src/lib/lttng-ctl/filter-parser.y

   1 %{
   2 /*
   3  * filter-parser.y
   4  *
   5  * LTTng filter expression parser
   6  *
   7  * Copyright 2012 - Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
   8  *
   9  * This library is free software; you can redistribute it and/or modify it
  10  * under the terms of the GNU Lesser General Public License, version 2.1 only,
  11  * as published by the Free Software Foundation.
  12  *
  13  * This library is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  16  * Lesser General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU Lesser General Public License
  19  * along with this library; if not, write to the Free Software Foundation,
  20  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  21  *
  22  * Grammar inspired from http://www.quut.com/c/ANSI-C-grammar-y.html
  23  */
  24
  25 #include <stdio.h>
  26 #include <unistd.h>
  27 #include <string.h>
  28 #include <stdlib.h>
  29 #include <assert.h>
  30 #include <errno.h>
  31 #include <inttypes.h>
  32 #include "filter-parser.h"
  33 #include "filter-ast.h"
  34
  35 int yydebug;
  36 int filter_parser_debug = 0;
  37
  38 int yyparse(struct filter_parser_ctx *parser_ctx);
  39 int yylex(union YYSTYPE *yyval, struct filter_parser_ctx *parser_ctx);
  40 int yylex_init_extra(struct filter_parser_ctx *parser_ctx, yyscan_t * ptr_yy_globals);
  41 int yylex_destroy(yyscan_t yyparser_ctx);
  42 void yyrestart(FILE * in_str, yyscan_t parser_ctx);
  43
  44 struct gc_string {
  45         struct cds_list_head gc;
  46         size_t alloclen;
  47         char s[];
  48 };
  49
  50 static const char *node_type_to_str[] = {
  51         [ NODE_UNKNOWN ] = "NODE_UNKNOWN",
  52         [ NODE_ROOT ] = "NODE_ROOT",
  53         [ NODE_EXPRESSION ] = "NODE_EXPRESSION",
  54         [ NODE_OP ] = "NODE_OP",
  55         [ NODE_UNARY_OP ] = "NODE_UNARY_OP",
  56 };
  57
  58 const char *node_type(struct filter_node *node)
  59 {
  60         if (node->type < NR_NODE_TYPES)
  61                 return node_type_to_str[node->type];
  62         else
  63                 return NULL;
  64 }
  65
  66 static struct gc_string *gc_string_alloc(struct filter_parser_ctx *parser_ctx,
  67                                          size_t len)
  68 {
  69         struct gc_string *gstr;
  70         size_t alloclen;
  71
  72         /* TODO: could be faster with find first bit or glib Gstring */
  73         /* sizeof long to account for malloc header (int or long ?) */
  74         for (alloclen = 8; alloclen < sizeof(long) + sizeof(*gstr) + len;
  75              alloclen *= 2);
  76
  77         gstr = malloc(alloclen);
  78         cds_list_add(&gstr->gc, &parser_ctx->allocated_strings);
  79         gstr->alloclen = alloclen;
  80         return gstr;
  81 }
  82
  83 /*
  84  * note: never use gc_string_append on a string that has external references.
  85  * gsrc will be garbage collected immediately, and gstr might be.
  86  * Should only be used to append characters to a string literal or constant.
