[lttng-tools.git] / src / lib / lttng-ctl / filter / filter-parser.y

%{
/*
 * filter-parser.y
 *
 * LTTng filter expression parser
 *
 * Copyright 2012 Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
 *
 * SPDX-License-Identifier: LGPL-2.1-only
 *
 * Grammar inspired from http://www.quut.com/c/ANSI-C-grammar-y.html
 */

#include <stdio.h>
#include <unistd.h>
#include <string.h>
#include <stdlib.h>
#include <assert.h>
#include <errno.h>
#include <inttypes.h>
#include "filter-ast.h"
#include "filter-parser.h"

#include <common/macros.h>

#define WIDTH_u64_SCANF_IS_A_BROKEN_API	"20"
#define WIDTH_o64_SCANF_IS_A_BROKEN_API	"22"
#define WIDTH_x64_SCANF_IS_A_BROKEN_API	"17"
#define WIDTH_lg_SCANF_IS_A_BROKEN_API	"4096"	/* Hugely optimistic approximation */

LTTNG_HIDDEN
int yydebug;
LTTNG_HIDDEN
int filter_parser_debug = 0;

LTTNG_HIDDEN
int yyparse(struct filter_parser_ctx *parser_ctx, yyscan_t scanner);
LTTNG_HIDDEN
int yylex(union YYSTYPE *yyval, yyscan_t scanner);
LTTNG_HIDDEN
int yylex_init_extra(struct filter_parser_ctx *parser_ctx, yyscan_t * ptr_yy_globals);
LTTNG_HIDDEN
int yylex_destroy(yyscan_t yyparser_ctx);
LTTNG_HIDDEN
void yyrestart(FILE * in_str, yyscan_t parser_ctx);

struct gc_string {
	struct cds_list_head gc;
	size_t alloclen;
	char s[];
};

static const char *node_type_to_str[] = {
	[ NODE_UNKNOWN ] = "NODE_UNKNOWN",
	[ NODE_ROOT ] = "NODE_ROOT",
	[ NODE_EXPRESSION ] = "NODE_EXPRESSION",
	[ NODE_OP ] = "NODE_OP",
	[ NODE_UNARY_OP ] = "NODE_UNARY_OP",
};

LTTNG_HIDDEN
const char *node_type(struct filter_node *node)
{
	if (node->type < NR_NODE_TYPES)
		return node_type_to_str[node->type];
	else
		return NULL;
}

static struct gc_string *gc_string_alloc(struct filter_parser_ctx *parser_ctx,
					 size_t len)
{
	struct gc_string *gstr;
	size_t alloclen;

	/* TODO: could be faster with find first bit or glib Gstring */
	/* sizeof long to account for malloc header (int or long ?) */
	for (alloclen = 8; alloclen < sizeof(long) + sizeof(*gstr) + len;
	     alloclen *= 2);

	gstr = zmalloc(alloclen);
	if (!gstr) {
		goto end;
	}
	cds_list_add(&gstr->gc, &parser_ctx->allocated_strings);
	gstr->alloclen = alloclen;
end:
	return gstr;
}

/*
 * note: never use gc_string_append on a string that has external references.
 * gsrc will be garbage collected immediately, and gstr might be.
 * Should only be used to append characters to a string literal or constant.
 */
LTTNG_HIDDEN
struct gc_string *gc_string_append(struct filter_parser_ctx *parser_ctx,
				   struct gc_string *gstr,
				   struct gc_string *gsrc)
{
	size_t newlen = strlen(gsrc->s) + strlen(gstr->s) + 1;
	size_t alloclen;

	/* TODO: could be faster with find first bit or glib Gstring */
	/* sizeof long to account for malloc header (int or long ?) */
	for (alloclen = 8; alloclen < sizeof(long) + sizeof(*gstr) + newlen;
	     alloclen *= 2);

	if (alloclen > gstr->alloclen) {
		struct gc_string *newgstr;

		newgstr = gc_string_alloc(parser_ctx, newlen);
		strcpy(newgstr->s, gstr->s);
		strcat(newgstr->s, gsrc->s);
		cds_list_del(&gstr->gc);
		free(gstr);
		gstr = newgstr;
	} else {
		strcat(gstr->s, gsrc->s);
	}
	cds_list_del(&gsrc->gc);
	free(gsrc);
	return gstr;
}

LTTNG_HIDDEN
void setstring(struct filter_parser_ctx *parser_ctx, YYSTYPE *lvalp, const char *src)
{
	lvalp->gs = gc_string_alloc(parser_ctx, strlen(src) + 1);
	strcpy(lvalp->gs->s, src);
}

static struct filter_node *make_node(struct filter_parser_ctx *scanner,
				  enum node_type type)
{
	struct filter_ast *ast = filter_parser_get_ast(scanner);
	struct filter_node *node;

	node = zmalloc(sizeof(*node));
	if (!node)
		return NULL;
	memset(node, 0, sizeof(*node));
	node->type = type;
	cds_list_add(&node->gc, &ast->allocated_nodes);

	switch (type) {
	case NODE_ROOT:
		fprintf(stderr, "[error] %s: trying to create root node\n", __func__);
		break;

	case NODE_EXPRESSION:
		break;
	case NODE_OP:
		break;
	case NODE_UNARY_OP:
		break;

	case NODE_UNKNOWN:
	default:
		fprintf(stderr, "[error] %s: unknown node type %d\n", __func__,
			(int) type);
		break;
	}

	return node;
}

static struct filter_node *make_op_node(struct filter_parser_ctx *scanner,
			enum op_type type,
			struct filter_node *lchild,
			struct filter_node *rchild)
{
	struct filter_ast *ast = filter_parser_get_ast(scanner);
	struct filter_node *node;

	node = zmalloc(sizeof(*node));
	if (!node)
		return NULL;
	memset(node, 0, sizeof(*node));
	node->type = NODE_OP;
	cds_list_add(&node->gc, &ast->allocated_nodes);
	node->u.op.type = type;
	node->u.op.lchild = lchild;
	node->u.op.rchild = rchild;
	return node;
}

