral.h> |
| 16 | #include <Parsers/ASTAsterisk.h> |
| 17 | #include <Parsers/ASTQualifiedAsterisk.h> |
| 18 | #include <Parsers/ASTQueryParameter.h> |
| 19 | #include <Parsers/ASTTTLElement.h> |
| 20 | #include <Parsers/ASTOrderByElement.h> |
| 21 | #include <Parsers/ASTSubquery.h> |
| 22 | #include <Parsers/ASTFunctionWithKeyValueArguments.h> |
| 23 | |
| 24 | #include <Parsers/parseIntervalKind.h> |
| 25 | #include <Parsers/ExpressionListParsers.h> |
| 26 | #include <Parsers/ParserSelectWithUnionQuery.h> |
| 27 | #include <Parsers/ParserCase.h> |
| 28 | |
| 29 | #include <Parsers/ExpressionElementParsers.h> |
| 30 | #include <Parsers/ParserCreateQuery.h> |
| 31 | |
| 32 | #include <Parsers/queryToString.h> |
| 33 | #include <boost/algorithm/string.hpp> |
| 34 | #include "ASTColumnsMatcher.h" |
| 35 | |
| 36 | |
| 37 | namespace DB |
| 38 | { |
| 39 | |
| 40 | namespace ErrorCodes |
| 41 | { |
| 42 | extern const int SYNTAX_ERROR; |
| 43 | extern const int LOGICAL_ERROR; |
| 44 | } |
| 45 | |
| 46 | |
| 47 | bool ParserArray::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 48 | { |
| 49 | ASTPtr contents_node; |
| 50 | ParserExpressionList contents(false); |
| 51 | |
| 52 | if (pos->type != TokenType::OpeningSquareBracket) |
| 53 | return false; |
| 54 | ++pos; |
| 55 | |
| 56 | if (!contents.parse(pos, contents_node, expected)) |
| 57 | return false; |
| 58 | |
| 59 | if (pos->type != TokenType::ClosingSquareBracket) |
| 60 | return false; |
| 61 | ++pos; |
| 62 | |
| 63 | auto function_node = std::make_shared<ASTFunction>(); |
| 64 | function_node->name = "array" ; |
| 65 | function_node->arguments = contents_node; |
| 66 | function_node->children.push_back(contents_node); |
| 67 | node = function_node; |
| 68 | |
| 69 | return true; |
| 70 | } |
| 71 | |
| 72 | |
| 73 | bool ParserParenthesisExpression::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 74 | { |
| 75 | ASTPtr contents_node; |
| 76 | ParserExpressionList contents(false); |
| 77 | |
| 78 | if (pos->type != TokenType::OpeningRoundBracket) |
| 79 | return false; |
| 80 | ++pos; |
| 81 | |
| 82 | if (!contents.parse(pos, contents_node, expected)) |
| 83 | return false; |
| 84 | |
| 85 | bool is_elem = true; |
| 86 | if (pos->type == TokenType::Comma) |
| 87 | { |
| 88 | is_elem = false; |
| 89 | ++pos; |
| 90 | } |
| 91 | |
| 92 | if (pos->type != TokenType::ClosingRoundBracket) |
| 93 | return false; |
| 94 | ++pos; |
| 95 | |
| 96 | const auto & expr_list = contents_node->as<ASTExpressionList &>(); |
| 97 | |
| 98 | /// empty expression in parentheses is not allowed |
| 99 | if (expr_list.children.empty()) |
| 100 | { |
| 101 | expected.add(pos, "non-empty parenthesized list of expressions" ); |
| 102 | return false; |
| 103 | } |
| 104 | |
| 105 | if (expr_list.children.size() == 1 && is_elem) |
| 106 | { |
| 107 | node = expr_list.children.front(); |
| 108 | } |
| 109 | else |
| 110 | { |
| 111 | auto function_node = std::make_shared<ASTFunction>(); |
| 112 | function_node->name = "tuple" ; |
| 113 | function_node->arguments = contents_node; |
| 114 | function_node->children.push_back(contents_node); |
| 115 | node = function_node; |
| 116 | } |
| 117 | |
| 118 | return true; |
| 119 | } |
| 120 | |
| 121 | |
| 122 | bool ParserSubquery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 123 | { |
| 124 | ASTPtr select_node; |
| 125 | ParserSelectWithUnionQuery select; |
| 126 | |
| 127 | if (pos->type != TokenType::OpeningRoundBracket) |
| 128 | return false; |
| 129 | ++pos; |
| 130 | |
| 131 | if (!select.parse(pos, select_node, expected)) |
| 132 | return false; |
| 133 | |
| 134 | if (pos->type != TokenType::ClosingRoundBracket) |
| 135 | return false; |
| 136 | ++pos; |
| 137 | |
| 138 | node = std::make_shared<ASTSubquery>(); |
| 139 | node->children.push_back(select_node); |
| 140 | return true; |
| 141 | } |
| 142 | |
| 143 | |
| 144 | bool ParserIdentifier::parseImpl(Pos & pos, ASTPtr & node, Expected &) |
| 145 | { |
| 146 | /// Identifier in backquotes or in double quotes |
| 147 | if (pos->type == TokenType::QuotedIdentifier) |
| 148 | { |
| 149 | ReadBufferFromMemory buf(pos->begin, pos->size()); |
| 150 | String s; |
| 151 | |
| 152 | if (*pos->begin == '`') |
| 153 | readBackQuotedStringWithSQLStyle(s, buf); |
| 154 | else |
| 155 | readDoubleQuotedStringWithSQLStyle(s, buf); |
| 156 | |
| 157 | if (s.empty()) /// Identifiers "empty string" are not allowed. |
| 158 | return false; |
| 159 | |
| 160 | node = std::make_shared<ASTIdentifier>(s); |
| 161 | ++pos; |
| 162 | return true; |
| 163 | } |
| 164 | else if (pos->type == TokenType::BareWord) |
| 165 | { |
| 166 | node = std::make_shared<ASTIdentifier>(String(pos->begin, pos->end)); |
| 167 | ++pos; |
| 168 | return true; |
| 169 | } |
| 170 | |
| 171 | return false; |
| 172 | } |
| 173 | |
| 174 | |
| 175 | bool ParserCompoundIdentifier::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 176 | { |
| 177 | ASTPtr id_list; |
| 178 | if (!ParserList(std::make_unique<ParserIdentifier>(), std::make_unique<ParserToken>(TokenType::Dot), false) |
| 179 | .parse(pos, id_list, expected)) |
| 180 | return false; |
| 181 | |
| 182 | String name; |
| 183 | std::vector<String> parts; |
| 184 | const auto & list = id_list->as<ASTExpressionList &>(); |
| 185 | for (const auto & child : list.children) |
| 186 | { |
| 187 | if (!name.empty()) |
| 188 | name += '.'; |
| 189 | parts.emplace_back(getIdentifierName(child)); |
| 190 | name += parts.back(); |
| 191 | } |
| 192 | |
| 193 | if (parts.size() == 1) |
| 194 | parts.clear(); |
| 195 | node = std::make_shared<ASTIdentifier>(name, std::move(parts)); |
| 196 | |
| 197 | return true; |
| 198 | } |
| 199 | |
| 200 | |
| 201 | bool ParserFunction::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 202 | { |
| 203 | ParserIdentifier id_parser; |
| 204 | ParserKeyword distinct("DISTINCT" ); |
| 205 | ParserExpressionList contents(false); |
| 206 | |
| 207 | bool has_distinct_modifier = false; |
| 208 | |
| 209 | ASTPtr identifier; |
| 210 | ASTPtr expr_list_args; |
| 211 | ASTPtr expr_list_params; |
| 212 | |
| 213 | if (!id_parser.parse(pos, identifier, expected)) |
| 214 | return false; |
| 215 | |
| 216 | if (pos->type != TokenType::OpeningRoundBracket) |
