summaryrefslogtreecommitdiff
path: root/src/analysis/scan/grammar.y
diff options
context:
space:
mode:
Diffstat (limited to 'src/analysis/scan/grammar.y')
-rw-r--r--src/analysis/scan/grammar.y865
1 files changed, 670 insertions, 195 deletions
diff --git a/src/analysis/scan/grammar.y b/src/analysis/scan/grammar.y
index e1f0e9e..2d985a7 100644
--- a/src/analysis/scan/grammar.y
+++ b/src/analysis/scan/grammar.y
@@ -6,10 +6,10 @@
/* Affiche un message d'erreur suite à l'analyse en échec. */
-static int yyerror(GContentScanner *, yyscan_t, GScanRule **, sized_string_t *, sized_string_t *, void/*GBytesPattern*/ **, char **, size_t *, size_t *, char *);
+static int yyerror(GContentScanner *, yyscan_t, GScanRule **, sized_string_t *, sized_string_t *, char *);
#define raise_error(msg) \
- yyerror(scanner, yyscanner, built_rule, tmp_0, tmp_1, NULL, buf, allocated, used, msg)
+ yyerror(scanner, yyscanner, built_rule, tmp_0, tmp_1, msg)
%}
@@ -30,7 +30,7 @@ typedef void *yyscan_t;
#include "exprs/access.h"
#include "exprs/arithmetic.h"
#include "exprs/call.h"
-#include "exprs/counter.h"
+#include "exprs/extract.h"
#include "exprs/handler.h"
#include "exprs/intersect.h"
#include "exprs/item.h"
@@ -40,8 +40,10 @@ typedef void *yyscan_t;
#include "exprs/setcounter.h"
#include "exprs/relational.h"
#include "exprs/strop.h"
+#include "patterns/customizer.h"
#include "patterns/modifier.h"
#include "patterns/modifiers/list.h"
+#include "patterns/modifiers/pipe.h"
#include "patterns/tokens/hex.h"
#include "patterns/tokens/plain.h"
#include "patterns/tokens/nodes/any.h"
@@ -74,15 +76,14 @@ typedef void *yyscan_t;
} masked;
ScanRuleFlags rule_flags; /* Fanions pour règle */
- GScanRule *rule; /* Nouvelle règle à intégrer */
GScanTokenNode *node; /* Bribe de motif à intégrer */
GSearchPattern *pattern; /* Nouveau motif à considérer */
GScanTokenModifier *modifier; /* Modificateur pour texte */
+ modifier_arg_t mod_arg; /* Argument pour modificateur */
ScanPlainNodeFlags str_flags; /* Fanions pour texte */
-
GScanExpression *expr; /* Expression de condition */
struct {
@@ -100,13 +101,13 @@ typedef void *yyscan_t;
%define api.pure full
-%parse-param { GContentScanner *scanner } { yyscan_t yyscanner } { GScanRule **built_rule } { sized_string_t *tmp_0} { sized_string_t *tmp_1} { void /*GBytesPattern*/ **built_pattern } { char **buf } { size_t *allocated } { size_t *used }
-%lex-param { yyscan_t yyscanner } { sized_string_t *tmp_0} { sized_string_t *tmp_1} { void/*GBytesPattern*/ **built_pattern } { char **buf } { size_t *allocated } { size_t *used }
+%parse-param { GContentScanner *scanner } { yyscan_t yyscanner } { GScanRule **built_rule } { sized_string_t *tmp_0} { sized_string_t *tmp_1}
+%lex-param { yyscan_t yyscanner } { sized_string_t *tmp_0} { sized_string_t *tmp_1}
%code provides {
#define YY_DECL \
- int rost_lex(YYSTYPE *yylval_param, yyscan_t yyscanner, sized_string_t *tmp_0, sized_string_t *tmp_1, void/*GBytesPattern*/ **built_pattern, char **buf, size_t *allocated, size_t *used)
+ int rost_lex(YYSTYPE *yylval_param, yyscan_t yyscanner, sized_string_t *tmp_0, sized_string_t *tmp_1)
YY_DECL;
@@ -115,8 +116,8 @@ YY_DECL;
%token INCLUDE "include"
-%token RAW_RULE
-%token RULE_NAME
+%token RAW_RULE "rule"
+%token RULE_IDENTIFIER
%token META "meta"
%token BYTES "bytes"
@@ -129,9 +130,13 @@ YY_DECL;
%token BYTES_ID
%token BYTES_FUZZY_ID
%token BYTES_ID_COUNTER
+%token BYTES_FUZZY_ID_COUNTER
%token BYTES_ID_START
+%token BYTES_FUZZY_ID_START
%token BYTES_ID_LENGTH
+%token BYTES_FUZZY_ID_LENGTH
%token BYTES_ID_END
+%token BYTES_FUZZY_ID_END
%token NAME
@@ -152,7 +157,8 @@ YY_DECL;
-%token BRACE_IN BRACE_OUT
+%token BRACE_IN "{"
+%token BRACE_OUT "}"
%token ASSIGN "="
%token COLON ":"
@@ -164,7 +170,6 @@ YY_DECL;
%token FALSE_ "false"
%token SIGNED_INTEGER
%token UNSIGNED_INTEGER
-%token STRING
%token KB MB GB
@@ -198,8 +203,6 @@ YY_DECL;
%token HOOK_O "["
%token HOOK_C "]"
-%token BRACKET_O "{"
-%token BRACKET_C "}"
%token QUESTION "?"
%token PAREN_O "("
@@ -208,6 +211,9 @@ YY_DECL;
%token DOT "."