  87  */
  88 struct gc_string *gc_string_append(struct filter_parser_ctx *parser_ctx,
  89                                    struct gc_string *gstr,
  90                                    struct gc_string *gsrc)
  91 {
  92         size_t newlen = strlen(gsrc->s) + strlen(gstr->s) + 1;
  93         size_t alloclen;
  94
  95         /* TODO: could be faster with find first bit or glib Gstring */
  96         /* sizeof long to account for malloc header (int or long ?) */
  97         for (alloclen = 8; alloclen < sizeof(long) + sizeof(*gstr) + newlen;
  98              alloclen *= 2);
  99
 100         if (alloclen > gstr->alloclen) {
 101                 struct gc_string *newgstr;
 102
 103                 newgstr = gc_string_alloc(parser_ctx, newlen);
 104                 strcpy(newgstr->s, gstr->s);
 105                 strcat(newgstr->s, gsrc->s);
 106                 cds_list_del(&gstr->gc);
 107                 free(gstr);
 108                 gstr = newgstr;
 109         } else {
 110                 strcat(gstr->s, gsrc->s);
 111         }
 112         cds_list_del(&gsrc->gc);
 113         free(gsrc);
 114         return gstr;
 115 }
 116
 117 void setstring(struct filter_parser_ctx *parser_ctx, YYSTYPE *lvalp, const char *src)
 118 {
 119         lvalp->gs = gc_string_alloc(parser_ctx, strlen(src) + 1);
 120         strcpy(lvalp->gs->s, src);
 121 }
 122
 123 static struct filter_node *make_node(struct filter_parser_ctx *scanner,
 124                                   enum node_type type)
 125 {
 126         struct filter_ast *ast = filter_parser_get_ast(scanner);
 127         struct filter_node *node;
 128
 129         node = malloc(sizeof(*node));
 130         if (!node)
 131                 return NULL;
 132         memset(node, 0, sizeof(*node));
 133         node->type = type;
 134         cds_list_add(&node->gc, &ast->allocated_nodes);
 135
 136         switch (type) {
 137         case NODE_ROOT:
 138                 fprintf(stderr, "[error] %s: trying to create root node\n", __func__);
 139                 break;
 140
 141         case NODE_EXPRESSION:
 142                 break;
 143         case NODE_OP:
 144                 break;
 145         case NODE_UNARY_OP:
 146                 break;
 147
 148         case NODE_UNKNOWN:
 149         default:
 150                 fprintf(stderr, "[error] %s: unknown node type %d\n", __func__,
 151                         (int) type);
 152                 break;
 153         }
 154
 155         return node;
 156 }
 157
 158 static struct filter_node *make_op_node(struct filter_parser_ctx *scanner,
 159                         enum op_type type,
 160                         struct filter_node *lchild,
 161                         struct filter_node *rchild)
 162 {
 163         struct filter_ast *ast = filter_parser_get_ast(scanner);
 164         struct filter_node *node;
 165
 166         node = malloc(sizeof(*node));
 167         if (!node)
 168                 return NULL;
 169         memset(node, 0, sizeof(*node));
 170         node->type = NODE_OP;
 171         cds_list_add(&node->gc, &ast->allocated_nodes);
 172         node->u.op.type = type;
 173         node->u.op.lchild = lchild;
 174         node->u.op.rchild = rchild;
 175         return node;
 176 }
 177
 178 void yyerror(struct filter_parser_ctx *parser_ctx, const char *str)
 179 {
 180         fprintf(stderr, "error %s\n", str);
 181 }
 182
 183 int yywrap(void)
 184 {
 185         return 1;
 186 }
 187
 188 #define parse_error(parser_ctx, str)                            \
 189 do {                                                            \
 190         yyerror(parser_ctx, YY_("parse error: " str "\n"));     \
 191         YYERROR;                                                \
 192 } while (0)
 193
 194 static void free_strings(struct cds_list_head *list)
 195 {
 196         struct gc_string *gstr, *tmp;
 197
 198         cds_list_for_each_entry_safe(gstr, tmp, list, gc)
 199                 free(gstr);
 200 }
 201
 202 static struct filter_ast *filter_ast_alloc(void)
 203 {
 204         struct filter_ast *ast;
 205
 206         ast = malloc(sizeof(*ast));
 207         if (!ast)
 208                 return NULL;
 209         memset(ast, 0, sizeof(*ast));
 210         CDS_INIT_LIST_HEAD(&ast->allocated_nodes);
 211         ast->root.type = NODE_ROOT;
 212         return ast;
 213 }
 214
 215 static void filter_ast_free(struct filter_ast *ast)
 216 {
 217         struct filter_node *node, *tmp;
 218
 219         cds_list_for_each_entry_safe(node, tmp, &ast->allocated_nodes, gc)
 220                 free(node);
 221 }
 222
 223 int filter_parser_ctx_append_ast(struct filter_parser_ctx *parser_ctx)
 224 {
 225         return yyparse(parser_ctx);
 226 }
 227
 228 struct filter_parser_ctx *filter_parser_ctx_alloc(FILE *input)
 229 {
