32#if defined(__GNUC__) && __GNUC__ >= 12
33#pragma GCC diagnostic ignored "-Warray-bounds"
34#pragma GCC diagnostic ignored "-Wstringop-overflow"
35#pragma GCC diagnostic ignored "-Wrestrict"
40#include "internal/sanitizers.h"
42#define WARN_BUFSIZE 256
44#define CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS
48 (( SYN_GNU_REGEX_OP | ONIG_SYN_OP_QMARK_NON_GREEDY |
49 ONIG_SYN_OP_ESC_OCTAL3 | ONIG_SYN_OP_ESC_X_HEX2 |
50 ONIG_SYN_OP_ESC_X_BRACE_HEX8 | ONIG_SYN_OP_ESC_CONTROL_CHARS |
51 ONIG_SYN_OP_ESC_C_CONTROL )
52 & ~ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END )
53 , ( ONIG_SYN_OP2_QMARK_GROUP_EFFECT |
54 ONIG_SYN_OP2_OPTION_RUBY |
55 ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP | ONIG_SYN_OP2_ESC_K_NAMED_BACKREF |
56 ONIG_SYN_OP2_ESC_G_SUBEXP_CALL |
57 ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY |
58 ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT |
59 ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT |
60 ONIG_SYN_OP2_CCLASS_SET_OP | ONIG_SYN_OP2_ESC_CAPITAL_C_BAR_CONTROL |
61 ONIG_SYN_OP2_ESC_CAPITAL_M_BAR_META | ONIG_SYN_OP2_ESC_V_VTAB |
62 ONIG_SYN_OP2_ESC_H_XDIGIT |
64 ONIG_SYN_OP2_ESC_U_HEX4 |
66 ONIG_SYN_OP2_ESC_CAPITAL_X_EXTENDED_GRAPHEME_CLUSTER |
67 ONIG_SYN_OP2_QMARK_LPAREN_CONDITION |
68 ONIG_SYN_OP2_ESC_CAPITAL_R_LINEBREAK |
69 ONIG_SYN_OP2_ESC_CAPITAL_K_KEEP |
70 ONIG_SYN_OP2_QMARK_TILDE_ABSENT )
71 , ( SYN_GNU_REGEX_BV |
72 ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV |
73 ONIG_SYN_DIFFERENT_LEN_ALT_LOOK_BEHIND |
74 ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP |
75 ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME |
76 ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY |
77 ONIG_SYN_WARN_CC_OP_NOT_ESCAPED |
78 ONIG_SYN_WARN_CC_DUP |
79 ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT )
80 , ( ONIG_OPTION_ASCII_RANGE | ONIG_OPTION_POSIX_BRACKET_ALL_RANGE |
81 ONIG_OPTION_WORD_BOUND_ALL_RANGE )
85 , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR
86 , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR
87 , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR
88 , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR
89 , (OnigCodePoint )ONIG_INEFFECTIVE_META_CHAR
95extern void onig_null_warn(
const char* s ARG_UNUSED) { }
97#ifdef DEFAULT_WARN_FUNCTION
98static OnigWarnFunc onig_warn = (OnigWarnFunc )DEFAULT_WARN_FUNCTION;
100static OnigWarnFunc onig_warn = onig_null_warn;
103#ifdef DEFAULT_VERB_WARN_FUNCTION
104static OnigWarnFunc onig_verb_warn = (OnigWarnFunc )DEFAULT_VERB_WARN_FUNCTION;
106static OnigWarnFunc onig_verb_warn = onig_null_warn;
109extern void onig_set_warn_func(OnigWarnFunc f)
114extern void onig_set_verb_warn_func(OnigWarnFunc f)
119static void CC_DUP_WARN(
ScanEnv *env, OnigCodePoint from, OnigCodePoint to);
122static unsigned int ParseDepthLimit = DEFAULT_PARSE_DEPTH_LIMIT;
125onig_get_parse_depth_limit(
void)
127 return ParseDepthLimit;
131onig_set_parse_depth_limit(
unsigned int depth)
134 ParseDepthLimit = DEFAULT_PARSE_DEPTH_LIMIT;
136 ParseDepthLimit = depth;
144 if (IS_NOT_NULL(bbuf)) {
145 if (IS_NOT_NULL(bbuf->p))
xfree(bbuf->p);
157 CHECK_NULL_RETURN_MEMERR(to);
158 r = BBUF_INIT(to, from->alloc);
159 if (r != 0)
return r;
160 to->used = from->used;
161 xmemcpy(to->p, from->p, from->used);
165#define BACKREF_REL_TO_ABS(rel_no, env) \
166 ((env)->num_mem + 1 + (rel_no))
168#define ONOFF(v,f,negative) (negative) ? ((v) &= ~(f)) : ((v) |= (f))
170#define MBCODE_START_POS(enc) \
171 (OnigCodePoint )(ONIGENC_MBC_MINLEN(enc) > 1 ? 0 : 0x80)
173#define SET_ALL_MULTI_BYTE_RANGE(enc, pbuf) \
174 add_code_range_to_buf(pbuf, env, MBCODE_START_POS(enc), ONIG_LAST_CODE_POINT)
176#define ADD_ALL_MULTI_BYTE_RANGE(enc, mbuf) do {\
177 if (! ONIGENC_IS_SINGLEBYTE(enc)) {\
178 r = SET_ALL_MULTI_BYTE_RANGE(enc, &(mbuf));\
184#define BITSET_SET_BIT_CHKDUP(bs, pos) do { \
185 if (BITSET_AT(bs, pos)) CC_DUP_WARN(env, pos, pos); \
186 BS_ROOM(bs, pos) |= BS_BIT(pos); \
189#define BITSET_IS_EMPTY(bs,empty) do {\
192 for (i = 0; i < BITSET_SIZE; i++) {\
200bitset_set_range(
ScanEnv *env, BitSetRef bs,
int from,
int to)
203 for (i = from; i <= to && i < SINGLE_BYTE_SIZE; i++) {
204 BITSET_SET_BIT_CHKDUP(bs, i);
210bitset_set_all(BitSetRef bs)
213 for (i = 0; i < BITSET_SIZE; i++) { bs[i] = ~((Bits )0); }
218bitset_invert(BitSetRef bs)
221 for (i = 0; i < BITSET_SIZE; i++) { bs[i] = ~(bs[i]); }
225bitset_invert_to(BitSetRef from, BitSetRef to)
228 for (i = 0; i < BITSET_SIZE; i++) { to[i] = ~(from[i]); }
232bitset_and(BitSetRef dest, BitSetRef bs)
235 for (i = 0; i < BITSET_SIZE; i++) { dest[i] &= bs[i]; }
239bitset_or(BitSetRef dest, BitSetRef bs)
242 for (i = 0; i < BITSET_SIZE; i++) { dest[i] |= bs[i]; }
246bitset_copy(BitSetRef dest, BitSetRef bs)
249 for (i = 0; i < BITSET_SIZE; i++) { dest[i] = bs[i]; }
252#if defined(USE_NAMED_GROUP) && !defined(USE_ST_LIBRARY)
254onig_strncmp(
const UChar* s1,
const UChar* s2,
int n)
267onig_strcpy(UChar* dest,
const UChar* src,
const UChar* end)
269 ptrdiff_t len = end - src;
271 xmemcpy(dest, src, len);
272 dest[len] = (UChar )0;
276#ifdef USE_NAMED_GROUP
285 term_len = ONIGENC_MBC_MINLEN(enc);
287 r = (UChar* )
xmalloc(slen + term_len);
288 CHECK_NULL_RETURN(r);
291 for (i = 0; i < term_len; i++)
292 r[slen + i] = (UChar )0;
303# define PFETCH_READY UChar* pfetch_prev = NULL; (void)pfetch_prev
305# define PFETCH_READY UChar* pfetch_prev
307#define PEND (p < end ? 0 : 1)
308#define PUNFETCH p = pfetch_prev
311 p += enclen(enc, p, end); \
313#define PFETCH(c) do { \
314 c = ((enc->max_enc_len == 1) ? *p : ONIGENC_MBC_TO_CODE(enc, p, end)); \
316 p += enclen(enc, p, end); \
320 p += enclen(enc, p, end); \
322#define PFETCH_S(c) do { \
323 c = ((enc->max_enc_len == 1) ? *p : ONIGENC_MBC_TO_CODE(enc, p, end)); \
324 p += enclen(enc, p, end); \
327#define PPEEK (p < end ? ONIGENC_MBC_TO_CODE(enc, p, end) : PEND_VALUE)
328#define PPEEK_IS(c) (PPEEK == (OnigCodePoint )c)
331strcat_capa(UChar* dest, UChar* dest_end,
const UChar* src,
const UChar* src_end,
337 r = (UChar* )
xrealloc(dest, capa + 1);
339 r = (UChar* )
xmalloc(capa + 1);
341 CHECK_NULL_RETURN(r);
342 onig_strcpy(r + (dest_end - dest), src, src_end);
348strcat_capa_from_static(UChar* dest, UChar* dest_end,
349 const UChar* src,
const UChar* src_end,
size_t capa)
353 r = (UChar* )
xmalloc(capa + 1);
354 CHECK_NULL_RETURN(r);
355 onig_strcpy(r, dest, dest_end);
356 onig_strcpy(r + (dest_end - dest), src, src_end);
375str_end_cmp(st_data_t xp, st_data_t yp)
383 if ((x->end - x->s) != (y->end - y->s))
389 c = (int )*p - (
int )*q;
390 if (c != 0)
return c;
398NO_SANITIZE(
"unsigned-integer-overflow",
static st_index_t str_end_hash(st_data_t xp));
401str_end_hash(st_data_t xp)
409 val = val * 997 + (int )*p++;
412 return val + (val >> 5);
415extern hash_table_type*
416onig_st_init_strend_table_with_size(st_index_t size)
423 return (hash_table_type* )
424 onig_st_init_table_with_size(&hashType, size);
428onig_st_lookup_strend(hash_table_type* table,
const UChar* str_key,
429 const UChar* end_key, hash_data_type *value)
433 key.s = (UChar* )str_key;
434 key.end = (UChar* )end_key;
436 return onig_st_lookup(table, (st_data_t )(&key), value);
440onig_st_insert_strend(hash_table_type* table,
const UChar* str_key,
441 const UChar* end_key, hash_data_type value)
447 key->s = (UChar* )str_key;
448 key->end = (UChar* )end_key;
449 result = onig_st_insert(table, (st_data_t )key, value);
459#ifdef USE_NAMED_GROUP
461# define INIT_NAME_BACKREFS_ALLOC_NUM 8
472# ifdef USE_ST_LIBRARY
475typedef st_data_t HashDataType;
479i_print_name_entry(UChar* key,
NameEntry* e,
void* arg)
484 fprintf(fp,
"%s: ", e->name);
485 if (e->back_num == 0)
487 else if (e->back_num == 1)
488 fprintf(fp,
"%d", e->back_ref1);
490 for (i = 0; i < e->back_num; i++) {
491 if (i > 0) fprintf(fp,
", ");
492 fprintf(fp,
"%d", e->back_refs[i]);
504 if (IS_NOT_NULL(t)) {
505 fprintf(fp,
"name table\n");
506 onig_st_foreach(t, (st_foreach_callback_func *)i_print_name_entry, (HashDataType )fp);
514i_free_name_entry(UChar* key,
NameEntry* e,
void* arg ARG_UNUSED)
517 if (IS_NOT_NULL(e->back_refs))
xfree(e->back_refs);
528 if (IS_NOT_NULL(t)) {
529 onig_st_foreach(t, (st_foreach_callback_func *)i_free_name_entry, 0);
540 r = names_clear(reg);
544 if (IS_NOT_NULL(t)) onig_st_free_table(t);
545 reg->name_table = (
void* )NULL;
550name_find(
regex_t* reg,
const UChar* name,
const UChar* name_end)
556 if (IS_NOT_NULL(t)) {
557 onig_st_lookup_strend(t, name, name_end, (HashDataType* )((
void* )(&e)));
563 int (*func)(
const UChar*,
const UChar*,int,
int*,
regex_t*,
void*);
573 int r = (*(arg->func))(e->name,
574 e->name + e->name_len,
576 (e->back_num > 1 ? e->back_refs : &(e->back_ref1)),
587 int (*func)(
const UChar*,
const UChar*,
int,
int*,
regex_t*,
void*),
void* arg)
593 if (IS_NOT_NULL(t)) {
598 onig_st_foreach(t, (st_foreach_callback_func *)i_names, (HashDataType )&narg);
608 if (e->back_num > 1) {
609 for (i = 0; i < e->back_num; i++) {
610 e->back_refs[i] = map[e->back_refs[i]].new_val;
613 else if (e->back_num == 1) {
614 e->back_ref1 = map[e->back_ref1].new_val;
625 if (IS_NOT_NULL(t)) {
626 onig_st_foreach(t, (st_foreach_callback_func *)i_renumber_name, (HashDataType )map);
633onig_number_of_names(
const regex_t* reg)
638 return (
int )t->num_entries;
645# define INIT_NAMES_ALLOC_NUM 8
661 if (IS_NOT_NULL(t) && t->num > 0) {
662 fprintf(fp,
"name table\n");
663 for (i = 0; i < t->num; i++) {
665 fprintf(fp,
"%s: ", e->name);
666 if (e->back_num == 0) {
669 else if (e->back_num == 1) {
670 fprintf(fp,
"%d", e->back_ref1);
673 for (j = 0; j < e->back_num; j++) {
674 if (j > 0) fprintf(fp,
", ");
675 fprintf(fp,
"%d", e->back_refs[j]);
693 if (IS_NOT_NULL(t)) {
694 for (i = 0; i < t->num; i++) {
696 if (IS_NOT_NULL(e->name)) {
702 if (IS_NOT_NULL(e->back_refs))
xfree(e->back_refs);
703 e->back_refs = (
int* )NULL;
706 if (IS_NOT_NULL(t->e)) {
721 r = names_clear(reg);
725 if (IS_NOT_NULL(t))
xfree(t);
726 reg->name_table = NULL;
731name_find(
regex_t* reg,
const UChar* name,
const UChar* name_end)
737 if (IS_NOT_NULL(t)) {
738 len = name_end - name;
739 for (i = 0; i < t->num; i++) {
741 if (len == e->name_len && onig_strncmp(name, e->name, len) == 0)
750 int (*func)(
const UChar*,
const UChar*,
int,
int*,
regex_t*,
void*),
void* arg)
756 if (IS_NOT_NULL(t)) {
757 for (i = 0; i < t->num; i++) {
759 r = (*func)(e->name, e->name + e->name_len, e->back_num,
760 (e->back_num > 1 ? e->back_refs : &(e->back_ref1)),
762 if (r != 0)
return r;
769onig_number_of_names(
const regex_t* reg)
782name_add(
regex_t* reg, UChar* name, UChar* name_end,
int backref,
ScanEnv* env)
788 if (name_end - name <= 0)
789 return ONIGERR_EMPTY_GROUP_NAME;
791 e = name_find(reg, name, name_end);
793# ifdef USE_ST_LIBRARY
795 t = onig_st_init_strend_table_with_size(5);
796 reg->name_table = (
void* )t;
799 CHECK_NULL_RETURN_MEMERR(e);
801 e->name = strdup_with_null(reg->enc, name, name_end);
802 if (IS_NULL(e->name)) {
804 return ONIGERR_MEMORY;
806 onig_st_insert_strend(t, e->name, (e->name + (name_end - name)),
809 e->name_len = name_end - name;
812 e->back_refs = (
int* )NULL;
817 alloc = INIT_NAMES_ALLOC_NUM;
819 CHECK_NULL_RETURN_MEMERR(t);
827 return ONIGERR_MEMORY;
833 else if (t->num == t->alloc) {
837 alloc = t->alloc * 2;
839 CHECK_NULL_RETURN_MEMERR(p);
844 for (i = t->num; i < t->alloc; i++) {
846 t->e[i].name_len = 0;
847 t->e[i].back_num = 0;
848 t->e[i].back_alloc = 0;
849 t->e[i].back_refs = (
int* )NULL;
854 e->name = strdup_with_null(reg->enc, name, name_end);
855 if (IS_NULL(e->name))
return ONIGERR_MEMORY;
856 e->name_len = name_end - name;
860 if (e->back_num >= 1 &&
861 ! IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME)) {
862 onig_scan_env_set_error_string(env, ONIGERR_MULTIPLEX_DEFINED_NAME,
864 return ONIGERR_MULTIPLEX_DEFINED_NAME;
868 if (e->back_num == 1) {
869 e->back_ref1 = backref;
872 if (e->back_num == 2) {
873 alloc = INIT_NAME_BACKREFS_ALLOC_NUM;
874 e->back_refs = (
int* )
xmalloc(
sizeof(
int) * alloc);
875 CHECK_NULL_RETURN_MEMERR(e->back_refs);
876 e->back_alloc = alloc;
877 e->back_refs[0] = e->back_ref1;
878 e->back_refs[1] = backref;
881 if (e->back_num > e->back_alloc) {
883 alloc = e->back_alloc * 2;
884 p = (
int* )
xrealloc(e->back_refs,
sizeof(
int) * alloc);
885 CHECK_NULL_RETURN_MEMERR(p);
887 e->back_alloc = alloc;
889 e->back_refs[e->back_num - 1] = backref;
897onig_name_to_group_numbers(
regex_t* reg,
const UChar* name,
898 const UChar* name_end,
int** nums)
900 NameEntry* e = name_find(reg, name, name_end);
902 if (IS_NULL(e))
return ONIGERR_UNDEFINED_NAME_REFERENCE;
904 switch (e->back_num) {
909 *nums = &(e->back_ref1);
912 *nums = e->back_refs;
919onig_name_to_backref_number(
regex_t* reg,
const UChar* name,
920 const UChar* name_end,
const OnigRegion *region)
924 n = onig_name_to_group_numbers(reg, name, name_end, &nums);
928 return ONIGERR_PARSER_BUG;
932 if (IS_NOT_NULL(region)) {
933 for (i = n - 1; i >= 0; i--) {
934 if (region->beg[nums[i]] != ONIG_REGION_NOTPOS)
945onig_name_to_group_numbers(
regex_t* reg,
const UChar* name,
946 const UChar* name_end,
int** nums)
948 return ONIG_NO_SUPPORT_CONFIG;
952onig_name_to_backref_number(
regex_t* reg,
const UChar* name,
953 const UChar* name_end,
const OnigRegion* region)
955 return ONIG_NO_SUPPORT_CONFIG;
960 int (*func)(
const UChar*,
const UChar*,
int,
int*,
regex_t*,
void*),
void* arg)
962 return ONIG_NO_SUPPORT_CONFIG;
966onig_number_of_names(