%token PIPE "|"
+%token MOD_GROUP_O "(("
+%token MOD_GROUP_C "))"
+
%token NONE "none"
%token ANY "any"
%token ALL "all"
@@ -216,26 +222,28 @@ YY_DECL;
%token IN "in"
-%type <sized_cstring> RULE_NAME
+%type <sized_cstring> RULE_IDENTIFIER
%type <sized_cstring> INFO_KEY
%type <sized_cstring> BYTES_ID
%type <sized_cstring> BYTES_FUZZY_ID
%type <sized_cstring> BYTES_ID_COUNTER
+%type <sized_cstring> BYTES_FUZZY_ID_COUNTER
%type <sized_cstring> BYTES_ID_START
+%type <sized_cstring> BYTES_FUZZY_ID_START
%type <sized_cstring> BYTES_ID_LENGTH
+%type <sized_cstring> BYTES_FUZZY_ID_LENGTH
%type <sized_cstring> BYTES_ID_END
+%type <sized_cstring> BYTES_FUZZY_ID_END
%type <sized_cstring> NAME
%type <signed_integer> SIGNED_INTEGER
%type <unsigned_integer> UNSIGNED_INTEGER
-%type <sized_cstring> STRING
%type <rule_flags> rule_flags
%type <rule_flags> rule_flag
-%type <rule> rule
%type <sized_cstring> PLAIN_TEXT
%type <tmp_cstring> ESCAPED_TEXT
@@ -252,15 +260,18 @@ YY_DECL;
%type <modifier> modifiers
%type <modifier> _modifiers
-%type <modifier> chained_modifiers
+ //%type <modifier> chained_modifiers
%type <modifier> mod_stage
%type <modifier> modifier
+%type <modifier> modifier_args
+%type <mod_arg> modifier_arg
%type <str_flags> str_flags
%type <pattern> hex_pattern
%type <node> hex_tokens
+%type <node> _hex_tokens
%type <node> hex_token
%type <node> hex_range
%type <node> hex_choices
@@ -270,7 +281,8 @@ YY_DECL;
%type <expr> cexpression _cexpression
%type <expr> literal
-%type <expr> item_chain
+%type <expr> chain_items
+%type <expr> chain_item
%type <args_list> call_args
%type <expr> logical_expr
%type <expr> relational_expr
@@ -281,9 +293,9 @@ YY_DECL;
%type <expr> pattern_set_items
%type <expr> set
%type <expr> set_items
-%type <expr> set_access
%type <expr> intersection
%type <expr> pattern_handler
+%type <expr> _pattern_handler
@@ -308,16 +320,36 @@ YY_DECL;
%left HOOK_O HOOK_C
+ %destructor { g_object_unref(G_OBJECT($$)); } <node>
+
+ %destructor { g_object_unref(G_OBJECT($$)); } <pattern>
+
+ %destructor { g_object_unref(G_OBJECT($$)); } <modifier>
+
+ %destructor { g_object_unref(G_OBJECT($$)); } <expr>
+
+ %destructor {
+ size_t __i;
+
+ for (__i = 0; __i < $$.count; __i++)
+ g_object_unref(G_OBJECT($$.args[__i]));
+ if ($$.args != NULL)
+ free($$.args);
-%destructor { printf("-------- Discarding symbol %p.\n", $$); } <rule>
+ } <args_list>
%%
rules : /* empty */
| external rules
- | rule rules { g_content_scanner_add_rule(scanner, $1); }
+ | rule
+ {
+ g_content_scanner_add_rule(scanner, *built_rule);
+ g_clear_object(built_rule);
+ }
+ rules
;
@@ -346,15 +378,11 @@ YY_DECL;
* Définition de règle.
*/
- rule : rule_flags RAW_RULE RULE_NAME
+ rule : rule_flags "rule" RULE_IDENTIFIER
{
*built_rule = g_scan_rule_new($1, $3.data);
- $<rule>$ = *built_rule;
- }
- BRACE_IN meta bytes condition BRACE_OUT
- {
- $$ = $<rule>4;
}
+ tags "{" meta bytes condition "}"
;
@@ -379,6 +407,21 @@ YY_DECL;
;
+ tags : /* empty */
+ | ":" tag_list
+ ;
+
+ tag_list : RULE_IDENTIFIER
+ {
+ g_scan_rule_add_tag(*built_rule, $1.data);
+ }
+ | tag_list RULE_IDENTIFIER
+ {
+ g_scan_rule_add_tag(*built_rule, $2.data);
+ }
+ ;
+
+
/**
* Section "meta:" d'une définition de règle.
*/
@@ -449,7 +492,7 @@ YY_DECL;
* Définition de motif en texte brut.
