regcomp.c - Oniguruma (regular expression library)\r
**********************************************************************/\r
/*-\r
- * Copyright (c) 2002-2013 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>\r
+ * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>\r
* All rights reserved.\r
*\r
- * (C) Copyright 2015 Hewlett Packard Enterprise Development LP<BR>\r
- *\r
* Redistribution and use in source and binary forms, with or without\r
* modification, are permitted provided that the following conditions\r
* are met:\r
\r
OnigCaseFoldType OnigDefaultCaseFoldFlag = ONIGENC_CASE_FOLD_MIN;\r
\r
+#if 0\r
+typedef struct {\r
+ int n;\r
+ int alloc;\r
+ int* v;\r
+} int_stack;\r
+\r
+static int\r
+make_int_stack(int_stack** rs, int init_size)\r
+{\r
+ int_stack* s;\r
+ int* v;\r
+\r
+ *rs = 0;\r
+\r
+ s = xmalloc(sizeof(*s));\r
+ if (IS_NULL(s)) return ONIGERR_MEMORY;\r
+\r
+ v = (int* )xmalloc(sizeof(int) * init_size);\r
+ if (IS_NULL(v)) {\r
+ xfree(s);\r
+ return ONIGERR_MEMORY;\r
+ }\r
+\r
+ s->n = 0;\r
+ s->alloc = init_size;\r
+ s->v = v;\r
+\r
+ *rs = s;\r
+ return ONIG_NORMAL;\r
+}\r
+\r
+static void\r
+free_int_stack(int_stack* s)\r
+{\r
+ if (IS_NOT_NULL(s)) {\r
+ if (IS_NOT_NULL(s->v))\r
+ xfree(s->v);\r
+ xfree(s);\r
+ }\r
+}\r
+\r
+static int\r
+int_stack_push(int_stack* s, int v)\r
+{\r
+ if (s->n >= s->alloc) {\r
+ int new_size = s->alloc * 2;\r
+ int* nv = (int* )xrealloc(s->v, sizeof(int) * new_size);\r
+ if (IS_NULL(nv)) return ONIGERR_MEMORY;\r
+\r
+ s->alloc = new_size;\r
+ s->v = nv;\r
+ }\r
+\r
+ s->v[s->n] = v;\r
+ s->n++;\r
+ return ONIG_NORMAL;\r
+}\r
+\r
+static int\r
+int_stack_pop(int_stack* s)\r
+{\r
+ int v;\r
+\r
+#ifdef ONIG_DEBUG\r
+ if (s->n <= 0) {\r
+ fprintf(stderr, "int_stack_pop: fail empty. %p\n", s);\r
+ return 0;\r
+ }\r
+#endif\r
+\r
+ v = s->v[s->n];\r
+ s->n--;\r
+ return v;\r
+}\r
+#endif\r
+\r
extern OnigCaseFoldType\r
onig_get_default_case_fold_flag(void)\r
{\r
return 0;\r
}\r
\r
-\r
-#ifndef PLATFORM_UNALIGNED_WORD_ACCESS\r
-static unsigned char PadBuf[WORD_ALIGNMENT_SIZE];\r
-#endif\r
-\r
-static UChar*\r
-str_dup(UChar* s, UChar* end)\r
+static int\r
+int_multiply_cmp(int x, int y, int v)\r
{\r
- int len = (int)(end - s);\r
+ if (x == 0 || y == 0) return -1;\r
\r
- if (len > 0) {\r
- UChar* r = (UChar* )xmalloc(len + 1);\r
- CHECK_NULL_RETURN(r);\r
- xmemcpy(r, s, len);\r
- r[len] = (UChar )0;\r
- return r;\r
+ if (x < INT_MAX / y) {\r
+ int xy = x * y;\r
+ if (xy > v) return 1;\r
+ else {\r
+ if (xy == v) return 0;\r
+ else return -1;\r
+ }\r
}\r
- else return NULL;\r
+ else\r
+ return 1;\r
}\r
\r
+\r
+#ifndef PLATFORM_UNALIGNED_WORD_ACCESS\r
+static unsigned char PadBuf[WORD_ALIGNMENT_SIZE];\r
+#endif\r
+\r
static void\r
swap_node(Node* a, Node* b)\r
{\r
Node c;\r
+\r
c = *a; *a = *b; *b = c;\r
\r
- if (NTYPE(a) == NT_STR) {\r
- StrNode* sn = NSTR(a);\r
+ if (NODE_TYPE(a) == NODE_STRING) {\r
+ StrNode* sn = STR_(a);\r
if (sn->capa == 0) {\r
- int len = (int)(sn->end - sn->s);\r
+ int len = (int )(sn->end - sn->s);\r
sn->s = sn->buf;\r
sn->end = sn->s + len;\r