const regex_t* reg)
973onig_noname_group_capture_is_active(
const regex_t* reg)
975 if (ONIG_IS_OPTION_ON(reg->options, ONIG_OPTION_DONT_CAPTURE_GROUP))
978#ifdef USE_NAMED_GROUP
979 if (onig_number_of_names(reg) > 0 &&
980 IS_SYNTAX_BV(reg->syntax, ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP) &&
981 !ONIG_IS_OPTION_ON(reg->options, ONIG_OPTION_CAPTURE_GROUP)) {
990#define INIT_SCANENV_MEMNODES_ALLOC_SIZE 16
997 BIT_STATUS_CLEAR(env->capture_history);
998 BIT_STATUS_CLEAR(env->bt_mem_start);
999 BIT_STATUS_CLEAR(env->bt_mem_end);
1000 BIT_STATUS_CLEAR(env->backrefed_mem);
1001 env->error = (UChar* )NULL;
1002 env->error_end = (UChar* )NULL;
1005#ifdef USE_NAMED_GROUP
1009 env->mem_nodes_dynamic = (
Node** )NULL;
1011 for (i = 0; i < SCANENV_MEMNODES_SIZE; i++)
1012 env->mem_nodes_static[i] = NULL_NODE;
1014#ifdef USE_COMBINATION_EXPLOSION_CHECK
1015 env->num_comb_exp_check = 0;
1016 env->comb_exp_max_regnum = 0;
1017 env->curr_max_regnum = 0;
1018 env->has_recursion = 0;
1020 env->parse_depth = 0;
1021 env->warnings_flag = 0;
1025scan_env_add_mem_entry(
ScanEnv* env)
1030 need = env->num_mem + 1;
1031 if (need > ONIG_MAX_CAPTURE_GROUP_NUM)
1032 return ONIGERR_TOO_MANY_CAPTURE_GROUPS;
1033 if (need >= SCANENV_MEMNODES_SIZE) {
1034 if (env->mem_alloc <= need) {
1035 if (IS_NULL(env->mem_nodes_dynamic)) {
1036 alloc = INIT_SCANENV_MEMNODES_ALLOC_SIZE;
1038 CHECK_NULL_RETURN_MEMERR(p);
1039 xmemcpy(p, env->mem_nodes_static,
1040 sizeof(
Node*) * SCANENV_MEMNODES_SIZE);
1043 alloc = env->mem_alloc * 2;
1045 CHECK_NULL_RETURN_MEMERR(p);
1048 for (i = env->num_mem + 1; i < alloc; i++)
1051 env->mem_nodes_dynamic = p;
1052 env->mem_alloc = alloc;
1057 return env->num_mem;
1061scan_env_set_mem_node(
ScanEnv* env,
int num,
Node* node)
1063 if (env->num_mem >= num)
1064 SCANENV_MEM_NODES(env)[num] = node;
1066 return ONIGERR_PARSER_BUG;
1072onig_node_free(
Node* node)
1075 if (IS_NULL(node)) return ;
1077 switch (NTYPE(node)) {
1079 if (NSTR(node)->capa != 0 &&
1080 IS_NOT_NULL(NSTR(node)->s) && NSTR(node)->s != NSTR(node)->buf) {
1081 xfree(NSTR(node)->s);
1087 onig_node_free(NCAR(node));
1089 Node* next_node = NCDR(node);
1102 bbuf_free(cc->mbuf);
1107 if (NQTFR(node)->target)
1108 onig_node_free(NQTFR(node)->target);
1112 if (NENCLOSE(node)->target)
1113 onig_node_free(NENCLOSE(node)->target);
1117 if (IS_NOT_NULL(NBREF(node)->back_dynamic))
1118 xfree(NBREF(node)->back_dynamic);
1122 if (NANCHOR(node)->target)
1123 onig_node_free(NANCHOR(node)->target);
1143 BITSET_CLEAR(cc->bs);
1150node_new_cclass(
void)
1152 Node* node = node_new();
1153 CHECK_NULL_RETURN(node);
1155 SET_NTYPE(node, NT_CCLASS);
1156 initialize_cclass(NCCLASS(node));
1161node_new_ctype(
int type,
int not,
int ascii_range)
1163 Node* node = node_new();
1164 CHECK_NULL_RETURN(node);
1166 SET_NTYPE(node, NT_CTYPE);
1167 NCTYPE(node)->ctype =
type;
1168 NCTYPE(node)->not = not;
1169 NCTYPE(node)->ascii_range = ascii_range;
1174node_new_anychar(
void)
1176 Node* node = node_new();
1177 CHECK_NULL_RETURN(node);
1179 SET_NTYPE(node, NT_CANY);
1184node_new_list(
Node* left,
Node* right)
1186 Node* node = node_new();
1187 CHECK_NULL_RETURN(node);
1189 SET_NTYPE(node, NT_LIST);
1196onig_node_new_list(
Node* left,
Node* right)
1198 return node_new_list(left, right);
1202onig_node_list_add(
Node* list,
Node* x)
1206 n = onig_node_new_list(x, NULL);
1207 if (IS_NULL(n))
return NULL_NODE;
1209 if (IS_NOT_NULL(list)) {
1210 while (IS_NOT_NULL(NCDR(list)))
1220onig_node_new_alt(
Node* left,
Node* right)
1222 Node* node = node_new();
1223 CHECK_NULL_RETURN(node);
1225 SET_NTYPE(node, NT_ALT);
1232onig_node_new_anchor(
int type)
1234 Node* node = node_new();
1235 CHECK_NULL_RETURN(node);
1237 SET_NTYPE(node, NT_ANCHOR);
1238 NANCHOR(node)->type =
type;
1239 NANCHOR(node)->target = NULL;
1240 NANCHOR(node)->char_len = -1;
1241 NANCHOR(node)->ascii_range = 0;
1246node_new_backref(
int back_num,
int* backrefs,
int by_name,
1247#ifdef USE_BACKREF_WITH_LEVEL
1248 int exist_level,
int nest_level,
1253 Node* node = node_new();
1255 CHECK_NULL_RETURN(node);
1257 SET_NTYPE(node, NT_BREF);
1258 NBREF(node)->state = 0;
1259 NBREF(node)->back_num = back_num;
1260 NBREF(node)->back_dynamic = (
int* )NULL;
1262 NBREF(node)->state |= NST_NAME_REF;
1264#ifdef USE_BACKREF_WITH_LEVEL
1265 if (exist_level != 0) {
1266 NBREF(node)->state |= NST_NEST_LEVEL;
1267 NBREF(node)->nest_level = nest_level;
1271 for (i = 0; i < back_num; i++) {
1272 if (backrefs[i] <= env->num_mem &&
1273 IS_NULL(SCANENV_MEM_NODES(env)[backrefs[i]])) {
1274 NBREF(node)->state |= NST_RECURSION;
1279 if (back_num <= NODE_BACKREFS_SIZE) {
1280 for (i = 0; i < back_num; i++)
1281 NBREF(node)->back_static[i] = backrefs[i];
1284 int* p = (
int* )
xmalloc(
sizeof(
int) * back_num);
1286 onig_node_free(node);
1289 NBREF(node)->back_dynamic = p;
1290 for (i = 0; i < back_num; i++)
1296#ifdef USE_SUBEXP_CALL
1298node_new_call(UChar* name, UChar* name_end,
int gnum)
1300 Node* node = node_new();
1301 CHECK_NULL_RETURN(node);
1303 SET_NTYPE(node, NT_CALL);
1304 NCALL(node)->state = 0;
1305 NCALL(node)->target = NULL_NODE;
1306 NCALL(node)->name = name;
1307 NCALL(node)->name_end = name_end;
1308 NCALL(node)->group_num = gnum;
1314node_new_quantifier(
int lower,
int upper,
int by_number)
1316 Node* node = node_new();
1317 CHECK_NULL_RETURN(node);
1319 SET_NTYPE(node, NT_QTFR);
1320 NQTFR(node)->state = 0;
1321 NQTFR(node)->target = NULL;
1322 NQTFR(node)->lower = lower;
1323 NQTFR(node)->upper = upper;
1324 NQTFR(node)->greedy = 1;
1325 NQTFR(node)->target_empty_info = NQ_TARGET_ISNOT_EMPTY;
1326 NQTFR(node)->head_exact = NULL_NODE;
1327 NQTFR(node)->next_head_exact = NULL_NODE;
1328 NQTFR(node)->is_referred = 0;
1330 NQTFR(node)->state |= NST_BY_NUMBER;
1332#ifdef USE_COMBINATION_EXPLOSION_CHECK
1333 NQTFR(node)->comb_exp_check_num = 0;
1340node_new_enclose(
int type)
1342 Node* node = node_new();
1343 CHECK_NULL_RETURN(node);
1345 SET_NTYPE(node, NT_ENCLOSE);
1346 NENCLOSE(node)->type =
type;
1347 NENCLOSE(node)->state = 0;
1348 NENCLOSE(node)->regnum = 0;
1349 NENCLOSE(node)->option = 0;
1350 NENCLOSE(node)->target = NULL;
1351 NENCLOSE(node)->call_addr = -1;
1352 NENCLOSE(node)->opt_count = 0;
1357onig_node_new_enclose(
int type)
1359 return node_new_enclose(
type);
1363node_new_enclose_memory(OnigOptionType option,
int is_named)
1365 Node* node = node_new_enclose(ENCLOSE_MEMORY);
1366 CHECK_NULL_RETURN(node);
1368 SET_ENCLOSE_STATUS(node, NST_NAMED_GROUP);
1370#ifdef USE_SUBEXP_CALL
1371 NENCLOSE(node)->option = option;
1377node_new_option(OnigOptionType option)
1379 Node* node = node_new_enclose(ENCLOSE_OPTION);
1380 CHECK_NULL_RETURN(node);
1381 NENCLOSE(node)->option = option;
1386onig_node_str_cat(
Node* node,
const UChar* s,
const UChar* end)
1388 ptrdiff_t addlen = end - s;
1391 ptrdiff_t len = NSTR(node)->end - NSTR(node)->s;
1393 if (NSTR(node)->capa > 0 || (len + addlen > NODE_STR_BUF_SIZE - 1)) {
1395 ptrdiff_t capa = len + addlen + NODE_STR_MARGIN;
1397 if (capa <= NSTR(node)->capa) {
1398 onig_strcpy(NSTR(node)->s + len, s, end);
1401 if (NSTR(node)->s == NSTR(node)->buf)
1402 p = strcat_capa_from_static(NSTR(node)->s, NSTR(node)->end,
1405 p = strcat_capa(NSTR(node)->s, NSTR(node)->end, s, end, capa);
1407 CHECK_NULL_RETURN_MEMERR(p);
1409 NSTR(node)->capa = (int )capa;
1413 onig_strcpy(NSTR(node)->s + len, s, end);
1415 NSTR(node)->end = NSTR(node)->s + len + addlen;
1422onig_node_str_set(
Node* node,
const UChar* s,
const UChar* end)
1424 onig_node_str_clear(node);
1425 return onig_node_str_cat(node, s, end);
1429node_str_cat_char(
Node* node, UChar c)
1434 return onig_node_str_cat(node, s, s + 1);
1440 UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];
1441 int num = ONIGENC_CODE_TO_MBC(enc, c, buf);
1442 if (num < 0)
return num;
1443 return onig_node_str_cat(node, buf, buf + num);
1448onig_node_conv_to_str_node(
Node* node,
int flag)
1450 SET_NTYPE(node, NT_STR);
1451 NSTR(node)->flag = flag;
1452 NSTR(node)->capa = 0;
1453 NSTR(node)->s = NSTR(node)->buf;
1454 NSTR(node)->end = NSTR(node)->buf;
1459onig_node_str_clear(
Node* node)
1461 if (NSTR(node)->capa != 0 &&
1462 IS_NOT_NULL(NSTR(node)->s) && NSTR(node)->s != NSTR(node)->buf) {
1463 xfree(NSTR(node)->s);
1466 NSTR(node)->capa = 0;
1467 NSTR(node)->flag = 0;
1468 NSTR(node)->s = NSTR(node)->buf;
1469 NSTR(node)->end = NSTR(node)->buf;
1473node_new_str(
const UChar* s,
const UChar* end)
1475 Node* node = node_new();
1476 CHECK_NULL_RETURN(node);
1478 SET_NTYPE(node, NT_STR);
1479 NSTR(node)->capa = 0;
1480 NSTR(node)->flag = 0;
1481 NSTR(node)->s = NSTR(node)->buf;
1482 NSTR(node)->end = NSTR(node)->buf;
1483 if (onig_node_str_cat(node, s, end)) {
1484 onig_node_free(node);
1491onig_node_new_str(
const UChar* s,
const UChar* end)
1493 return node_new_str(s, end);
1497node_new_str_raw(UChar* s, UChar* end)
1499 Node* node = node_new_str(s, end);
1500 if (IS_NOT_NULL(node))
1501 NSTRING_SET_RAW(node);
1508 return node_new_str(NULL, NULL);
1512node_new_str_raw_char(UChar c)
1517 return node_new_str_raw(p, p + 1);
1524 Node* n = NULL_NODE;
1526 if (sn->end > sn->s) {
1527 p = onigenc_get_prev_char_head(enc, sn->s, sn->end, sn->end);
1528 if (p && p > sn->s) {
1529 n = node_new_str(p, sn->end);
1530 if (IS_NOT_NULL(n) && (sn->flag & NSTR_RAW) != 0)
1532 sn->end = (UChar* )p;
1541 if (sn->end > sn->s) {
1542 return ((enclen(enc, sn->s, sn->end) < sn->end - sn->s) ? 1 : 0);
1547#ifdef USE_PAD_TO_SHORT_BYTE_CHAR
1549node_str_head_pad(
StrNode* sn,
int num, UChar val)
1551 UChar buf[NODE_STR_BUF_SIZE];
1554 len = sn->end - sn->s;
1555 onig_strcpy(buf, sn->s, sn->end);
1556 onig_strcpy(&(sn->s[num]), buf, buf + len);
1559 for (i = 0; i < num; i++) {
1566onig_scan_unsigned_number(UChar** src,
const UChar* end,
OnigEncoding enc)
1568 unsigned int num, val;
1576 if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
1577 val = (
unsigned int )DIGITVAL(c);
1578 if ((INT_MAX_LIMIT - val) / 10UL < num)
1581 num = num * 10 + val;
1593scan_unsigned_hexadecimal_number(UChar** src, UChar* end,
int minlen,
1597 unsigned int num, val;
1602 restlen = maxlen - minlen;
1604 while (!PEND && maxlen-- != 0) {
1606 if (ONIGENC_IS_CODE_XDIGIT(enc, c)) {
1607 val = (
unsigned int )XDIGITVAL(enc,c);
1608 if ((INT_MAX_LIMIT - val) / 16UL < num)
1611 num = (num << 4) + XDIGITVAL(enc,c);
1619 if (maxlen > restlen)
1626scan_unsigned_octal_number(UChar** src, UChar* end,
int maxlen,
1630 unsigned int num, val;
1635 while (!PEND && maxlen-- != 0) {
1637 if (ONIGENC_IS_CODE_DIGIT(enc, c) && c <
'8') {
1639 if ((INT_MAX_LIMIT - val) / 8UL < num)
1642 num = (num << 3) + val;
1654#define BBUF_WRITE_CODE_POINT(bbuf,pos,code) \
1655 BBUF_WRITE(bbuf, pos, &(code), SIZE_CODE_POINT)
1662new_code_range(
BBuf** pbuf)
1664#define INIT_MULTI_BYTE_RANGE_SIZE (SIZE_CODE_POINT * 5)
1670 CHECK_NULL_RETURN_MEMERR(*pbuf);
1671 r = BBUF_INIT(*pbuf, INIT_MULTI_BYTE_RANGE_SIZE);
1675 BBUF_WRITE_CODE_POINT(bbuf, 0, n);
1680add_code_range_to_buf0(
BBuf** pbuf,
ScanEnv* env, OnigCodePoint from, OnigCodePoint to,
1684 OnigCodePoint low, high, bound, x;
1685 OnigCodePoint n, *data;
1689 n = from; from = to; to = n;
1692 if (IS_NULL(*pbuf)) {
1693 r = new_code_range(pbuf);
1700 GET_CODE_POINT(n, bbuf->p);
1702 data = (OnigCodePoint* )(bbuf->p);
1705 bound = (from == 0) ? 0 : n;
1706 for (low = 0; low < bound; ) {
1707 x = (low + bound) >> 1;
1708 if (from - 1 > data[x*2 + 1])
1714 high = (to == ONIG_LAST_CODE_POINT) ? n : low;
1715 for (bound = n; high < bound; ) {
1716 x = (high + bound) >> 1;
1717 if (to + 1 >= data[x*2])
1726 inc_n = low + 1 - high;
1727 if (n + inc_n > ONIG_MAX_MULTI_BYTE_RANGES_NUM)
1728 return ONIGERR_TOO_MANY_MULTI_BYTE_RANGES;
1731 if (checkdup && from <= data[low*2+1]
1732 && (data[low*2] <= from || data[low*2+1] <= to))
1733 CC_DUP_WARN(env, from, to);
1734 if (from > data[low*2])
1736 if (to < data[(high - 1)*2 + 1])
1737 to = data[(high - 1)*2 + 1];
1741 int from_pos = SIZE_CODE_POINT * (1 + high * 2);
1742 int to_pos = SIZE_CODE_POINT * (1 + (low + 1) * 2);
1746 int size = (n - high) * 2 * SIZE_CODE_POINT;
1747 BBUF_MOVE_RIGHT(bbuf, from_pos, to_pos, size);
1751 BBUF_MOVE_LEFT_REDUCE(bbuf, from_pos, to_pos);
1755 pos = SIZE_CODE_POINT * (1 + low * 2);
1756 BBUF_ENSURE_SIZE(bbuf, pos + SIZE_CODE_POINT * 2);
1757 BBUF_WRITE_CODE_POINT(bbuf, pos, from);
1758 BBUF_WRITE_CODE_POINT(bbuf, pos + SIZE_CODE_POINT, to);
1760 BBUF_WRITE_CODE_POINT(bbuf, 0, n);
1766add_code_range_to_buf(
BBuf** pbuf,
ScanEnv* env, OnigCodePoint from, OnigCodePoint to)
1768 return add_code_range_to_buf0(pbuf, env, from, to, 1);
1772add_code_range0(
BBuf** pbuf,
ScanEnv* env, OnigCodePoint from, OnigCodePoint to,
int checkdup)
1775 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC))
1778 return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS;
1781 return add_code_range_to_buf0(pbuf, env, from, to, checkdup);
1785add_code_range(
BBuf** pbuf,
ScanEnv* env, OnigCodePoint from, OnigCodePoint to)
1787 return add_code_range0(pbuf, env, from, to, 1);
1794 OnigCodePoint pre, from, *data, to = 0;
1796 *pbuf = (
BBuf* )NULL;
1797 if (IS_NULL(bbuf)) {
1799 return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);
1802 data = (OnigCodePoint* )(bbuf->p);
1803 GET_CODE_POINT(n, data);
1805 if (n <= 0)
goto set_all;
1808 pre = MBCODE_START_POS(enc);
1809 for (i = 0; i < n; i++) {
1812 if (pre <= from - 1) {
1813 r = add_code_range_to_buf(pbuf, env, pre, from - 1);
1814 if (r != 0)
return r;