LTTNG_HIDDEN
void yyerror(struct filter_parser_ctx *parser_ctx, yyscan_t scanner, const char *str)
{
	fprintf(stderr, "error %s\n", str);
}
 
LTTNG_HIDDEN
int yywrap(void)
{
	return 1;
} 

#define parse_error(parser_ctx, str)				\
do {								\
	yyerror(parser_ctx, parser_ctx->scanner, YY_("parse error: " str "\n"));	\
	YYERROR;						\
} while (0)

static void free_strings(struct cds_list_head *list)
{
	struct gc_string *gstr, *tmp;

	cds_list_for_each_entry_safe(gstr, tmp, list, gc)
		free(gstr);
}

static struct filter_ast *filter_ast_alloc(void)
{
	struct filter_ast *ast;

	ast = zmalloc(sizeof(*ast));
	if (!ast)
		return NULL;
	memset(ast, 0, sizeof(*ast));
	CDS_INIT_LIST_HEAD(&ast->allocated_nodes);
	ast->root.type = NODE_ROOT;
	return ast;
}

static void filter_ast_free(struct filter_ast *ast)
{
	struct filter_node *node, *tmp;

	cds_list_for_each_entry_safe(node, tmp, &ast->allocated_nodes, gc)
		free(node);
	free(ast);
}

LTTNG_HIDDEN
int filter_parser_ctx_append_ast(struct filter_parser_ctx *parser_ctx)
{
	return yyparse(parser_ctx, parser_ctx->scanner);
}

LTTNG_HIDDEN
struct filter_parser_ctx *filter_parser_ctx_alloc(FILE *input)
{
	struct filter_parser_ctx *parser_ctx;
	int ret;

	yydebug = filter_parser_debug;

	parser_ctx = zmalloc(sizeof(*parser_ctx));
	if (!parser_ctx)
		return NULL;
	memset(parser_ctx, 0, sizeof(*parser_ctx));

	ret = yylex_init_extra(parser_ctx, &parser_ctx->scanner);
	if (ret) {
		fprintf(stderr, "yylex_init error\n");
		goto cleanup_parser_ctx;
	}
	/* Start processing new stream */
	yyrestart(input, parser_ctx->scanner);

	parser_ctx->ast = filter_ast_alloc();
	if (!parser_ctx->ast)
		goto cleanup_lexer;
	CDS_INIT_LIST_HEAD(&parser_ctx->allocated_strings);

	if (yydebug)
		fprintf(stdout, "parser_ctx input is a%s.\n",
			isatty(fileno(input)) ? "n interactive tty" :
						" noninteractive file");

	return parser_ctx;

cleanup_lexer:
	ret = yylex_destroy(parser_ctx->scanner);
	if (!ret)
		fprintf(stderr, "yylex_destroy error\n");
cleanup_parser_ctx:
	free(parser_ctx);
	return NULL;
}

LTTNG_HIDDEN
void filter_parser_ctx_free(struct filter_parser_ctx *parser_ctx)
{
	int ret;

	free_strings(&parser_ctx->allocated_strings);
	filter_ast_free(parser_ctx->ast);
	ret = yylex_destroy(parser_ctx->scanner);
	if (ret)
		fprintf(stderr, "yylex_destroy error\n");
	free(parser_ctx);
}

%}

%define api.pure
	/* %locations */
%parse-param {struct filter_parser_ctx *parser_ctx}
%parse-param {yyscan_t scanner}
%lex-param {yyscan_t scanner}
%start translation_unit
%token CHARACTER_CONSTANT_START SQUOTE STRING_LITERAL_START DQUOTE
%token ESCSEQ CHAR_STRING_TOKEN
%token DECIMAL_CONSTANT OCTAL_CONSTANT HEXADECIMAL_CONSTANT FLOAT_CONSTANT
%token LSBRAC RSBRAC LPAREN RPAREN LBRAC RBRAC RARROW
%token STAR PLUS MINUS
%token MOD_OP DIV_OP RIGHT_OP LEFT_OP
%token EQ_OP NE_OP LE_OP GE_OP LT_OP GT_OP AND_OP OR_OP NOT_OP
%token ASSIGN COLON SEMICOLON DOTDOTDOT DOT EQUAL COMMA
%token XOR_BIN AND_BIN OR_BIN NOT_BIN

%token <gs> IDENTIFIER GLOBAL_IDENTIFIER
%token ERROR
%union
{
	long long ll;
	char c;
	struct gc_string *gs;
	struct filter_node *n;
}

