chore: update c, h, md files

include/ast.h

@@ -1,6 +1,6 @@
 /* retoor <retoor@molodetz.nl> */
-#ifndef LOREG_AST_H
-#define LOREG_AST_H
+#ifndef LOREX_AST_H
+#define LOREX_AST_H
 
 #include <stdbool.h>
 #include <stddef.h>
@@ -36,6 +36,8 @@ typedef struct {
     size_t count;
     size_t capacity;
     bool negated;
+    unsigned char bitmap[32];
+    bool bitmap_valid;
 } bracket_class_t;
 
 typedef struct {

include/lexer.h

@@ -1,6 +1,6 @@
 /* retoor <retoor@molodetz.nl> */
-#ifndef LOREG_LEXER_H
-#define LOREG_LEXER_H
+#ifndef LOREX_LEXER_H
+#define LOREX_LEXER_H
 
 #include <stddef.h>
 #include <stdbool.h>

include/loreg.h

@@ -1,45 +0,0 @@
-/* retoor <retoor@molodetz.nl> */
-#ifndef LOREG_H
-#define LOREG_H
-
-#include <stddef.h>
-#include <stdbool.h>
-
-#define LOREG_VERSION "1.0.0"
-#define LOREG_MAX_STATES 4096
-#define LOREG_MAX_GROUPS 32
-
-typedef enum {
-    LOREG_OK = 0,
-    LOREG_ERR_INVALID_PATTERN,
-    LOREG_ERR_UNBALANCED_PAREN,
-    LOREG_ERR_EMPTY_GROUP,
-    LOREG_ERR_INVALID_QUANTIFIER,
-    LOREG_ERR_INVALID_ESCAPE,
-    LOREG_ERR_OUT_OF_MEMORY,
-    LOREG_ERR_STATE_OVERFLOW
-} loreg_error_t;
-
-typedef struct {
-    size_t start;
-    size_t end;
-    bool matched;
-} loreg_group_t;
-
-typedef struct {
-    bool matched;
-    size_t match_start;
-    size_t match_end;
-    loreg_group_t groups[LOREG_MAX_GROUPS];
-    size_t group_count;
-} loreg_match_t;
-
-typedef struct loreg_regex loreg_regex_t;
-
-loreg_regex_t *loreg_compile(const char *pattern, loreg_error_t *error);
-void loreg_free(loreg_regex_t *regex);
-bool loreg_match(loreg_regex_t *regex, const char *text, loreg_match_t *result);
-bool loreg_search(loreg_regex_t *regex, const char *text, loreg_match_t *result);
-const char *loreg_error_string(loreg_error_t error);
-
-#endif

include/lorex.h

@@ -0,0 +1,45 @@
+/* retoor <retoor@molodetz.nl> */
+#ifndef LOREX_H
+#define LOREX_H
+
+#include <stddef.h>
+#include <stdbool.h>
+
+#define LOREX_VERSION "1.0.0"
+#define LOREX_MAX_STATES 4096
+#define LOREX_MAX_GROUPS 32
+
+typedef enum {
+    LOREX_OK = 0,
+    LOREX_ERR_INVALID_PATTERN,
+    LOREX_ERR_UNBALANCED_PAREN,
+    LOREX_ERR_EMPTY_GROUP,
+    LOREX_ERR_INVALID_QUANTIFIER,
+    LOREX_ERR_INVALID_ESCAPE,
+    LOREX_ERR_OUT_OF_MEMORY,
+    LOREX_ERR_STATE_OVERFLOW
+} lorex_error_t;
+
+typedef struct {
+    size_t start;
+    size_t end;
+    bool matched;
+} lorex_group_t;
+
+typedef struct {
+    bool matched;
+    size_t match_start;
+    size_t match_end;
+    lorex_group_t groups[LOREX_MAX_GROUPS];
+    size_t group_count;
+} lorex_match_t;
+
+typedef struct lorex_regex lorex_regex_t;
+
+lorex_regex_t *lorex_compile(const char *pattern, lorex_error_t *error);
+void lorex_free(lorex_regex_t *regex);
+bool lorex_match(lorex_regex_t *regex, const char *text, lorex_match_t *result);
+bool lorex_search(lorex_regex_t *regex, const char *text, lorex_match_t *result);
+const char *lorex_error_string(lorex_error_t error);
+
+#endif

include/matcher.h

@@ -1,9 +1,9 @@
 /* retoor <retoor@molodetz.nl> */
-#ifndef LOREG_MATCHER_H
-#define LOREG_MATCHER_H
+#ifndef LOREX_MATCHER_H
+#define LOREX_MATCHER_H
 