1816 if (to == ONIG_LAST_CODE_POINT)
break;
1819 if (to < ONIG_LAST_CODE_POINT) {
1820 r = add_code_range_to_buf(pbuf, env, to + 1, ONIG_LAST_CODE_POINT);
1825#define SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2) do {\
1828 tnot = not1; not1 = not2; not2 = tnot; \
1829 tbuf = bbuf1; bbuf1 = bbuf2; bbuf2 = tbuf; \
1837 OnigCodePoint i, n1, *data1;
1838 OnigCodePoint from, to;
1840 *pbuf = (
BBuf* )NULL;
1841 if (IS_NULL(bbuf1) && IS_NULL(bbuf2)) {
1842 if (not1 != 0 || not2 != 0)
1843 return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);
1849 SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);
1851 if (IS_NULL(bbuf1)) {
1853 return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);
1857 return bbuf_clone(pbuf, bbuf2);
1860 return not_code_range_buf(enc, bbuf2, pbuf, env);
1866 SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);
1868 data1 = (OnigCodePoint* )(bbuf1->p);
1869 GET_CODE_POINT(n1, data1);
1872 if (not2 == 0 && not1 == 0) {
1873 r = bbuf_clone(pbuf, bbuf2);
1875 else if (not1 == 0) {
1876 r = not_code_range_buf(enc, bbuf2, pbuf, env);
1878 if (r != 0)
return r;
1880 for (i = 0; i < n1; i++) {
1883 r = add_code_range_to_buf(pbuf, env, from, to);
1884 if (r != 0)
return r;
1890and_code_range1(
BBuf** pbuf,
ScanEnv* env, OnigCodePoint from1, OnigCodePoint to1,
1891 OnigCodePoint* data,
int n)
1894 OnigCodePoint from2, to2;
1896 for (i = 0; i < n; i++) {
1899 if (from2 < from1) {
1900 if (to2 < from1)
continue;
1905 else if (from2 <= to1) {
1907 if (from1 <= from2 - 1) {
1908 r = add_code_range_to_buf(pbuf, env, from1, from2-1);
1909 if (r != 0)
return r;
1920 if (from1 > to1)
break;
1923 r = add_code_range_to_buf(pbuf, env, from1, to1);
1924 if (r != 0)
return r;
1930and_code_range_buf(
BBuf* bbuf1,
int not1,
BBuf* bbuf2,
int not2,
BBuf** pbuf,
ScanEnv* env)
1933 OnigCodePoint i, j, n1, n2, *data1, *data2;
1934 OnigCodePoint from, to, from1, to1, from2, to2;
1936 *pbuf = (
BBuf* )NULL;
1937 if (IS_NULL(bbuf1)) {
1938 if (not1 != 0 && IS_NOT_NULL(bbuf2))
1939 return bbuf_clone(pbuf, bbuf2);
1942 else if (IS_NULL(bbuf2)) {
1944 return bbuf_clone(pbuf, bbuf1);
1949 SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);
1951 data1 = (OnigCodePoint* )(bbuf1->p);
1952 data2 = (OnigCodePoint* )(bbuf2->p);
1953 GET_CODE_POINT(n1, data1);
1954 GET_CODE_POINT(n2, data2);
1958 if (not2 == 0 && not1 == 0) {
1959 for (i = 0; i < n1; i++) {
1962 for (j = 0; j < n2; j++) {
1965 if (from2 > to1)
break;
1966 if (to2 < from1)
continue;
1967 from = MAX(from1, from2);
1969 r = add_code_range_to_buf(pbuf, env, from, to);
1970 if (r != 0)
return r;
1974 else if (not1 == 0) {
1975 for (i = 0; i < n1; i++) {
1978 r = and_code_range1(pbuf, env, from1, to1, data2, n2);
1979 if (r != 0)
return r;
1991 BBuf *buf1, *buf2, *pbuf = 0;
1992 BitSetRef bsr1, bsr2;
1995 not1 = IS_NCCLASS_NOT(dest);
1998 not2 = IS_NCCLASS_NOT(cc);
2003 bitset_invert_to(bsr1, bs1);
2007 bitset_invert_to(bsr2, bs2);
2010 bitset_and(bsr1, bsr2);
2011 if (bsr1 != dest->bs) {
2012 bitset_copy(dest->bs, bsr1);
2016 bitset_invert(dest->bs);
2019 if (! ONIGENC_IS_SINGLEBYTE(enc)) {
2020 if (not1 != 0 && not2 != 0) {
2021 r = or_code_range_buf(enc, buf1, 0, buf2, 0, &pbuf, env);
2024 r = and_code_range_buf(buf1, not1, buf2, not2, &pbuf, env);
2025 if (r == 0 && not1 != 0) {
2027 r = not_code_range_buf(enc, pbuf, &tbuf, env);
2049 BBuf *buf1, *buf2, *pbuf = 0;
2050 BitSetRef bsr1, bsr2;
2053 not1 = IS_NCCLASS_NOT(dest);
2056 not2 = IS_NCCLASS_NOT(cc);
2061 bitset_invert_to(bsr1, bs1);
2065 bitset_invert_to(bsr2, bs2);
2068 bitset_or(bsr1, bsr2);
2069 if (bsr1 != dest->bs) {
2070 bitset_copy(dest->bs, bsr1);
2074 bitset_invert(dest->bs);
2077 if (! ONIGENC_IS_SINGLEBYTE(enc)) {
2078 if (not1 != 0 && not2 != 0) {
2079 r = and_code_range_buf(buf1, 0, buf2, 0, &pbuf, env);
2082 r = or_code_range_buf(enc, buf1, not1, buf2, not2, &pbuf, env);
2083 if (r == 0 && not1 != 0) {
2085 r = not_code_range_buf(enc, pbuf, &tbuf, env);
2103static void UNKNOWN_ESC_WARN(
ScanEnv *env,
int c);
2106conv_backslash_value(OnigCodePoint c,
ScanEnv* env)
2108 if (IS_SYNTAX_OP(env->syntax, ONIG_SYN_OP_ESC_CONTROL_CHARS)) {
2110 case 'n':
return '\n';
2111 case 't':
return '\t';
2112 case 'r':
return '\r';
2113 case 'f':
return '\f';
2114 case 'a':
return '\007';
2115 case 'b':
return '\010';
2116 case 'e':
return '\033';
2118 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ESC_V_VTAB))
2123 if ((
'a' <= c && c <=
'z') || (
'A' <= c && c <=
'Z'))
2124 UNKNOWN_ESC_WARN(env, c);
2131#ifdef USE_NO_INVALID_QUANTIFIER
2132# define is_invalid_quantifier_target(node) 0
2135is_invalid_quantifier_target(
Node* node)
2137 switch (NTYPE(node)) {
2149 if (! is_invalid_quantifier_target(NCAR(node)))
return 0;
2150 }
while (IS_NOT_NULL(node = NCDR(node)));
2156 if (is_invalid_quantifier_target(NCAR(node)))
return 1;
2157 }
while (IS_NOT_NULL(node = NCDR(node)));
2172 if (q->lower == 0) {
2173 if (q->upper == 1)
return 0;
2174 else if (IS_REPEAT_INFINITE(q->upper))
return 1;
2176 else if (q->lower == 1) {
2177 if (IS_REPEAT_INFINITE(q->upper))
return 2;
2181 if (q->lower == 0) {
2182 if (q->upper == 1)
return 3;
2183 else if (IS_REPEAT_INFINITE(q->upper))
return 4;
2185 else if (q->lower == 1) {
2186 if (IS_REPEAT_INFINITE(q->upper))
return 5;
2202static enum ReduceType
const ReduceTypeTable[6][6] = {
2204 {RQ_DEL, RQ_A, RQ_A, RQ_QQ, RQ_AQ, RQ_ASIS},
2205 {RQ_DEL, RQ_DEL, RQ_DEL, RQ_P_QQ, RQ_P_QQ, RQ_DEL},
2206 {RQ_A, RQ_A, RQ_DEL, RQ_ASIS, RQ_P_QQ, RQ_DEL},
2207 {RQ_DEL, RQ_AQ, RQ_AQ, RQ_DEL, RQ_AQ, RQ_AQ},
2208 {RQ_DEL, RQ_DEL, RQ_DEL, RQ_DEL, RQ_DEL, RQ_DEL},
2209 {RQ_ASIS, RQ_ASIS, RQ_ASIS, RQ_AQ, RQ_AQ, RQ_DEL}
2213onig_reduce_nested_quantifier(
Node* pnode,
Node* cnode)
2220 pnum = popular_quantifier_num(p);
2221 cnum = popular_quantifier_num(c);
2222 if (pnum < 0 || cnum < 0) return ;
2224 switch (ReduceTypeTable[cnum][pnum]) {
2229 p->target = c->target;
2230 p->lower = 0; p->upper = REPEAT_INFINITE; p->greedy = 1;
2233 p->target = c->target;
2234 p->lower = 0; p->upper = REPEAT_INFINITE; p->greedy = 0;
2237 p->target = c->target;
2238 p->lower = 0; p->upper = 1; p->greedy = 0;
2242 p->lower = 0; p->upper = 1; p->greedy = 0;
2243 c->lower = 1; c->upper = REPEAT_INFINITE; c->greedy = 1;
2252 c->target = NULL_NODE;
2253 onig_node_free(cnode);
2278 TK_EXTENDED_GRAPHEME_CLUSTER,
2283 TK_POSIX_BRACKET_OPEN,
2289 enum TokenSyms
type;
2312#ifdef USE_BACKREF_WITH_LEVEL
2332fetch_range_quantifier(UChar** src, UChar* end,
OnigToken* tok,
ScanEnv* env)
2334 int low, up, syn_allow, non_low = 0;
2341 syn_allow = IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_INVALID_INTERVAL);
2347 return ONIGERR_END_PATTERN_AT_LEFT_BRACE;
2352 if (c ==
')' || c ==
'(' || c ==
'|') {
2353 return ONIGERR_END_PATTERN_AT_LEFT_BRACE;
2357 low = onig_scan_unsigned_number(&p, end, env->enc);
2358 if (low < 0)
return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
2359 if (low > ONIG_MAX_REPEAT_NUM)
2360 return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
2363 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV)) {
2372 if (PEND)
goto invalid;
2376 up = onig_scan_unsigned_number(&p, end, env->enc);
2377 if (up < 0)
return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
2378 if (up > ONIG_MAX_REPEAT_NUM)
2379 return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
2384 up = REPEAT_INFINITE;
2396 if (PEND)
goto invalid;
2398 if (IS_SYNTAX_OP(env->syntax, ONIG_SYN_OP_ESC_BRACE_INTERVAL)) {
2399 if (c != MC_ESC(env->syntax))
goto invalid;
2400 if (PEND)
goto invalid;
2403 if (c !=
'}')
goto invalid;
2405 if (!IS_REPEAT_INFINITE(up) && low > up) {
2406 return ONIGERR_UPPER_SMALLER_THAN_LOWER_IN_REPEAT_RANGE;
2409 tok->type = TK_INTERVAL;
2410 tok->u.repeat.lower = low;
2411 tok->u.repeat.upper = up;
2419 return ONIGERR_INVALID_REPEAT_RANGE_PATTERN;
2424fetch_escaped_value(UChar** src, UChar* end,
ScanEnv* env, OnigCodePoint* val)
2431 if (PEND)
return ONIGERR_END_PATTERN_AT_ESCAPE;
2436 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ESC_CAPITAL_M_BAR_META)) {
2437 if (PEND)
return ONIGERR_END_PATTERN_AT_META;
2439 if (c !=
'-')
return ONIGERR_META_CODE_SYNTAX;
2440 if (PEND)
return ONIGERR_END_PATTERN_AT_META;
2442 if (c == MC_ESC(env->syntax)) {
2443 v = fetch_escaped_value(&p, end, env, &c);
2444 if (v < 0)
return v;
2446 c = ((c & 0xff) | 0x80);
2453 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ESC_CAPITAL_C_BAR_CONTROL)) {
2454 if (PEND)
return ONIGERR_END_PATTERN_AT_CONTROL;
2456 if (c !=
'-')
return ONIGERR_CONTROL_CODE_SYNTAX;
2463 if (IS_SYNTAX_OP(env->syntax, ONIG_SYN_OP_ESC_C_CONTROL)) {
2465 if (PEND)
return ONIGERR_END_PATTERN_AT_CONTROL;
2471 if (c == MC_ESC(env->syntax)) {
2472 v = fetch_escaped_value(&p, end, env, &c);
2473 if (v < 0)
return v;
2484 c = conv_backslash_value(c, env);
2494static int fetch_token(
OnigToken* tok, UChar** src, UChar* end,
ScanEnv* env);
2497get_name_end_code_point(OnigCodePoint start)
2500 case '<':
return (OnigCodePoint )
'>';
break;
2501 case '\'':
return (OnigCodePoint )
'\'';
break;
2502 case '(':
return (OnigCodePoint )
')';
break;
2503 case '{':
return (OnigCodePoint )
'}';
break;
2508 return (OnigCodePoint )0;
2511#ifdef USE_NAMED_GROUP
2513# define ONIGENC_IS_CODE_NAME(enc, c) TRUE
2515# define ONIGENC_IS_CODE_NAME(enc, c) ONIGENC_IS_CODE_WORD(enc, c)
2518# ifdef USE_BACKREF_WITH_LEVEL
2525fetch_name_with_level(OnigCodePoint start_code, UChar** src, UChar* end,
2526 UChar** rname_end,
ScanEnv* env,
2527 int* rback_num,
int* rlevel)
2529 int r, sign, is_num, exist_level;
2530 OnigCodePoint end_code;
2531 OnigCodePoint c = 0;
2539 is_num = exist_level = 0;
2543 end_code = get_name_end_code_point(start_code);
2548 return ONIGERR_EMPTY_GROUP_NAME;
2553 return ONIGERR_EMPTY_GROUP_NAME;
2555 if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
2558 else if (c ==
'-') {
2563 else if (!ONIGENC_IS_CODE_NAME(enc, c)) {
2564 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2571 if (c == end_code || c ==
')' || c ==
'+' || c ==
'-') {
2572 if (is_num == 2) r = ONIGERR_INVALID_GROUP_NAME;
2577 if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
2581 r = ONIGERR_INVALID_GROUP_NAME;
2585 else if (!ONIGENC_IS_CODE_NAME(enc, c)) {
2586 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2590 if (r == 0 && c != end_code) {
2591 if (c ==
'+' || c ==
'-') {
2593 int flag = (c ==
'-' ? -1 : 1);
2596 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2600 if (! ONIGENC_IS_CODE_DIGIT(enc, c))
goto err;
2602 level = onig_scan_unsigned_number(&p, end, enc);
2603 if (level < 0)
return ONIGERR_TOO_BIG_NUMBER;
2604 *rlevel = (level * flag);
2615 r = ONIGERR_INVALID_GROUP_NAME;
2622 *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc);
2623 if (*rback_num < 0)
return ONIGERR_TOO_BIG_NUMBER;
2624 else if (*rback_num == 0)
goto err;
2629 *rname_end = name_end;
2631 return (exist_level ? 1 : 0);
2634 onig_scan_env_set_error_string(env, r, *src, name_end);
2645fetch_name(OnigCodePoint start_code, UChar** src, UChar* end,
2646 UChar** rname_end,
ScanEnv* env,
int* rback_num,
int ref)
2648 int r, is_num, sign;
2649 OnigCodePoint end_code;
2650 OnigCodePoint c = 0;
2658 end_code = get_name_end_code_point(start_code);
2666 return ONIGERR_EMPTY_GROUP_NAME;
2671 return ONIGERR_EMPTY_GROUP_NAME;
2673 if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
2677 r = ONIGERR_INVALID_GROUP_NAME;
2681 else if (c ==
'-') {
2688 r = ONIGERR_INVALID_GROUP_NAME;
2692 else if (!ONIGENC_IS_CODE_NAME(enc, c)) {
2693 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2701 if (c == end_code || c ==
')') {
2703 r = ONIGERR_INVALID_GROUP_NAME;
2710 if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
2714 if (!ONIGENC_IS_CODE_WORD(enc, c))
2715 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2717 r = ONIGERR_INVALID_GROUP_NAME;
2722 if (!ONIGENC_IS_CODE_NAME(enc, c)) {
2723 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2729 if (c != end_code) {
2730 r = ONIGERR_INVALID_GROUP_NAME;
2736 *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc);
2737 if (*rback_num < 0)
return ONIGERR_TOO_BIG_NUMBER;
2738 else if (*rback_num == 0) {
2739 r = ONIGERR_INVALID_GROUP_NAME;
2746 *rname_end = name_end;
2755 if (c == end_code || c ==
')')
2762 onig_scan_env_set_error_string(env, r, *src, name_end);
2768fetch_name(OnigCodePoint start_code, UChar** src, UChar* end,
2769 UChar** rname_end,
ScanEnv* env,
int* rback_num,
int ref)
2771 int r, is_num, sign;
2772 OnigCodePoint end_code;
2773 OnigCodePoint c = 0;
2782 end_code = get_name_end_code_point(start_code);
2784 *rname_end = name_end = end;
2791 return ONIGERR_EMPTY_GROUP_NAME;
2796 return ONIGERR_EMPTY_GROUP_NAME;
2798 if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
2801 else if (c ==
'-') {
2807 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2815 if (c == end_code || c ==
')')
break;
2816 if (! ONIGENC_IS_CODE_DIGIT(enc, c))
2817 r = ONIGERR_INVALID_CHAR_IN_GROUP_NAME;
2819 if (r == 0 && c != end_code) {
2820 r = ONIGERR_INVALID_GROUP_NAME;
2825 *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc);
2826 if (*rback_num < 0)
return ONIGERR_TOO_BIG_NUMBER;
2827 else if (*rback_num == 0) {
2828 r = ONIGERR_INVALID_GROUP_NAME;
2833 *rname_end = name_end;
2839 onig_scan_env_set_error_string(env, r, *src, name_end);
2847PRINTF_ARGS(
static void onig_syntax_warn(
ScanEnv *env,
const char *fmt, ...), 2, 3);
2851onig_syntax_warn(
ScanEnv *env,
const char *fmt, ...)