%type <gs> s_char s_char_sequence c_char c_char_sequence

%type <n> primary_expression
%type <n> prefix_expression
%type <n> prefix_expression_rec
%type <n> postfix_expression
%type <n> unary_expression
%type <n> unary_operator
%type <n> multiplicative_expression
%type <n> additive_expression
%type <n> shift_expression
%type <n> relational_expression
%type <n> equality_expression
%type <n> and_expression
%type <n> exclusive_or_expression
%type <n> inclusive_or_expression
%type <n> logical_and_expression
%type <n> logical_or_expression
%type <n> expression
%type <n> identifiers

%%


/* 1.5 Constants */

c_char_sequence:
		c_char
		{	$$ = $1;					}
	|	c_char_sequence c_char
		{	$$ = gc_string_append(parser_ctx, $1, $2);		}
	;

c_char:
		CHAR_STRING_TOKEN
		{	$$ = yylval.gs;					}
	|	ESCSEQ
		{
			parse_error(parser_ctx, "escape sequences not supported yet");
		}
	;

/* 1.6 String literals */

s_char_sequence:
		s_char
		{	$$ = $1;					}
	|	s_char_sequence s_char
		{	$$ = gc_string_append(parser_ctx, $1, $2);		}
	;

s_char:
		CHAR_STRING_TOKEN
		{	$$ = yylval.gs;					}
	|	ESCSEQ
		{
			parse_error(parser_ctx, "escape sequences not supported yet");
		}
	;

primary_expression:
		DECIMAL_CONSTANT
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_CONSTANT;
			if (sscanf(yylval.gs->s, "%" WIDTH_u64_SCANF_IS_A_BROKEN_API SCNu64,
					&$$->u.expression.u.constant) != 1) {
				parse_error(parser_ctx, "cannot scanf decimal constant");
			}
		}
	|	OCTAL_CONSTANT
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_CONSTANT;
			if (!strcmp(yylval.gs->s, "0")) {
				$$->u.expression.u.constant = 0;
			} else if (sscanf(yylval.gs->s, "0%" WIDTH_o64_SCANF_IS_A_BROKEN_API SCNo64,
					&$$->u.expression.u.constant) != 1) {
				parse_error(parser_ctx, "cannot scanf octal constant");
			}
		}
	|	HEXADECIMAL_CONSTANT
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_CONSTANT;
			if (sscanf(yylval.gs->s, "0x%" WIDTH_x64_SCANF_IS_A_BROKEN_API SCNx64,
					&$$->u.expression.u.constant) != 1) {
				parse_error(parser_ctx, "cannot scanf hexadecimal constant");
			}
		}
	|	FLOAT_CONSTANT
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_FLOAT_CONSTANT;
			if (sscanf(yylval.gs->s, "%" WIDTH_lg_SCANF_IS_A_BROKEN_API "lg",
					&$$->u.expression.u.float_constant) != 1) {
				parse_error(parser_ctx, "cannot scanf float constant");
			}
		}
	|	STRING_LITERAL_START DQUOTE
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_STRING;
			$$->u.expression.u.string = "";
		}
	|	STRING_LITERAL_START s_char_sequence DQUOTE
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_STRING;
			$$->u.expression.u.string = $2->s;
		}
	|	CHARACTER_CONSTANT_START c_char_sequence SQUOTE
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_STRING;
			$$->u.expression.u.string = $2->s;
		}
	|	LPAREN expression RPAREN
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_NESTED;
			$$->u.expression.u.child = $2;
		}
	;

identifiers
	:	IDENTIFIER
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_IDENTIFIER;
			$$->u.expression.u.identifier = yylval.gs->s;
		}
	|	GLOBAL_IDENTIFIER
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_GLOBAL_IDENTIFIER;
			$$->u.expression.u.identifier = yylval.gs->s;
		}
	;

prefix_expression_rec
	: LSBRAC unary_expression RSBRAC
		{
			$$ = $2;
		}
	| LSBRAC unary_expression RSBRAC prefix_expression_rec
		{
			$$ = $2;
			$$->u.expression.pre_op = AST_LINK_BRACKET;
			$$->u.expression.prev = $4;
		}
	;

prefix_expression
	: identifiers
		{
			$$ = $1;
		}
	| identifiers prefix_expression_rec
		{
			$$ = $1;
			$$->u.expression.pre_op = AST_LINK_BRACKET;
			$$->u.expression.next_bracket = $2;
		}
	;

postfix_expression
	: prefix_expression
		{
			$$ = $1;
		}
	| postfix_expression DOT prefix_expression
		{
			$$ = $3;
			$$->u.expression.post_op = AST_LINK_DOT;
			$$->u.expression.prev = $1;
		}
	| postfix_expression RARROW prefix_expression
		{
			$$ = $3;
			$$->u.expression.post_op = AST_LINK_RARROW;
			$$->u.expression.prev = $1;
		}
	;

unary_expression
	: postfix_expression
		{	$$ = $1;					}
	| primary_expression
		{	$$ = $1;					}
	| unary_operator unary_expression
		{
			$$ = $1;
			$$->u.unary_op.child = $2;
		}
	;