| 217 | return false; |
| 218 | ++pos; |
| 219 | |
| 220 | if (distinct.ignore(pos, expected)) |
| 221 | has_distinct_modifier = true; |
| 222 | |
| 223 | const char * contents_begin = pos->begin; |
| 224 | if (!contents.parse(pos, expr_list_args, expected)) |
| 225 | return false; |
| 226 | const char * contents_end = pos->begin; |
| 227 | |
| 228 | if (pos->type != TokenType::ClosingRoundBracket) |
| 229 | return false; |
| 230 | ++pos; |
| 231 | |
| 232 | /** Check for a common error case - often due to the complexity of quoting command-line arguments, |
| 233 | * an expression of the form toDate(2014-01-01) appears in the query instead of toDate('2014-01-01'). |
| 234 | * If you do not report that the first option is an error, then the argument will be interpreted as 2014 - 01 - 01 - some number, |
| 235 | * and the query silently returns an unexpected result. |
| 236 | */ |
| 237 | if (getIdentifierName(identifier) == "toDate" |
| 238 | && contents_end - contents_begin == strlen("2014-01-01" ) |
| 239 | && contents_begin[0] >= '2' && contents_begin[0] <= '3' |
| 240 | && contents_begin[1] >= '0' && contents_begin[1] <= '9' |
| 241 | && contents_begin[2] >= '0' && contents_begin[2] <= '9' |
| 242 | && contents_begin[3] >= '0' && contents_begin[3] <= '9' |
| 243 | && contents_begin[4] == '-' |
| 244 | && contents_begin[5] >= '0' && contents_begin[5] <= '9' |
| 245 | && contents_begin[6] >= '0' && contents_begin[6] <= '9' |
| 246 | && contents_begin[7] == '-' |
| 247 | && contents_begin[8] >= '0' && contents_begin[8] <= '9' |
| 248 | && contents_begin[9] >= '0' && contents_begin[9] <= '9') |
| 249 | { |
| 250 | std::string contents_str(contents_begin, contents_end - contents_begin); |
| 251 | throw Exception("Argument of function toDate is unquoted: toDate(" + contents_str + "), must be: toDate('" + contents_str + "')" |
| 252 | , ErrorCodes::SYNTAX_ERROR); |
| 253 | } |
| 254 | |
| 255 | /// The parametric aggregate function has two lists (parameters and arguments) in parentheses. Example: quantile(0.9)(x). |
| 256 | if (allow_function_parameters && pos->type == TokenType::OpeningRoundBracket) |
| 257 | { |
| 258 | ++pos; |
| 259 | |
| 260 | /// Parametric aggregate functions cannot have DISTINCT in parameters list. |
| 261 | if (has_distinct_modifier) |
| 262 | return false; |
| 263 | |
| 264 | expr_list_params = expr_list_args; |
| 265 | expr_list_args = nullptr; |
| 266 | |
| 267 | if (distinct.ignore(pos, expected)) |
| 268 | has_distinct_modifier = true; |
| 269 | |
| 270 | if (!contents.parse(pos, expr_list_args, expected)) |
| 271 | return false; |
| 272 | |
| 273 | if (pos->type != TokenType::ClosingRoundBracket) |
| 274 | return false; |
| 275 | ++pos; |
| 276 | } |
| 277 | |
| 278 | auto function_node = std::make_shared<ASTFunction>(); |
| 279 | tryGetIdentifierNameInto(identifier, function_node->name); |
| 280 | |
| 281 | /// func(DISTINCT ...) is equivalent to funcDistinct(...) |
| 282 | if (has_distinct_modifier) |
| 283 | function_node->name += "Distinct" ; |
| 284 | |
| 285 | function_node->arguments = expr_list_args; |
| 286 | function_node->children.push_back(function_node->arguments); |
| 287 | |
| 288 | if (expr_list_params) |
| 289 | { |
| 290 | function_node->parameters = expr_list_params; |
| 291 | function_node->children.push_back(function_node->parameters); |
| 292 | } |
| 293 | |
| 294 | node = function_node; |
| 295 | return true; |
| 296 | } |
| 297 | |
| 298 | bool ParserCodecDeclarationList::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 299 | { |
| 300 | return ParserList(std::make_unique<ParserIdentifierWithOptionalParameters>(), |
| 301 | std::make_unique<ParserToken>(TokenType::Comma), false).parse(pos, node, expected); |
| 302 | } |
| 303 | |
| 304 | bool ParserCodec::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 305 | { |
| 306 | ParserCodecDeclarationList codecs; |
| 307 | ASTPtr expr_list_args; |
| 308 | |
| 309 | if (pos->type != TokenType::OpeningRoundBracket) |
| 310 | return false; |
| 311 | |
| 312 | ++pos; |
| 313 | if (!codecs.parse(pos, expr_list_args, expected)) |
| 314 | return false; |
| 315 | |
| 316 | if (pos->type != TokenType::ClosingRoundBracket) |
| 317 | return false; |
| 318 | ++pos; |
| 319 | |
| 320 | auto function_node = std::make_shared<ASTFunction>(); |
| 321 | function_node->name = "CODEC" ; |
| 322 | function_node->arguments = expr_list_args; |
| 323 | function_node->children.push_back(function_node->arguments); |
| 324 | |
| 325 | node = function_node; |
| 326 | return true; |
| 327 | } |
| 328 | |
| 329 | bool ParserCastExpression::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 330 | { |
| 331 | /// Either CAST(expr AS type) or CAST(expr, 'type') |
| 332 | /// The latter will be parsed normally as a function later. |
| 333 | |
| 334 | ASTPtr expr_node; |
| 335 | ASTPtr type_node; |
| 336 | |
| 337 | if (ParserKeyword("CAST" ).ignore(pos, expected) |
| 338 | && ParserToken(TokenType::OpeningRoundBracket).ignore(pos, expected) |
| 339 | && ParserExpression().parse(pos, expr_node, expected) |
| 340 | && ParserKeyword("AS" ).ignore(pos, expected) |
| 341 | && ParserIdentifierWithOptionalParameters().parse(pos, type_node, expected) |
| 342 | && ParserToken(TokenType::ClosingRoundBracket).ignore(pos, expected)) |
| 343 | { |
| 344 | /// Convert to canonical representation in functional form: CAST(expr, 'type') |
| 345 | |
| 346 | auto type_literal = std::make_shared<ASTLiteral>(queryToString(type_node)); |
| 347 | |
| 348 | auto expr_list_args = std::make_shared<ASTExpressionList>(); |
| 349 | expr_list_args->children.push_back(expr_node); |
| 350 | expr_list_args->children.push_back(std::move(type_literal)); |
| 351 | |
| 352 | auto func_node = std::make_shared<ASTFunction>(); |
| 353 | func_node->name = "CAST" ; |
| 354 | func_node->arguments = std::move(expr_list_args); |
| 355 | func_node->children.push_back(func_node->arguments); |
| 356 | |
| 357 | node = std::move(func_node); |
| 358 | return true; |
| 359 | } |
| 360 | |
| 361 | return false; |
| 362 | } |
| 363 | |
| 364 | bool ParserSubstringExpression::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 365 | { |