2854 UChar buf[WARN_BUFSIZE];
2855 va_start(args, fmt);
2856 onig_vsnprintf_with_pattern(buf, WARN_BUFSIZE, env->enc,
2857 env->pattern, env->pattern_end,
2861 if (env->sourcefile == NULL)
2866 (*onig_warn)((
char* )buf);
2871CC_ESC_WARN(
ScanEnv *env, UChar *c)
2873 if (onig_warn == onig_null_warn) return ;
2875 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_CC_OP_NOT_ESCAPED) &&
2876 IS_SYNTAX_BV(env->syntax, ONIG_SYN_BACKSLASH_ESCAPE_IN_CC)) {
2877 onig_syntax_warn(env,
"character class has '%s' without escape", c);
2882CLOSE_BRACKET_WITHOUT_ESC_WARN(
ScanEnv* env, UChar* c)
2884 if (onig_warn == onig_null_warn) return ;
2886 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_CC_OP_NOT_ESCAPED)) {
2887 onig_syntax_warn(env,
"regular expression has '%s' without escape", c);
2896CC_DUP_WARN(
ScanEnv *env, OnigCodePoint from ARG_UNUSED, OnigCodePoint to ARG_UNUSED)
2900 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_CC_DUP) &&
2901 !(env->warnings_flag & ONIG_SYN_WARN_CC_DUP)) {
2902#ifdef WARN_ALL_CC_DUP
2903 onig_syntax_warn(env,
"character class has duplicated range: %04x-%04x", from, to);
2905 env->warnings_flag |= ONIG_SYN_WARN_CC_DUP;
2906 onig_syntax_warn(env,
"character class has duplicated range");
2912UNKNOWN_ESC_WARN(
ScanEnv *env,
int c)
2915 onig_syntax_warn(env,
"Unknown escape \\%c is ignored", c);
2919find_str_position(OnigCodePoint s[],
int n, UChar* from, UChar* to,
2928 x = ONIGENC_MBC_TO_CODE(enc, p, to);
2929 q = p + enclen(enc, p, to);
2931 for (i = 1; i < n && q < to; i++) {
2932 x = ONIGENC_MBC_TO_CODE(enc, q, to);
2933 if (x != s[i])
break;
2934 q += enclen(enc, q, to);
2937 if (IS_NOT_NULL(next))
2948str_exist_check_with_esc(OnigCodePoint s[],
int n, UChar* from, UChar* to,
2960 p += enclen(enc, p, to);
2963 x = ONIGENC_MBC_TO_CODE(enc, p, to);
2964 q = p + enclen(enc, p, to);
2966 for (i = 1; i < n && q < to; i++) {
2967 x = ONIGENC_MBC_TO_CODE(enc, q, to);
2968 if (x != s[i])
break;
2969 q += enclen(enc, q, to);
2971 if (i >= n)
return 1;
2972 p += enclen(enc, p, to);
2975 x = ONIGENC_MBC_TO_CODE(enc, p, to);
2976 if (x == bad)
return 0;
2977 else if (x == MC_ESC(syn)) in_esc = 1;
2989 OnigCodePoint c, c2;
3002 tok->type = TK_CHAR;
3008 tok->type = TK_CC_CLOSE;
3010 else if (c ==
'-') {
3011 tok->type = TK_CC_RANGE;
3013 else if (c == MC_ESC(syn)) {
3014 if (! IS_SYNTAX_BV(syn, ONIG_SYN_BACKSLASH_ESCAPE_IN_CC))
3017 if (PEND)
return ONIGERR_END_PATTERN_AT_ESCAPE;
3024 tok->type = TK_CHAR_TYPE;
3025 tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
3026 tok->u.prop.not = 0;
3029 tok->type = TK_CHAR_TYPE;
3030 tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
3031 tok->u.prop.not = 1;
3034 tok->type = TK_CHAR_TYPE;
3035 tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
3036 tok->u.prop.not = 0;
3039 tok->type = TK_CHAR_TYPE;
3040 tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
3041 tok->u.prop.not = 1;
3044 tok->type = TK_CHAR_TYPE;
3045 tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
3046 tok->u.prop.not = 0;
3049 tok->type = TK_CHAR_TYPE;
3050 tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
3051 tok->u.prop.not = 1;
3054 if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT))
break;
3055 tok->type = TK_CHAR_TYPE;
3056 tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;
3057 tok->u.prop.not = 0;
3060 if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT))
break;
3061 tok->type = TK_CHAR_TYPE;
3062 tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;
3063 tok->u.prop.not = 1;
3072 IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY)) {
3074 tok->type = TK_CHAR_PROPERTY;
3075 tok->u.prop.not = (c ==
'P' ? 1 : 0);
3077 if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT)) {
3080 tok->u.prop.not = (tok->u.prop.not == 0 ? 1 : 0);
3087 onig_syntax_warn(env,
"invalid Unicode Property \\%c", c);
3095 if (PPEEK_IS(
'{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_BRACE_HEX8)) {
3097 num = scan_unsigned_hexadecimal_number(&p, end, 0, 8, enc);
3098 if (num < 0)
return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;
3101 if (ONIGENC_IS_CODE_XDIGIT(enc, c2))
3102 return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE;
3105 if (p > prev + enclen(enc, prev, end) && !PEND && (PPEEK_IS(
'}'))) {
3107 tok->type = TK_CODE_POINT;
3109 tok->u.code = (OnigCodePoint )num;
3116 else if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_HEX2)) {
3117 num = scan_unsigned_hexadecimal_number(&p, end, 0, 2, enc);
3118 if (num < 0)
return ONIGERR_TOO_BIG_NUMBER;
3122 tok->type = TK_RAW_BYTE;
3132 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_U_HEX4)) {
3133 num = scan_unsigned_hexadecimal_number(&p, end, 4, 4, enc);
3134 if (num < -1)
return ONIGERR_TOO_SHORT_DIGITS;
3135 else if (num < 0)
return ONIGERR_TOO_BIG_NUMBER;
3139 tok->type = TK_CODE_POINT;
3141 tok->u.code = (OnigCodePoint )num;
3149 if (PPEEK_IS(
'{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_O_BRACE_OCTAL)) {
3151 num = scan_unsigned_octal_number(&p, end, 11, enc);
3152 if (num < 0)
return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;
3155 if (ONIGENC_IS_CODE_DIGIT(enc, c2) && c2 <
'8')
3156 return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE;
3159 if (p > prev + enclen(enc, prev, end) && !PEND && (PPEEK_IS(
'}'))) {
3161 tok->type = TK_CODE_POINT;
3163 tok->u.code = (OnigCodePoint )num;
3173 case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
3174 if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_OCTAL3)) {
3177 num = scan_unsigned_octal_number(&p, end, 3, enc);
3178 if (num < 0 || 0xff < num)
return ONIGERR_TOO_BIG_NUMBER;
3182 tok->type = TK_RAW_BYTE;
3190 num = fetch_escaped_value(&p, end, env, &c2);
3191 if (num < 0)
return num;
3192 if ((OnigCodePoint )tok->u.c != c2) {
3193 tok->u.code = (OnigCodePoint )c2;
3194 tok->type = TK_CODE_POINT;
3199 else if (c ==
'[') {
3200 if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_POSIX_BRACKET) && (PPEEK_IS(
':'))) {
3201 OnigCodePoint send[] = { (OnigCodePoint )
':', (OnigCodePoint )
']' };
3204 if (str_exist_check_with_esc(send, 2, p, end,
3205 (OnigCodePoint )
']', enc, syn)) {
3206 tok->type = TK_POSIX_BRACKET_OPEN;
3215 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_CCLASS_SET_OP)) {
3216 tok->type = TK_CC_CC_OPEN;
3219 CC_ESC_WARN(env, (UChar* )
"[");
3223 else if (c ==
'&') {
3224 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_CCLASS_SET_OP) &&
3225 !PEND && (PPEEK_IS(
'&'))) {
3227 tok->type = TK_CC_AND;
3236#ifdef USE_NAMED_GROUP
3238fetch_named_backref_token(OnigCodePoint c,
OnigToken* tok, UChar** src,
3251# ifdef USE_BACKREF_WITH_LEVEL
3252 name_end = NULL_UCHARP;
3253 r = fetch_name_with_level(c, &p, end, &name_end,
3254 env, &back_num, &tok->u.backref.level);
3255 if (r == 1) tok->u.backref.exist_level = 1;
3256 else tok->u.backref.exist_level = 0;
3258 r = fetch_name(&p, end, &name_end, env, &back_num, 1);
3260 if (r < 0)
return r;
3262 if (back_num != 0) {
3264 back_num = BACKREF_REL_TO_ABS(back_num, env);
3266 return ONIGERR_INVALID_BACKREF;
3269 if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) {
3270 if (back_num > env->num_mem ||
3271 IS_NULL(SCANENV_MEM_NODES(env)[back_num]))
3272 return ONIGERR_INVALID_BACKREF;
3274 tok->type = TK_BACKREF;
3275 tok->u.backref.by_name = 0;
3276 tok->u.backref.num = 1;
3277 tok->u.backref.ref1 = back_num;
3280 num = onig_name_to_group_numbers(env->reg, prev, name_end, &backs);
3282 onig_scan_env_set_error_string(env,
3283 ONIGERR_UNDEFINED_NAME_REFERENCE, prev, name_end);
3284 return ONIGERR_UNDEFINED_NAME_REFERENCE;
3286 if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) {
3288 for (i = 0; i < num; i++) {
3289 if (backs[i] > env->num_mem ||
3290 IS_NULL(SCANENV_MEM_NODES(env)[backs[i]]))
3291 return ONIGERR_INVALID_BACKREF;
3295 tok->type = TK_BACKREF;
3296 tok->u.backref.by_name = 1;
3297 if (num == 1 || IS_SYNTAX_BV(syn, ONIG_SYN_USE_LEFT_MOST_NAMED_GROUP)) {
3298 tok->u.backref.num = 1;
3299 tok->u.backref.ref1 = backs[0];
3302 tok->u.backref.num = num;
3303 tok->u.backref.refs = backs;
3328 tok->type = TK_STRING;
3333 if (IS_MC_ESC_CODE(c, syn)) {
3334 if (PEND)
return ONIGERR_END_PATTERN_AT_ESCAPE;
3343 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_ASTERISK_ZERO_INF))
break;
3344 tok->type = TK_OP_REPEAT;
3345 tok->u.repeat.lower = 0;
3346 tok->u.repeat.upper = REPEAT_INFINITE;
3351 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_PLUS_ONE_INF))
break;
3352 tok->type = TK_OP_REPEAT;
3353 tok->u.repeat.lower = 1;
3354 tok->u.repeat.upper = REPEAT_INFINITE;
3359 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_QMARK_ZERO_ONE))
break;
3360 tok->type = TK_OP_REPEAT;
3361 tok->u.repeat.lower = 0;
3362 tok->u.repeat.upper = 1;
3364 if (!PEND && PPEEK_IS(
'?') &&
3365 IS_SYNTAX_OP(syn, ONIG_SYN_OP_QMARK_NON_GREEDY)) {
3367 tok->u.repeat.greedy = 0;
3368 tok->u.repeat.possessive = 0;
3372 if (!PEND && PPEEK_IS(
'+') &&
3373 ((IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT) &&
3374 tok->type != TK_INTERVAL) ||
3375 (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_PLUS_POSSESSIVE_INTERVAL) &&
3376 tok->type == TK_INTERVAL))) {
3378 tok->u.repeat.greedy = 1;
3379 tok->u.repeat.possessive = 1;
3382 tok->u.repeat.greedy = 1;
3383 tok->u.repeat.possessive = 0;
3389 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_BRACE_INTERVAL))
break;
3390 r = fetch_range_quantifier(&p, end, tok, env);
3391 if (r < 0)
return r;
3392 if (r == 0)
goto greedy_check;
3394 if (IS_SYNTAX_BV(syn, ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY))
3395 goto possessive_check;
3403 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_VBAR_ALT))
break;
3408 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LPAREN_SUBEXP))
break;
3409 tok->type = TK_SUBEXP_OPEN;
3413 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LPAREN_SUBEXP))
break;
3414 tok->type = TK_SUBEXP_CLOSE;
3418 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_W_WORD))
break;
3419 tok->type = TK_CHAR_TYPE;
3420 tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
3421 tok->u.prop.not = 0;
3425 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_W_WORD))
break;
3426 tok->type = TK_CHAR_TYPE;
3427 tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
3428 tok->u.prop.not = 1;
3432 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_B_WORD_BOUND))
break;
3433 tok->type = TK_ANCHOR;
3434 tok->u.anchor.subtype = ANCHOR_WORD_BOUND;
3435 tok->u.anchor.ascii_range = IS_ASCII_RANGE(env->option)
3436 && ! IS_WORD_BOUND_ALL_RANGE(env->option);
3440 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_B_WORD_BOUND))
break;
3441 tok->type = TK_ANCHOR;
3442 tok->u.anchor.subtype = ANCHOR_NOT_WORD_BOUND;
3443 tok->u.anchor.ascii_range = IS_ASCII_RANGE(env->option)
3444 && ! IS_WORD_BOUND_ALL_RANGE(env->option);
3447#ifdef USE_WORD_BEGIN_END
3449 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END))
break;
3450 tok->type = TK_ANCHOR;
3451 tok->u.anchor.subtype = ANCHOR_WORD_BEGIN;
3452 tok->u.anchor.ascii_range = IS_ASCII_RANGE(env->option);
3456 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END))
break;
3457 tok->type = TK_ANCHOR;
3458 tok->u.anchor.subtype = ANCHOR_WORD_END;
3459 tok->u.anchor.ascii_range = IS_ASCII_RANGE(env->option);
3464 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_S_WHITE_SPACE))
break;
3465 tok->type = TK_CHAR_TYPE;
3466 tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
3467 tok->u.prop.not = 0;
3471 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_S_WHITE_SPACE))
break;
3472 tok->type = TK_CHAR_TYPE;
3473 tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
3474 tok->u.prop.not = 1;
3478 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_D_DIGIT))
break;
3479 tok->type = TK_CHAR_TYPE;
3480 tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
3481 tok->u.prop.not = 0;
3485 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_D_DIGIT))
break;
3486 tok->type = TK_CHAR_TYPE;
3487 tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
3488 tok->u.prop.not = 1;
3492 if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT))
break;
3493 tok->type = TK_CHAR_TYPE;
3494 tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;
3495 tok->u.prop.not = 0;
3499 if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_H_XDIGIT))
break;
3500 tok->type = TK_CHAR_TYPE;
3501 tok->u.prop.ctype = ONIGENC_CTYPE_XDIGIT;
3502 tok->u.prop.not = 1;
3506 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR))
break;
3508 tok->type = TK_ANCHOR;
3509 tok->u.anchor.subtype = ANCHOR_BEGIN_BUF;
3513 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR))
break;
3514 tok->type = TK_ANCHOR;
3515 tok->u.anchor.subtype = ANCHOR_SEMI_END_BUF;
3519 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR))
break;
3521 tok->type = TK_ANCHOR;
3522 tok->u.anchor.subtype = ANCHOR_END_BUF;
3526 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR))
break;
3527 tok->type = TK_ANCHOR;
3528 tok->u.anchor.subtype = ANCHOR_BEGIN_POSITION;
3532 if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_GNU_BUF_ANCHOR))
break;
3537 if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_GNU_BUF_ANCHOR))
break;
3545 if (PPEEK_IS(
'{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_BRACE_HEX8)) {
3547 num = scan_unsigned_hexadecimal_number(&p, end, 0, 8, enc);
3548 if (num < 0)
return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;
3550 if (ONIGENC_IS_CODE_XDIGIT(enc, PPEEK))
3551 return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE;
3554 if ((p > prev + enclen(enc, prev, end)) && !PEND && PPEEK_IS(
'}')) {
3556 tok->type = TK_CODE_POINT;
3557 tok->u.code = (OnigCodePoint )num;
3564 else if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_HEX2)) {
3565 num = scan_unsigned_hexadecimal_number(&p, end, 0, 2, enc);
3566 if (num < 0)
return ONIGERR_TOO_BIG_NUMBER;
3570 tok->type = TK_RAW_BYTE;
3580 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_U_HEX4)) {