unary_operator
	: PLUS
		{
			$$ = make_node(parser_ctx, NODE_UNARY_OP);
			$$->u.unary_op.type = AST_UNARY_PLUS;
		}
	| MINUS
		{
			$$ = make_node(parser_ctx, NODE_UNARY_OP);
			$$->u.unary_op.type = AST_UNARY_MINUS;
		}
	| NOT_OP
		{
			$$ = make_node(parser_ctx, NODE_UNARY_OP);
			$$->u.unary_op.type = AST_UNARY_NOT;
		}
	| NOT_BIN
		{
			$$ = make_node(parser_ctx, NODE_UNARY_OP);
			$$->u.unary_op.type = AST_UNARY_BIT_NOT;
		}
	;

multiplicative_expression
	: unary_expression
		{	$$ = $1;					}
	| multiplicative_expression STAR unary_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_MUL, $1, $3);
		}
	| multiplicative_expression DIV_OP unary_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_DIV, $1, $3);
		}
	| multiplicative_expression MOD_OP unary_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_MOD, $1, $3);
		}
	;

additive_expression
	: multiplicative_expression
		{	$$ = $1;					}
	| additive_expression PLUS multiplicative_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_PLUS, $1, $3);
		}
	| additive_expression MINUS multiplicative_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_MINUS, $1, $3);
		}
	;

shift_expression
	: additive_expression
		{	$$ = $1;					}
	| shift_expression LEFT_OP additive_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_BIT_LSHIFT, $1, $3);
		}
	| shift_expression RIGHT_OP additive_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_BIT_RSHIFT, $1, $3);
		}
	;

and_expression
	: shift_expression
		{	$$ = $1;					}
	| and_expression AND_BIN shift_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_BIT_AND, $1, $3);
		}
	;

exclusive_or_expression
	: and_expression
		{	$$ = $1;					}
	| exclusive_or_expression XOR_BIN and_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_BIT_XOR, $1, $3);
		}
	;

inclusive_or_expression
	: exclusive_or_expression
		{	$$ = $1;					}
	| inclusive_or_expression OR_BIN exclusive_or_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_BIT_OR, $1, $3);
		}
	;

relational_expression
	: inclusive_or_expression
		{	$$ = $1;					}
	| relational_expression LT_OP inclusive_or_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_LT, $1, $3);
		}
	| relational_expression GT_OP inclusive_or_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_GT, $1, $3);
		}
	| relational_expression LE_OP inclusive_or_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_LE, $1, $3);
		}
	| relational_expression GE_OP inclusive_or_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_GE, $1, $3);
		}
	;

equality_expression
	: relational_expression
		{	$$ = $1;					}
	| equality_expression EQ_OP relational_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_EQ, $1, $3);
		}
	| equality_expression NE_OP relational_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_NE, $1, $3);
		}
	;

logical_and_expression
	: equality_expression
		{	$$ = $1;					}
	| logical_and_expression AND_OP equality_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_AND, $1, $3);
		}
	;

logical_or_expression
	: logical_and_expression
		{	$$ = $1;					}
	| logical_or_expression OR_OP logical_and_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_OR, $1, $3);
		}
	;

expression
	: logical_or_expression
		{	$$ = $1;					}
	;