| 366 | /// Either SUBSTRING(expr FROM start) or SUBSTRING(expr FROM start FOR length) or SUBSTRING(expr, start, length) |
| 367 | /// The latter will be parsed normally as a function later. |
| 368 | |
| 369 | ASTPtr expr_node; |
| 370 | ASTPtr start_node; |
| 371 | ASTPtr length_node; |
| 372 | |
| 373 | if (!ParserKeyword("SUBSTRING" ).ignore(pos, expected)) |
| 374 | return false; |
| 375 | |
| 376 | if (pos->type != TokenType::OpeningRoundBracket) |
| 377 | return false; |
| 378 | ++pos; |
| 379 | |
| 380 | if (!ParserExpression().parse(pos, expr_node, expected)) |
| 381 | return false; |
| 382 | |
| 383 | if (pos->type != TokenType::Comma) |
| 384 | { |
| 385 | if (!ParserKeyword("FROM" ).ignore(pos, expected)) |
| 386 | return false; |
| 387 | } |
| 388 | else |
| 389 | { |
| 390 | ++pos; |
| 391 | } |
| 392 | |
| 393 | if (!ParserExpression().parse(pos, start_node, expected)) |
| 394 | return false; |
| 395 | |
| 396 | if (pos->type == TokenType::ClosingRoundBracket) |
| 397 | { |
| 398 | ++pos; |
| 399 | } |
| 400 | else |
| 401 | { |
| 402 | if (pos->type != TokenType::Comma) |
| 403 | { |
| 404 | if (!ParserKeyword("FOR" ).ignore(pos, expected)) |
| 405 | return false; |
| 406 | } |
| 407 | else |
| 408 | { |
| 409 | ++pos; |
| 410 | } |
| 411 | |
| 412 | if (!ParserExpression().parse(pos, length_node, expected)) |
| 413 | return false; |
| 414 | |
| 415 | ParserToken(TokenType::ClosingRoundBracket).ignore(pos, expected); |
| 416 | } |
| 417 | |
| 418 | /// Convert to canonical representation in functional form: SUBSTRING(expr, start, length) |
| 419 | |
| 420 | auto expr_list_args = std::make_shared<ASTExpressionList>(); |
| 421 | expr_list_args->children = {expr_node, start_node}; |
| 422 | |
| 423 | if (length_node) |
| 424 | expr_list_args->children.push_back(length_node); |
| 425 | |
| 426 | auto func_node = std::make_shared<ASTFunction>(); |
| 427 | func_node->name = "substring" ; |
| 428 | func_node->arguments = std::move(expr_list_args); |
| 429 | func_node->children.push_back(func_node->arguments); |
| 430 | |
| 431 | node = std::move(func_node); |
| 432 | return true; |
| 433 | } |
| 434 | |
| 435 | bool ParserTrimExpression::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 436 | { |
| 437 | /// Handles all possible TRIM/LTRIM/RTRIM call variants |
| 438 | |
| 439 | std::string func_name; |
| 440 | bool trim_left = false; |
| 441 | bool trim_right = false; |
| 442 | bool char_override = false; |
| 443 | ASTPtr expr_node; |
| 444 | ASTPtr pattern_node; |
| 445 | ASTPtr to_remove; |
| 446 | |
| 447 | if (ParserKeyword("LTRIM" ).ignore(pos, expected)) |
| 448 | { |
| 449 | if (pos->type != TokenType::OpeningRoundBracket) |
| 450 | return false; |
| 451 | ++pos; |
| 452 | trim_left = true; |
| 453 | } |
| 454 | else if (ParserKeyword("RTRIM" ).ignore(pos, expected)) |
| 455 | { |
| 456 | if (pos->type != TokenType::OpeningRoundBracket) |
| 457 | return false; |
| 458 | ++pos; |
| 459 | trim_right = true; |
| 460 | } |
| 461 | else if (ParserKeyword("TRIM" ).ignore(pos, expected)) |
| 462 | { |
| 463 | if (pos->type != TokenType::OpeningRoundBracket) |
| 464 | return false; |
| 465 | ++pos; |
| 466 | |
| 467 | if (ParserKeyword("BOTH" ).ignore(pos, expected)) |
| 468 | { |
| 469 | trim_left = true; |
| 470 | trim_right = true; |
| 471 | char_override = true; |
| 472 | } |
| 473 | else if (ParserKeyword("LEADING" ).ignore(pos, expected)) |
| 474 | { |
| 475 | trim_left = true; |
| 476 | char_override = true; |
| 477 | } |
| 478 | else if (ParserKeyword("TRAILING" ).ignore(pos, expected)) |
| 479 | { |
| 480 | trim_right = true; |
| 481 | char_override = true; |
| 482 | } |
| 483 | else |
| 484 | { |
| 485 | trim_left = true; |
| 486 | trim_right = true; |
| 487 | } |
| 488 | |
| 489 | if (char_override) |
| 490 | { |
| 491 | if (!ParserExpression().parse(pos, to_remove, expected)) |
| 492 | return false; |
| 493 | if (!ParserKeyword("FROM" ).ignore(pos, expected)) |
| 494 | return false; |
| 495 | |
| 496 | auto quote_meta_func_node = std::make_shared<ASTFunction>(); |
| 497 | auto quote_meta_list_args = std::make_shared<ASTExpressionList>(); |
| 498 | quote_meta_list_args->children = {to_remove}; |
| 499 | |
| 500 | quote_meta_func_node->name = "regexpQuoteMeta" ; |
| 501 | quote_meta_func_node->arguments = std::move(quote_meta_list_args); |
| 502 | quote_meta_func_node->children.push_back(quote_meta_func_node->arguments); |
| 503 | |
| 504 | to_remove = std::move(quote_meta_func_node); |
| 505 | } |
| 506 | } |
| 507 | |
| 508 | if (!(trim_left || trim_right)) |
| 509 | return false; |
| 510 | |
| 511 | if (!ParserExpression().parse(pos, expr_node, expected)) |
| 512 | return false; |
| 513 | |
| 514 | if (pos->type != TokenType::ClosingRoundBracket) |
| 515 | return false; |
| 516 | ++pos; |
| 517 | |
| 518 | /// Convert to regexp replace function call |
| 519 | |
| 520 | if (char_override) |
| 521 | { |
| 522 | auto pattern_func_node = std::make_shared<ASTFunction>(); |
| 523 | auto pattern_list_args = std::make_shared<ASTExpressionList>(); |
| 524 | if (trim_left && trim_right) |
| 525 | { |
| 526 | pattern_list_args->children = { |
| 527 | std::make_shared<ASTLiteral>("^[" ), |
| 528 | to_remove, |
| 529 | std::make_shared<ASTLiteral>("]*|[" ), |
| 530 | to_remove, |
| 531 | std::make_shared<ASTLiteral>("]*$" ) |
| 532 | }; |
| 533 | func_name = "replaceRegexpAll" ; |
| 534 | } |
| 535 | else |
| 536 | { |
| 537 | if (trim_left) |
| 538 | { |
| 539 | pattern_list_args->children = { |
| 540 | std::make_shared<ASTLiteral>("^[" ), |
| 541 | to_remove, |
| 542 | std::make_shared<ASTLiteral>("]*" ) |
| 543 | }; |
| 544 | } |
| 545 | else |
| 546 | { |
| 547 | /// trim_right == false not possible |
| 548 | pattern_list_args->children = { |
| 549 | std::make_shared<ASTLiteral>("[" ), |
| 550 | to_remove, |
| 551 | std::make_shared<ASTLiteral>("]*$" ) |
| 552 | }; |
| 553 | } |
| 554 | func_name = "replaceRegexpOne" ; |
| 555 | } |
| 556 | |
| 557 | pattern_func_node->name = "concat" ; |
| 558 | pattern_func_node->arguments = std::move(pattern_list_args); |
| 559 | pattern_func_node->children.push_back(pattern_func_node->arguments); |
| 560 | |
| 561 | pattern_node = std::move(pattern_func_node); |
| 562 | } |
| 563 | else |
| 564 | { |
| 565 | if (trim_left && trim_right) |