*/
- str_pattern : BYTES_ID ASSIGN PLAIN_TEXT modifiers str_flags
+ str_pattern : BYTES_ID "=" PLAIN_TEXT modifiers str_flags
{
GScanTokenNode *node;
@@ -461,7 +504,7 @@ YY_DECL;
g_object_unref(G_OBJECT(node));
}
- | BYTES_ID ASSIGN ESCAPED_TEXT modifiers str_flags
+ | BYTES_ID "=" ESCAPED_TEXT modifiers str_flags
{
GScanTokenNode *node;
@@ -488,7 +531,9 @@ YY_DECL;
{
$$ = $1;
+
// if (...) useless
+ // ex : xxx | { yyy zzz }
}
;
@@ -497,19 +542,45 @@ YY_DECL;
{
$$ = $1;
}
- | chained_modifiers
+ | _modifiers "|" mod_stage
{
- $$ = $1;
+ bool status;
+
+ if (G_IS_SCAN_MODIFIER_PIPE($1))
+ $$ = $1;
+ else
+ {
+ $$ = g_scan_modifier_pipe_new();
+ g_scan_modifier_pipe_add(G_SCAN_MODIFIER_PIPE($$), $1);
+ g_object_unref(G_OBJECT($1));
+ }
+
+ g_scan_modifier_pipe_add(G_SCAN_MODIFIER_PIPE($$), $3);
+ g_object_unref(G_OBJECT($3));
+
}
;
- chained_modifiers : _modifiers "|" _modifiers
+/*
+ chained_modifiers : modifiers "|" modifiers
+ {
+ printf("need chains....\n");
+
+ $$ = NULL;
+
+ }
;
+*/
mod_stage : modifier
{
$$ = $1;
}
+ | "((" _modifiers "))"
+ {
+ $$ = NULL;
+ YYERROR; /* TODO */
+ }
| mod_stage modifier
{
bool status;
@@ -520,6 +591,7 @@ YY_DECL;
{
$$ = g_scan_modifier_list_new();
g_scan_modifier_list_add(G_SCAN_MODIFIER_LIST($$), $1);
+ g_object_unref(G_OBJECT($1));
}
status = g_scan_modifier_list_add(G_SCAN_MODIFIER_LIST($$), $2);
@@ -527,21 +599,22 @@ YY_DECL;
{
if (1)
log_simple_message(LMT_WARNING, "modifier already taken into account!");
- g_object_unref(G_OBJECT($2));
}
+ g_object_unref(G_OBJECT($2));
+
}
;
modifier : NAME
{
- $$ = find_scan_token_modifiers_for_name($1.data);
+ $$ = find_scan_token_modifiers_for_name(&$1);
if ($$ == NULL)
{
char *_msg;
int _ret;
- _ret = asprintf(&_msg, _("Unknown modifier: \"%s\""), $1.data);
+ _ret = asprintf(&_msg, _("Unknown modifier: \"%.*s\""), (int)$1.len, $1.data);
if (_ret != -1)
{
@@ -552,9 +625,74 @@ YY_DECL;
YYERROR;
}
}
- | "(" chained_modifiers ")"
+ | NAME "(" modifier_args ")"
{
- $$ = $2;
+ GScanTokenModifier *_mod;
+ bool _status;
+
+ $$ = $3;
+
+ _mod = find_scan_token_modifiers_for_name(&$1);
+ if (_mod == NULL)
+ {
+ char *_msg;
+ int _ret;
+
+ _ret = asprintf(&_msg, _("Unknown modifier: \"%.*s\""), (int)$1.len, $1.data);
+
+ if (_ret != -1)
+ {
+ raise_error(_msg);
+ free(_msg);
+ }
+
+ g_object_unref(G_OBJECT($$));
+
+ YYERROR;
+ }
+
+ _status = g_scan_token_customizer_attach_modifier(G_SCAN_TOKEN_CUSTOMIZER($$), _mod);
+
+ g_object_unref(G_OBJECT(_mod));
+
+ if (!_status)
+ {
+ char *_msg;
+ int _ret;
+
+ _ret = asprintf(&_msg,
+ _("Unsupported argument for modifier: \"%.*s\""),
+ (int)$1.len, $1.data);
+ if (_ret != -1)
+ {
+ raise_error(_msg);
+ free(_msg);
+ }
+
+ g_object_unref(G_OBJECT($$));
+
+ YYERROR;
+ }
+
+ }
+ ;
+
+
+ modifier_args : modifier_arg
+ {
+ $$ = g_scan_token_customizer_new(&$1);
+ }
+ | modifier_args "," modifier_arg
+ {
+ $$ = $1;
+ g_scan_token_customizer_add_extra_arg(G_SCAN_TOKEN_CUSTOMIZER($$), &$3);
+ }
+ ;
+
+ modifier_arg : PLAIN_TEXT
+ {
+ $$.type = MAT_STRING;
+ $$.value.string = $1;
}
;
@@ -586,26 +724,49 @@ YY_DECL;
* Définition de motif en hexadécimal.
*/
- hex_pattern : BYTES_ID ASSIGN hex_tokens
+ hex_pattern : BYTES_ID "=" hex_tokens
{
$$ = g_scan_hex_bytes_new($3, false);
g_search_pattern_set_name($$, $1.data, $1.len);
}
- | BYTES_ID ASSIGN hex_tokens "private"
+ | BYTES_ID "=" hex_tokens "private"
{
$$ = g_scan_hex_bytes_new($3, true);
g_search_pattern_set_name($$, $1.data, $1.len);
}
;
- hex_tokens : hex_token
+ hex_tokens : _hex_tokens
+ {
+ $$ = $1;
+
+ if (G_IS_SCAN_TOKEN_NODE_SEQUENCE($$))
+ {
+ if (g_scan_token_node_sequence_count(G_SCAN_TOKEN_NODE_SEQUENCE($$)) == 1)
+ {
+ GScanTokenNode *node;
+
+ node = g_scan_token_node_sequence_get(G_SCAN_TOKEN_NODE_SEQUENCE($$), 0);
+
+ g_object_unref(G_OBJECT($$));
+
+ $$ = node;
+
+ }
+
+ }
+
+ }
+ ;
+
+ _hex_tokens : hex_token
{
if ($1 == NULL) YYERROR;
$$ = $1;
}
- | hex_tokens hex_token
+ | _hex_tokens hex_token
{
if ($2 == NULL) YYERROR;
@@ -665,10 +826,18 @@ YY_DECL;
}
| hex_range
{
+ if ($1 == NULL)
+ {
+ raise_error(_("Unable to build hexadecimal range"));
+ YYERROR;
+ }
+
$$ = $1;
+
}
| "~" hex_token
{
+ if ($2 == NULL) YYERROR;
$$ = g_scan_token_node_not_new($2);
}
@@ -726,6 +895,9 @@ YY_DECL;
hex_choices : hex_token "|" hex_token
{
+ if ($1 == NULL) YYERROR;
+ if ($3 == NULL) YYERROR;
+
$$ = g_scan_token_node_choice_new();
g_scan_token_node_choice_add(G_SCAN_TOKEN_NODE_CHOICE($$), $1);
g_object_unref(G_OBJECT($1));
@@ -734,6 +906,8 @@ YY_DECL;
}
| hex_choices "|" hex_token
{
+ if ($3 == NULL) YYERROR;
+
$$ = $1;
g_scan_token_node_choice_add(G_SCAN_TOKEN_NODE_CHOICE($$), $3);
g_object_unref(G_OBJECT($3));
@@ -745,7 +919,7 @@ YY_DECL;
* Définition de motif sous forme d'expression régulière
*/
- regex_pattern : BYTES_ID ASSIGN regex_tokens
+ regex_pattern : BYTES_ID "=" regex_tokens
{
}
@@ -789,7 +963,7 @@ YY_DECL;
}
;
- _regex_token : DOT
+ _regex_token : "."