3581 num = scan_unsigned_hexadecimal_number(&p, end, 4, 4, enc);
3582 if (num < -1)
return ONIGERR_TOO_SHORT_DIGITS;
3583 else if (num < 0)
return ONIGERR_TOO_BIG_NUMBER;
3587 tok->type = TK_CODE_POINT;
3589 tok->u.code = (OnigCodePoint )num;
3597 if (PPEEK_IS(
'{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_O_BRACE_OCTAL)) {
3599 num = scan_unsigned_octal_number(&p, end, 11, enc);
3600 if (num < 0)
return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;
3602 OnigCodePoint c = PPEEK;
3603 if (ONIGENC_IS_CODE_DIGIT(enc, c) && c <
'8')
3604 return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE;
3607 if ((p > prev + enclen(enc, prev, end)) && !PEND && PPEEK_IS(
'}')) {
3609 tok->type = TK_CODE_POINT;
3610 tok->u.code = (OnigCodePoint )num;
3619 case '1':
case '2':
case '3':
case '4':
3620 case '5':
case '6':
case '7':
case '8':
case '9':
3623 num = onig_scan_unsigned_number(&p, end, enc);
3624 if (num < 0 || num > ONIG_MAX_BACKREF_NUM) {
3628 if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_DECIMAL_BACKREF) &&
3629 (num <= env->num_mem || num <= 9)) {
3630 if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) {
3631 if (num > env->num_mem || IS_NULL(SCANENV_MEM_NODES(env)[num]))
3632 return ONIGERR_INVALID_BACKREF;
3635 tok->type = TK_BACKREF;
3636 tok->u.backref.num = 1;
3637 tok->u.backref.ref1 = num;
3638 tok->u.backref.by_name = 0;
3639#ifdef USE_BACKREF_WITH_LEVEL
3640 tok->u.backref.exist_level = 0;
3646 if (c ==
'8' || c ==
'9') {
3655 if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_OCTAL3)) {
3657 num = scan_unsigned_octal_number(&p, end, (c ==
'0' ? 2:3), enc);
3658 if (num < 0 || 0xff < num)
return ONIGERR_TOO_BIG_NUMBER;
3662 tok->type = TK_RAW_BYTE;
3666 else if (c !=
'0') {
3671#ifdef USE_NAMED_GROUP
3673 if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_K_NAMED_BACKREF)) {
3675 if (c ==
'<' || c ==
'\'') {
3676 r = fetch_named_backref_token(c, tok, &p, end, env);
3677 if (r < 0)
return r;
3681 onig_syntax_warn(env,
"invalid back reference");
3687#if defined(USE_SUBEXP_CALL) || defined(USE_NAMED_GROUP)
3689# ifdef USE_NAMED_GROUP
3690 if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_G_BRACE_BACKREF)) {
3693 r = fetch_named_backref_token(c, tok, &p, end, env);
3694 if (r < 0)
return r;
3700# ifdef USE_SUBEXP_CALL
3701 if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_G_SUBEXP_CALL)) {
3703 if (c ==
'<' || c ==
'\'') {
3704 int gnum = -1, rel = 0;
3706 OnigCodePoint cnext;
3711 if (PPEEK_IS(get_name_end_code_point(c))) {
3717 else if (cnext ==
'+') {
3723 r = fetch_name((OnigCodePoint )c, &p, end, &name_end, env, &gnum, 1);
3724 if (r < 0)
return r;
3727 tok->type = TK_CALL;
3728 tok->u.call.name = prev;
3729 tok->u.call.name_end = name_end;
3730 tok->u.call.gnum = gnum;
3731 tok->u.call.rel = rel;
3734 onig_syntax_warn(env,
"invalid subexp call");
3743 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE)) {
3744 tok->type = TK_QUOTE_OPEN;
3750 if (PPEEK_IS(
'{') &&
3751 IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY)) {
3753 tok->type = TK_CHAR_PROPERTY;
3754 tok->u.prop.not = (c ==
'P' ? 1 : 0);
3756 if (!PEND && IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT)) {
3759 tok->u.prop.not = (tok->u.prop.not == 0 ? 1 : 0);
3766 onig_syntax_warn(env,
"invalid Unicode Property \\%c", c);
3771 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_R_LINEBREAK)) {
3772 tok->type = TK_LINEBREAK;
3777 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_X_EXTENDED_GRAPHEME_CLUSTER)) {
3778 tok->type = TK_EXTENDED_GRAPHEME_CLUSTER;
3783 if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_K_KEEP)) {
3784 tok->type = TK_KEEP;
3793 num = fetch_escaped_value(&p, end, env, &c2);
3794 if (num < 0)
return num;
3796 if ((OnigCodePoint )tok->u.c != c2) {
3797 tok->type = TK_CODE_POINT;
3798 tok->u.code = (OnigCodePoint )c2;
3801 p = tok->backp + enclen(enc, tok->backp, end);
3811#ifdef USE_VARIABLE_META_CHARS
3812 if ((c != ONIG_INEFFECTIVE_META_CHAR) &&
3813 IS_SYNTAX_OP(syn, ONIG_SYN_OP_VARIABLE_META_CHARACTERS)) {
3814 if (c == MC_ANYCHAR(syn))
3816 else if (c == MC_ANYTIME(syn))
3818 else if (c == MC_ZERO_OR_ONE_TIME(syn))
3819 goto zero_or_one_time;
3820 else if (c == MC_ONE_OR_MORE_TIME(syn))
3821 goto one_or_more_time;
3822 else if (c == MC_ANYCHAR_ANYTIME(syn)) {
3823 tok->type = TK_ANYCHAR_ANYTIME;
3831 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_DOT_ANYCHAR))
break;
3832#ifdef USE_VARIABLE_META_CHARS
3835 tok->type = TK_ANYCHAR;
3839 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ASTERISK_ZERO_INF))
break;
3840#ifdef USE_VARIABLE_META_CHARS
3843 tok->type = TK_OP_REPEAT;
3844 tok->u.repeat.lower = 0;
3845 tok->u.repeat.upper = REPEAT_INFINITE;
3850 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_PLUS_ONE_INF))
break;
3851#ifdef USE_VARIABLE_META_CHARS
3854 tok->type = TK_OP_REPEAT;
3855 tok->u.repeat.lower = 1;
3856 tok->u.repeat.upper = REPEAT_INFINITE;
3861 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_QMARK_ZERO_ONE))
break;
3862#ifdef USE_VARIABLE_META_CHARS
3865 tok->type = TK_OP_REPEAT;
3866 tok->u.repeat.lower = 0;
3867 tok->u.repeat.upper = 1;
3872 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_BRACE_INTERVAL))
break;
3873 r = fetch_range_quantifier(&p, end, tok, env);
3874 if (r < 0)
return r;
3875 if (r == 0)
goto greedy_check;
3877 if (IS_SYNTAX_BV(syn, ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY))
3878 goto possessive_check;
3886 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_VBAR_ALT))
break;
3891 if (PPEEK_IS(
'?') &&
3892 IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_QMARK_GROUP_EFFECT)) {
3894 if (PPEEK_IS(
'#')) {
3897 if (PEND)
return ONIGERR_END_PATTERN_IN_GROUP;
3899 if (c == MC_ESC(syn)) {
3900 if (!PEND) PFETCH(c);
3903 if (c ==
')')
break;
3908#ifdef USE_PERL_SUBEXP_CALL
3911 if ((c ==
'&' || c ==
'R' || ONIGENC_IS_CODE_DIGIT(enc, c)) &&
3912 IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_SUBEXP_CALL)) {
3918 if (c ==
'R' || c ==
'0') {
3920 if (!PPEEK_IS(
')'))
return ONIGERR_INVALID_GROUP_NAME;
3922 name_end = name = p;
3932 r = fetch_name((OnigCodePoint )
'(', &p, end, &name_end, env, &gnum, numref);
3933 if (r < 0)
return r;
3936 tok->type = TK_CALL;
3937 tok->u.call.name = name;
3938 tok->u.call.name_end = name_end;
3939 tok->u.call.gnum = gnum;
3940 tok->u.call.rel = 0;
3943 else if ((c ==
'-' || c ==
'+') &&
3944 IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_SUBEXP_CALL)) {
3949 OnigCodePoint cnext;
3954 if (ONIGENC_IS_CODE_DIGIT(enc, cnext)) {
3955 if (c ==
'-') PUNFETCH;
3957 r = fetch_name((OnigCodePoint )
'(', &p, end, &name_end, env, &gnum, 1);
3958 if (r < 0)
return r;
3960 tok->type = TK_CALL;
3961 tok->u.call.name = name;
3962 tok->u.call.name_end = name_end;
3963 tok->u.call.gnum = gnum;
3964 tok->u.call.rel = 1;
3969#ifdef USE_CAPITAL_P_NAMED_GROUP
3970 if (PPEEK_IS(
'P') &&
3971 IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_CAPITAL_P_NAMED_GROUP)) {
3978 if (PEND)
return ONIGERR_UNDEFINED_GROUP_OPTION;
3981 r = fetch_named_backref_token((OnigCodePoint )
'(', tok, &p, end, env);
3982 if (r < 0)
return r;
3985 else if (c ==
'>') {
3987 r = fetch_name((OnigCodePoint )
'(', &p, end, &name_end, env, &gnum, 0);
3988 if (r < 0)
return r;
3990 tok->type = TK_CALL;
3991 tok->u.call.name = name;
3992 tok->u.call.name_end = name_end;
3993 tok->u.call.gnum = gnum;
3994 tok->u.call.rel = 0;
4002 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LPAREN_SUBEXP))
break;
4003 tok->type = TK_SUBEXP_OPEN;
4007 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LPAREN_SUBEXP))
break;
4008 tok->type = TK_SUBEXP_CLOSE;
4012 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LINE_ANCHOR))
break;
4013 tok->type = TK_ANCHOR;
4014 tok->u.anchor.subtype = (IS_SINGLELINE(env->option)
4015 ? ANCHOR_BEGIN_BUF : ANCHOR_BEGIN_LINE);
4019 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_LINE_ANCHOR))
break;
4020 tok->type = TK_ANCHOR;
4021 tok->u.anchor.subtype = (IS_SINGLELINE(env->option)
4022 ? ANCHOR_SEMI_END_BUF : ANCHOR_END_LINE);
4026 if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_BRACKET_CC))
break;
4027 tok->type = TK_CC_OPEN;
4031 if (*src > env->pattern)
4032 CLOSE_BRACKET_WITHOUT_ESC_WARN(env, (UChar* )
"]");
4036 if (IS_EXTEND(env->option)) {
4039 if (ONIGENC_IS_CODE_NEWLINE(enc, c))
4047 case ' ':
case '\t':
case '\n':
case '\r':
case '\f':
4048 if (IS_EXTEND(env->option))
4058#ifdef USE_VARIABLE_META_CHARS
4066add_ctype_to_cc_by_range(
CClassNode* cc,
int ctype ARG_UNUSED,
int not,
4068 OnigCodePoint sb_out,
const OnigCodePoint mbr[])
4073 int n = ONIGENC_CODE_RANGE_NUM(mbr);
4076 for (i = 0; i < n; i++) {
4077 for (j = ONIGENC_CODE_RANGE_FROM(mbr, i);
4078 j <= ONIGENC_CODE_RANGE_TO(mbr, i); j++) {
4080 if (j > ONIGENC_CODE_RANGE_FROM(mbr, i)) {
4081 r = add_code_range_to_buf(&(cc->mbuf), env, j,
4082 ONIGENC_CODE_RANGE_TO(mbr, i));
4083 if (r != 0)
return r;
4089 BITSET_SET_BIT_CHKDUP(cc->bs, j);
4094 for ( ; i < n; i++) {
4095 r = add_code_range_to_buf(&(cc->mbuf), env,
4096 ONIGENC_CODE_RANGE_FROM(mbr, i),
4097 ONIGENC_CODE_RANGE_TO(mbr, i));
4098 if (r != 0)
return r;
4102 OnigCodePoint prev = 0;
4104 for (i = 0; i < n; i++) {
4106 j < ONIGENC_CODE_RANGE_FROM(mbr, i); j++) {
4110 BITSET_SET_BIT_CHKDUP(cc->bs, j);
4112 prev = ONIGENC_CODE_RANGE_TO(mbr, i) + 1;
4114 for (j = prev; j < sb_out; j++) {
4115 BITSET_SET_BIT_CHKDUP(cc->bs, j);
4121 for (i = 0; i < n; i++) {
4122 if (prev < ONIGENC_CODE_RANGE_FROM(mbr, i)) {
4123 r = add_code_range_to_buf(&(cc->mbuf), env, prev,
4124 ONIGENC_CODE_RANGE_FROM(mbr, i) - 1);
4125 if (r != 0)
return r;
4127 prev = ONIGENC_CODE_RANGE_TO(mbr, i) + 1;
4129 if (prev < 0x7fffffff) {
4130 r = add_code_range_to_buf(&(cc->mbuf), env, prev, 0x7fffffff);
4131 if (r != 0)
return r;
4139add_ctype_to_cc(
CClassNode* cc,
int ctype,
int not,
int ascii_range,
ScanEnv* env)
4143 const OnigCodePoint *ranges;
4144 OnigCodePoint sb_out;
4147 r = ONIGENC_GET_CTYPE_CODE_RANGE(enc, ctype, &sb_out, &ranges);
4151 initialize_cclass(&ccwork);
4152 r = add_ctype_to_cc_by_range(&ccwork, ctype, not, env, sb_out,
4156 r = add_code_range_to_buf0(&(ccwork.mbuf), env, 0x80, ONIG_LAST_CODE_POINT, FALSE);
4160 initialize_cclass(&ccascii);
4161 if (ONIGENC_MBC_MINLEN(env->enc) > 1) {
4162 r = add_code_range(&(ccascii.mbuf), env, 0x00, 0x7F);
4165 bitset_set_range(env, ccascii.bs, 0x00, 0x7F);
4169 r = and_cclass(&ccwork, &ccascii, env);
4171 if (IS_NOT_NULL(ccascii.mbuf)) bbuf_free(ccascii.mbuf);
4174 r = or_cclass(cc, &ccwork, env);
4176 if (IS_NOT_NULL(ccwork.mbuf)) bbuf_free(ccwork.mbuf);
4180 r = add_ctype_to_cc_by_range(cc, ctype, not, env, sb_out, ranges);
4184 else if (r != ONIG_NO_SUPPORT_CONFIG) {
4188 maxcode = ascii_range ? 0x80 : SINGLE_BYTE_SIZE;
4191 case ONIGENC_CTYPE_ALPHA:
4192 case ONIGENC_CTYPE_BLANK:
4193 case ONIGENC_CTYPE_CNTRL:
4194 case ONIGENC_CTYPE_DIGIT:
4195 case ONIGENC_CTYPE_LOWER:
4196 case ONIGENC_CTYPE_PUNCT:
4197 case ONIGENC_CTYPE_SPACE:
4198 case ONIGENC_CTYPE_UPPER:
4199 case ONIGENC_CTYPE_XDIGIT:
4200 case ONIGENC_CTYPE_ASCII:
4201 case ONIGENC_CTYPE_ALNUM:
4203 for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
4204 if (! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
4205 BITSET_SET_BIT_CHKDUP(cc->bs, c);
4207 ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
4210 for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
4211 if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
4212 BITSET_SET_BIT_CHKDUP(cc->bs, c);
4217 case ONIGENC_CTYPE_GRAPH:
4218 case ONIGENC_CTYPE_PRINT:
4220 for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
4221 if (! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype)
4223 BITSET_SET_BIT_CHKDUP(cc->bs, c);
4226 ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
4229 for (c = 0; c < maxcode; c++) {
4230 if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
4231 BITSET_SET_BIT_CHKDUP(cc->bs, c);
4234 ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
4238 case ONIGENC_CTYPE_WORD:
4240 for (c = 0; c < maxcode; c++) {
4241 if (ONIGENC_IS_CODE_WORD(enc, c)) BITSET_SET_BIT_CHKDUP(cc->bs, c);
4244 ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
4247 for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
4248 if ((ONIGENC_CODE_TO_MBCLEN(enc, c) > 0)
4249 && (! ONIGENC_IS_CODE_WORD(enc, c) || c >= maxcode))
4250 BITSET_SET_BIT_CHKDUP(cc->bs, c);
4253 ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
4258 return ONIGERR_PARSER_BUG;
4267 UChar** src, UChar* end,
ScanEnv* env)
4269#define POSIX_BRACKET_CHECK_LIMIT_LENGTH 20
4270#define POSIX_BRACKET_NAME_MIN_LEN 4
4273 POSIX_BRACKET_ENTRY_INIT(
"alnum", ONIGENC_CTYPE_ALNUM),
4274 POSIX_BRACKET_ENTRY_INIT(
"alpha", ONIGENC_CTYPE_ALPHA),
4275 POSIX_BRACKET_ENTRY_INIT(
"blank", ONIGENC_CTYPE_BLANK),
4276 POSIX_BRACKET_ENTRY_INIT(
"cntrl", ONIGENC_CTYPE_CNTRL),
4277 POSIX_BRACKET_ENTRY_INIT(
"digit", ONIGENC_CTYPE_DIGIT),
4278 POSIX_BRACKET_ENTRY_INIT(
"graph", ONIGENC_CTYPE_GRAPH),
4279 POSIX_BRACKET_ENTRY_INIT(
"lower", ONIGENC_CTYPE_LOWER),
4280 POSIX_BRACKET_ENTRY_INIT(
"print", ONIGENC_CTYPE_PRINT),
4281 POSIX_BRACKET_ENTRY_INIT(
"punct", ONIGENC_CTYPE_PUNCT),
4282 POSIX_BRACKET_ENTRY_INIT(
"space", ONIGENC_CTYPE_SPACE),
4283 POSIX_BRACKET_ENTRY_INIT(
"upper", ONIGENC_CTYPE_UPPER),
4284 POSIX_BRACKET_ENTRY_INIT(
"xdigit", ONIGENC_CTYPE_XDIGIT),
4285 POSIX_BRACKET_ENTRY_INIT(
"ascii", ONIGENC_CTYPE_ASCII),
4286 POSIX_BRACKET_ENTRY_INIT(
"word", ONIGENC_CTYPE_WORD),
4296 if (PPEEK_IS(
'^')) {
4303 if (onigenc_strlen(enc, p, end) < POSIX_BRACKET_NAME_MIN_LEN + 3)
4304 goto not_posix_bracket;
4306 ascii_range = IS_ASCII_RANGE(env->option) &&
4307 ! IS_POSIX_BRACKET_ALL_RANGE(env->option);
4308 for (pb = PBS; pb < PBS + numberof(PBS); pb++) {
4309 if (onigenc_with_ascii_strncmp(enc, p, end, pb->name, pb->len) == 0) {
4310 p = (UChar* )onigenc_step(enc, p, end, pb->len);
4311 if (onigenc_with_ascii_strncmp(enc, p, end, (UChar* )