 #include "nfa.h"
-#include "loreg.h"
+#include "lorex.h"
 
 typedef struct {
     nfa_state_t **states;
@@ -20,7 +20,13 @@ void state_set_clear(state_set_t *set);
 void state_set_add(state_set_t *set, nfa_state_t *state);
 bool state_set_contains(state_set_t *set, nfa_state_t *state);
 
-bool nfa_match(nfa_t *nfa, const char *text, size_t start_pos, loreg_match_t *result);
-bool nfa_search(nfa_t *nfa, const char *text, loreg_match_t *result);
+typedef struct match_ctx match_ctx_t;
+
+match_ctx_t *match_ctx_create(nfa_t *nfa);
+void match_ctx_free(match_ctx_t *ctx);
+
+bool nfa_match(nfa_t *nfa, const char *text, size_t start_pos, lorex_match_t *result);
+bool nfa_match_with_ctx(nfa_t *nfa, const char *text, size_t start_pos, lorex_match_t *result, match_ctx_t *ctx);
+bool nfa_search(nfa_t *nfa, const char *text, lorex_match_t *result);
 
 #endif

include/nfa.h

@@ -1,9 +1,9 @@
 /* retoor <retoor@molodetz.nl> */
-#ifndef LOREG_NFA_H
-#define LOREG_NFA_H
+#ifndef LOREX_NFA_H
+#define LOREX_NFA_H
 
 #include "ast.h"
-#include "loreg.h"
+#include "lorex.h"
 #include <stdbool.h>
 #include <stddef.h>
 
@@ -30,19 +30,19 @@ typedef enum {
 } transition_type_t;
 
 typedef struct {
-    transition_type_t type;
-    char value;
     nfa_state_t *target;
     bracket_class_t *bracket;
+    transition_type_t type;
     int group_id;
+    char value;
 } transition_t;
 
 struct nfa_state {
-    int id;
-    bool accepting;
     transition_t *transitions;
     size_t trans_count;
     size_t trans_capacity;
+    int id;
+    bool accepting;
 };
 
 typedef struct {
@@ -52,10 +52,22 @@ typedef struct {
 
 typedef struct {
     nfa_state_t **states;
+    nfa_state_t *start;
+    char *literal_prefix;
+    char *literal_suffix;
     size_t state_count;
     size_t capacity;
-    nfa_state_t *start;
+    size_t prefix_len;
+    size_t suffix_len;
     int group_count;
+    char single_first_char;
+    bool anchored_start;
+    bool anchored_end;
+    bool first_chars_valid;
+    bool is_pure_literal;
+    bool has_alt_dispatch;
+    unsigned char first_chars[32];
+    unsigned char alt_dispatch[256];
 } nfa_t;
 
 nfa_t *nfa_create(void);
@@ -64,6 +76,6 @@ nfa_state_t *nfa_add_state(nfa_t *nfa);
 void nfa_add_transition(nfa_state_t *from, nfa_state_t *to, transition_type_t type, char value);
 void nfa_add_bracket_transition(nfa_state_t *from, nfa_state_t *to, bracket_class_t *bracket);
 void nfa_add_group_transition(nfa_state_t *from, nfa_state_t *to, transition_type_t type, int group_id);
-nfa_t *nfa_from_ast(ast_node_t *ast, loreg_error_t *error);
+nfa_t *nfa_from_ast(ast_node_t *ast, lorex_error_t *error);
 
 #endif

include/parser.h

@@ -1,20 +1,20 @@
 /* retoor <retoor@molodetz.nl> */
-#ifndef LOREG_PARSER_H
-#define LOREG_PARSER_H
+#ifndef LOREX_PARSER_H
+#define LOREX_PARSER_H
 
 #include "ast.h"
 #include "lexer.h"
-#include "loreg.h"
+#include "lorex.h"
 
 typedef struct {
     lexer_t lexer;
     token_t current;
-    loreg_error_t error;
+    lorex_error_t error;
     int group_count;
 } parser_t;
 
 void parser_init(parser_t *parser, const char *pattern);
 ast_node_t *parser_parse(parser_t *parser);
-loreg_error_t parser_get_error(parser_t *parser);
+lorex_error_t parser_get_error(parser_t *parser);
 
 #endif

include/repl.h

@@ -1,6 +1,6 @@
 /* retoor <retoor@molodetz.nl> */
-#ifndef LOREG_REPL_H
-#define LOREG_REPL_H
+#ifndef LOREX_REPL_H
+#define LOREX_REPL_H
 
 void repl_run(void);