5 * LTTng filter expression parser
7 * Copyright 2012 Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
9 * SPDX-License-Identifier: LGPL-2.1-only
11 * Grammar inspired from http://www.quut.com/c/ANSI-C-grammar-y.html
21 #include "filter-ast.h"
22 #include "filter-parser.h"
24 #include <common/macros.h>
26 #define WIDTH_u64_SCANF_IS_A_BROKEN_API "20"
27 #define WIDTH_o64_SCANF_IS_A_BROKEN_API "22"
28 #define WIDTH_x64_SCANF_IS_A_BROKEN_API "17"
29 #define WIDTH_lg_SCANF_IS_A_BROKEN_API "4096" /* Hugely optimistic approximation */
34 int filter_parser_debug = 0;
37 int yyparse(struct filter_parser_ctx *parser_ctx, yyscan_t scanner);
39 int yylex(union YYSTYPE *yyval, yyscan_t scanner);
41 int yylex_init_extra(struct filter_parser_ctx *parser_ctx, yyscan_t * ptr_yy_globals);
43 int yylex_destroy(yyscan_t yyparser_ctx);
45 void yyrestart(FILE * in_str, yyscan_t parser_ctx);
48 struct cds_list_head gc;
53 static const char *node_type_to_str[] = {
54 [ NODE_UNKNOWN ] = "NODE_UNKNOWN",
55 [ NODE_ROOT ] = "NODE_ROOT",
56 [ NODE_EXPRESSION ] = "NODE_EXPRESSION",
57 [ NODE_OP ] = "NODE_OP",
58 [ NODE_UNARY_OP ] = "NODE_UNARY_OP",
62 const char *node_type(struct filter_node *node)
64 if (node->type < NR_NODE_TYPES)
65 return node_type_to_str[node->type];
70 static struct gc_string *gc_string_alloc(struct filter_parser_ctx *parser_ctx,
73 struct gc_string *gstr;
76 /* TODO: could be faster with find first bit or glib Gstring */
77 /* sizeof long to account for malloc header (int or long ?) */
78 for (alloclen = 8; alloclen < sizeof(long) + sizeof(*gstr) + len;
81 gstr = zmalloc(alloclen);
85 cds_list_add(&gstr->gc, &parser_ctx->allocated_strings);
86 gstr->alloclen = alloclen;
92 * note: never use gc_string_append on a string that has external references.
93 * gsrc will be garbage collected immediately, and gstr might be.
94 * Should only be used to append characters to a string literal or constant.
97 struct gc_string *gc_string_append(struct filter_parser_ctx *parser_ctx,
98 struct gc_string *gstr,
99 struct gc_string *gsrc)
101 size_t newlen = strlen(gsrc->s) + strlen(gstr->s) + 1;
104 /* TODO: could be faster with find first bit or glib Gstring */
105 /* sizeof long to account for malloc header (int or long ?) */
106 for (alloclen = 8; alloclen < sizeof(long) + sizeof(*gstr) + newlen;
109 if (alloclen > gstr->alloclen) {
110 struct gc_string *newgstr;
112 newgstr = gc_string_alloc(parser_ctx, newlen);
113 strcpy(newgstr->s, gstr->s);
114 strcat(newgstr->s, gsrc->s);
115 cds_list_del(&gstr->gc);
119 strcat(gstr->s, gsrc->s);
121 cds_list_del(&gsrc->gc);
127 void setstring(struct filter_parser_ctx *parser_ctx, YYSTYPE *lvalp, const char *src)
129 lvalp->gs = gc_string_alloc(parser_ctx, strlen(src) + 1);
130 strcpy(lvalp->gs->s, src);
133 static struct filter_node *make_node(struct filter_parser_ctx *scanner,
136 struct filter_ast *ast = filter_parser_get_ast(scanner);
137 struct filter_node *node;
139 node = zmalloc(sizeof(*node));
142 memset(node, 0, sizeof(*node));
144 cds_list_add(&node->gc, &ast->allocated_nodes);
148 fprintf(stderr, "[error] %s: trying to create root node\n", __func__);
151 case NODE_EXPRESSION:
160 fprintf(stderr, "[error] %s: unknown node type %d\n", __func__,