":]", 2) != 0)
4312 return ONIGERR_INVALID_POSIX_BRACKET_TYPE;
4314 r = add_ctype_to_cc(cc, pb->ctype, not, ascii_range, env);
4315 if (r != 0)
return r;
4317 if (IS_NOT_NULL(asc_cc)) {
4318 if (pb->ctype != ONIGENC_CTYPE_WORD &&
4319 pb->ctype != ONIGENC_CTYPE_ASCII &&
4321 r = add_ctype_to_cc(asc_cc, pb->ctype, not, ascii_range, env);
4322 if (r != 0)
return r;
4334 while (!PEND && ((c = PPEEK) !=
':') && c !=
']') {
4336 if (++i > POSIX_BRACKET_CHECK_LIMIT_LENGTH)
break;
4338 if (c ==
':' && ! PEND) {
4343 return ONIGERR_INVALID_POSIX_BRACKET_TYPE;
4351fetch_char_property_to_ctype(UChar** src, UChar* end,
ScanEnv* env)
4356 UChar *prev, *start, *p = *src;
4358 r = ONIGERR_INVALID_CHAR_PROPERTY_NAME;
4365 r = ONIGENC_PROPERTY_NAME_TO_CTYPE(enc, start, prev);
4371 else if (c ==
'(' || c ==
')' || c ==
'{' || c ==
'|') {
4376 onig_scan_env_set_error_string(env, r, *src, prev);
4383parse_char_property(
Node** np,
OnigToken* tok, UChar** src, UChar* end,
4389 ctype = fetch_char_property_to_ctype(src, end, env);
4390 if (ctype < 0)
return ctype;
4392 *np = node_new_cclass();
4393 CHECK_NULL_RETURN_MEMERR(*np);
4395 r = add_ctype_to_cc(cc, ctype, 0, 0, env);
4396 if (r != 0)
return r;
4397 if (tok->u.prop.not != 0) NCCLASS_SET_NOT(cc);
4399 if (IS_IGNORECASE(env->option)) {
4400 if (ctype != ONIGENC_CTYPE_ASCII)
4401 r = cclass_case_fold(np, cc, cc, env);
4422 OnigCodePoint* vs,
enum CCVALTYPE*
type,
4423 enum CCSTATE* state,
ScanEnv* env)
4427 if (*state == CCS_RANGE)
4428 return ONIGERR_CHAR_CLASS_VALUE_AT_END_OF_RANGE;
4430 if (*state == CCS_VALUE && *
type != CCV_CLASS) {
4431 if (*
type == CCV_SB) {
4432 BITSET_SET_BIT_CHKDUP(cc->bs, (
int )(*vs));
4433 if (IS_NOT_NULL(asc_cc))
4434 BITSET_SET_BIT(asc_cc->bs, (
int )(*vs));
4436 else if (*
type == CCV_CODE_POINT) {
4437 r = add_code_range(&(cc->mbuf), env, *vs, *vs);
4438 if (r < 0)
return r;
4439 if (IS_NOT_NULL(asc_cc)) {
4440 r = add_code_range0(&(asc_cc->mbuf), env, *vs, *vs, 0);
4441 if (r < 0)
return r;
4453 OnigCodePoint *from, OnigCodePoint to,
4454 int* from_israw,
int to_israw,
4455 enum CCVALTYPE intype,
enum CCVALTYPE*
type,
4456 enum CCSTATE* state,
ScanEnv* env)
4462 if (*
type == CCV_SB) {
4463 BITSET_SET_BIT_CHKDUP(cc->bs, (
int )(*from));
4464 if (IS_NOT_NULL(asc_cc))
4465 BITSET_SET_BIT(asc_cc->bs, (
int )(*from));
4467 else if (*
type == CCV_CODE_POINT) {
4468 r = add_code_range(&(cc->mbuf), env, *from, *from);
4469 if (r < 0)
return r;
4470 if (IS_NOT_NULL(asc_cc)) {
4471 r = add_code_range0(&(asc_cc->mbuf), env, *from, *from, 0);
4472 if (r < 0)
return r;
4478 if (intype == *
type) {
4479 if (intype == CCV_SB) {
4480 if (*from > 0xff || to > 0xff)
4481 return ONIGERR_INVALID_CODE_POINT_VALUE;
4484 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC))
4487 return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS;
4489 bitset_set_range(env, cc->bs, (
int )*from, (
int )to);
4490 if (IS_NOT_NULL(asc_cc))
4491 bitset_set_range(env, asc_cc->bs, (
int )*from, (
int )to);
4494 r = add_code_range(&(cc->mbuf), env, *from, to);
4495 if (r < 0)
return r;
4496 if (IS_NOT_NULL(asc_cc)) {
4497 r = add_code_range0(&(asc_cc->mbuf), env, *from, to, 0);
4498 if (r < 0)
return r;
4504 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC))
4507 return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS;
4509 bitset_set_range(env, cc->bs, (
int )*from, (
int )(to < 0xff ? to : 0xff));
4510 r = add_code_range(&(cc->mbuf), env, (OnigCodePoint )*from, to);
4511 if (r < 0)
return r;
4512 if (IS_NOT_NULL(asc_cc)) {
4513 bitset_set_range(env, asc_cc->bs, (
int )*from, (
int )(to < 0xff ? to : 0xff));
4514 r = add_code_range0(&(asc_cc->mbuf), env, (OnigCodePoint )*from, to, 0);
4515 if (r < 0)
return r;
4519 *state = CCS_COMPLETE;
4531 *from_israw = to_israw;
4538code_exist_check(OnigCodePoint c, UChar* from, UChar* end,
int ignore_escaped,
4548 if (ignore_escaped && in_esc) {
4553 if (code == c)
return 1;
4554 if (code == MC_ESC(env->syntax)) in_esc = 1;
4561parse_char_class(
Node** np,
Node** asc_np,
OnigToken* tok, UChar** src, UChar* end,
4564 int r, neg, len, fetched, and_start;
4565 OnigCodePoint v, vs;
4574 enum CCVALTYPE val_type, in_type;
4575 int val_israw, in_israw;
4577 *np = *asc_np = NULL_NODE;
4579 if (env->parse_depth > ParseDepthLimit)
4580 return ONIGERR_PARSE_DEPTH_LIMIT_OVER;
4582 r = fetch_token_in_cc(tok, src, end, env);
4583 if (r == TK_CHAR && tok->u.c ==
'^' && tok->escaped == 0) {
4585 r = fetch_token_in_cc(tok, src, end, env);
4591 if (r < 0)
return r;
4592 if (r == TK_CC_CLOSE) {
4593 if (! code_exist_check((OnigCodePoint )
']',
4594 *src, env->pattern_end, 1, env))
4595 return ONIGERR_EMPTY_CHAR_CLASS;
4597 CC_ESC_WARN(env, (UChar* )
"]");
4598 r = tok->type = TK_CHAR;
4601 *np = node = node_new_cclass();
4602 CHECK_NULL_RETURN_MEMERR(node);
4605 if (IS_IGNORECASE(env->option)) {
4606 *asc_np = asc_node = node_new_cclass();
4607 CHECK_NULL_RETURN_MEMERR(asc_node);
4608 asc_cc = NCCLASS(asc_node);
4611 asc_node = NULL_NODE;
4618 while (r != TK_CC_CLOSE) {
4622 if ((tok->u.code >= SINGLE_BYTE_SIZE) ||
4623 (len = ONIGENC_CODE_TO_MBCLEN(env->enc, tok->u.c)) > 1) {
4624 in_type = CCV_CODE_POINT;
4634 v = (OnigCodePoint )tok->u.c;
4641 if (! ONIGENC_IS_SINGLEBYTE(env->enc) && tok->base != 0) {
4642 UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];
4643 UChar* bufe = buf + ONIGENC_CODE_TO_MBC_MAXLEN;
4645 int i, base = tok->base;
4647 buf[0] = (UChar )tok->u.c;
4648 for (i = 1; i < ONIGENC_MBC_MAXLEN(env->enc); i++) {
4649 r = fetch_token_in_cc(tok, &p, end, env);
4650 if (r < 0)
goto err;
4651 if (r != TK_RAW_BYTE || tok->base != base) {
4655 buf[i] = (UChar )tok->u.c;
4658 if (i < ONIGENC_MBC_MINLEN(env->enc)) {
4659 r = ONIGERR_TOO_SHORT_MULTI_BYTE_STRING;
4663 len = enclen(env->enc, buf, buf + i);
4665 r = ONIGERR_TOO_SHORT_MULTI_BYTE_STRING;
4670 for (i = 1; i < len; i++) {
4671 (void)fetch_token_in_cc(tok, &p, end, env);
4678 v = (OnigCodePoint )buf[0];
4682 v = ONIGENC_MBC_TO_CODE(env->enc, buf, bufe);
4683 in_type = CCV_CODE_POINT;
4687 v = (OnigCodePoint )tok->u.c;
4699 len = ONIGENC_CODE_TO_MBCLEN(env->enc, v);
4704 in_type = (len == 1 ? CCV_SB : CCV_CODE_POINT);
4706 r = next_state_val(cc, asc_cc, &vs, v, &val_israw, in_israw, in_type, &val_type,
4708 if (r != 0)
goto err;
4711 case TK_POSIX_BRACKET_OPEN:
4712 r = parse_posix_bracket(cc, asc_cc, &p, end, env);
4713 if (r < 0)
goto err;
4715 CC_ESC_WARN(env, (UChar* )
"[");
4717 v = (OnigCodePoint )tok->u.c;
4725 r = add_ctype_to_cc(cc, tok->u.prop.ctype, tok->u.prop.not,
4726 IS_ASCII_RANGE(env->option), env);
4727 if (r != 0)
return r;
4728 if (IS_NOT_NULL(asc_cc)) {
4729 if (tok->u.prop.ctype != ONIGENC_CTYPE_WORD)
4730 r = add_ctype_to_cc(asc_cc, tok->u.prop.ctype, tok->u.prop.not,
4731 IS_ASCII_RANGE(env->option), env);
4732 if (r != 0)
return r;
4736 r = next_state_class(cc, asc_cc, &vs, &val_type, &state, env);
4737 if (r != 0)
goto err;
4740 case TK_CHAR_PROPERTY:
4744 ctype = fetch_char_property_to_ctype(&p, end, env);
4745 if (ctype < 0)
return ctype;
4746 r = add_ctype_to_cc(cc, ctype, tok->u.prop.not, 0, env);
4747 if (r != 0)
return r;
4748 if (IS_NOT_NULL(asc_cc)) {
4749 if (ctype != ONIGENC_CTYPE_ASCII)
4750 r = add_ctype_to_cc(asc_cc, ctype, tok->u.prop.not, 0, env);
4751 if (r != 0)
return r;
4758 if (state == CCS_VALUE) {
4759 r = fetch_token_in_cc(tok, &p, end, env);
4760 if (r < 0)
goto err;
4762 if (r == TK_CC_CLOSE) {
4764 v = (OnigCodePoint )
'-';
4768 else if (r == TK_CC_AND) {
4769 CC_ESC_WARN(env, (UChar* )
"-");
4773 if (val_type == CCV_CLASS) {
4774 r = ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS;
4780 else if (state == CCS_START) {
4782 v = (OnigCodePoint )tok->u.c;
4785 r = fetch_token_in_cc(tok, &p, end, env);
4786 if (r < 0)
goto err;
4789 if (r == TK_CC_RANGE || and_start != 0)
4790 CC_ESC_WARN(env, (UChar* )
"-");
4794 else if (state == CCS_RANGE) {
4795 CC_ESC_WARN(env, (UChar* )
"-");
4799 r = fetch_token_in_cc(tok, &p, end, env);
4800 if (r < 0)
goto err;
4802 if (r == TK_CC_CLOSE)
goto range_end_val;
4803 else if (r == TK_CC_AND) {
4804 CC_ESC_WARN(env, (UChar* )
"-");
4808 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC)) {
4809 CC_ESC_WARN(env, (UChar* )
"-");
4812 r = ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS;
4819 Node *anode, *aasc_node;
4822 r = parse_char_class(&anode, &aasc_node, tok, &p, end, env);
4824 acc = NCCLASS(anode);
4825 r = or_cclass(cc, acc, env);
4827 if (r == 0 && IS_NOT_NULL(aasc_node)) {
4828 acc = NCCLASS(aasc_node);
4829 r = or_cclass(asc_cc, acc, env);
4831 onig_node_free(anode);
4832 onig_node_free(aasc_node);
4833 if (r != 0)
goto err;
4839 if (state == CCS_VALUE) {
4840 r = next_state_val(cc, asc_cc, &vs, 0, &val_israw, 0, val_type,
4841 &val_type, &state, env);
4842 if (r != 0)
goto err;
4848 if (IS_NOT_NULL(prev_cc)) {
4849 r = and_cclass(prev_cc, cc, env);
4850 if (r != 0)
goto err;
4851 bbuf_free(cc->mbuf);
4852 if (IS_NOT_NULL(asc_cc)) {
4853 r = and_cclass(asc_prev_cc, asc_cc, env);
4854 if (r != 0)
goto err;
4855 bbuf_free(asc_cc->mbuf);
4861 if (IS_NOT_NULL(asc_cc)) {
4862 asc_prev_cc = asc_cc;
4863 asc_cc = &asc_work_cc;
4866 initialize_cclass(cc);
4867 if (IS_NOT_NULL(asc_cc))
4868 initialize_cclass(asc_cc);
4873 r = ONIGERR_PREMATURE_END_OF_CHAR_CLASS;
4877 r = ONIGERR_PARSER_BUG;
4885 r = fetch_token_in_cc(tok, &p, end, env);
4886 if (r < 0)
goto err;
4890 if (state == CCS_VALUE) {
4891 r = next_state_val(cc, asc_cc, &vs, 0, &val_israw, 0, val_type,
4892 &val_type, &state, env);
4893 if (r != 0)
goto err;
4896 if (IS_NOT_NULL(prev_cc)) {
4897 r = and_cclass(prev_cc, cc, env);
4898 if (r != 0)
goto err;
4899 bbuf_free(cc->mbuf);
4901 if (IS_NOT_NULL(asc_cc)) {
4902 r = and_cclass(asc_prev_cc, asc_cc, env);
4903 if (r != 0)
goto err;
4904 bbuf_free(asc_cc->mbuf);
4905 asc_cc = asc_prev_cc;
4910 NCCLASS_SET_NOT(cc);
4911 if (IS_NOT_NULL(asc_cc))
4912 NCCLASS_SET_NOT(asc_cc);
4915 NCCLASS_CLEAR_NOT(cc);
4916 if (IS_NOT_NULL(asc_cc))
4917 NCCLASS_CLEAR_NOT(asc_cc);
4919 if (IS_NCCLASS_NOT(cc) &&
4920 IS_SYNTAX_BV(env->syntax, ONIG_SYN_NOT_NEWLINE_IN_NEGATIVE_CC)) {
4923 is_empty = (IS_NULL(cc->mbuf) ? 1 : 0);
4925 BITSET_IS_EMPTY(cc->bs, is_empty);
4927 if (is_empty == 0) {
4928#define NEWLINE_CODE 0x0a
4930 if (ONIGENC_IS_CODE_NEWLINE(env->enc, NEWLINE_CODE)) {
4931 if (ONIGENC_CODE_TO_MBCLEN(env->enc, NEWLINE_CODE) == 1)
4932 BITSET_SET_BIT_CHKDUP(cc->bs, NEWLINE_CODE);
4934 r = add_code_range(&(cc->mbuf), env, NEWLINE_CODE, NEWLINE_CODE);
4935 if (r < 0)
goto err;
4945 if (cc != NCCLASS(*np))
4946 bbuf_free(cc->mbuf);
4947 if (IS_NOT_NULL(asc_cc) && (asc_cc != NCCLASS(*asc_np)))
4948 bbuf_free(asc_cc->mbuf);
4952static int parse_subexp(
Node** top,
OnigToken* tok,
int term,
4953 UChar** src, UChar* end,
ScanEnv* env);
4956parse_enclose(
Node** np,
OnigToken* tok,
int term, UChar** src, UChar* end,
4960 Node *target, *work1 = NULL, *work2 = NULL;
4961 OnigOptionType option;
4965#ifdef USE_NAMED_GROUP
4973 if (PEND)
return ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS;
4975 option = env->option;
4976 if (PPEEK_IS(
'?') &&
4977 IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_GROUP_EFFECT)) {
4979 if (PEND)
return ONIGERR_END_PATTERN_IN_GROUP;
4985 r = fetch_token(tok, &p, end, env);
4986 if (r < 0)
return r;
4987 r = parse_subexp(np, tok, term, &p, end, env);
4988 if (r < 0)
return r;
4994 *np = onig_node_new_anchor(ANCHOR_PREC_READ);
4997 *np = onig_node_new_anchor(ANCHOR_PREC_READ_NOT);
5000 *np = node_new_enclose(ENCLOSE_STOP_BACKTRACK);
5003 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_TILDE_ABSENT)) {
5004 *np = node_new_enclose(ENCLOSE_ABSENT);
5007 return ONIGERR_UNDEFINED_GROUP_OPTION;
5011#ifdef USE_NAMED_GROUP
5013 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) {
5017 return ONIGERR_UNDEFINED_GROUP_OPTION;
5020# ifdef USE_CAPITAL_P_NAMED_GROUP
5023 IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_CAPITAL_P_NAMED_GROUP)) {
5025 if (c ==
'<')
goto named_group1;
5027 return ONIGERR_UNDEFINED_GROUP_OPTION;
5033 if (PEND)
return ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS;
5036 *np = onig_node_new_anchor(ANCHOR_LOOK_BEHIND);
5038 *np = onig_node_new_anchor(ANCHOR_LOOK_BEHIND_NOT);
5039#ifdef USE_NAMED_GROUP
5041 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) {
5051# ifdef USE_CAPTURE_HISTORY
5055 r = fetch_name((OnigCodePoint )c, &p, end, &name_end, env, &num, 0);
5056 if (r < 0)
return r;
5058 num = scan_env_add_mem_entry(env);
5059 if (num < 0)
return num;
5060 if (list_capture != 0 && num >= (
int )BIT_STATUS_BITS_NUM)
5061 return ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY;
5063 r = name_add(env->reg, name, name_end, num, env);
5064 if (r != 0)
return r;
5065 *np = node_new_enclose_memory(env->option, 1);
5066 CHECK_NULL_RETURN_MEMERR(*np);
5067 NENCLOSE(*np)->regnum = num;
5068 if (list_capture != 0)
5069 BIT_STATUS_ON_AT_SIMPLE(env->capture_history, num);
5073 return ONIGERR_UNDEFINED_GROUP_OPTION;
5078 return ONIGERR_UNDEFINED_GROUP_OPTION;
5083#ifdef USE_CAPTURE_HISTORY
5085 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_ATMARK_CAPTURE_HISTORY)) {
5086# ifdef USE_NAMED_GROUP
5088 IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) {
5090 if (c ==
'<' || c ==
'\'') {
5097 *np = node_new_enclose_memory(env->option, 0);
5098 CHECK_NULL_RETURN_MEMERR(*np);
5099 num = scan_env_add_mem_entry(env);
5100 if (num < 0)
return num;
5101 if (num >= (
int )BIT_STATUS_BITS_NUM)
5102 return ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY;
5104 NENCLOSE(*np)->regnum = num;
5105 BIT_STATUS_ON_AT_SIMPLE(env->capture_history, num);
5108 return ONIGERR_UNDEFINED_GROUP_OPTION;
5115 IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LPAREN_CONDITION)) {
5119 if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
5121 r = fetch_name((OnigCodePoint )
'(', &p, end, &name_end, env, &num, 1);
5122 if (r < 0)
return r;
5126 num = BACKREF_REL_TO_ABS(num, env);