| 566 | { |
| 567 | func_name = "trimBoth" ; |
| 568 | } |
| 569 | else |
| 570 | { |
| 571 | if (trim_left) |
| 572 | { |
| 573 | func_name = "trimLeft" ; |
| 574 | } |
| 575 | else |
| 576 | { |
| 577 | /// trim_right == false not possible |
| 578 | func_name = "trimRight" ; |
| 579 | } |
| 580 | } |
| 581 | } |
| 582 | |
| 583 | auto expr_list_args = std::make_shared<ASTExpressionList>(); |
| 584 | if (char_override) |
| 585 | expr_list_args->children = {expr_node, pattern_node, std::make_shared<ASTLiteral>("" )}; |
| 586 | else |
| 587 | expr_list_args->children = {expr_node}; |
| 588 | |
| 589 | auto func_node = std::make_shared<ASTFunction>(); |
| 590 | func_node->name = func_name; |
| 591 | func_node->arguments = std::move(expr_list_args); |
| 592 | func_node->children.push_back(func_node->arguments); |
| 593 | |
| 594 | node = std::move(func_node); |
| 595 | return true; |
| 596 | } |
| 597 | |
| 598 | bool ParserLeftExpression::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 599 | { |
| 600 | /// Rewrites left(expr, length) to SUBSTRING(expr, 1, length) |
| 601 | |
| 602 | ASTPtr expr_node; |
| 603 | ASTPtr start_node; |
| 604 | ASTPtr length_node; |
| 605 | |
| 606 | if (!ParserKeyword("LEFT" ).ignore(pos, expected)) |
| 607 | return false; |
| 608 | |
| 609 | if (pos->type != TokenType::OpeningRoundBracket) |
| 610 | return false; |
| 611 | ++pos; |
| 612 | |
| 613 | if (!ParserExpression().parse(pos, expr_node, expected)) |
| 614 | return false; |
| 615 | |
| 616 | ParserToken(TokenType::Comma).ignore(pos, expected); |
| 617 | |
| 618 | if (!ParserExpression().parse(pos, length_node, expected)) |
| 619 | return false; |
| 620 | |
| 621 | if (pos->type != TokenType::ClosingRoundBracket) |
| 622 | return false; |
| 623 | ++pos; |
| 624 | |
| 625 | auto expr_list_args = std::make_shared<ASTExpressionList>(); |
| 626 | start_node = std::make_shared<ASTLiteral>(1); |
| 627 | expr_list_args->children = {expr_node, start_node, length_node}; |
| 628 | |
| 629 | auto func_node = std::make_shared<ASTFunction>(); |
| 630 | func_node->name = "substring" ; |
| 631 | func_node->arguments = std::move(expr_list_args); |
| 632 | func_node->children.push_back(func_node->arguments); |
| 633 | |
| 634 | node = std::move(func_node); |
| 635 | return true; |
| 636 | } |
| 637 | |
| 638 | bool ParserRightExpression::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 639 | { |
| 640 | /// Rewrites RIGHT(expr, length) to substring(expr, -length) |
| 641 | |
| 642 | ASTPtr expr_node; |
| 643 | ASTPtr length_node; |
| 644 | |
| 645 | if (!ParserKeyword("RIGHT" ).ignore(pos, expected)) |
| 646 | return false; |
| 647 | |
| 648 | if (pos->type != TokenType::OpeningRoundBracket) |
| 649 | return false; |
| 650 | ++pos; |
| 651 | |
| 652 | if (!ParserExpression().parse(pos, expr_node, expected)) |
| 653 | return false; |
| 654 | |
| 655 | ParserToken(TokenType::Comma).ignore(pos, expected); |
| 656 | |
| 657 | if (!ParserExpression().parse(pos, length_node, expected)) |
| 658 | return false; |
| 659 | |
| 660 | if (pos->type != TokenType::ClosingRoundBracket) |
| 661 | return false; |
| 662 | ++pos; |
| 663 | |
| 664 | auto start_expr_list_args = std::make_shared<ASTExpressionList>(); |
| 665 | start_expr_list_args->children = {length_node}; |
| 666 | |
| 667 | auto start_node = std::make_shared<ASTFunction>(); |
| 668 | start_node->name = "negate" ; |
| 669 | start_node->arguments = std::move(start_expr_list_args); |
| 670 | start_node->children.push_back(start_node->arguments); |
| 671 | |
| 672 | auto expr_list_args = std::make_shared<ASTExpressionList>(); |
| 673 | expr_list_args->children = {expr_node, start_node}; |
| 674 | |
| 675 | auto func_node = std::make_shared<ASTFunction>(); |
| 676 | func_node->name = "substring" ; |
| 677 | func_node->arguments = std::move(expr_list_args); |
| 678 | func_node->children.push_back(func_node->arguments); |
| 679 | |
| 680 | node = std::move(func_node); |
| 681 | return true; |
| 682 | } |
| 683 | |
| 684 | bool ParserExtractExpression::(Pos & pos, ASTPtr & node, Expected & expected) |
| 685 | { |
| 686 | if (!ParserKeyword("EXTRACT" ).ignore(pos, expected)) |
| 687 | return false; |
| 688 | |
| 689 | if (pos->type != TokenType::OpeningRoundBracket) |
| 690 | return false; |
| 691 | ++pos; |
| 692 | |
| 693 | ASTPtr expr; |
| 694 | |
| 695 | IntervalKind interval_kind; |
| 696 | if (!parseIntervalKind(pos, expected, interval_kind)) |
| 697 | return false; |
| 698 | |
| 699 | ParserKeyword s_from("FROM" ); |
| 700 | if (!s_from.ignore(pos, expected)) |
| 701 | return false; |
| 702 | |
| 703 | ParserExpression elem_parser; |
| 704 | if (!elem_parser.parse(pos, expr, expected)) |
| 705 | return false; |
| 706 | |
| 707 | if (pos->type != TokenType::ClosingRoundBracket) |
| 708 | return false; |
| 709 | ++pos; |
| 710 | |
| 711 | auto function = std::make_shared<ASTFunction>(); |
| 712 | auto exp_list = std::make_shared<ASTExpressionList>(); |
| 713 | function->name = interval_kind.toNameOfFunctionExtractTimePart(); |
| 714 | function->arguments = exp_list; |
| 715 | function->children.push_back(exp_list); |
| 716 | exp_list->children.push_back(expr); |
| 717 | node = function; |
| 718 | |
| 719 | return true; |
| 720 | } |
| 721 | |
| 722 | bool ParserDateAddExpression::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 723 | { |
| 724 | const char * function_name = nullptr; |
| 725 | ASTPtr timestamp_node; |
| 726 | ASTPtr offset_node; |
| 727 | |
| 728 | if (ParserKeyword("DATEADD" ).ignore(pos, expected) || ParserKeyword("DATE_ADD" ).ignore(pos, expected) |
| 729 | || ParserKeyword("TIMESTAMPADD" ).ignore(pos, expected) || ParserKeyword("TIMESTAMP_ADD" ).ignore(pos, expected)) |
| 730 | function_name = "plus" ; |
| 731 | else if (ParserKeyword("DATESUB" ).ignore(pos, expected) || ParserKeyword("DATE_SUB" ).ignore(pos, expected) |
| 732 | || ParserKeyword("TIMESTAMPSUB" ).ignore(pos, expected) || ParserKeyword("TIMESTAMP_SUB" ).ignore(pos, expected)) |
| 733 | function_name = "minus" ; |
| 734 | else |
| 735 | return false; |
| 736 | |
| 737 | if (pos->type != TokenType::OpeningRoundBracket) |
| 738 | return false; |
| 739 | ++pos; |
| 740 | |
| 741 | IntervalKind interval_kind; |