{
printf("reg dot!\n");
}
@@ -851,24 +1025,24 @@ YY_DECL;
* Définition des conditions.
*/
- condition : CONDITION COLON cexpression
- {
- g_scan_rule_set_match_condition(*built_rule, $3);
- g_object_unref(G_OBJECT($3));
- }
- ;
+ condition : "condition" ":" cexpression
+ {
+ g_scan_rule_set_match_condition(*built_rule, $3);
+ g_object_unref(G_OBJECT($3));
+ }
+ ;
- cexpression : _cexpression { $$ = $1; if ($$ == NULL) { printf("ERROR !!!\n"); YYERROR; } }
+ cexpression : _cexpression { $$ = $1; if ($$ == NULL) { printf("ERROR !!!\n"); YYERROR; } }
+ ;
_cexpression : literal { $$ = $1; }
- | item_chain { $$ = $1; }
+ | chain_items { $$ = $1; }
| logical_expr { $$ = $1; }
| relational_expr { $$ = $1; }
| string_op { $$ = $1; }
| arithm_expr { $$ = $1; }
| set_match_counter { $$ = $1; }
| set { $$ = $1; }
- | set_access { $$ = $1; }
| intersection { $$ = $1; }
| pattern_handler { $$ = $1; }
| "(" cexpression ")" { $$ = $2; }
@@ -908,46 +1082,65 @@ YY_DECL;
__converted = $1 * 1073741824;
$$ = g_scan_literal_expression_new(LVT_UNSIGNED_INTEGER, &__converted);
}
- | STRING
+ | PLAIN_TEXT
{
$$ = g_scan_literal_expression_new(LVT_STRING, &$1);
}
+ | PLAIN_TEXT "[" cexpression "]"
+ {
+ GScanExpression *__src;
+ __src = g_scan_literal_expression_new(LVT_STRING, &$1);
+ $$ = g_scan_set_item_new(__src, $3);
+ g_object_unref(G_OBJECT(__src));
+ g_object_unref(G_OBJECT($3));
+ }
+ | ESCAPED_TEXT
+ {
+ $$ = g_scan_literal_expression_new(LVT_STRING, $1);
+ }
+ | ESCAPED_TEXT "[" cexpression "]"
+ {
+ GScanExpression *__src;
+ __src = g_scan_literal_expression_new(LVT_STRING, $1);
+ $$ = g_scan_set_item_new(__src, $3);
+ g_object_unref(G_OBJECT(__src));
+ g_object_unref(G_OBJECT($3));
+ }
;
- item_chain : NAME { $$ = g_scan_named_access_new(&$1); }
- | NAME "(" ")" { $$ = g_scan_pending_call_new(&$1, NULL, 0); }
- | NAME "(" call_args ")"
+
+ chain_items : chain_item
{
- size_t __i;
- $$ = g_scan_pending_call_new(&$1, $3.args, $3.count);
- for (__i = 0; __i < $3.count; __i++)
- g_object_unref(G_OBJECT($3.args[__i]));
- free($3.args);
+ $$ = $1;
}
- | item_chain "." NAME
+ | chain_items "." chain_item
{
- GScanExpression *__next;
- __next = g_scan_named_access_new(&$3);
- g_scan_named_access_attach_next(G_SCAN_NAMED_ACCESS($1), G_SCAN_NAMED_ACCESS(__next));
+ g_scan_named_access_attach_next(G_SCAN_NAMED_ACCESS($1), G_SCAN_NAMED_ACCESS($3));
+ g_object_unref(G_OBJECT($3));
$$ = $1;
}
- | item_chain "." NAME "(" ")"
+ ;
+
+ chain_item : NAME
{
- GScanExpression *__next;
- __next = g_scan_pending_call_new(&$3, NULL, 0);
- g_scan_named_access_attach_next(G_SCAN_NAMED_ACCESS($1), G_SCAN_NAMED_ACCESS(__next));
- $$ = $1;
+ $$ = g_scan_named_access_new(&$1);
}
- | item_chain "." NAME "(" call_args ")"
+ | NAME "(" ")"
+ {
+ $$ = g_scan_pending_call_new(&$1, NULL, 0);
+ }
+ | NAME "(" call_args ")"
{
- GScanExpression *__next;
size_t __i;
- __next = g_scan_pending_call_new(&$3, $5.args, $5.count);
- for (__i = 0; __i < $5.count; __i++)
- g_object_unref(G_OBJECT($5.args[__i]));
- free($5.args);
- g_scan_named_access_attach_next(G_SCAN_NAMED_ACCESS($1), G_SCAN_NAMED_ACCESS(__next));
- $$ = $1;
+ $$ = g_scan_pending_call_new(&$1, $3.args, $3.count);
+ for (__i = 0; __i < $3.count; __i++)
+ g_object_unref(G_OBJECT($3.args[__i]));
+ free($3.args);
+ }
+ | NAME "[" cexpression "]"
+ {
+ $$ = g_scan_pending_extraction_new(&$1, $3);
+ g_object_unref(G_OBJECT($3));
}
;
@@ -966,58 +1159,143 @@ YY_DECL;
}
;
- logical_expr : cexpression "and" cexpression { $$ = g_scan_logical_operation_new(BOT_AND, $1, $3); }
- | cexpression "or" cexpression { $$ = g_scan_logical_operation_new(BOT_OR, $1, $3); }
- | "not" "(" cexpression ")" { $$ = g_scan_logical_operation_new(BOT_NOT, $3, NULL); }
+ logical_expr : cexpression "and" cexpression
+ {
+ $$ = g_scan_logical_operation_new(BOT_AND, $1, $3);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
+ }
+ | cexpression "or" cexpression
+ {
+ $$ = g_scan_logical_operation_new(BOT_OR, $1, $3);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
+ }
+ | "not" "(" cexpression ")"
+ {
+ $$ = g_scan_logical_operation_new(BOT_NOT, $3, NULL);
+ g_object_unref(G_OBJECT($3));