 230         struct filter_parser_ctx *parser_ctx;
 231         int ret;
 232
 233         yydebug = filter_parser_debug;
 234
 235         parser_ctx = malloc(sizeof(*parser_ctx));
 236         if (!parser_ctx)
 237                 return NULL;
 238         memset(parser_ctx, 0, sizeof(*parser_ctx));
 239
 240         ret = yylex_init_extra(parser_ctx, &parser_ctx->scanner);
 241         if (ret) {
 242                 fprintf(stderr, "yylex_init error\n");
 243                 goto cleanup_parser_ctx;
 244         }
 245         /* Start processing new stream */
 246         yyrestart(input, parser_ctx->scanner);
 247
 248         parser_ctx->ast = filter_ast_alloc();
 249         if (!parser_ctx->ast)
 250                 goto cleanup_lexer;
 251         CDS_INIT_LIST_HEAD(&parser_ctx->allocated_strings);
 252
 253         if (yydebug)
 254                 fprintf(stdout, "parser_ctx input is a%s.\n",
 255                         isatty(fileno(input)) ? "n interactive tty" :
 256                                                 " noninteractive file");
 257
 258         return parser_ctx;
 259
 260 cleanup_lexer:
 261         ret = yylex_destroy(parser_ctx->scanner);
 262         if (!ret)
 263                 fprintf(stderr, "yylex_destroy error\n");
 264 cleanup_parser_ctx:
 265         free(parser_ctx);
 266         return NULL;
 267 }
 268
 269 void filter_parser_ctx_free(struct filter_parser_ctx *parser_ctx)
 270 {
 271         int ret;
 272
 273         free_strings(&parser_ctx->allocated_strings);
 274         filter_ast_free(parser_ctx->ast);
 275         ret = yylex_destroy(parser_ctx->scanner);
 276         if (ret)
 277                 fprintf(stderr, "yylex_destroy error\n");
 278         free(parser_ctx);
 279 }
 280
 281 %}
 282
 283 %define api.pure
 284         /* %locations */
 285 %parse-param {struct filter_parser_ctx *parser_ctx}
 286 %lex-param {struct filter_parser_ctx *parser_ctx}
 287 %start translation_unit
 288 %token CHARACTER_CONSTANT_START SQUOTE STRING_LITERAL_START DQUOTE
 289 %token ESCSEQ CHAR_STRING_TOKEN
 290 %token DECIMAL_CONSTANT OCTAL_CONSTANT HEXADECIMAL_CONSTANT
 291 %token LSBRAC RSBRAC LPAREN RPAREN LBRAC RBRAC RARROW
 292 %token STAR PLUS MINUS
 293 %token MOD_OP DIV_OP RIGHT_OP LEFT_OP
 294 %token EQ_OP NE_OP LE_OP GE_OP LT_OP GT_OP AND_OP OR_OP NOT_OP
 295 %token ASSIGN COLON SEMICOLON DOTDOTDOT DOT EQUAL COMMA
 296 %token XOR_BIN AND_BIN OR_BIN NOT_BIN
 297
 298 %token <gs> IDENTIFIER
 299 %token ERROR
 300 %union
 301 {
 302         long long ll;
 303         char c;
 304         struct gc_string *gs;
 305         struct filter_node *n;
 306 }
 307
 308 %type <gs> s_char s_char_sequence c_char c_char_sequence
 309
 310 %type <n> primary_expression
 311 %type <n> postfix_expression
 312 %type <n> unary_expression
 313 %type <n> unary_operator
 314 %type <n> multiplicative_expression
 315 %type <n> additive_expression
 316 %type <n> shift_expression
 317 %type <n> relational_expression
 318 %type <n> equality_expression
 319 %type <n> and_expression
 320 %type <n> exclusive_or_expression
 321 %type <n> inclusive_or_expression
 322 %type <n> logical_and_expression
 323 %type <n> logical_or_expression
 324 %type <n> expression
 325
 326 %%
 327
 328
 329 /* 1.5 Constants */
 330
 331 c_char_sequence:
 332                 c_char
 333                 {       $$ = $1;                                        }
 334         |       c_char_sequence c_char
 335                 {       $$ = gc_string_append(parser_ctx, $1, $2);              }
 336         ;
 337
 338 c_char:
 339                 CHAR_STRING_TOKEN
 340                 {       $$ = yylval.gs;                                 }
 341         |       ESCSEQ
 342                 {
 343                         parse_error(parser_ctx, "escape sequences not supported yet");
 344                 }
 345         ;
 346
 347 /* 1.6 String literals */
 348
 349 s_char_sequence:
 350                 s_char
 351                 {       $$ = $1;                                        }
 352         |       s_char_sequence s_char
 353                 {       $$ = gc_string_append(parser_ctx, $1, $2);              }
 354         ;
 355
 356 s_char:
 357                 CHAR_STRING_TOKEN
 358                 {       $$ = yylval.gs;                                 }
 359         |       ESCSEQ
 360                 {
 361                         parse_error(parser_ctx, "escape sequences not supported yet");
 362                 }
 363         ;
 364
 365 primary_expression
 366         :       IDENTIFIER
 367                 {
 368                         $$ = make_node(parser_ctx, NODE_EXPRESSION);
 369                         $$->u.expression.type = AST_EXP_IDENTIFIER;
 370                         $$->u.expression.u.identifier = yylval.gs->s;
 371                 }
 372         |       DECIMAL_CONSTANT