| 742 | if (parseIntervalKind(pos, expected, interval_kind)) |
| 743 | { |
| 744 | /// function(unit, offset, timestamp) |
| 745 | if (pos->type != TokenType::Comma) |
| 746 | return false; |
| 747 | ++pos; |
| 748 | |
| 749 | if (!ParserExpression().parse(pos, offset_node, expected)) |
| 750 | return false; |
| 751 | |
| 752 | if (pos->type != TokenType::Comma) |
| 753 | return false; |
| 754 | ++pos; |
| 755 | |
| 756 | if (!ParserExpression().parse(pos, timestamp_node, expected)) |
| 757 | return false; |
| 758 | } |
| 759 | else |
| 760 | { |
| 761 | /// function(timestamp, INTERVAL offset unit) |
| 762 | if (!ParserExpression().parse(pos, timestamp_node, expected)) |
| 763 | return false; |
| 764 | |
| 765 | if (pos->type != TokenType::Comma) |
| 766 | return false; |
| 767 | ++pos; |
| 768 | |
| 769 | if (!ParserKeyword("INTERVAL" ).ignore(pos, expected)) |
| 770 | return false; |
| 771 | |
| 772 | if (!ParserExpression().parse(pos, offset_node, expected)) |
| 773 | return false; |
| 774 | |
| 775 | if (!parseIntervalKind(pos, expected, interval_kind)) |
| 776 | return false; |
| 777 | } |
| 778 | if (pos->type != TokenType::ClosingRoundBracket) |
| 779 | return false; |
| 780 | ++pos; |
| 781 | |
| 782 | auto interval_expr_list_args = std::make_shared<ASTExpressionList>(); |
| 783 | interval_expr_list_args->children = {offset_node}; |
| 784 | |
| 785 | auto interval_func_node = std::make_shared<ASTFunction>(); |
| 786 | interval_func_node->name = interval_kind.toNameOfFunctionToIntervalDataType(); |
| 787 | interval_func_node->arguments = std::move(interval_expr_list_args); |
| 788 | interval_func_node->children.push_back(interval_func_node->arguments); |
| 789 | |
| 790 | auto expr_list_args = std::make_shared<ASTExpressionList>(); |
| 791 | expr_list_args->children = {timestamp_node, interval_func_node}; |
| 792 | |
| 793 | auto func_node = std::make_shared<ASTFunction>(); |
| 794 | func_node->name = function_name; |
| 795 | func_node->arguments = std::move(expr_list_args); |
| 796 | func_node->children.push_back(func_node->arguments); |
| 797 | |
| 798 | node = std::move(func_node); |
| 799 | |
| 800 | return true; |
| 801 | } |
| 802 | |
| 803 | bool ParserDateDiffExpression::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 804 | { |
| 805 | ASTPtr left_node; |
| 806 | ASTPtr right_node; |
| 807 | |
| 808 | if (!(ParserKeyword("DATEDIFF" ).ignore(pos, expected) || ParserKeyword("DATE_DIFF" ).ignore(pos, expected) |
| 809 | || ParserKeyword("TIMESTAMPDIFF" ).ignore(pos, expected) || ParserKeyword("TIMESTAMP_DIFF" ).ignore(pos, expected))) |
| 810 | return false; |
| 811 | |
| 812 | if (pos->type != TokenType::OpeningRoundBracket) |
| 813 | return false; |
| 814 | ++pos; |
| 815 | |
| 816 | IntervalKind interval_kind; |
| 817 | if (!parseIntervalKind(pos, expected, interval_kind)) |
| 818 | return false; |
| 819 | |
| 820 | if (pos->type != TokenType::Comma) |
| 821 | return false; |
| 822 | ++pos; |
| 823 | |
| 824 | if (!ParserExpression().parse(pos, left_node, expected)) |
| 825 | return false; |
| 826 | |
| 827 | if (pos->type != TokenType::Comma) |
| 828 | return false; |
| 829 | ++pos; |
| 830 | |
| 831 | if (!ParserExpression().parse(pos, right_node, expected)) |
| 832 | return false; |
| 833 | |
| 834 | if (pos->type != TokenType::ClosingRoundBracket) |
| 835 | return false; |
| 836 | ++pos; |
| 837 | |
| 838 | auto expr_list_args = std::make_shared<ASTExpressionList>(); |
| 839 | expr_list_args->children = {std::make_shared<ASTLiteral>(interval_kind.toDateDiffUnit()), left_node, right_node}; |
| 840 | |
| 841 | auto func_node = std::make_shared<ASTFunction>(); |
| 842 | func_node->name = "dateDiff" ; |
| 843 | func_node->arguments = std::move(expr_list_args); |
| 844 | func_node->children.push_back(func_node->arguments); |
| 845 | |
| 846 | node = std::move(func_node); |
| 847 | |
| 848 | return true; |
| 849 | } |
| 850 | |
| 851 | |
| 852 | bool ParserNull::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 853 | { |
| 854 | ParserKeyword nested_parser("NULL" ); |
| 855 | if (nested_parser.parse(pos, node, expected)) |
| 856 | { |
| 857 | node = std::make_shared<ASTLiteral>(Null()); |
| 858 | return true; |
| 859 | } |
| 860 | else |
| 861 | return false; |
| 862 | } |
| 863 | |
| 864 | |
| 865 | bool ParserNumber::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 866 | { |
| 867 | Pos literal_begin = pos; |
| 868 | bool negative = false; |
| 869 | |
| 870 | if (pos->type == TokenType::Minus) |
| 871 | { |
| 872 | ++pos; |
| 873 | negative = true; |
| 874 | } |
| 875 | else if (pos->type == TokenType::Plus) /// Leading plus is simply ignored. |
| 876 | ++pos; |
| 877 | |
| 878 | Field res; |
| 879 | |
| 880 | if (!pos.isValid()) |
| 881 | return false; |
| 882 | |
| 883 | /** Maximum length of number. 319 symbols is enough to write maximum double in decimal form. |
| 884 | * Copy is needed to use strto* functions, which require 0-terminated string. |
| 885 | */ |
| 886 | static constexpr size_t MAX_LENGTH_OF_NUMBER = 319; |
| 887 | |
| 888 | if (pos->size() > MAX_LENGTH_OF_NUMBER) |
| 889 | { |
| 890 | expected.add(pos, "number" ); |
| 891 | return false; |
| 892 | } |
| 893 | |
| 894 | char buf[MAX_LENGTH_OF_NUMBER + 1]; |
| 895 | |
| 896 | memcpy(buf, pos->begin, pos->size()); |
| 897 | buf[pos->size()] = 0; |
| 898 | |
| 899 | char * pos_double = buf; |
| 900 | errno = 0; /// Functions strto* don't clear errno. |
| 901 | Float64 float_value = std::strtod(buf, &pos_double); |
| 902 | if (pos_double != buf + pos->size() || errno == ERANGE) |
| 903 | { |
| 904 | expected.add(pos, "number" ); |
| 905 | return false; |
| 906 | } |
| 907 | |
| 908 | if (float_value < 0) |
| 909 | throw Exception("Logical error: token number cannot begin with minus, but parsed float number is less than zero." , ErrorCodes::LOGICAL_ERROR); |
| 910 | |
| 911 | if (negative) |
| 912 | float_value = -float_value; |
| 913 | |
| 914 | res = float_value; |
| 915 | |
| 916 | /// try to use more exact type: UInt64 |
| 917 | |
| 918 | char * pos_integer = buf; |
| 919 | |
| 920 | errno = 0; |
| 921 | UInt64 uint_value = std::strtoull(buf, &pos_integer, 0); |
| 922 | if (pos_integer == pos_double && errno != ERANGE && (!negative || uint_value <= (1ULL << 63))) |