translation_unit
	: expression
		{
			parser_ctx->ast->root.u.root.child = $1;
		}
	;
Commit	Line	Data
953192ba MD	1	%{
	2	/*
	3	* filter-parser.y
	4	*
	5	* LTTng filter expression parser
	6	*
ab5be9fa	7	* Copyright 2012 Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
953192ba	8	*
ab5be9fa	9	* SPDX-License-Identifier: LGPL-2.1-only
953192ba MD	10	*
	11	* Grammar inspired from http://www.quut.com/c/ANSI-C-grammar-y.html
	12	*/
	13
	14	#include <stdio.h>
	15	#include <unistd.h>
	16	#include <string.h>
	17	#include <stdlib.h>
	18	#include <assert.h>
	19	#include <errno.h>
	20	#include <inttypes.h>
953192ba	21	#include "filter-ast.h"
95b9bd90	22	#include "filter-parser.h"
953192ba	23
a187da1a DG	24	#include <common/macros.h>
a187da1a DG	25
8ab7c0d9 MD	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 */
	30
a187da1a	31	LTTNG_HIDDEN
953192ba	32	int yydebug;
a187da1a	33	LTTNG_HIDDEN
953192ba MD	34	int filter_parser_debug = 0;
953192ba MD	35
a187da1a	36	LTTNG_HIDDEN
9039edd4	37	int yyparse(struct filter_parser_ctx *parser_ctx, yyscan_t scanner);
a187da1a	38	LTTNG_HIDDEN
9039edd4	39	int yylex(union YYSTYPE *yyval, yyscan_t scanner);
a187da1a	40	LTTNG_HIDDEN
953192ba	41	int yylex_init_extra(struct filter_parser_ctx parser_ctx, yyscan_t ptr_yy_globals);
a187da1a	42	LTTNG_HIDDEN
953192ba	43	int yylex_destroy(yyscan_t yyparser_ctx);
a187da1a	44	LTTNG_HIDDEN
953192ba MD	45	void yyrestart(FILE * in_str, yyscan_t parser_ctx);
	46
	47	struct gc_string {
	48	struct cds_list_head gc;
	49	size_t alloclen;
	50	char s[];
	51	};
	52
	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",
	59	};
	60
a187da1a	61	LTTNG_HIDDEN
953192ba MD	62	const char node_type(struct filter_node node)
	63	{
	64	if (node->type < NR_NODE_TYPES)
	65	return node_type_to_str[node->type];
	66	else
	67	return NULL;
	68	}
	69
	70	static struct gc_string gc_string_alloc(struct filter_parser_ctx parser_ctx,
	71	size_t len)
	72	{
	73	struct gc_string *gstr;
	74	size_t alloclen;
	75
	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;
	79	alloclen *= 2);
	80
be61ee34 MD	81	gstr = zmalloc(alloclen);
	82	if (!gstr) {
	83	goto end;
	84	}
953192ba MD	85	cds_list_add(&gstr->gc, &parser_ctx->allocated_strings);
953192ba MD	86	gstr->alloclen = alloclen;
be61ee34	87	end:
953192ba MD	88	return gstr;
	89	}
	90
	91	/*
	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.
	95	*/
a187da1a	96	LTTNG_HIDDEN
953192ba MD	97	struct gc_string gc_string_append(struct filter_parser_ctx parser_ctx,
	98	struct gc_string *gstr,
	99	struct gc_string *gsrc)
	100	{
	101	size_t newlen = strlen(gsrc->s) + strlen(gstr->s) + 1;
	102	size_t alloclen;
	103
	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;
	107	alloclen *= 2);
	108
	109	if (alloclen > gstr->alloclen) {
	110	struct gc_string *newgstr;
	111
	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);
	116	free(gstr);
	117	gstr = newgstr;
	118	} else {
	119	strcat(gstr->s, gsrc->s);
	120	}
	121	cds_list_del(&gsrc->gc);
	122	free(gsrc);
	123	return gstr;
	124	}
	125
a187da1a	126	LTTNG_HIDDEN
953192ba MD	127	void setstring(struct filter_parser_ctx parser_ctx, YYSTYPE lvalp, const char *src)
	128	{
	129	lvalp->gs = gc_string_alloc(parser_ctx, strlen(src) + 1);
	130	strcpy(lvalp->gs->s, src);
	131	}
	132
	133	static struct filter_node make_node(struct filter_parser_ctx scanner,
	134	enum node_type type)
	135	{
	136	struct filter_ast *ast = filter_parser_get_ast(scanner);
	137	struct filter_node *node;
	138
be61ee34	139	node = zmalloc(sizeof(*node));
953192ba MD	140	if (!node)
	141	return NULL;
	142	memset(node, 0, sizeof(*node));
	143	node->type = type;
	144	cds_list_add(&node->gc, &ast->allocated_nodes);
	145
	146	switch (type) {
	147	case NODE_ROOT:
	148	fprintf(stderr, "[error] %s: trying to create root node\n", __func__);
	149	break;
	150
	151	case NODE_EXPRESSION:
	152	break;
	153	case NODE_OP:
	154	break;
	155	case NODE_UNARY_OP:
	156	break;
	157
	158	case NODE_UNKNOWN:
	159	default:
	160	fprintf(stderr, "[error] %s: unknown node type %d\n", __func__,
	161	(int) type);
	162	break;
	163	}
	164
	165	return node;
	166	}
	167
	168	static struct filter_node make_op_node(struct filter_parser_ctx scanner,
	169	enum op_type type,
	170	struct filter_node *lchild,
	171	struct filter_node *rchild)
	172	{
	173	struct filter_ast *ast = filter_parser_get_ast(scanner);
	174	struct filter_node *node;
	175