5128 return ONIGERR_INVALID_BACKREF;
5131 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_STRICT_CHECK_BACKREF)) {
5132 if (num > env->num_mem ||
5133 IS_NULL(SCANENV_MEM_NODES(env)[num]))
5134 return ONIGERR_INVALID_BACKREF;
5137#ifdef USE_NAMED_GROUP
5138 else if (c ==
'<' || c ==
'\'') {
5140 r = fetch_named_backref_token(c, tok, &p, end, env);
5141 if (r < 0)
return r;
5142 if (!PPEEK_IS(
')'))
return ONIGERR_UNDEFINED_GROUP_OPTION;
5145 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_USE_LEFT_MOST_NAMED_GROUP)) {
5146 num = tok->u.backref.ref1;
5153 int len = tok->u.backref.num;
5154 num = len > 1 ? tok->u.backref.refs[0] : tok->u.backref.ref1;
5159 return ONIGERR_INVALID_CONDITION_PATTERN;
5160 *np = node_new_enclose(ENCLOSE_CONDITION);
5161 CHECK_NULL_RETURN_MEMERR(*np);
5162 NENCLOSE(*np)->regnum = num;
5163 if (IS_NOT_NULL(name)) NENCLOSE(*np)->state |= NST_NAME_REF;
5166 return ONIGERR_UNDEFINED_GROUP_OPTION;
5171 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_VBAR_BRANCH_RESET)) {
5175 return ONIGERR_UNDEFINED_GROUP_OPTION;
5180 if (!PEND && IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) {
5182 ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1);
5183 ONOFF(option, ONIG_OPTION_IGNORECASE, 1);
5184 ONOFF(option, ONIG_OPTION_SINGLELINE, 0);
5185 ONOFF(option, ONIG_OPTION_MULTILINE, 1);
5186 ONOFF(option, ONIG_OPTION_EXTEND, 1);
5190 else if (!PEND && IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) {
5192 ONOFF(option, ONIG_OPTION_ASCII_RANGE, 0);
5193 ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 0);
5194 ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 0);
5195 ONOFF(option, ONIG_OPTION_IGNORECASE, 1);
5196 ONOFF(option, ONIG_OPTION_MULTILINE, 1);
5197 ONOFF(option, ONIG_OPTION_EXTEND, 1);
5202 return ONIGERR_UNDEFINED_GROUP_OPTION;
5205#ifdef USE_POSIXLINE_OPTION
5208 case '-':
case 'i':
case 'm':
case 's':
case 'x':
5209 case 'a':
case 'd':
case 'l':
case 'u':
5219 case '-': neg = 1;
break;
5220 case 'x': ONOFF(option, ONIG_OPTION_EXTEND, neg);
break;
5221 case 'i': ONOFF(option, ONIG_OPTION_IGNORECASE, neg);
break;
5223 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) {
5224 ONOFF(option, ONIG_OPTION_MULTILINE, neg);
5227 return ONIGERR_UNDEFINED_GROUP_OPTION;
5231 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) {
5232 ONOFF(option, ONIG_OPTION_SINGLELINE, (neg == 0 ? 1 : 0));
5234 else if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) {
5235 ONOFF(option, ONIG_OPTION_MULTILINE, neg);
5238 return ONIGERR_UNDEFINED_GROUP_OPTION;
5240#ifdef USE_POSIXLINE_OPTION
5242 ONOFF(option, ONIG_OPTION_MULTILINE|ONIG_OPTION_SINGLELINE, neg);
5247 if ((IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) ||
5248 IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) &&
5250 ONOFF(option, ONIG_OPTION_ASCII_RANGE, 0);
5251 ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 1);
5252 ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 1);
5255 return ONIGERR_UNDEFINED_GROUP_OPTION;
5259 if ((IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) ||
5260 IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY)) &&
5262 ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1);
5263 ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 1);
5264 ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 1);
5267 return ONIGERR_UNDEFINED_GROUP_OPTION;
5271 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) &&
5273 ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1);
5275 else if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_RUBY) &&
5277 ONOFF(option, ONIG_OPTION_ASCII_RANGE, 0);
5278 ONOFF(option, ONIG_OPTION_POSIX_BRACKET_ALL_RANGE, 0);
5279 ONOFF(option, ONIG_OPTION_WORD_BOUND_ALL_RANGE, 0);
5282 return ONIGERR_UNDEFINED_GROUP_OPTION;
5286 if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL) && (neg == 0)) {
5287 ONOFF(option, ONIG_OPTION_ASCII_RANGE, 1);
5290 return ONIGERR_UNDEFINED_GROUP_OPTION;
5294 return ONIGERR_UNDEFINED_GROUP_OPTION;
5298 *np = node_new_option(option);
5299 CHECK_NULL_RETURN_MEMERR(*np);
5303 else if (c ==
':') {
5304 OnigOptionType prev = env->option;
5306 env->option = option;
5307 r = fetch_token(tok, &p, end, env);
5312 r = parse_subexp(&target, tok, term, &p, end, env);
5314 if (r < 0)
return r;
5315 *np = node_new_option(option);
5316 CHECK_NULL_RETURN_MEMERR(*np);
5317 NENCLOSE(*np)->target = target;
5322 if (PEND)
return ONIGERR_END_PATTERN_IN_GROUP;
5329 return ONIGERR_UNDEFINED_GROUP_OPTION;
5333 if (ONIG_IS_OPTION_ON(env->option, ONIG_OPTION_DONT_CAPTURE_GROUP))
5336 *np = node_new_enclose_memory(env->option, 0);
5337 CHECK_NULL_RETURN_MEMERR(*np);
5338 num = scan_env_add_mem_entry(env);
5339 if (num < 0)
return num;
5340 NENCLOSE(*np)->regnum = num;
5343 CHECK_NULL_RETURN_MEMERR(*np);
5344 r = fetch_token(tok, &p, end, env);
5345 if (r < 0)
return r;
5346 r = parse_subexp(&target, tok, term, &p, end, env);
5348 onig_node_free(target);
5352 if (NTYPE(*np) == NT_ANCHOR)
5353 NANCHOR(*np)->target = target;
5355 NENCLOSE(*np)->target = target;
5356 if (NENCLOSE(*np)->type == ENCLOSE_MEMORY) {
5358 r = scan_env_set_mem_node(env, NENCLOSE(*np)->regnum, *np);
5359 if (r != 0)
return r;
5361 else if (NENCLOSE(*np)->type == ENCLOSE_CONDITION) {
5362 if (NTYPE(target) != NT_ALT) {
5364 work1 = node_new_empty();
5365 if (IS_NULL(work1))
goto err;
5366 work2 = onig_node_new_alt(work1, NULL_NODE);
5367 if (IS_NULL(work2))
goto err;
5368 work1 = onig_node_new_alt(target, work2);
5369 if (IS_NULL(work1))
goto err;
5370 NENCLOSE(*np)->target = work1;
5379 onig_node_free(work1);
5380 onig_node_free(work2);
5381 onig_node_free(*np);
5383 return ONIGERR_MEMORY;
5386static const char*
const PopularQStr[] = {
5387 "?",
"*",
"+",
"??",
"*?",
"+?"
5390static const char*
const ReduceQStr[] = {
5391 "",
"",
"*",
"*?",
"??",
"+ and ??",
"+? and ?"
5400 if (qn->lower == 1 && qn->upper == 1) {
5404 switch (NTYPE(target)) {
5408 if (str_node_can_be_split(sn, env->enc)) {
5409 Node* n = str_node_split_last_char(sn, env->enc);
5410 if (IS_NOT_NULL(n)) {
5422 int nestq_num = popular_quantifier_num(qn);
5423 int targetq_num = popular_quantifier_num(qnt);
5425#ifdef USE_WARNING_REDUNDANT_NESTED_REPEAT_OPERATOR
5426 if (nestq_num >= 0 && targetq_num >= 0 &&
5427 IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT)) {
5428 switch (ReduceTypeTable[targetq_num][nestq_num]) {
5433 if (onig_warn != onig_null_warn) {
5434 onig_syntax_warn(env,
"regular expression has redundant nested repeat operator '%s'",
5435 PopularQStr[targetq_num]);
5441 if (onig_warn != onig_null_warn) {
5442 onig_syntax_warn(env,
"nested repeat operator '%s' and '%s' was replaced with '%s' in regular expression",
5443 PopularQStr[targetq_num], PopularQStr[nestq_num],
5444 ReduceQStr[ReduceTypeTable[targetq_num][nestq_num]]);
5453 if (targetq_num >= 0) {
5454 if (nestq_num >= 0) {
5455 onig_reduce_nested_quantifier(qnode, target);
5458 else if (targetq_num == 1 || targetq_num == 2) {
5460 if (! IS_REPEAT_INFINITE(qn->upper) && qn->upper > 1 && qn->greedy) {
5461 qn->upper = (qn->lower == 0 ? 1 : qn->lower);
5472 qn->target = target;
5478#ifndef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS
5485 if (IS_NCCLASS_NOT(cc)) {
5486 bitset_invert(cc->bs);
5488 if (! ONIGENC_IS_SINGLEBYTE(enc)) {
5489 r = not_code_range_buf(enc, cc->mbuf, &tbuf);
5490 if (r != 0)
return r;
5492 bbuf_free(cc->mbuf);
5496 NCCLASS_CLEAR_NOT(cc);
5512i_apply_case_fold(OnigCodePoint from, OnigCodePoint to[],
5513 int to_len,
void* arg)
5525 asc_cc = iarg->asc_cc;
5528 if (IS_NULL(asc_cc)) {
5531 else if (ONIGENC_IS_ASCII_CODE(from) == ONIGENC_IS_ASCII_CODE(*to)) {
5535 add_flag = onig_is_code_in_cc(env->enc, from, asc_cc);
5536 if (IS_NCCLASS_NOT(asc_cc))
5537 add_flag = !add_flag;
5541 int is_in = onig_is_code_in_cc(env->enc, from, cc);
5542#ifdef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS
5543 if ((is_in != 0 && !IS_NCCLASS_NOT(cc)) ||
5544 (is_in == 0 && IS_NCCLASS_NOT(cc))) {
5546 if (ONIGENC_MBC_MINLEN(env->enc) > 1 || *to >= SINGLE_BYTE_SIZE) {
5547 r = add_code_range0(&(cc->mbuf), env, *to, *to, 0);
5548 if (r < 0)
return r;
5551 BITSET_SET_BIT(bs, *to);
5558 if (ONIGENC_MBC_MINLEN(env->enc) > 1 || *to >= SINGLE_BYTE_SIZE) {
5559 if (IS_NCCLASS_NOT(cc)) clear_not_flag_cclass(cc, env->enc);
5560 r = add_code_range0(&(cc->mbuf), env, *to, *to, 0);
5561 if (r < 0)
return r;
5564 if (IS_NCCLASS_NOT(cc)) {
5565 BITSET_CLEAR_BIT(bs, *to);
5568 BITSET_SET_BIT(bs, *to);
5577 UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];
5578 Node *snode = NULL_NODE;
5580 if (onig_is_code_in_cc(env->enc, from, cc)
5581#ifdef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS
5582 && !IS_NCCLASS_NOT(cc)
5585 for (i = 0; i < to_len; i++) {
5586 len = ONIGENC_CODE_TO_MBC(env->enc, to[i], buf);
5588 snode = onig_node_new_str(buf, buf + len);
5589 CHECK_NULL_RETURN_MEMERR(snode);
5593 NSTRING_SET_AMBIG(snode);
5596 r = onig_node_str_cat(snode, buf, buf + len);
5598 onig_node_free(snode);
5604 *(iarg->ptail) = onig_node_new_alt(snode, NULL_NODE);
5605 CHECK_NULL_RETURN_MEMERR(*(iarg->ptail));
5606 iarg->ptail = &(NCDR((*(iarg->ptail))));
5621 iarg.asc_cc = asc_cc;
5622 iarg.alt_root = NULL_NODE;
5623 iarg.ptail = &(iarg.alt_root);
5625 r = ONIGENC_APPLY_ALL_CASE_FOLD(env->enc, env->case_fold_flag,
5626 i_apply_case_fold, &iarg);
5628 onig_node_free(iarg.alt_root);
5631 if (IS_NOT_NULL(iarg.alt_root)) {
5632 Node* work = onig_node_new_alt(*np, iarg.alt_root);
5633 if (IS_NULL(work)) {
5634 onig_node_free(iarg.alt_root);
5635 return ONIGERR_MEMORY;
5648 Node* target1 = NULL;
5649 Node* target2 = NULL;
5652 UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN * 2];
5655 num1 = ONIGENC_CODE_TO_MBC(env->enc, 0x0D, buf);
5656 if (num1 < 0)
return num1;
5657 num2 = ONIGENC_CODE_TO_MBC(env->enc, 0x0A, buf + num1);
5658 if (num2 < 0)
return num2;
5659 left = node_new_str_raw(buf, buf + num1 + num2);
5660 if (IS_NULL(left))
goto err;
5663 right = node_new_cclass();
5664 if (IS_NULL(right))
goto err;
5665 cc = NCCLASS(right);
5666 if (ONIGENC_MBC_MINLEN(env->enc) > 1) {
5667 r = add_code_range(&(cc->mbuf), env, 0x0A, 0x0D);
5668 if (r != 0)
goto err;
5671 bitset_set_range(env, cc->bs, 0x0A, 0x0D);
5675 if (ONIGENC_IS_UNICODE(env->enc)) {
5677 r = add_code_range(&(cc->mbuf), env, 0x85, 0x85);
5678 if (r != 0)
goto err;
5679 r = add_code_range(&(cc->mbuf), env, 0x2028, 0x2029);
5680 if (r != 0)
goto err;
5684 target1 = onig_node_new_alt(right, NULL_NODE);
5685 if (IS_NULL(target1))
goto err;
5687 target2 = onig_node_new_alt(left, target1);
5688 if (IS_NULL(target2))
goto err;
5693 *np = node_new_enclose(ENCLOSE_STOP_BACKTRACK);
5694 if (IS_NULL(*np))
goto err;
5695 NENCLOSE(*np)->target = target2;
5699 onig_node_free(left);
5700 onig_node_free(right);
5701 onig_node_free(target1);
5702 onig_node_free(target2);
5703 return ONIGERR_MEMORY;
5707propname2ctype(
ScanEnv* env,
const char* propname)
5709 UChar* name = (UChar* )propname;
5710 UChar* name_end = name + strlen(propname);
5711 int ctype = env->enc->property_name_to_ctype(ONIG_ENCODING_ASCII,
5714 onig_scan_env_set_error_string(env, ctype, name, name_end);
5722 int ctype = propname2ctype(env, propname);
5723 if (ctype < 0)
return ctype;
5724 return add_ctype_to_cc(cc, ctype, not, 0, env);
5731create_property_node(
Node **np,
ScanEnv* env,
const char* propname)
5736 *np = node_new_cclass();
5737 if (IS_NULL(*np))
return ONIGERR_MEMORY;
5739 r = add_property_to_cc(cc, propname, 0, env);
5741 onig_node_free(*np);
5746quantify_node(
Node **np,
int lower,
int upper)
5748 Node* tmp = node_new_quantifier(lower, upper, 0);
5749 if (IS_NULL(tmp))
return ONIGERR_MEMORY;
5750 NQTFR(tmp)->target = *np;
5756quantify_property_node(
Node **np,
ScanEnv* env,
const char* propname,
char repetitions)
5760 int upper = REPEAT_INFINITE;
5762 r = create_property_node(np, env, propname);
5763 if (r != 0)
return r;
5764 switch (repetitions) {
5765 case '?': upper = 1;
break;
5766 case '+': lower = 1;
break;
5768 case '2': lower = upper = 2;
break;
5769 default :
return ONIGERR_PARSER_BUG;
5771 return quantify_node(np, lower, upper);
5779create_node_from_array(
int kind,
Node **np,
Node **node_array)
5781 Node* tmp = NULL_NODE;
5784 while (node_array[i] != NULL_NODE) i++;
5786 *np = kind==LIST ? node_new_list(node_array[i], tmp)
5787 : onig_node_new_alt(node_array[i], tmp);
5790 onig_node_free(node_array[i]);
5791 node_array[i--] = NULL_NODE;
5793 onig_node_free(tmp);
5794 return ONIGERR_MEMORY;
5797 node_array[i] = NULL_NODE;
5803#define R_ERR(call) r=(call);if(r!=0)goto err
5826#define NODE_COMMON_SIZE 16
5829node_extended_grapheme_cluster(
Node** np,
ScanEnv* env)
5833 Node* top_alt = NULL;
5837 int any_target_position;
5838 UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN * 2];
5839 OnigOptionType option;
5842 Node *node_common[NODE_COMMON_SIZE];
5843 Node **alts = node_common+0;
5845 for (i=0; i<NODE_COMMON_SIZE; i++)
5846 node_common[i] = NULL_NODE;
5850 r = ONIGENC_CODE_TO_MBC(env->enc, 0x0D, buf);
5851 if (r < 0)
goto err;
5853 r = ONIGENC_CODE_TO_MBC(env->enc, 0x0A, buf + num1);
5854 if (r < 0)
goto err;
5855 alts[0] = node_new_str_raw(buf, buf + num1 + r);
5856 if (IS_NULL(alts[0]))
goto err;
5858#ifdef USE_UNICODE_PROPERTIES
5859 if (ONIGENC_IS_UNICODE(env->enc)) {
5862 if (propname2ctype(env,
"Grapheme_Cluster_Break=Extend") < 0)
goto err;
5870 alts[1] = node_new_cclass();
5871 if (IS_NULL(alts[1]))
goto err;
5872 cc = NCCLASS(alts[1]);
5873 R_ERR(add_property_to_cc(cc,
"Grapheme_Cluster_Break=Control", 0, env));
5874 if (ONIGENC_MBC_MINLEN(env->enc) > 1) {
5875 R_ERR(add_code_range(&(cc->mbuf), env, 0x000A, 0x000A));
5876 R_ERR(add_code_range(&(cc->mbuf), env, 0x000D, 0x000D));
5879 BITSET_SET_BIT(cc->bs, 0x0a);
5880 BITSET_SET_BIT(cc->bs, 0x0d);
5885 Node **list = alts + 3;
5888 R_ERR(quantify_property_node(list+0, env,
"Grapheme_Cluster_Break=Prepend",
'*'));
5895 Node **core_alts = list + 2;