 373                 {
 374                         $$ = make_node(parser_ctx, NODE_EXPRESSION);
 375                         $$->u.expression.type = AST_EXP_CONSTANT;
 376                         sscanf(yylval.gs->s, "%" PRIu64,
 377                                &$$->u.expression.u.constant);
 378                 }
 379         |       OCTAL_CONSTANT
 380                 {
 381                         $$ = make_node(parser_ctx, NODE_EXPRESSION);
 382                         $$->u.expression.type = AST_EXP_CONSTANT;
 383                         sscanf(yylval.gs->s, "0%" PRIo64,
 384                                &$$->u.expression.u.constant);
 385                 }
 386         |       HEXADECIMAL_CONSTANT
 387                 {
 388                         $$ = make_node(parser_ctx, NODE_EXPRESSION);
 389                         $$->u.expression.type = AST_EXP_CONSTANT;
 390                         sscanf(yylval.gs->s, "0x%" PRIx64,
 391                                &$$->u.expression.u.constant);
 392                 }
 393         |       STRING_LITERAL_START DQUOTE
 394                 {
 395                         $$ = make_node(parser_ctx, NODE_EXPRESSION);
 396                         $$->u.expression.type = AST_EXP_STRING;
 397                         $$->u.expression.u.string = "";
 398                 }
 399         |       STRING_LITERAL_START s_char_sequence DQUOTE
 400                 {
 401                         $$ = make_node(parser_ctx, NODE_EXPRESSION);
 402                         $$->u.expression.type = AST_EXP_STRING;
 403                         $$->u.expression.u.string = $2->s;
 404                 }
 405         |       CHARACTER_CONSTANT_START c_char_sequence SQUOTE
 406                 {
 407                         $$ = make_node(parser_ctx, NODE_EXPRESSION);
 408                         $$->u.expression.type = AST_EXP_STRING;
 409                         $$->u.expression.u.string = $2->s;
 410                 }
 411         |       LPAREN expression RPAREN
 412                 {
 413                         $$ = make_node(parser_ctx, NODE_EXPRESSION);
 414                         $$->u.expression.type = AST_EXP_NESTED;
 415                         $$->u.expression.u.child = $2;
 416                 }
 417         ;
 418
 419 postfix_expression
 420         : primary_expression
 421                 {       $$ = $1;                                        }
 422         | postfix_expression DOT IDENTIFIER
 423                 {
 424                         $$ = make_node(parser_ctx, NODE_EXPRESSION);
 425                         $$->u.expression.type = AST_EXP_IDENTIFIER;
 426                         $$->u.expression.post_op = AST_LINK_DOT;
 427                         $$->u.expression.u.identifier = $3->s;
 428                         $$->u.expression.prev = $1;
 429                 }
 430         | postfix_expression RARROW IDENTIFIER
 431                 {
 432                         $$ = make_node(parser_ctx, NODE_EXPRESSION);
 433                         $$->u.expression.type = AST_EXP_IDENTIFIER;
 434                         $$->u.expression.post_op = AST_LINK_RARROW;
 435                         $$->u.expression.u.identifier = $3->s;
 436                         $$->u.expression.prev = $1;
 437                 }
 438         ;
 439
 440 unary_expression
 441         : postfix_expression
 442                 {       $$ = $1;                                        }
 443         | unary_operator unary_expression
 444                 {
 445                         $$ = $1;
 446                         $$->u.unary_op.child = $2;
 447                 }
 448         ;
 449
 450 unary_operator
 451         : PLUS
 452                 {
 453                         $$ = make_node(parser_ctx, NODE_UNARY_OP);
 454                         $$->u.unary_op.type = AST_UNARY_PLUS;
 455                 }
 456         | MINUS
 457                 {
 458                         $$ = make_node(parser_ctx, NODE_UNARY_OP);
 459                         $$->u.unary_op.type = AST_UNARY_MINUS;
 460                 }
 461         | NOT_OP
 462                 {
 463                         $$ = make_node(parser_ctx, NODE_UNARY_OP);
 464                         $$->u.unary_op.type = AST_UNARY_NOT;
 465                 }
 466         ;
 467
 468 multiplicative_expression
 469         : unary_expression
 470                 {       $$ = $1;                                        }
 471         | multiplicative_expression STAR unary_expression
 472                 {
 473                         $$ = make_op_node(parser_ctx, AST_OP_MUL, $1, $3);
 474                 }
 475         | multiplicative_expression DIV_OP unary_expression
 476                 {
 477                         $$ = make_op_node(parser_ctx, AST_OP_DIV, $1, $3);