168 static struct filter_node *make_op_node(struct filter_parser_ctx *scanner,
170 struct filter_node *lchild,
171 struct filter_node *rchild)
173 struct filter_ast *ast = filter_parser_get_ast(scanner);
174 struct filter_node *node;
176 node = zmalloc(sizeof(*node));
179 memset(node, 0, sizeof(*node));
180 node->type = NODE_OP;
181 cds_list_add(&node->gc, &ast->allocated_nodes);
182 node->u.op.type = type;
183 node->u.op.lchild = lchild;
184 node->u.op.rchild = rchild;
189 void yyerror(struct filter_parser_ctx *parser_ctx, yyscan_t scanner, const char *str)
191 fprintf(stderr, "error %s\n", str);
200 #define parse_error(parser_ctx, str) \
202 yyerror(parser_ctx, parser_ctx->scanner, YY_("parse error: " str "\n")); \
206 static void free_strings(struct cds_list_head *list)
208 struct gc_string *gstr, *tmp;
210 cds_list_for_each_entry_safe(gstr, tmp, list, gc)
214 static struct filter_ast *filter_ast_alloc(void)
216 struct filter_ast *ast;
218 ast = zmalloc(sizeof(*ast));
221 memset(ast, 0, sizeof(*ast));
222 CDS_INIT_LIST_HEAD(&ast->allocated_nodes);
223 ast->root.type = NODE_ROOT;
227 static void filter_ast_free(struct filter_ast *ast)
229 struct filter_node *node, *tmp;
231 cds_list_for_each_entry_safe(node, tmp, &ast->allocated_nodes, gc)
237 int filter_parser_ctx_append_ast(struct filter_parser_ctx *parser_ctx)
239 return yyparse(parser_ctx, parser_ctx->scanner);
243 struct filter_parser_ctx *filter_parser_ctx_alloc(FILE *input)
245 struct filter_parser_ctx *parser_ctx;
248 yydebug = filter_parser_debug;
250 parser_ctx = zmalloc(sizeof(*parser_ctx));
253 memset(parser_ctx, 0, sizeof(*parser_ctx));
255 ret = yylex_init_extra(parser_ctx, &parser_ctx->scanner);
257 fprintf(stderr, "yylex_init error\n");
258 goto cleanup_parser_ctx;
260 /* Start processing new stream */
261 yyrestart(input, parser_ctx->scanner);
263 parser_ctx->ast = filter_ast_alloc();
264 if (!parser_ctx->ast)
266 CDS_INIT_LIST_HEAD(&parser_ctx->allocated_strings);
269 fprintf(stdout, "parser_ctx input is a%s.\n",
270 isatty(fileno(input)) ? "n interactive tty" :
271 " noninteractive file");
276 ret = yylex_destroy(parser_ctx->scanner);
278 fprintf(stderr, "yylex_destroy error\n");
285 void filter_parser_ctx_free(struct filter_parser_ctx *parser_ctx)
289 free_strings(&parser_ctx->allocated_strings);
290 filter_ast_free(parser_ctx->ast);
291 ret = yylex_destroy(parser_ctx->scanner);
293 fprintf(stderr, "yylex_destroy error\n");
301 %parse-param {struct filter_parser_ctx *parser_ctx}
302 %parse-param {yyscan_t scanner}
303 %lex-param {yyscan_t scanner}
304 %start translation_unit
305 %token CHARACTER_CONSTANT_START SQUOTE STRING_LITERAL_START DQUOTE
306 %token ESCSEQ CHAR_STRING_TOKEN
307 %token DECIMAL_CONSTANT OCTAL_CONSTANT HEXADECIMAL_CONSTANT FLOAT_CONSTANT
308 %token LSBRAC RSBRAC LPAREN RPAREN LBRAC RBRAC RARROW
309 %token STAR PLUS MINUS
310 %token MOD_OP DIV_OP RIGHT_OP LEFT_OP
311 %token EQ_OP NE_OP LE_OP GE_OP LT_OP GT_OP AND_OP OR_OP NOT_OP
312 %token ASSIGN COLON SEMICOLON DOTDOTDOT DOT EQUAL COMMA
313 %token XOR_BIN AND_BIN OR_BIN NOT_BIN
315 %token <gs> IDENTIFIER GLOBAL_IDENTIFIER
321 struct gc_string *gs;
322 struct filter_node *n;
325 %type <gs> s_char s_char_sequence c_char c_char_sequence
327 %type <n> primary_expression