+ }
;
-relational_expr : cexpression "<" cexpression { $$ = g_scan_relational_operation_new(RCO_LT, $1, $3); }
- | cexpression "<=" cexpression { $$ = g_scan_relational_operation_new(RCO_LE, $1, $3); }
- | cexpression "==" cexpression { $$ = g_scan_relational_operation_new(RCO_EQ, $1, $3); }
- | cexpression "!=" cexpression { $$ = g_scan_relational_operation_new(RCO_NE, $1, $3); }
- | cexpression ">" cexpression { $$ = g_scan_relational_operation_new(RCO_GT, $1, $3); }
- | cexpression ">=" cexpression { $$ = g_scan_relational_operation_new(RCO_GE, $1, $3); }
+relational_expr : cexpression "<" cexpression
+ {
+ $$ = g_scan_relational_operation_new(RCO_LT, $1, $3);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
+ }
+ | cexpression "<=" cexpression
+ {
+ $$ = g_scan_relational_operation_new(RCO_LE, $1, $3);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
+ }
+ | cexpression "==" cexpression
+ {
+ $$ = g_scan_relational_operation_new(RCO_EQ, $1, $3);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
+ }
+ | cexpression "!=" cexpression
+ {
+ $$ = g_scan_relational_operation_new(RCO_NE, $1, $3);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
+ }
+ | cexpression ">" cexpression
+ {
+ $$ = g_scan_relational_operation_new(RCO_GT, $1, $3);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
+ }
+ | cexpression ">=" cexpression
+ {
+ $$ = g_scan_relational_operation_new(RCO_GE, $1, $3);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
+ }
;
string_op : cexpression "contains" cexpression
{
$$ = g_scan_string_operation_new(SOT_CONTAINS, $1, $3, true);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
}
| cexpression "startswith" cexpression
{
$$ = g_scan_string_operation_new(SOT_STARTSWITH, $1, $3, true);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
}
| cexpression "endswith" cexpression
{
$$ = g_scan_string_operation_new(SOT_ENDSWITH, $1, $3, true);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
}
| cexpression "matches" cexpression
{
$$ = g_scan_string_operation_new(SOT_MATCHES, $1, $3, true);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
}
| cexpression "icontains" cexpression
{
$$ = g_scan_string_operation_new(SOT_CONTAINS, $1, $3, false);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
}
| cexpression "istartswith" cexpression
{
$$ = g_scan_string_operation_new(SOT_STARTSWITH, $1, $3, false);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
}
| cexpression "iendswith" cexpression
{
$$ = g_scan_string_operation_new(SOT_ENDSWITH, $1, $3, false);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
}
| cexpression "iequals" cexpression
{
$$ = g_scan_string_operation_new(SOT_IEQUALS, $1, $3, false);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
}
;
- arithm_expr : cexpression "+" cexpression { $$ = g_scan_arithmetic_operation_new(AEO_PLUS, $1, $3); }
- | cexpression "-" cexpression { $$ = g_scan_arithmetic_operation_new(AEO_MINUS, $1, $3); }
- | cexpression "*" cexpression { $$ = g_scan_arithmetic_operation_new(AEO_MUL, $1, $3); }
- | cexpression "/" cexpression { $$ = g_scan_arithmetic_operation_new(AEO_DIV, $1, $3); }
- | cexpression "%" cexpression { $$ = g_scan_arithmetic_operation_new(AEO_MOD, $1, $3); }
+ arithm_expr : cexpression "+" cexpression
+ {
+ $$ = g_scan_arithmetic_operation_new(AEO_PLUS, $1, $3);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
+ }
+ | cexpression "-" cexpression
+ {
+ $$ = g_scan_arithmetic_operation_new(AEO_MINUS, $1, $3);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
+ }
+ | cexpression "*" cexpression
+ {
+ $$ = g_scan_arithmetic_operation_new(AEO_MUL, $1, $3);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
+ }
+ | cexpression "/" cexpression
+ {
+ $$ = g_scan_arithmetic_operation_new(AEO_DIV, $1, $3);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
+ }
+ | cexpression "%" cexpression
+ {
+ $$ = g_scan_arithmetic_operation_new(AEO_MOD, $1, $3);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
+ }
;
@@ -1078,15 +1356,17 @@ relational_expr : cexpression "<" cexpression { $$ = g_scan_relational_operatio
pattern_set : "them"
{
size_t __count;