5906 Node **H_list = core_alts + 1;
5907 R_ERR(quantify_property_node(H_list+0, env,
"Grapheme_Cluster_Break=L",
'*'));
5911 Node **H_alt2 = H_list + 2;
5912 R_ERR(quantify_property_node(H_alt2+0, env,
"Grapheme_Cluster_Break=V",
'+'));
5916 Node **H_list2 = H_alt2 + 2;
5918 R_ERR(create_property_node(H_list2+0, env,
"Grapheme_Cluster_Break=LV"));
5919 R_ERR(quantify_property_node(H_list2+1, env,
"Grapheme_Cluster_Break=V",
'*'));
5920 R_ERR(create_node_from_array(LIST, H_alt2+1, H_list2));
5923 R_ERR(create_property_node(H_alt2+2, env,
"Grapheme_Cluster_Break=LVT"));
5924 R_ERR(create_node_from_array(ALT, H_list+1, H_alt2));
5927 R_ERR(quantify_property_node(H_list+2, env,
"Grapheme_Cluster_Break=T",
'*'));
5928 R_ERR(create_node_from_array(LIST, core_alts+0, H_list));
5931 R_ERR(quantify_property_node(core_alts+1, env,
"Grapheme_Cluster_Break=L",
'+'));
5932 R_ERR(quantify_property_node(core_alts+2, env,
"Grapheme_Cluster_Break=T",
'+'));
5936 R_ERR(quantify_property_node(core_alts+3, env,
"Regional_Indicator",
'2'));
5940 Node **XP_list = core_alts + 5;
5941 R_ERR(create_property_node(XP_list+0, env,
"Extended_Pictographic"));
5945 Node **Ex_list = XP_list + 2;
5947 R_ERR(quantify_property_node(Ex_list+0, env,
"Grapheme_Cluster_Break=Extend",
'*'));
5950 r = ONIGENC_CODE_TO_MBC(env->enc, 0x200D, buf);
5951 if (r < 0)
goto err;
5952 Ex_list[1] = node_new_str_raw(buf, buf + r);
5953 if (IS_NULL(Ex_list[1]))
goto err;
5955 R_ERR(create_property_node(Ex_list+2, env,
"Extended_Pictographic"));
5956 R_ERR(create_node_from_array(LIST, XP_list+1, Ex_list));
5958 R_ERR(quantify_node(XP_list+1, 0, REPEAT_INFINITE));
5960 R_ERR(create_node_from_array(LIST, core_alts+4, XP_list));
5964 core_alts[5] = node_new_cclass();
5965 if (IS_NULL(core_alts[5]))
goto err;
5966 cc = NCCLASS(core_alts[5]);
5967 if (ONIGENC_MBC_MINLEN(env->enc) > 1) {
5968 BBuf *inverted_buf = NULL;
5971 const int dup_not_warned = env->warnings_flag | ~ONIG_SYN_WARN_CC_DUP;
5972 env->warnings_flag |= ONIG_SYN_WARN_CC_DUP;
5976 R_ERR(add_property_to_cc(cc,
"Grapheme_Cluster_Break=Control", 0, env));
5977 R_ERR(add_code_range(&(cc->mbuf), env, 0x000A, 0x000A));
5978 R_ERR(add_code_range(&(cc->mbuf), env, 0x000D, 0x000D));
5979 R_ERR(not_code_range_buf(env->enc, cc->mbuf, &inverted_buf, env));
5980 cc->mbuf = inverted_buf;
5982 env->warnings_flag &= dup_not_warned;
5985 R_ERR(add_property_to_cc(cc,
"Grapheme_Cluster_Break=Control", 1, env));
5986 BITSET_CLEAR_BIT(cc->bs, 0x0a);
5987 BITSET_CLEAR_BIT(cc->bs, 0x0d);
5990 R_ERR(create_node_from_array(ALT, list+1, core_alts));
5994 R_ERR(create_property_node(list+2, env,
"Grapheme_Cluster_Break=Extend"));
5995 cc = NCCLASS(list[2]);
5996 R_ERR(add_property_to_cc(cc,
"Grapheme_Cluster_Break=SpacingMark", 0, env));
5997 R_ERR(add_code_range(&(cc->mbuf), env, 0x200D, 0x200D));
5998 R_ERR(quantify_node(list+2, 0, REPEAT_INFINITE));
6000 R_ERR(create_node_from_array(LIST, alts+2, list));
6003 any_target_position = 3;
6008 any_target_position = 1;
6014 np1 = node_new_anychar();
6015 if (IS_NULL(np1))
goto err;
6017 option = env->option;
6018 ONOFF(option, ONIG_OPTION_MULTILINE, 0);
6019 tmp = node_new_option(option);
6020 if (IS_NULL(tmp))
goto err;
6021 NENCLOSE(tmp)->target = np1;
6022 alts[any_target_position] = tmp;
6025 R_ERR(create_node_from_array(ALT, &top_alt, alts));
6030 tmp = node_new_enclose(ENCLOSE_STOP_BACKTRACK);
6031 if (IS_NULL(tmp))
goto err;
6032 NENCLOSE(tmp)->target = top_alt;
6035#ifdef USE_UNICODE_PROPERTIES
6036 if (ONIGENC_IS_UNICODE(env->enc)) {
6038 option = env->option;
6039 ONOFF(option, ONIG_OPTION_IGNORECASE, 1);
6040 *np = node_new_option(option);
6041 if (IS_NULL(*np))
goto err;
6042 NENCLOSE(*np)->target = np1;
6052 onig_node_free(np1);
6053 for (i=0; i<NODE_COMMON_SIZE; i++)
6054 onig_node_free(node_common[i]);
6055 return (r == 0) ? ONIGERR_MEMORY : r;
6060countbits(
unsigned int bits)
6062 bits = (bits & 0x55555555) + ((bits >> 1) & 0x55555555);
6063 bits = (bits & 0x33333333) + ((bits >> 2) & 0x33333333);
6064 bits = (bits & 0x0f0f0f0f) + ((bits >> 4) & 0x0f0f0f0f);
6065 bits = (bits & 0x00ff00ff) + ((bits >> 8) & 0x00ff00ff);
6066 return (bits & 0x0000ffff) + ((bits >>16) & 0x0000ffff);
6070is_onechar_cclass(
CClassNode* cc, OnigCodePoint* code)
6072 const OnigCodePoint not_found = ONIG_LAST_CODE_POINT;
6073 OnigCodePoint c = not_found;
6075 BBuf *bbuf = cc->mbuf;
6077 if (IS_NCCLASS_NOT(cc))
return 0;
6080 if (IS_NOT_NULL(bbuf)) {
6081 OnigCodePoint n, *data;
6082 GET_CODE_POINT(n, bbuf->p);
6083 data = (OnigCodePoint* )(bbuf->p) + 1;
6084 if ((n == 1) && (data[0] == data[1])) {
6087 if (((c < SINGLE_BYTE_SIZE) && BITSET_AT(cc->bs, c))) {
6098 for (i = 0; i < BITSET_SIZE; i++) {
6099 Bits b1 = cc->bs[i];
6101 if (((b1 & (b1 - 1)) == 0) && (c == not_found)) {
6102 c = BITS_IN_ROOM * i + countbits(b1 - 1);
6109 if (c != not_found) {
6121 UChar** src, UChar* end,
ScanEnv* env)
6123 int r, len, group = 0;
6128 if (tok->type == (
enum TokenSyms )term)
6131 switch (tok->type) {
6135 *np = node_new_empty();
6139 case TK_SUBEXP_OPEN:
6140 r = parse_enclose(np, tok, TK_SUBEXP_CLOSE, src, end, env);
6141 if (r < 0)
return r;
6142 if (r == 1) group = 1;
6145 OnigOptionType prev = env->option;
6147 env->option = NENCLOSE(*np)->option;
6148 r = fetch_token(tok, src, end, env);
6153 r = parse_subexp(&target, tok, term, src, end, env);
6156 onig_node_free(target);
6159 NENCLOSE(*np)->target = target;
6164 case TK_SUBEXP_CLOSE:
6165 if (! IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_UNMATCHED_CLOSE_SUBEXP))
6166 return ONIGERR_UNMATCHED_CLOSE_PARENTHESIS;
6168 if (tok->escaped)
goto tk_raw_byte;
6173 r = node_linebreak(np, env);
6174 if (r < 0)
return r;
6177 case TK_EXTENDED_GRAPHEME_CLUSTER:
6178 r = node_extended_grapheme_cluster(np, env);
6179 if (r < 0)
return r;
6183 *np = onig_node_new_anchor(ANCHOR_KEEP);
6184 CHECK_NULL_RETURN_MEMERR(*np);
6190 *np = node_new_str(tok->backp, *src);
6191 CHECK_NULL_RETURN_MEMERR(*np);
6195 r = fetch_token(tok, src, end, env);
6196 if (r < 0)
return r;
6197 if (r == TK_STRING) {
6198 r = onig_node_str_cat(*np, tok->backp, *src);
6200#ifndef NUMBERED_CHAR_IS_NOT_CASE_AMBIG
6201 else if (r == TK_CODE_POINT) {
6202 r = node_str_cat_codepoint(*np, env->enc, tok->u.code);
6208 if (r < 0)
return r;
6220 *np = node_new_str_raw_char((UChar )tok->u.c);
6221 CHECK_NULL_RETURN_MEMERR(*np);
6224 if (len >= ONIGENC_MBC_MINLEN(env->enc)) {
6225 if (len == enclen(env->enc, NSTR(*np)->s, NSTR(*np)->end)) {
6226 r = fetch_token(tok, src, end, env);
6227 NSTRING_CLEAR_RAW(*np);
6232 r = fetch_token(tok, src, end, env);
6233 if (r < 0)
return r;
6234 if (r != TK_RAW_BYTE) {
6236#ifdef USE_PAD_TO_SHORT_BYTE_CHAR
6238 if (len < ONIGENC_MBC_MINLEN(env->enc)) {
6239 rem = ONIGENC_MBC_MINLEN(env->enc) - len;
6240 (void )node_str_head_pad(NSTR(*np), rem, (UChar )0);
6241 if (len + rem == enclen(env->enc, NSTR(*np)->s)) {
6242 NSTRING_CLEAR_RAW(*np);
6247 return ONIGERR_TOO_SHORT_MULTI_BYTE_STRING;
6250 r = node_str_cat_char(*np, (UChar )tok->u.c);
6251 if (r < 0)
return r;
6260 *np = node_new_empty();
6261 CHECK_NULL_RETURN_MEMERR(*np);
6262 r = node_str_cat_codepoint(*np, env->enc, tok->u.code);
6263 if (r != 0)
return r;
6264#ifdef NUMBERED_CHAR_IS_NOT_CASE_AMBIG
6265 NSTRING_SET_RAW(*np);
6274 OnigCodePoint end_op[2];
6275 UChar *qstart, *qend, *nextp;
6277 end_op[0] = (OnigCodePoint )MC_ESC(env->syntax);
6278 end_op[1] = (OnigCodePoint )
'E';
6280 qend = find_str_position(end_op, 2, qstart, end, &nextp, env->enc);
6281 if (IS_NULL(qend)) {
6284 *np = node_new_str(qstart, qend);
6285 CHECK_NULL_RETURN_MEMERR(*np);
6292 switch (tok->u.prop.ctype) {
6293 case ONIGENC_CTYPE_WORD:
6294 *np = node_new_ctype(tok->u.prop.ctype, tok->u.prop.not,
6295 IS_ASCII_RANGE(env->option));
6296 CHECK_NULL_RETURN_MEMERR(*np);
6299 case ONIGENC_CTYPE_SPACE:
6300 case ONIGENC_CTYPE_DIGIT:
6301 case ONIGENC_CTYPE_XDIGIT:
6305 *np = node_new_cclass();
6306 CHECK_NULL_RETURN_MEMERR(*np);
6308 r = add_ctype_to_cc(cc, tok->u.prop.ctype, 0,
6309 IS_ASCII_RANGE(env->option), env);
6310 if (r != 0)
return r;
6311 if (tok->u.prop.not != 0) NCCLASS_SET_NOT(cc);
6316 return ONIGERR_PARSER_BUG;
6322 case TK_CHAR_PROPERTY:
6323 r = parse_char_property(np, tok, src, end, env);
6324 if (r != 0)
return r;
6333 r = parse_char_class(np, &asc_node, tok, src, end, env);
6335 onig_node_free(asc_node);
6340 if (is_onechar_cclass(cc, &code)) {
6341 onig_node_free(*np);
6342 onig_node_free(asc_node);
6343 *np = node_new_empty();
6344 CHECK_NULL_RETURN_MEMERR(*np);
6345 r = node_str_cat_codepoint(*np, env->enc, code);
6346 if (r != 0)
return r;
6349 if (IS_IGNORECASE(env->option)) {
6350 r = cclass_case_fold(np, cc, NCCLASS(asc_node), env);
6352 onig_node_free(asc_node);
6356 onig_node_free(asc_node);
6361 *np = node_new_anychar();
6362 CHECK_NULL_RETURN_MEMERR(*np);
6365 case TK_ANYCHAR_ANYTIME:
6366 *np = node_new_anychar();
6367 CHECK_NULL_RETURN_MEMERR(*np);
6368 qn = node_new_quantifier(0, REPEAT_INFINITE, 0);
6369 CHECK_NULL_RETURN_MEMERR(qn);
6370 NQTFR(qn)->target = *np;
6375 len = tok->u.backref.num;
6376 *np = node_new_backref(len,
6377 (len > 1 ? tok->u.backref.refs : &(tok->u.backref.ref1)),
6378 tok->u.backref.by_name,
6379#ifdef USE_BACKREF_WITH_LEVEL
6380 tok->u.backref.exist_level,
6381 tok->u.backref.level,
6384 CHECK_NULL_RETURN_MEMERR(*np);
6387#ifdef USE_SUBEXP_CALL
6390 int gnum = tok->u.call.gnum;
6392 if (gnum < 0 || tok->u.call.rel != 0) {
6393 if (gnum > 0) gnum--;
6394 gnum = BACKREF_REL_TO_ABS(gnum, env);
6396 return ONIGERR_INVALID_BACKREF;
6398 *np = node_new_call(tok->u.call.name, tok->u.call.name_end, gnum);
6399 CHECK_NULL_RETURN_MEMERR(*np);
6406 *np = onig_node_new_anchor(tok->u.anchor.subtype);
6407 CHECK_NULL_RETURN_MEMERR(*np);
6408 NANCHOR(*np)->ascii_range = tok->u.anchor.ascii_range;
6413 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS)) {
6414 if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS))
6415 return ONIGERR_TARGET_OF_REPEAT_OPERATOR_NOT_SPECIFIED;
6417 *np = node_new_empty();
6425 return ONIGERR_PARSER_BUG;
6433 r = fetch_token(tok, src, end, env);
6434 if (r < 0)
return r;
6437 if (r == TK_OP_REPEAT || r == TK_INTERVAL) {
6438 if (is_invalid_quantifier_target(*targetp))
6439 return ONIGERR_TARGET_OF_REPEAT_OPERATOR_INVALID;
6441 qn = node_new_quantifier(tok->u.repeat.lower, tok->u.repeat.upper,
6442 (r == TK_INTERVAL ? 1 : 0));
6443 CHECK_NULL_RETURN_MEMERR(qn);
6444 NQTFR(qn)->greedy = tok->u.repeat.greedy;
6445 r = set_quantifier(qn, *targetp, group, env);
6451 if (tok->u.repeat.possessive != 0) {
6453 en = node_new_enclose(ENCLOSE_STOP_BACKTRACK);
6456 return ONIGERR_MEMORY;
6458 NENCLOSE(en)->target = qn;
6471 *targetp = node_new_list(*targetp, NULL);
6472 if (IS_NULL(*targetp)) {
6474 return ONIGERR_MEMORY;
6476 tmp = NCDR(*targetp) = node_new_list(qn, NULL);
6479 return ONIGERR_MEMORY;
6481 targetp = &(NCAR(tmp));
6492 UChar** src, UChar* end,
ScanEnv* env)
6495 Node *node, **headp;
6498 r = parse_exp(&node, tok, term, src, end, env);
6500 onig_node_free(node);
6504 if (r == TK_EOT || r == term || r == TK_ALT) {
6508 *top = node_new_list(node, NULL);
6509 headp = &(NCDR(*top));
6510 while (r != TK_EOT && r != term && r != TK_ALT) {
6511 r = parse_exp(&node, tok, term, src, end, env);
6513 onig_node_free(node);
6517 if (NTYPE(node) == NT_LIST) {
6519 while (IS_NOT_NULL(NCDR(node))) node = NCDR(node);
6520 headp = &(NCDR(node));
6523 *headp = node_new_list(node, NULL);
6524 headp = &(NCDR(*headp));
6535 UChar** src, UChar* end,
ScanEnv* env)
6538 Node *node, **headp;
6542 if (env->parse_depth > ParseDepthLimit)
6543 return ONIGERR_PARSE_DEPTH_LIMIT_OVER;
6544 r = parse_branch(&node, tok, term, src, end, env);
6546 onig_node_free(node);
6553 else if (r == TK_ALT) {
6554 *top = onig_node_new_alt(node, NULL);
6555 headp = &(NCDR(*top));
6556 while (r == TK_ALT) {
6557 r = fetch_token(tok, src, end, env);
6558 if (r < 0)
return r;
6559 r = parse_branch(&node, tok, term, src, end, env);
6561 onig_node_free(node);
6565 *headp = onig_node_new_alt(node, NULL);
6566 headp = &(NCDR(*headp));
6569 if (tok->type != (
enum TokenSyms )term)
6573 onig_node_free(node);
6575 if (term == TK_SUBEXP_CLOSE)
6576 return ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS;
6578 return ONIGERR_PARSER_BUG;
6586parse_regexp(
Node** top, UChar** src, UChar* end,
ScanEnv* env)
6591 r = fetch_token(&tok, src, end, env);
6592 if (r < 0)
return r;
6593 r = parse_subexp(top, &tok, TK_EOT, src, end, env);
6594 if (r < 0)
return r;
6596#ifdef USE_SUBEXP_CALL
6597 if (env->num_call > 0) {
6601 np = node_new_enclose_memory(env->option, 0);
6602 CHECK_NULL_RETURN_MEMERR(np);
6603 NENCLOSE(np)->regnum = num;
6604 NENCLOSE(np)->target = *top;
6605 r = scan_env_set_mem_node(env, num, np);
6617onig_parse_make_tree(
Node** root,
const UChar* pattern,
const UChar* end,
6623#ifdef USE_NAMED_GROUP
6627 scan_env_clear(env);
6628 env->option = reg->options;
6629 env->case_fold_flag = reg->case_fold_flag;
6630 env->enc = reg->enc;
6631 env->syntax = reg->syntax;
6632 env->pattern = (UChar* )pattern;
6633 env->pattern_end = (UChar* )end;
6637 p = (UChar* )pattern;
6638 r = parse_regexp(root, &p, (UChar* )end, env);
6639 reg->num_mem = env->num_mem;
6644onig_scan_env_set_error_string(
ScanEnv* env,
int ecode ARG_UNUSED,
6645 UChar* arg, UChar* arg_end)
6648 env->error_end = arg_end;
#define xfree
Old name of ruby_xfree.
#define xrealloc
Old name of ruby_xrealloc.
#define xmalloc
Old name of ruby_xmalloc.
void rb_compile_warn(const char *file, int line, const char *fmt,...)
Identical to rb_compile_warning(), except it reports always regardless of runtime -W flag.
#define ruby_verbose
This variable controls whether the interpreter is in debug mode.
void rb_warn(const char *fmt,...)
Identical to rb_warning(), except it reports always regardless of runtime -W flag.
VALUE type(ANYARGS)
ANYARGS-ed function type.
#define RTEST
This is an old name of RB_TEST.