be61ee34	176	node = zmalloc(sizeof(*node));
953192ba MD	177	if (!node)
	178	return NULL;
	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;
	185	return node;
	186	}
	187
a187da1a	188	LTTNG_HIDDEN
9039edd4	189	void yyerror(struct filter_parser_ctx parser_ctx, yyscan_t scanner, const char str)
953192ba MD	190	{
	191	fprintf(stderr, "error %s\n", str);
	192	}
	193
a187da1a	194	LTTNG_HIDDEN
953192ba MD	195	int yywrap(void)
	196	{
	197	return 1;
	198	}
	199
	200	#define parse_error(parser_ctx, str) \
	201	do { \
9039edd4	202	yyerror(parser_ctx, parser_ctx->scanner, YY_("parse error: " str "\n")); \
953192ba MD	203	YYERROR; \
	204	} while (0)
	205
	206	static void free_strings(struct cds_list_head *list)
	207	{
	208	struct gc_string gstr, tmp;
	209
	210	cds_list_for_each_entry_safe(gstr, tmp, list, gc)
	211	free(gstr);
	212	}
	213
	214	static struct filter_ast *filter_ast_alloc(void)
	215	{
	216	struct filter_ast *ast;
	217
be61ee34	218	ast = zmalloc(sizeof(*ast));
953192ba MD	219	if (!ast)
	220	return NULL;
	221	memset(ast, 0, sizeof(*ast));
	222	CDS_INIT_LIST_HEAD(&ast->allocated_nodes);
	223	ast->root.type = NODE_ROOT;
	224	return ast;
	225	}
	226
	227	static void filter_ast_free(struct filter_ast *ast)
	228	{
	229	struct filter_node node, tmp;
	230
	231	cds_list_for_each_entry_safe(node, tmp, &ast->allocated_nodes, gc)
	232	free(node);
37600d79	233	free(ast);
953192ba MD	234	}
953192ba MD	235
a187da1a	236	LTTNG_HIDDEN
953192ba MD	237	int filter_parser_ctx_append_ast(struct filter_parser_ctx *parser_ctx)
953192ba MD	238	{
9039edd4	239	return yyparse(parser_ctx, parser_ctx->scanner);
953192ba MD	240	}
953192ba MD	241
a187da1a	242	LTTNG_HIDDEN
953192ba MD	243	struct filter_parser_ctx filter_parser_ctx_alloc(FILE input)
	244	{
	245	struct filter_parser_ctx *parser_ctx;
	246	int ret;
	247
	248	yydebug = filter_parser_debug;
	249
be61ee34	250	parser_ctx = zmalloc(sizeof(*parser_ctx));
953192ba MD	251	if (!parser_ctx)
	252	return NULL;
	253	memset(parser_ctx, 0, sizeof(*parser_ctx));
	254
	255	ret = yylex_init_extra(parser_ctx, &parser_ctx->scanner);
	256	if (ret) {
	257	fprintf(stderr, "yylex_init error\n");
	258	goto cleanup_parser_ctx;
	259	}
	260	/* Start processing new stream */
	261	yyrestart(input, parser_ctx->scanner);
	262
	263	parser_ctx->ast = filter_ast_alloc();
	264	if (!parser_ctx->ast)
	265	goto cleanup_lexer;
	266	CDS_INIT_LIST_HEAD(&parser_ctx->allocated_strings);
	267
	268	if (yydebug)
	269	fprintf(stdout, "parser_ctx input is a%s.\n",
	270	isatty(fileno(input)) ? "n interactive tty" :
	271	" noninteractive file");
	272
	273	return parser_ctx;
	274
	275	cleanup_lexer:
	276	ret = yylex_destroy(parser_ctx->scanner);
	277	if (!ret)
	278	fprintf(stderr, "yylex_destroy error\n");
	279	cleanup_parser_ctx:
	280	free(parser_ctx);
	281	return NULL;
	282	}
	283
a187da1a	284	LTTNG_HIDDEN
953192ba MD	285	void filter_parser_ctx_free(struct filter_parser_ctx *parser_ctx)
	286	{
	287	int ret;
	288
	289	free_strings(&parser_ctx->allocated_strings);
	290	filter_ast_free(parser_ctx->ast);
	291	ret = yylex_destroy(parser_ctx->scanner);
	292	if (ret)
	293	fprintf(stderr, "yylex_destroy error\n");
	294	free(parser_ctx);
	295	}
	296
	297	%}
	298
	299	%define api.pure
	300	/* %locations */
	301	%parse-param {struct filter_parser_ctx *parser_ctx}
9039edd4 ZT	302	%parse-param {yyscan_t scanner}
9039edd4 ZT	303	%lex-param {yyscan_t scanner}
953192ba MD	304	%start translation_unit
	305	%token CHARACTER_CONSTANT_START SQUOTE STRING_LITERAL_START DQUOTE
	306	%token ESCSEQ CHAR_STRING_TOKEN
e90d8561	307	%token DECIMAL_CONSTANT OCTAL_CONSTANT HEXADECIMAL_CONSTANT FLOAT_CONSTANT
953192ba MD	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
	314
586dc72f	315	%token <gs> IDENTIFIER GLOBAL_IDENTIFIER
953192ba MD	316	%token ERROR
	317	%union
	318	{
	319	long long ll;
	320	char c;
	321	struct gc_string *gs;
	322	struct filter_node *n;
	323	}
	324
	325	%type <gs> s_char s_char_sequence c_char c_char_sequence
	326
	327	%type <n> primary_expression
bff988fa MD	328	%type <n> prefix_expression
bff988fa MD	329	%type <n> prefix_expression_rec
953192ba MD	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
	343	%type <n> expression
bff988fa	344	%type <n> identifiers
953192ba MD	345
	346	%%
	347
	348
	349	/* 1.5 Constants */
	350
	351	c_char_sequence:
	352	c_char
	353	{ $$ = $1; }
	354	\| c_char_sequence c_char
	355	{ $$ = gc_string_append(parser_ctx, $1, $2); }