- GSearchPattern **__patterns;
- size_t __i;
+ const GSearchPattern **__patterns;
__patterns = g_scan_rule_get_local_variables(*built_rule, NULL, &__count);
- $$ = g_scan_set_match_counter_new(__patterns, __count);
+ if (__patterns == NULL)
+ {
+ raise_error(_("No pattern found for \"them\""));
+ YYERROR;
+ }
- for (__i = 0; __i < __count; __i++)
- g_object_unref(G_OBJECT(__patterns[__i]));
+ $$ = g_scan_set_match_counter_new_shared(__patterns, __count);
free(__patterns);
@@ -1099,7 +1379,7 @@ relational_expr : cexpression "<" cexpression { $$ = g_scan_relational_operatio
pattern_set_items : BYTES_ID
{
- GSearchPattern *__pat;
+ const GSearchPattern *__pat;
__pat = g_scan_rule_get_local_variable(*built_rule, $1.data);
@@ -1119,20 +1399,17 @@ relational_expr : cexpression "<" cexpression { $$ = g_scan_relational_operatio
YYERROR;
}
- $$ = g_scan_set_match_counter_new((GSearchPattern *[]) { __pat }, 1);
-
- g_object_unref(G_OBJECT(__pat));
+ $$ = g_scan_set_match_counter_new_shared((const GSearchPattern *[]) { __pat }, 1);
}
| BYTES_FUZZY_ID
{
size_t __count;
- GSearchPattern **__patterns;
- size_t __i;
+ const GSearchPattern **__patterns;
__patterns = g_scan_rule_get_local_variables(*built_rule, $1.data, &__count);
- if (__count == 0)
+ if (__patterns == NULL)
{
char *_msg;
int _ret;
@@ -1148,17 +1425,14 @@ relational_expr : cexpression "<" cexpression { $$ = g_scan_relational_operatio
YYERROR;
}
- $$ = g_scan_set_match_counter_new(__patterns, __count);
-
- for (__i = 0; __i < __count; __i++)
- g_object_unref(G_OBJECT(__patterns[__i]));
+ $$ = g_scan_set_match_counter_new_shared(__patterns, __count);
free(__patterns);
}
| pattern_set_items "," BYTES_ID
{
- GSearchPattern *__pat;
+ const GSearchPattern *__pat;
GScanSetMatchCounter *__counter;
__pat = g_scan_rule_get_local_variable(*built_rule, $3.data);
@@ -1180,9 +1454,8 @@ relational_expr : cexpression "<" cexpression { $$ = g_scan_relational_operatio
}
__counter = G_SCAN_SET_MATCH_COUNTER($1);
- g_scan_set_match_counter_add_extra_patterns(__counter, (GSearchPattern *[]) { __pat }, 1);
-
- g_object_unref(G_OBJECT(__pat));
+ g_scan_set_match_counter_add_extra_shared_patterns(__counter,
+ (const GSearchPattern *[]) { __pat }, 1);
$$ = $1;
@@ -1190,13 +1463,12 @@ relational_expr : cexpression "<" cexpression { $$ = g_scan_relational_operatio
| pattern_set_items "," BYTES_FUZZY_ID
{
size_t __count;
- GSearchPattern **__patterns;
+ const GSearchPattern **__patterns;
GScanSetMatchCounter *__counter;
- size_t __i;
__patterns = g_scan_rule_get_local_variables(*built_rule, $3.data, &__count);
- if (__count == 0)
+ if (__patterns == NULL)
{
char *_msg;
int _ret;
@@ -1213,10 +1485,7 @@ relational_expr : cexpression "<" cexpression { $$ = g_scan_relational_operatio
}
__counter = G_SCAN_SET_MATCH_COUNTER($1);
- g_scan_set_match_counter_add_extra_patterns(__counter, __patterns, __count);
-
- for (__i = 0; __i < __count; __i++)
- g_object_unref(G_OBJECT(__patterns[__i]));
+ g_scan_set_match_counter_add_extra_shared_patterns(__counter, __patterns, __count);
free(__patterns);
@@ -1226,46 +1495,39 @@ relational_expr : cexpression "<" cexpression { $$ = g_scan_relational_operatio
;
- set : "(" ")"
- {
- $$ = g_scan_generic_set_new();
- }
- | "(" cexpression "," ")"
- {
- $$ = g_scan_generic_set_new();
- g_scan_generic_set_add_item(G_SCAN_GENERIC_SET($$), $2);
- g_object_unref(G_OBJECT($2));
- }
- | "(" set_items ")"
- {
- $$ = $2;
- }
- ;
-
- set_items : cexpression "," cexpression
- {
- $$ = g_scan_generic_set_new();
- g_scan_generic_set_add_item(G_SCAN_GENERIC_SET($$), $1);
- g_object_unref(G_OBJECT($1));
- g_scan_generic_set_add_item(G_SCAN_GENERIC_SET($$), $3);
- g_object_unref(G_OBJECT($3));
- }
- | set_items "," cexpression
- {
- $$ = $1;
- g_scan_generic_set_add_item(G_SCAN_GENERIC_SET($$), $3);
- g_object_unref(G_OBJECT($3));
- }
- ;
+ set : "(" ")"
+ {
+ $$ = g_scan_generic_set_new();
+ }
+ | "(" cexpression "," ")"
+ {
+ $$ = g_scan_generic_set_new();
+ g_scan_generic_set_add_item(G_SCAN_GENERIC_SET($$), $2);
+ g_object_unref(G_OBJECT($2));
+ }
+ | "(" set_items ")"
+ {
+ $$ = $2;
+ }
+ ;
- set_access : cexpression "[" cexpression "]"