| 923 | { |
| 924 | if (negative) |
| 925 | res = static_cast<Int64>(-uint_value); |
| 926 | else |
| 927 | res = uint_value; |
| 928 | } |
| 929 | |
| 930 | auto literal = std::make_shared<ASTLiteral>(res); |
| 931 | literal->begin = literal_begin; |
| 932 | literal->end = ++pos; |
| 933 | node = literal; |
| 934 | return true; |
| 935 | } |
| 936 | |
| 937 | |
| 938 | bool ParserUnsignedInteger::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 939 | { |
| 940 | Field res; |
| 941 | |
| 942 | if (!pos.isValid()) |
| 943 | return false; |
| 944 | |
| 945 | UInt64 x = 0; |
| 946 | ReadBufferFromMemory in(pos->begin, pos->size()); |
| 947 | if (!tryReadIntText(x, in) || in.count() != pos->size()) |
| 948 | { |
| 949 | expected.add(pos, "unsigned integer" ); |
| 950 | return false; |
| 951 | } |
| 952 | |
| 953 | res = x; |
| 954 | auto literal = std::make_shared<ASTLiteral>(res); |
| 955 | literal->begin = pos; |
| 956 | literal->end = ++pos; |
| 957 | node = literal; |
| 958 | return true; |
| 959 | } |
| 960 | |
| 961 | |
| 962 | bool ParserStringLiteral::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 963 | { |
| 964 | if (pos->type != TokenType::StringLiteral) |
| 965 | return false; |
| 966 | |
| 967 | String s; |
| 968 | ReadBufferFromMemory in(pos->begin, pos->size()); |
| 969 | |
| 970 | try |
| 971 | { |
| 972 | readQuotedStringWithSQLStyle(s, in); |
| 973 | } |
| 974 | catch (const Exception &) |
| 975 | { |
| 976 | expected.add(pos, "string literal" ); |
| 977 | return false; |
| 978 | } |
| 979 | |
| 980 | if (in.count() != pos->size()) |
| 981 | { |
| 982 | expected.add(pos, "string literal" ); |
| 983 | return false; |
| 984 | } |
| 985 | |
| 986 | auto literal = std::make_shared<ASTLiteral>(s); |
| 987 | literal->begin = pos; |
| 988 | literal->end = ++pos; |
| 989 | node = literal; |
| 990 | return true; |
| 991 | } |
| 992 | |
| 993 | |
| 994 | bool ParserArrayOfLiterals::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 995 | { |
| 996 | if (pos->type != TokenType::OpeningSquareBracket) |
| 997 | return false; |
| 998 | |
| 999 | Pos literal_begin = pos; |
| 1000 | |
| 1001 | Array arr; |
| 1002 | |
| 1003 | ParserLiteral literal_p; |
| 1004 | |
| 1005 | ++pos; |
| 1006 | |
| 1007 | while (pos.isValid()) |
| 1008 | { |
| 1009 | if (!arr.empty()) |
| 1010 | { |
| 1011 | if (pos->type == TokenType::ClosingSquareBracket) |
| 1012 | { |
| 1013 | auto literal = std::make_shared<ASTLiteral>(arr); |
| 1014 | literal->begin = literal_begin; |
| 1015 | literal->end = ++pos; |
| 1016 | node = literal; |
| 1017 | return true; |
| 1018 | } |
| 1019 | else if (pos->type == TokenType::Comma) |
| 1020 | { |
| 1021 | ++pos; |
| 1022 | } |
| 1023 | else |
| 1024 | { |
| 1025 | expected.add(pos, "comma or closing square bracket" ); |
| 1026 | return false; |
| 1027 | } |
| 1028 | } |
| 1029 | |
| 1030 | ASTPtr literal_node; |
| 1031 | if (!literal_p.parse(pos, literal_node, expected)) |
| 1032 | return false; |
| 1033 | |
| 1034 | arr.push_back(literal_node->as<ASTLiteral &>().value); |
| 1035 | } |
| 1036 | |
| 1037 | expected.add(pos, "closing square bracket" ); |
| 1038 | return false; |
| 1039 | } |
| 1040 | |
| 1041 | |
| 1042 | bool ParserLiteral::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 1043 | { |
| 1044 | ParserNull null_p; |
| 1045 | ParserNumber num_p; |
| 1046 | ParserStringLiteral str_p; |
| 1047 | |
| 1048 | if (null_p.parse(pos, node, expected)) |
| 1049 | return true; |
| 1050 | |
| 1051 | if (num_p.parse(pos, node, expected)) |
| 1052 | return true; |
| 1053 | |
| 1054 | if (str_p.parse(pos, node, expected)) |
| 1055 | return true; |
| 1056 | |
| 1057 | return false; |
| 1058 | } |
| 1059 | |
| 1060 | |
| 1061 | const char * ParserAlias::restricted_keywords[] = |
| 1062 | { |
| 1063 | "FROM" , |
| 1064 | "FINAL" , |
| 1065 | "SAMPLE" , |
| 1066 | "ARRAY" , |
| 1067 | "LEFT" , |
| 1068 | "RIGHT" , |
| 1069 | "INNER" , |
| 1070 | "FULL" , |
| 1071 | "CROSS" , |
| 1072 | "JOIN" , |
| 1073 | "GLOBAL" , |
| 1074 | "ANY" , |
| 1075 | "ALL" , |
| 1076 | "ASOF" , |
| 1077 | "SEMI" , |
| 1078 | "ANTI" , |
| 1079 | "ONLY" , /// YQL synonym for ANTI |
| 1080 | "ON" , |
| 1081 | "USING" , |
| 1082 | "PREWHERE" , |
| 1083 | "WHERE" , |
| 1084 | "GROUP" , |
| 1085 | "WITH" , |
| 1086 | "HAVING" , |
| 1087 | "ORDER" , |
| 1088 | "LIMIT" , |
| 1089 | "SETTINGS" , |
| 1090 | "FORMAT" , |
| 1091 | "UNION" , |
| 1092 | "INTO" , |
| 1093 | "NOT" , |
| 1094 | "BETWEEN" , |
| 1095 | "LIKE" , |
| 1096 | nullptr |
| 1097 | }; |
| 1098 | |
| 1099 | bool ParserAlias::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 1100 | { |
| 1101 | ParserKeyword s_as("AS" ); |
| 1102 | ParserIdentifier id_p; |
| 1103 | |
| 1104 | bool has_as_word = s_as.ignore(pos, expected); |
| 1105 | if (!allow_alias_without_as_keyword && !has_as_word) |
| 1106 | return false; |
| 1107 | |
| 1108 | if (!id_p.parse(pos, node, expected)) |
| 1109 | return false; |
| 1110 | |
| 1111 | if (!has_as_word) |
| 1112 | { |
| 1113 | /** In this case, the alias can not match the keyword - |
| 1114 | * so that in the query "SELECT x FROM t", the word FROM was not considered an alias, |
| 1115 | * and in the query "SELECT x FRO FROM t", the word FRO was considered an alias. |
| 1116 | */ |
| 1117 | |
| 1118 | const String name = getIdentifierName(node); |
| 1119 | |
| 1120 | for (const char ** keyword = restricted_keywords; *keyword != nullptr; ++keyword) |
| 1121 | if (0 == strcasecmp(name.data(), *keyword)) |
| 1122 | return false; |
| 1123 | } |
| 1124 | |
| 1125 | return true; |
| 1126 | } |
| 1127 | |
| 1128 | |
| 1129 | bool ParserColumnsMatcher::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 1130 | { |
| 1131 | ParserKeyword columns("COLUMNS" ); |
| 1132 | ParserStringLiteral regex; |
| 1133 | |
| 1134 | if (!columns.ignore(pos, expected)) |
| 1135 | return false; |
| 1136 | |
| 1137 | if (pos->type != TokenType::OpeningRoundBracket) |
| 1138 | return false; |
| 1139 | ++pos; |
| 1140 | |
| 1141 | ASTPtr regex_node; |
| 1142 | if (!regex.parse(pos, regex_node, expected)) |
| 1143 | return false; |
| 1144 | |
| 1145 | if (pos->type != TokenType::ClosingRoundBracket) |
| 1146 | return false; |
| 1147 | ++pos; |