	356	;
	357
	358	c_char:
	359	CHAR_STRING_TOKEN
	360	{ $$ = yylval.gs; }
	361	\| ESCSEQ
	362	{
	363	parse_error(parser_ctx, "escape sequences not supported yet");
	364	}
	365	;
	366
	367	/* 1.6 String literals */
	368
	369	s_char_sequence:
	370	s_char
	371	{ $$ = $1; }
	372	\| s_char_sequence s_char
	373	{ $$ = gc_string_append(parser_ctx, $1, $2); }
	374	;
	375
	376	s_char:
	377	CHAR_STRING_TOKEN
	378	{ $$ = yylval.gs; }
	379	\| ESCSEQ
	380	{
	381	parse_error(parser_ctx, "escape sequences not supported yet");
	382	}
	383	;
	384
bff988fa MD	385	primary_expression:
bff988fa MD	386	DECIMAL_CONSTANT
953192ba MD	387	{
	388	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	389	$$->u.expression.type = AST_EXP_CONSTANT;
8ab7c0d9 MD	390	if (sscanf(yylval.gs->s, "%" WIDTH_u64_SCANF_IS_A_BROKEN_API SCNu64,
	391	&$$->u.expression.u.constant) != 1) {
	392	parse_error(parser_ctx, "cannot scanf decimal constant");
	393	}
953192ba MD	394	}
	395	\| OCTAL_CONSTANT
	396	{
	397	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	398	$$->u.expression.type = AST_EXP_CONSTANT;
8ab7c0d9 MD	399	if (!strcmp(yylval.gs->s, "0")) {
	400	$$->u.expression.u.constant = 0;
	401	} else if (sscanf(yylval.gs->s, "0%" WIDTH_o64_SCANF_IS_A_BROKEN_API SCNo64,
	402	&$$->u.expression.u.constant) != 1) {
	403	parse_error(parser_ctx, "cannot scanf octal constant");
	404	}
953192ba MD	405	}
	406	\| HEXADECIMAL_CONSTANT
	407	{
	408	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	409	$$->u.expression.type = AST_EXP_CONSTANT;
8ab7c0d9 MD	410	if (sscanf(yylval.gs->s, "0x%" WIDTH_x64_SCANF_IS_A_BROKEN_API SCNx64,
	411	&$$->u.expression.u.constant) != 1) {
	412	parse_error(parser_ctx, "cannot scanf hexadecimal constant");
	413	}
953192ba	414	}
e90d8561 MD	415	\| FLOAT_CONSTANT
	416	{
	417	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	418	$$->u.expression.type = AST_EXP_FLOAT_CONSTANT;
8ab7c0d9 MD	419	if (sscanf(yylval.gs->s, "%" WIDTH_lg_SCANF_IS_A_BROKEN_API "lg",
	420	&$$->u.expression.u.float_constant) != 1) {
	421	parse_error(parser_ctx, "cannot scanf float constant");
	422	}
e90d8561	423	}
953192ba MD	424	\| STRING_LITERAL_START DQUOTE
	425	{
	426	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	427	$$->u.expression.type = AST_EXP_STRING;
	428	$$->u.expression.u.string = "";
	429	}
	430	\| STRING_LITERAL_START s_char_sequence DQUOTE
	431	{
	432	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	433	$$->u.expression.type = AST_EXP_STRING;
	434	$$->u.expression.u.string = $2->s;
	435	}
	436	\| CHARACTER_CONSTANT_START c_char_sequence SQUOTE
	437	{
	438	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	439	$$->u.expression.type = AST_EXP_STRING;
	440	$$->u.expression.u.string = $2->s;
	441	}
	442	\| LPAREN expression RPAREN
	443	{
	444	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	445	$$->u.expression.type = AST_EXP_NESTED;
	446	$$->u.expression.u.child = $2;
	447	}
	448	;
	449
bff988fa MD	450	identifiers
	451	: IDENTIFIER
	452	{
	453	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	454	$$->u.expression.type = AST_EXP_IDENTIFIER;
	455	$$->u.expression.u.identifier = yylval.gs->s;
	456	}
	457	\| GLOBAL_IDENTIFIER
	458	{
	459	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	460	$$->u.expression.type = AST_EXP_GLOBAL_IDENTIFIER;
	461	$$->u.expression.u.identifier = yylval.gs->s;
	462	}
	463	;
	464
	465	prefix_expression_rec
	466	: LSBRAC unary_expression RSBRAC
	467	{
	468	$$ = $2;
	469	}
	470	\| LSBRAC unary_expression RSBRAC prefix_expression_rec
	471	{
	472	$$ = $2;
	473	$$->u.expression.pre_op = AST_LINK_BRACKET;
	474	$$->u.expression.prev = $4;
	475	}
	476	;
	477
	478	prefix_expression
	479	: identifiers
	480	{
	481	$$ = $1;
	482	}
	483	\| identifiers prefix_expression_rec
661dfdd1 MD	484	{
	485	$$ = $1;
	486	$$->u.expression.pre_op = AST_LINK_BRACKET;
bff988fa	487	$$->u.expression.next_bracket = $2;
661dfdd1	488	}
bff988fa MD	489	;
	490
	491	postfix_expression
	492	: prefix_expression
953192ba	493	{
bff988fa MD	494	$$ = $1;
	495	}
	496	\| postfix_expression DOT prefix_expression
	497	{
	498	$$ = $3;
953192ba	499	$$->u.expression.post_op = AST_LINK_DOT;
953192ba MD	500	$$->u.expression.prev = $1;
953192ba MD	501	}
bff988fa	502	\| postfix_expression RARROW prefix_expression
953192ba	503	{
bff988fa	504	$$ = $3;
953192ba	505	$$->u.expression.post_op = AST_LINK_RARROW;