+ set_items : cexpression "," cexpression
{
- $$ = g_scan_set_item_new($1, $3);
+ $$ = g_scan_generic_set_new();
+ g_scan_generic_set_add_item(G_SCAN_GENERIC_SET($$), $1);
g_object_unref(G_OBJECT($1));
+ g_scan_generic_set_add_item(G_SCAN_GENERIC_SET($$), $3);
+ g_object_unref(G_OBJECT($3));
+ }
+ | set_items "," cexpression
+ {
+ $$ = $1;
+ g_scan_generic_set_add_item(G_SCAN_GENERIC_SET($$), $3);
g_object_unref(G_OBJECT($3));
}
;
+
intersection : cexpression "in" cexpression
{
$$ = g_scan_sets_intersection_new($1, $3);
@@ -1274,65 +1536,290 @@ relational_expr : cexpression "<" cexpression { $$ = g_scan_relational_operatio
}
;
- pattern_handler : BYTES_ID
+
+ pattern_handler : _pattern_handler
+ {
+ $$ = $1;
+ }
+ | _pattern_handler "[" cexpression "]"
+ {
+ if (g_scan_pattern_handler_get_handler_type(G_SCAN_PATTERN_HANDLER($1)) == SHT_COUNTER)
+ {
+ raise_error("Match counts can not get indexed");
+ YYERROR;
+ }
+
+ $$ = g_scan_set_item_new($1, $3);
+ g_object_unref(G_OBJECT($1));
+ g_object_unref(G_OBJECT($3));
+
+ }
+ ;
+
+ _pattern_handler : BYTES_ID
{
- GSearchPattern *__pat;
+ const GSearchPattern *__pat;
+
__pat = g_scan_rule_get_local_variable(*built_rule, $1.data);
+
if (__pat == NULL)
- $$ = NULL;
- else
{
- $$ = g_scan_pattern_handler_new(__pat, SHT_RAW);
- g_object_unref(G_OBJECT(__pat));
+ char *_msg;
+ int _ret;
+
+ _ret = asprintf(&_msg, _("Pattern not found: \"%s\""), $1.data);
+
+ if (_ret != -1)
+ {
+ raise_error(_msg);
+ free(_msg);
+ }
+
+ YYERROR;
+ }
+
+ $$ = g_scan_pattern_handler_new_shared((const GSearchPattern *[]) { __pat }, 1, SHT_RAW);
+
+ }
+ | BYTES_FUZZY_ID
+ {
+ size_t __count;
+ const GSearchPattern **__patterns;
+
+ __patterns = g_scan_rule_get_local_variables(*built_rule, $1.data, &__count);
+
+ if (__patterns == NULL)
+ {
+ char *_msg;
+ int _ret;
+
+ _ret = asprintf(&_msg, _("Patterns not found: \"%s\""), $1.data);
+
+ if (_ret != -1)
+ {
+ raise_error(_msg);
+ free(_msg);
+ }
+
+ YYERROR;
}
+
+ $$ = g_scan_pattern_handler_new_shared(__patterns, __count, SHT_RAW);
+
+ free(__patterns);
+
}
| BYTES_ID_COUNTER
{
- GSearchPattern *__pat;
+ const GSearchPattern *__pat;
+
__pat = g_scan_rule_get_local_variable(*built_rule, $1.data);
+
if (__pat == NULL)
- $$ = NULL;
- else
{
- $$ = g_scan_match_counter_new(__pat);
- g_object_unref(G_OBJECT(__pat));
+ char *_msg;
+ int _ret;
+
+ _ret = asprintf(&_msg, _("Pattern not found: \"%s\""), $1.data);
+
+ if (_ret != -1)
+ {
+ raise_error(_msg);
+ free(_msg);
+ }
+
+ YYERROR;
+ }
+
+ $$ = g_scan_pattern_handler_new_shared((const GSearchPattern *[]) { __pat }, 1, SHT_COUNTER);
+
+ }
+ | BYTES_FUZZY_ID_COUNTER
+ {
+ size_t __count;
+ const GSearchPattern **__patterns;
+
+ __patterns = g_scan_rule_get_local_variables(*built_rule, $1.data, &__count);
+
+ if (__patterns == NULL)
+ {
+ char *_msg;
+ int _ret;
+
+ _ret = asprintf(&_msg, _("Patterns not found: \"%s\""), $1.data);
+
+ if (_ret != -1)
+ {
+ raise_error(_msg);
+ free(_msg);
+ }
+
+ YYERROR;
}
+
+ $$ = g_scan_pattern_handler_new_shared(__patterns, __count, SHT_COUNTER);
+
+ free(__patterns);
+
}
| BYTES_ID_START
{
- GSearchPattern *__pat;
+ const GSearchPattern *__pat;
+
__pat = g_scan_rule_get_local_variable(*built_rule, $1.data);
+
if (__pat == NULL)
- $$ = NULL;
- else
{
- $$ = g_scan_pattern_handler_new(__pat, SHT_START);
- g_object_unref(G_OBJECT(__pat));
+ char *_msg;
+ int _ret;
+
+ _ret = asprintf(&_msg, _("Pattern not found: \"%s\""), $1.data);
+
+ if (_ret != -1)
+ {
+ raise_error(_msg);
+ free(_msg);
+ }
+
+ YYERROR;
+ }
+
+ $$ = g_scan_pattern_handler_new_shared((const GSearchPattern *[]) { __pat }, 1, SHT_START);
+
+ }
+ | BYTES_FUZZY_ID_START
+ {
+ size_t __count;
+ const GSearchPattern **__patterns;
+
+ __patterns = g_scan_rule_get_local_variables(*built_rule, $1.data, &__count);
+
+ if (__patterns == NULL)
+ {
+ char *_msg;
+ int _ret;
+
+ _ret = asprintf(&_msg, _("Patterns not found: \"%s\""), $1.data);
+
+ if (_ret != -1)
+ {
+ raise_error(_msg);
+ free(_msg);
+ }
+
+ YYERROR;
}
+
+ $$ = g_scan_pattern_handler_new_shared(__patterns, __count, SHT_START);
+
+ free(__patterns);
+
}
| BYTES_ID_LENGTH
{
- GSearchPattern *__pat;