| 1148 | |
| 1149 | auto res = std::make_shared<ASTColumnsMatcher>(); |
| 1150 | res->setPattern(regex_node->as<ASTLiteral &>().value.get<String>()); |
| 1151 | res->children.push_back(regex_node); |
| 1152 | node = std::move(res); |
| 1153 | return true; |
| 1154 | } |
| 1155 | |
| 1156 | |
| 1157 | bool ParserAsterisk::parseImpl(Pos & pos, ASTPtr & node, Expected &) |
| 1158 | { |
| 1159 | if (pos->type == TokenType::Asterisk) |
| 1160 | { |
| 1161 | ++pos; |
| 1162 | node = std::make_shared<ASTAsterisk>(); |
| 1163 | return true; |
| 1164 | } |
| 1165 | return false; |
| 1166 | } |
| 1167 | |
| 1168 | |
| 1169 | bool ParserQualifiedAsterisk::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 1170 | { |
| 1171 | if (!ParserCompoundIdentifier().parse(pos, node, expected)) |
| 1172 | return false; |
| 1173 | |
| 1174 | if (pos->type != TokenType::Dot) |
| 1175 | return false; |
| 1176 | ++pos; |
| 1177 | |
| 1178 | if (pos->type != TokenType::Asterisk) |
| 1179 | return false; |
| 1180 | ++pos; |
| 1181 | |
| 1182 | auto res = std::make_shared<ASTQualifiedAsterisk>(); |
| 1183 | res->children.push_back(node); |
| 1184 | node = std::move(res); |
| 1185 | return true; |
| 1186 | } |
| 1187 | |
| 1188 | |
| 1189 | bool ParserSubstitution::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 1190 | { |
| 1191 | if (pos->type != TokenType::OpeningCurlyBrace) |
| 1192 | return false; |
| 1193 | |
| 1194 | ++pos; |
| 1195 | |
| 1196 | if (pos->type != TokenType::BareWord) |
| 1197 | { |
| 1198 | expected.add(pos, "substitution name (identifier)" ); |
| 1199 | return false; |
| 1200 | } |
| 1201 | |
| 1202 | String name(pos->begin, pos->end); |
| 1203 | ++pos; |
| 1204 | |
| 1205 | if (pos->type != TokenType::Colon) |
| 1206 | { |
| 1207 | expected.add(pos, "colon between name and type" ); |
| 1208 | return false; |
| 1209 | } |
| 1210 | |
| 1211 | ++pos; |
| 1212 | |
| 1213 | auto old_pos = pos; |
| 1214 | ParserIdentifierWithOptionalParameters type_parser; |
| 1215 | if (!type_parser.ignore(pos, expected)) |
| 1216 | { |
| 1217 | expected.add(pos, "substitution type" ); |
| 1218 | return false; |
| 1219 | } |
| 1220 | |
| 1221 | String type(old_pos->begin, pos->begin); |
| 1222 | |
| 1223 | if (pos->type != TokenType::ClosingCurlyBrace) |
| 1224 | { |
| 1225 | expected.add(pos, "closing curly brace" ); |
| 1226 | return false; |
| 1227 | } |
| 1228 | |
| 1229 | ++pos; |
| 1230 | node = std::make_shared<ASTQueryParameter>(name, type); |
| 1231 | return true; |
| 1232 | } |
| 1233 | |
| 1234 | |
| 1235 | bool ParserExpressionElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 1236 | { |
| 1237 | return ParserSubquery().parse(pos, node, expected) |
| 1238 | || ParserParenthesisExpression().parse(pos, node, expected) |
| 1239 | || ParserArrayOfLiterals().parse(pos, node, expected) |
| 1240 | || ParserArray().parse(pos, node, expected) |
| 1241 | || ParserLiteral().parse(pos, node, expected) |
| 1242 | || ParserCastExpression().parse(pos, node, expected) |
| 1243 | || ParserExtractExpression().parse(pos, node, expected) |
| 1244 | || ParserDateAddExpression().parse(pos, node, expected) |
| 1245 | || ParserDateDiffExpression().parse(pos, node, expected) |
| 1246 | || ParserSubstringExpression().parse(pos, node, expected) |
| 1247 | || ParserTrimExpression().parse(pos, node, expected) |
| 1248 | || ParserLeftExpression().parse(pos, node, expected) |
| 1249 | || ParserRightExpression().parse(pos, node, expected) |
| 1250 | || ParserCase().parse(pos, node, expected) |
| 1251 | || ParserColumnsMatcher().parse(pos, node, expected) /// before ParserFunction because it can be also parsed as a function. |
| 1252 | || ParserFunction().parse(pos, node, expected) |
| 1253 | || ParserQualifiedAsterisk().parse(pos, node, expected) |
| 1254 | || ParserAsterisk().parse(pos, node, expected) |
| 1255 | || ParserCompoundIdentifier().parse(pos, node, expected) |
| 1256 | || ParserSubstitution().parse(pos, node, expected); |
| 1257 | } |
| 1258 | |
| 1259 | |
| 1260 | bool ParserWithOptionalAlias::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 1261 | { |
| 1262 | if (!elem_parser->parse(pos, node, expected)) |
| 1263 | return false; |
| 1264 | |
| 1265 | /** Little hack. |
| 1266 | * |
| 1267 | * In the SELECT section, we allow parsing aliases without specifying the AS keyword. |
| 1268 | * These aliases can not be the same as the query keywords. |
| 1269 | * And the expression itself can be an identifier that matches the keyword. |
| 1270 | * For example, a column may be called where. And in the query it can be written `SELECT where AS x FROM table` or even `SELECT where x FROM table`. |
| 1271 | * Even can be written `SELECT where AS from FROM table`, but it can not be written `SELECT where from FROM table`. |
| 1272 | * See the ParserAlias implementation for details. |
| 1273 | * |
| 1274 | * But there is a small problem - an inconvenient error message if there is an extra comma in the SELECT section at the end. |
| 1275 | * Although this error is very common. Example: `SELECT x, y, z, FROM tbl` |
| 1276 | * If you do nothing, it's parsed as a column with the name FROM and alias tbl. |
| 1277 | * To avoid this situation, we do not allow the parsing of the alias without the AS keyword for the identifier with the name FROM. |
| 1278 | * |
| 1279 | * Note: this also filters the case when the identifier is quoted. |
| 1280 | * Example: SELECT x, y, z, `FROM` tbl. But such a case could be solved. |
| 1281 | * |
| 1282 | * In the future it would be easier to disallow unquoted identifiers that match the keywords. |
| 1283 | */ |
| 1284 | bool allow_alias_without_as_keyword_now = allow_alias_without_as_keyword; |
| 1285 | if (allow_alias_without_as_keyword) |
| 1286 | if (auto opt_id = tryGetIdentifierName(node)) |
| 1287 | if (0 == strcasecmp(opt_id->data(), "FROM" )) |
| 1288 | allow_alias_without_as_keyword_now = false; |
| 1289 | |
| 1290 | ASTPtr alias_node; |
| 1291 | if (ParserAlias(allow_alias_without_as_keyword_now).parse(pos, alias_node, expected)) |
| 1292 | { |
| 1293 | /// FIXME: try to prettify this cast using `as<>()` |
| 1294 | if (auto * ast_with_alias = dynamic_cast<ASTWithAlias *>(node.get())) |
| 1295 | { |