328 %type <n> prefix_expression
329 %type <n> prefix_expression_rec
330 %type <n> postfix_expression
331 %type <n> unary_expression
332 %type <n> unary_operator
333 %type <n> multiplicative_expression
334 %type <n> additive_expression
335 %type <n> shift_expression
336 %type <n> relational_expression
337 %type <n> equality_expression
338 %type <n> and_expression
339 %type <n> exclusive_or_expression
340 %type <n> inclusive_or_expression
341 %type <n> logical_and_expression
342 %type <n> logical_or_expression
344 %type <n> identifiers
354 | c_char_sequence c_char
355 { $$ = gc_string_append(parser_ctx, $1, $2); }
363 parse_error(parser_ctx, "escape sequences not supported yet");
367 /* 1.6 String literals */
372 | s_char_sequence s_char
373 { $$ = gc_string_append(parser_ctx, $1, $2); }
381 parse_error(parser_ctx, "escape sequences not supported yet");
388 $$ = make_node(parser_ctx, NODE_EXPRESSION);
389 $$->u.expression.type = AST_EXP_CONSTANT;
390 if (sscanf(yylval.gs->s, "%" WIDTH_u64_SCANF_IS_A_BROKEN_API SCNu64,
391 &$$->u.expression.u.constant) != 1) {
392 parse_error(parser_ctx, "cannot scanf decimal constant");
397 $$ = make_node(parser_ctx, NODE_EXPRESSION);
398 $$->u.expression.type = AST_EXP_CONSTANT;
399 if (!strcmp(yylval.gs->s, "0")) {
400 $$->u.expression.u.constant = 0;
401 } else if (sscanf(yylval.gs->s, "0%" WIDTH_o64_SCANF_IS_A_BROKEN_API SCNo64,
402 &$$->u.expression.u.constant) != 1) {
403 parse_error(parser_ctx, "cannot scanf octal constant");
406 | HEXADECIMAL_CONSTANT
408 $$ = make_node(parser_ctx, NODE_EXPRESSION);
409 $$->u.expression.type = AST_EXP_CONSTANT;
410 if (sscanf(yylval.gs->s, "0x%" WIDTH_x64_SCANF_IS_A_BROKEN_API SCNx64,
411 &$$->u.expression.u.constant) != 1) {
412 parse_error(parser_ctx, "cannot scanf hexadecimal constant");
417 $$ = make_node(parser_ctx, NODE_EXPRESSION);
418 $$->u.expression.type = AST_EXP_FLOAT_CONSTANT;
419 if (sscanf(yylval.gs->s, "%" WIDTH_lg_SCANF_IS_A_BROKEN_API "lg",
420 &$$->u.expression.u.float_constant) != 1) {
421 parse_error(parser_ctx, "cannot scanf float constant");
424 | STRING_LITERAL_START DQUOTE
426 $$ = make_node(parser_ctx, NODE_EXPRESSION);
427 $$->u.expression.type = AST_EXP_STRING;
428 $$->u.expression.u.string = "";
430 | STRING_LITERAL_START s_char_sequence DQUOTE
432 $$ = make_node(parser_ctx, NODE_EXPRESSION);
433 $$->u.expression.type = AST_EXP_STRING;
434 $$->u.expression.u.string = $2->s;
436 | CHARACTER_CONSTANT_START c_char_sequence SQUOTE
438 $$ = make_node(parser_ctx, NODE_EXPRESSION);
439 $$->u.expression.type = AST_EXP_STRING;
440 $$->u.expression.u.string = $2->s;
442 | LPAREN expression RPAREN
444 $$ = make_node(parser_ctx, NODE_EXPRESSION);
445 $$->u.expression.type = AST_EXP_NESTED;
446 $$->u.expression.u.child = $2;
453 $$ = make_node(parser_ctx, NODE_EXPRESSION);
454 $$->u.expression.type = AST_EXP_IDENTIFIER;
455 $$->u.expression.u.identifier = yylval.gs->s;
459 $$ = make_node(parser_ctx, NODE_EXPRESSION);
460 $$->u.expression.type = AST_EXP_GLOBAL_IDENTIFIER;
461 $$->u.expression.u.identifier = yylval.gs->s;
465 prefix_expression_rec
466 : LSBRAC unary_expression RSBRAC
470 | LSBRAC unary_expression RSBRAC prefix_expression_rec
473 $$->u.expression.pre_op = AST_LINK_BRACKET;
474 $$->u.expression.prev = $4;