+ const GSearchPattern *__pat;
+
__pat = g_scan_rule_get_local_variable(*built_rule, $1.data);
+
if (__pat == NULL)
- $$ = NULL;
- else
{
- $$ = g_scan_pattern_handler_new(__pat, SHT_LENGTH);
- g_object_unref(G_OBJECT(__pat));
+ char *_msg;
+ int _ret;
+
+ _ret = asprintf(&_msg, _("Pattern not found: \"%s\""), $1.data);
+
+ if (_ret != -1)
+ {
+ raise_error(_msg);
+ free(_msg);
+ }
+
+ YYERROR;
}
+
+ $$ = g_scan_pattern_handler_new_shared((const GSearchPattern *[]) { __pat }, 1, SHT_LENGTH);
+
+ }
+ | BYTES_FUZZY_ID_LENGTH
+ {
+ size_t __count;
+ const GSearchPattern **__patterns;
+
+ __patterns = g_scan_rule_get_local_variables(*built_rule, $1.data, &__count);
+
+ if (__patterns == NULL)
+ {
+ char *_msg;
+ int _ret;
+
+ _ret = asprintf(&_msg, _("Patterns not found: \"%s\""), $1.data);
+
+ if (_ret != -1)
+ {
+ raise_error(_msg);
+ free(_msg);
+ }
+
+ YYERROR;
+ }
+
+ $$ = g_scan_pattern_handler_new_shared(__patterns, __count, SHT_LENGTH);
+
+ free(__patterns);
+
}
| BYTES_ID_END
{
- GSearchPattern *__pat;
+ const GSearchPattern *__pat;
+
__pat = g_scan_rule_get_local_variable(*built_rule, $1.data);
+
if (__pat == NULL)
- $$ = NULL;
- else
{
- $$ = g_scan_pattern_handler_new(__pat, SHT_END);
- g_object_unref(G_OBJECT(__pat));
+ char *_msg;
+ int _ret;
+
+ _ret = asprintf(&_msg, _("Pattern not found: \"%s\""), $1.data);
+
+ if (_ret != -1)
+ {
+ raise_error(_msg);
+ free(_msg);
+ }
+
+ YYERROR;
+ }
+
+ $$ = g_scan_pattern_handler_new_shared((const GSearchPattern *[]) { __pat }, 1, SHT_END);
+
+ }
+ | BYTES_FUZZY_ID_END
+ {
+ size_t __count;
+ const GSearchPattern **__patterns;
+
+ __patterns = g_scan_rule_get_local_variables(*built_rule, $1.data, &__count);
+
+ if (__patterns == NULL)
+ {
+ char *_msg;
+ int _ret;
+
+ _ret = asprintf(&_msg, _("Patterns not found: \"%s\""), $1.data);
+
+ if (_ret != -1)
+ {
+ raise_error(_msg);
+ free(_msg);
+ }
+
+ YYERROR;
}
+
+ $$ = g_scan_pattern_handler_new_shared(__patterns, __count, SHT_END);
+
+ free(__patterns);
+
}
;
@@ -1353,9 +1840,9 @@ relational_expr : cexpression "<" cexpression { $$ = g_scan_relational_operatio
* *
******************************************************************************/
-static int yyerror(GContentScanner *scanner, yyscan_t yyscanner, GScanRule **built_rule, sized_string_t *tmp_0, sized_string_t *tmp_1, void/*GBytesPattern*/ **built_pattern, char **buf, size_t *allocated, size_t *used, char *msg)
+static int yyerror(GContentScanner *scanner, yyscan_t yyscanner, GScanRule **built_rule, sized_string_t *tmp_0, sized_string_t *tmp_1, char *msg)
{
- printf("YYERROR line %d: %s\n", yyget_lineno(yyscanner), msg);
+ //printf("YYERROR line %d: %s\n", yyget_lineno(yyscanner), msg);
return 0;
@@ -1382,10 +1869,6 @@ bool process_rules_definitions(GContentScanner *scanner, const char *text, size_
GScanRule *built_rule; /* Règle en construction */
sized_string_t tmp_0; /* Zone tampon #1 */
sized_string_t tmp_1; /* Zone tampon #2 */
- void /*GBytesPattern*/ *built_pattern; /* Motif en construction */
- char *buf; /* Zone de travail temporaire */
- size_t allocated; /* Taille de mémoire allouée */
- size_t used; /* Quantité utilisée */
yyscan_t lexstate; /* Gestion d'analyse lexicale */
YY_BUFFER_STATE state; /* Contexte d'analyse */
int status; /* Bilan d'une analyse */
@@ -1400,19 +1883,11 @@ bool process_rules_definitions(GContentScanner *scanner, const char *text, size_
tmp_1.data = malloc((length + 1) * sizeof(bin_t));
tmp_1.len = 0;
- built_pattern = NULL;
-
- allocated = 256;
- used = 0;
-
- buf = malloc(allocated * sizeof(char));
- buf[0] = '\0';
-
rost_lex_init(&lexstate);
state = rost__scan_bytes(text, length, lexstate);
- status = yyparse(scanner, lexstate, &built_rule, &tmp_0, &tmp_1, &built_pattern, &buf, &allocated, &used);
+ status = yyparse(scanner, lexstate, &built_rule, &tmp_0, &tmp_1);
result = (status == EXIT_SUCCESS);
@@ -1423,7 +1898,7 @@ bool process_rules_definitions(GContentScanner *scanner, const char *text, size_
exit_szstr(&tmp_0);
exit_szstr(&tmp_1);
- free(buf);
+ g_clear_object(&built_rule);
return result;
generated by cgit v0.11.2-87-g4458 at 2025-12-16 05:55:53 (GMT)