| 1296 | tryGetIdentifierNameInto(alias_node, ast_with_alias->alias); |
| 1297 | } |
| 1298 | else |
| 1299 | { |
| 1300 | expected.add(pos, "alias cannot be here" ); |
| 1301 | return false; |
| 1302 | } |
| 1303 | } |
| 1304 | |
| 1305 | return true; |
| 1306 | } |
| 1307 | |
| 1308 | |
| 1309 | bool ParserOrderByElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 1310 | { |
| 1311 | ParserExpressionWithOptionalAlias elem_p(false); |
| 1312 | ParserKeyword ascending("ASCENDING" ); |
| 1313 | ParserKeyword descending("DESCENDING" ); |
| 1314 | ParserKeyword asc("ASC" ); |
| 1315 | ParserKeyword desc("DESC" ); |
| 1316 | ParserKeyword nulls("NULLS" ); |
| 1317 | ParserKeyword first("FIRST" ); |
| 1318 | ParserKeyword last("LAST" ); |
| 1319 | ParserKeyword collate("COLLATE" ); |
| 1320 | ParserKeyword with_fill("WITH FILL" ); |
| 1321 | ParserKeyword from("FROM" ); |
| 1322 | ParserKeyword to("TO" ); |
| 1323 | ParserKeyword step("STEP" ); |
| 1324 | ParserStringLiteral collate_locale_parser; |
| 1325 | ParserExpressionWithOptionalAlias exp_parser(false); |
| 1326 | |
| 1327 | ASTPtr expr_elem; |
| 1328 | if (!elem_p.parse(pos, expr_elem, expected)) |
| 1329 | return false; |
| 1330 | |
| 1331 | int direction = 1; |
| 1332 | |
| 1333 | if (descending.ignore(pos) || desc.ignore(pos)) |
| 1334 | direction = -1; |
| 1335 | else |
| 1336 | ascending.ignore(pos) || asc.ignore(pos); |
| 1337 | |
| 1338 | int nulls_direction = direction; |
| 1339 | bool nulls_direction_was_explicitly_specified = false; |
| 1340 | |
| 1341 | if (nulls.ignore(pos)) |
| 1342 | { |
| 1343 | nulls_direction_was_explicitly_specified = true; |
| 1344 | |
| 1345 | if (first.ignore(pos)) |
| 1346 | nulls_direction = -direction; |
| 1347 | else if (last.ignore(pos)) |
| 1348 | ; |
| 1349 | else |
| 1350 | return false; |
| 1351 | } |
| 1352 | |
| 1353 | ASTPtr locale_node; |
| 1354 | if (collate.ignore(pos)) |
| 1355 | { |
| 1356 | if (!collate_locale_parser.parse(pos, locale_node, expected)) |
| 1357 | return false; |
| 1358 | } |
| 1359 | |
| 1360 | /// WITH FILL [FROM x] [TO y] [STEP z] |
| 1361 | bool has_with_fill = false; |
| 1362 | ASTPtr fill_from; |
| 1363 | ASTPtr fill_to; |
| 1364 | ASTPtr fill_step; |
| 1365 | if (with_fill.ignore(pos)) |
| 1366 | { |
| 1367 | has_with_fill = true; |
| 1368 | if (from.ignore(pos) && !exp_parser.parse(pos, fill_from, expected)) |
| 1369 | return false; |
| 1370 | |
| 1371 | if (to.ignore(pos) && !exp_parser.parse(pos, fill_to, expected)) |
| 1372 | return false; |
| 1373 | |
| 1374 | if (step.ignore(pos) && !exp_parser.parse(pos, fill_step, expected)) |
| 1375 | return false; |
| 1376 | } |
| 1377 | |
| 1378 | node = std::make_shared<ASTOrderByElement>( |
| 1379 | direction, nulls_direction, nulls_direction_was_explicitly_specified, locale_node, |
| 1380 | has_with_fill, fill_from, fill_to, fill_step); |
| 1381 | node->children.push_back(expr_elem); |
| 1382 | if (locale_node) |
| 1383 | node->children.push_back(locale_node); |
| 1384 | |
| 1385 | return true; |
| 1386 | } |
| 1387 | |
| 1388 | bool ParserFunctionWithKeyValueArguments::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 1389 | { |
| 1390 | ParserIdentifier id_parser; |
| 1391 | ParserKeyValuePairsList pairs_list_parser; |
| 1392 | |
| 1393 | ASTPtr identifier; |
| 1394 | ASTPtr expr_list_args; |
| 1395 | if (!id_parser.parse(pos, identifier, expected)) |
| 1396 | return false; |
| 1397 | |
| 1398 | if (pos.get().type != TokenType::OpeningRoundBracket) |
| 1399 | return false; |
| 1400 | |
| 1401 | ++pos; |
| 1402 | if (!pairs_list_parser.parse(pos, expr_list_args, expected)) |
| 1403 | return false; |
| 1404 | |
| 1405 | if (pos.get().type != TokenType::ClosingRoundBracket) |
| 1406 | return false; |
| 1407 | |
| 1408 | ++pos; |
| 1409 | auto function = std::make_shared<ASTFunctionWithKeyValueArguments>(); |
| 1410 | function->name = Poco::toLower(typeid_cast<ASTIdentifier &>(*identifier.get()).name); |
| 1411 | function->elements = expr_list_args; |
| 1412 | function->children.push_back(function->elements); |
| 1413 | node = function; |
| 1414 | |
| 1415 | return true; |
| 1416 | } |
| 1417 | |
| 1418 | bool ParserTTLElement::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 1419 | { |
| 1420 | ParserKeyword s_to_disk("TO DISK" ); |
| 1421 | ParserKeyword s_to_volume("TO VOLUME" ); |
| 1422 | ParserKeyword s_delete("DELETE" ); |
| 1423 | ParserStringLiteral parser_string_literal; |
| 1424 | ParserExpression parser_exp; |
| 1425 | |
| 1426 | ASTPtr expr_elem; |
| 1427 | if (!parser_exp.parse(pos, expr_elem, expected)) |
| 1428 | return false; |
| 1429 | |
| 1430 | PartDestinationType destination_type = PartDestinationType::DELETE; |
| 1431 | String destination_name; |
| 1432 | if (s_to_disk.ignore(pos)) |
| 1433 | destination_type = PartDestinationType::DISK; |
| 1434 | else if (s_to_volume.ignore(pos)) |
| 1435 | destination_type = PartDestinationType::VOLUME; |
| 1436 | else |
| 1437 | s_delete.ignore(pos); |
| 1438 | |
| 1439 | if (destination_type == PartDestinationType::DISK || destination_type == PartDestinationType::VOLUME) |
| 1440 | { |
| 1441 | ASTPtr ast_space_name; |
| 1442 | if (!parser_string_literal.parse(pos, ast_space_name, expected)) |
| 1443 | return false; |
| 1444 | |
| 1445 | destination_name = ast_space_name->as<ASTLiteral &>().value.get<const String &>(); |
| 1446 | } |
| 1447 | |
| 1448 | node = std::make_shared<ASTTTLElement>(destination_type, destination_name); |
| 1449 | node->children.push_back(expr_elem); |
| 1450 | |
| 1451 | return true; |
| 1452 | } |
| 1453 | |
| 1454 | bool ParserIdentifierWithOptionalParameters::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) |
| 1455 | { |
| 1456 | ParserIdentifier non_parametric; |
| 1457 | ParserIdentifierWithParameters parametric; |
| 1458 | |
| 1459 | if (parametric.parse(pos, node, expected)) |
| 1460 | return true; |
| 1461 | |
| 1462 | ASTPtr ident; |
| 1463 | if (non_parametric.parse(pos, ident, expected)) |
| 1464 | { |
| 1465 | auto func = std::make_shared<ASTFunction>(); |
| 1466 | tryGetIdentifierNameInto(ident, func->name); |
| 1467 | node = func; |
| 1468 | return true; |
| 1469 | } |
| 1470 | |
| 1471 | return false; |
| 1472 | } |
| 1473 | |
| 1474 | } |
| 1475 | |