483 | identifiers prefix_expression_rec
486 $$->u.expression.pre_op = AST_LINK_BRACKET;
487 $$->u.expression.next_bracket = $2;
496 | postfix_expression DOT prefix_expression
499 $$->u.expression.post_op = AST_LINK_DOT;
500 $$->u.expression.prev = $1;
502 | postfix_expression RARROW prefix_expression
505 $$->u.expression.post_op = AST_LINK_RARROW;
506 $$->u.expression.prev = $1;
515 | unary_operator unary_expression
518 $$->u.unary_op.child = $2;
525 $$ = make_node(parser_ctx, NODE_UNARY_OP);
526 $$->u.unary_op.type = AST_UNARY_PLUS;
530 $$ = make_node(parser_ctx, NODE_UNARY_OP);
531 $$->u.unary_op.type = AST_UNARY_MINUS;
535 $$ = make_node(parser_ctx, NODE_UNARY_OP);
536 $$->u.unary_op.type = AST_UNARY_NOT;
540 $$ = make_node(parser_ctx, NODE_UNARY_OP);
541 $$->u.unary_op.type = AST_UNARY_BIT_NOT;
545 multiplicative_expression
548 | multiplicative_expression STAR unary_expression
550 $$ = make_op_node(parser_ctx, AST_OP_MUL, $1, $3);
552 | multiplicative_expression DIV_OP unary_expression
554 $$ = make_op_node(parser_ctx, AST_OP_DIV, $1, $3);
556 | multiplicative_expression MOD_OP unary_expression
558 $$ = make_op_node(parser_ctx, AST_OP_MOD, $1, $3);
563 : multiplicative_expression
565 | additive_expression PLUS multiplicative_expression
567 $$ = make_op_node(parser_ctx, AST_OP_PLUS, $1, $3);
569 | additive_expression MINUS multiplicative_expression
571 $$ = make_op_node(parser_ctx, AST_OP_MINUS, $1, $3);
576 : additive_expression
578 | shift_expression LEFT_OP additive_expression
580 $$ = make_op_node(parser_ctx, AST_OP_BIT_LSHIFT, $1, $3);
582 | shift_expression RIGHT_OP additive_expression
584 $$ = make_op_node(parser_ctx, AST_OP_BIT_RSHIFT, $1, $3);
591 | and_expression AND_BIN shift_expression
593 $$ = make_op_node(parser_ctx, AST_OP_BIT_AND, $1, $3);
597 exclusive_or_expression
600 | exclusive_or_expression XOR_BIN and_expression
602 $$ = make_op_node(parser_ctx, AST_OP_BIT_XOR, $1, $3);
606 inclusive_or_expression
607 : exclusive_or_expression
609 | inclusive_or_expression OR_BIN exclusive_or_expression
611 $$ = make_op_node(parser_ctx, AST_OP_BIT_OR, $1, $3);
615 relational_expression
616 : inclusive_or_expression
618 | relational_expression LT_OP inclusive_or_expression
620 $$ = make_op_node(parser_ctx, AST_OP_LT, $1, $3);
622 | relational_expression GT_OP inclusive_or_expression
624 $$ = make_op_node(parser_ctx, AST_OP_GT, $1, $3);
626 | relational_expression LE_OP inclusive_or_expression
628 $$ = make_op_node(parser_ctx, AST_OP_LE, $1, $3);
630 | relational_expression GE_OP inclusive_or_expression
632 $$ = make_op_node(parser_ctx, AST_OP_GE, $1, $3);
637 : relational_expression
639 | equality_expression EQ_OP relational_expression
641 $$ = make_op_node(parser_ctx, AST_OP_EQ, $1, $3);
643 | equality_expression NE_OP relational_expression
645 $$ = make_op_node(parser_ctx, AST_OP_NE, $1, $3);
649 logical_and_expression
650 : equality_expression
652 | logical_and_expression AND_OP equality_expression
654 $$ = make_op_node(parser_ctx, AST_OP_AND, $1, $3);
658 logical_or_expression
659 : logical_and_expression
661 | logical_or_expression OR_OP logical_and_expression
663 $$ = make_op_node(parser_ctx, AST_OP_OR, $1, $3);
668 : logical_or_expression
675 parser_ctx->ast->root.u.root.child = $1;