 478                 }
 479         | multiplicative_expression MOD_OP unary_expression
 480                 {
 481                         $$ = make_op_node(parser_ctx, AST_OP_MOD, $1, $3);
 482                 }
 483         ;
 484
 485 additive_expression
 486         : multiplicative_expression
 487                 {       $$ = $1;                                        }
 488         | additive_expression PLUS multiplicative_expression
 489                 {
 490                         $$ = make_op_node(parser_ctx, AST_OP_PLUS, $1, $3);
 491                 }
 492         | additive_expression MINUS multiplicative_expression
 493                 {
 494                         $$ = make_op_node(parser_ctx, AST_OP_MINUS, $1, $3);
 495                 }
 496         ;
 497
 498 shift_expression
 499         : additive_expression
 500                 {       $$ = $1;                                        }
 501         | shift_expression LEFT_OP additive_expression
 502                 {
 503                         $$ = make_op_node(parser_ctx, AST_OP_LSHIFT, $1, $3);
 504                 }
 505         | shift_expression RIGHT_OP additive_expression
 506                 {
 507                         $$ = make_op_node(parser_ctx, AST_OP_RSHIFT, $1, $3);
 508                 }
 509         ;
 510
 511 relational_expression
 512         : shift_expression
 513                 {       $$ = $1;                                        }
 514         | relational_expression LT_OP shift_expression
 515                 {
 516                         $$ = make_op_node(parser_ctx, AST_OP_LT, $1, $3);
 517                 }
 518         | relational_expression GT_OP shift_expression
 519                 {
 520                         $$ = make_op_node(parser_ctx, AST_OP_GT, $1, $3);
 521                 }
 522         | relational_expression LE_OP shift_expression
 523                 {
 524                         $$ = make_op_node(parser_ctx, AST_OP_LE, $1, $3);
 525                 }
 526         | relational_expression GE_OP shift_expression
 527                 {
 528                         $$ = make_op_node(parser_ctx, AST_OP_GE, $1, $3);
 529                 }
 530         ;
 531
 532 equality_expression
 533         : relational_expression
 534                 {       $$ = $1;                                        }
 535         | equality_expression EQ_OP relational_expression
 536                 {
 537                         $$ = make_op_node(parser_ctx, AST_OP_EQ, $1, $3);
 538                 }
 539         | equality_expression NE_OP relational_expression
 540                 {
 541                         $$ = make_op_node(parser_ctx, AST_OP_NE, $1, $3);
 542                 }
 543         ;
 544
 545 and_expression
 546         : equality_expression
 547                 {       $$ = $1;                                        }
 548         | and_expression AND_BIN equality_expression
 549                 {
 550                         $$ = make_op_node(parser_ctx, AST_OP_BIN_AND, $1, $3);
 551                 }
 552         ;
 553
 554 exclusive_or_expression
 555         : and_expression
 556                 {       $$ = $1;                                        }
 557         | exclusive_or_expression XOR_BIN and_expression
 558                 {
 559                         $$ = make_op_node(parser_ctx, AST_OP_BIN_XOR, $1, $3);
 560                 }
 561         ;
 562
 563 inclusive_or_expression
 564         : exclusive_or_expression
 565                 {       $$ = $1;                                        }
 566         | inclusive_or_expression OR_BIN exclusive_or_expression
 567                 {
 568                         $$ = make_op_node(parser_ctx, AST_OP_BIN_OR, $1, $3);
 569                 }
 570         ;
 571
 572 logical_and_expression
 573         : inclusive_or_expression
 574                 {       $$ = $1;                                        }
 575         | logical_and_expression AND_OP inclusive_or_expression
 576                 {
 577                         $$ = make_op_node(parser_ctx, AST_OP_AND, $1, $3);
 578                 }
 579         ;
 580
 581 logical_or_expression
 582         : logical_and_expression
 583                 {       $$ = $1;                                        }
 584         | logical_or_expression OR_OP logical_and_expression
 585                 {
 586                         $$ = make_op_node(parser_ctx, AST_OP_OR, $1, $3);
 587                 }
 588         ;
 589
 590 expression
 591         : logical_or_expression
 592                 {       $$ = $1;                                        }
 593         ;
 594
 595 translation_unit
 596         : expression
 597                 {
 598                         parser_ctx->ast->root.u.root.child = $1;
 599                 }
 600         ;