953192ba MD	506	$$->u.expression.prev = $1;
	507	}
	508	;
	509
	510	unary_expression
	511	: postfix_expression
	512	{ $$ = $1; }
bff988fa MD	513	\| primary_expression
bff988fa MD	514	{ $$ = $1; }
953192ba MD	515	\| unary_operator unary_expression
	516	{
	517	$$ = $1;
	518	$$->u.unary_op.child = $2;
	519	}
	520	;
	521
	522	unary_operator
	523	: PLUS
	524	{
	525	$$ = make_node(parser_ctx, NODE_UNARY_OP);
	526	$$->u.unary_op.type = AST_UNARY_PLUS;
	527	}
	528	\| MINUS
	529	{
	530	$$ = make_node(parser_ctx, NODE_UNARY_OP);
	531	$$->u.unary_op.type = AST_UNARY_MINUS;
	532	}
	533	\| NOT_OP
	534	{
	535	$$ = make_node(parser_ctx, NODE_UNARY_OP);
	536	$$->u.unary_op.type = AST_UNARY_NOT;
	537	}
ab78f161 CB	538	\| NOT_BIN
	539	{
	540	$$ = make_node(parser_ctx, NODE_UNARY_OP);
bff988fa	541	$$->u.unary_op.type = AST_UNARY_BIT_NOT;
ab78f161	542	}
953192ba MD	543	;
	544
	545	multiplicative_expression
	546	: unary_expression
	547	{ $$ = $1; }
	548	\| multiplicative_expression STAR unary_expression
	549	{
	550	$$ = make_op_node(parser_ctx, AST_OP_MUL, $1, $3);
	551	}
	552	\| multiplicative_expression DIV_OP unary_expression
	553	{
	554	$$ = make_op_node(parser_ctx, AST_OP_DIV, $1, $3);
	555	}
	556	\| multiplicative_expression MOD_OP unary_expression
	557	{
	558	$$ = make_op_node(parser_ctx, AST_OP_MOD, $1, $3);
	559	}
	560	;
	561
	562	additive_expression
	563	: multiplicative_expression
	564	{ $$ = $1; }
	565	\| additive_expression PLUS multiplicative_expression
	566	{
	567	$$ = make_op_node(parser_ctx, AST_OP_PLUS, $1, $3);
	568	}
	569	\| additive_expression MINUS multiplicative_expression
	570	{
	571	$$ = make_op_node(parser_ctx, AST_OP_MINUS, $1, $3);
	572	}
	573	;
	574
	575	shift_expression
	576	: additive_expression
	577	{ $$ = $1; }
	578	\| shift_expression LEFT_OP additive_expression
	579	{
116d3c01	580	$$ = make_op_node(parser_ctx, AST_OP_BIT_LSHIFT, $1, $3);
953192ba MD	581	}
	582	\| shift_expression RIGHT_OP additive_expression
	583	{
116d3c01	584	$$ = make_op_node(parser_ctx, AST_OP_BIT_RSHIFT, $1, $3);
953192ba MD	585	}
	586	;
	587
831b702b	588	and_expression
953192ba MD	589	: shift_expression
953192ba MD	590	{ $$ = $1; }
831b702b MD	591	\| and_expression AND_BIN shift_expression
	592	{
	593	$$ = make_op_node(parser_ctx, AST_OP_BIT_AND, $1, $3);
	594	}
	595	;
	596
	597	exclusive_or_expression
	598	: and_expression
	599	{ $$ = $1; }
	600	\| exclusive_or_expression XOR_BIN and_expression
	601	{
	602	$$ = make_op_node(parser_ctx, AST_OP_BIT_XOR, $1, $3);
	603	}
	604	;
	605
	606	inclusive_or_expression
	607	: exclusive_or_expression
	608	{ $$ = $1; }
	609	\| inclusive_or_expression OR_BIN exclusive_or_expression
	610	{
	611	$$ = make_op_node(parser_ctx, AST_OP_BIT_OR, $1, $3);
	612	}
	613	;
	614
	615	relational_expression
	616	: inclusive_or_expression
	617	{ $$ = $1; }
	618	\| relational_expression LT_OP inclusive_or_expression
953192ba MD	619	{
	620	$$ = make_op_node(parser_ctx, AST_OP_LT, $1, $3);
	621	}
831b702b	622	\| relational_expression GT_OP inclusive_or_expression
953192ba MD	623	{
	624	$$ = make_op_node(parser_ctx, AST_OP_GT, $1, $3);
	625	}
831b702b	626	\| relational_expression LE_OP inclusive_or_expression
953192ba MD	627	{
	628	$$ = make_op_node(parser_ctx, AST_OP_LE, $1, $3);
	629	}
831b702b	630	\| relational_expression GE_OP inclusive_or_expression
953192ba MD	631	{
	632	$$ = make_op_node(parser_ctx, AST_OP_GE, $1, $3);
	633	}
	634	;
	635
	636	equality_expression
	637	: relational_expression
	638	{ $$ = $1; }
	639	\| equality_expression EQ_OP relational_expression
	640	{
	641	$$ = make_op_node(parser_ctx, AST_OP_EQ, $1, $3);
	642	}
	643	\| equality_expression NE_OP relational_expression
	644	{
	645	$$ = make_op_node(parser_ctx, AST_OP_NE, $1, $3);
	646	}
	647	;
	648
953192ba	649	logical_and_expression
831b702b	650	: equality_expression
953192ba	651	{ $$ = $1; }
831b702b	652	\| logical_and_expression AND_OP equality_expression
953192ba MD	653	{
	654	$$ = make_op_node(parser_ctx, AST_OP_AND, $1, $3);
	655	}
	656	;
	657
	658	logical_or_expression
	659	: logical_and_expression
	660	{ $$ = $1; }
	661	\| logical_or_expression OR_OP logical_and_expression
	662	{
	663	$$ = make_op_node(parser_ctx, AST_OP_OR, $1, $3);
	664	}
	665	;
	666
	667	expression
	668	: logical_or_expression
	669	{ $$ = $1; }
	670	;
	671
	672	translation_unit
	673	: expression
	674	{
	675	parser_ctx->ast->root.u.root.child = $1;
	676	}
	677	;