}\r
}\r
\r
- if (NTYPE(b) == NT_STR) {\r
- StrNode* sn = NSTR(b);\r
+ if (NODE_TYPE(b) == NODE_STRING) {\r
+ StrNode* sn = STR_(b);\r
if (sn->capa == 0) {\r
- int len = (int)(sn->end - sn->s);\r
+ int len = (int )(sn->end - sn->s);\r
sn->s = sn->buf;\r
sn->end = sn->s + len;\r
}\r
}\r
}\r
\r
-static OnigDistance\r
-distance_add(OnigDistance d1, OnigDistance d2)\r
+static OnigLen\r
+distance_add(OnigLen d1, OnigLen d2)\r
{\r
- if (d1 == ONIG_INFINITE_DISTANCE || d2 == ONIG_INFINITE_DISTANCE)\r
- return ONIG_INFINITE_DISTANCE;\r
+ if (d1 == INFINITE_LEN || d2 == INFINITE_LEN)\r
+ return INFINITE_LEN;\r
else {\r
- if (d1 <= ONIG_INFINITE_DISTANCE - d2) return d1 + d2;\r
- else return ONIG_INFINITE_DISTANCE;\r
+ if (d1 <= INFINITE_LEN - d2) return d1 + d2;\r
+ else return INFINITE_LEN;\r
}\r
}\r
\r
-static OnigDistance\r
-distance_multiply(OnigDistance d, int m)\r
+static OnigLen\r
+distance_multiply(OnigLen d, int m)\r
{\r
if (m == 0) return 0;\r
\r
- if (d < ONIG_INFINITE_DISTANCE / m)\r
+ if (d < INFINITE_LEN / m)\r
return d * m;\r
else\r
- return ONIG_INFINITE_DISTANCE;\r
+ return INFINITE_LEN;\r
}\r
\r
static int\r
bitset_is_empty(BitSetRef bs)\r
{\r
int i;\r
+\r
for (i = 0; i < (int )BITSET_SIZE; i++) {\r
if (bs[i] != 0) return 0;\r
}\r
return 1;\r
}\r
\r
-#ifdef ONIG_DEBUG\r
-static int\r
-bitset_on_num(BitSetRef bs)\r
-{\r
- int i, n;\r
-\r
- n = 0;\r
- for (i = 0; i < SINGLE_BYTE_SIZE; i++) {\r
- if (BITSET_AT(bs, i)) n++;\r
- }\r
- return n;\r
-}\r
-#endif\r
-\r
extern int\r
onig_bbuf_init(BBuf* buf, int size)\r
{\r
}\r
\r
\r
-#ifdef USE_SUBEXP_CALL\r
+#ifdef USE_CALL\r
\r
static int\r
-unset_addr_list_init(UnsetAddrList* uslist, int size)\r
+unset_addr_list_init(UnsetAddrList* list, int size)\r
{\r
- UnsetAddr* p;\r
-\r
- p = (UnsetAddr* )xmalloc(sizeof(UnsetAddr)* size);\r
+ UnsetAddr* p = (UnsetAddr* )xmalloc(sizeof(UnsetAddr)* size);\r
CHECK_NULL_RETURN_MEMERR(p);\r
- uslist->num = 0;\r
- uslist->alloc = size;\r
- uslist->us = p;\r
+\r
+ list->num = 0;\r
+ list->alloc = size;\r
+ list->us = p;\r
return 0;\r
}\r
\r
static void\r
-unset_addr_list_end(UnsetAddrList* uslist)\r
+unset_addr_list_end(UnsetAddrList* list)\r
{\r
- if (IS_NOT_NULL(uslist->us))\r
- xfree(uslist->us);\r
+ if (IS_NOT_NULL(list->us))\r
+ xfree(list->us);\r
}\r
\r
static int\r
-unset_addr_list_add(UnsetAddrList* uslist, int offset, struct _Node* node)\r
+unset_addr_list_add(UnsetAddrList* list, int offset, struct _Node* node)\r
{\r
UnsetAddr* p;\r
int size;\r
\r
- if (uslist->num >= uslist->alloc) {\r
- size = uslist->alloc * 2;\r
- p = (UnsetAddr* )xrealloc(uslist->us, sizeof(UnsetAddr) * size, sizeof(UnsetAddr) * uslist->alloc);\r
+ if (list->num >= list->alloc) {\r
+ size = list->alloc * 2;\r
+ p = (UnsetAddr* )xrealloc(list->us, sizeof(UnsetAddr) * size, sizeof(UnsetAddr)* list->alloc);\r
CHECK_NULL_RETURN_MEMERR(p);\r
- uslist->alloc = size;\r
- uslist->us = p;\r
+ list->alloc = size;\r
+ list->us = p;\r
}\r
\r
- uslist->us[uslist->num].offset = offset;\r
- uslist->us[uslist->num].target = node;\r
- uslist->num++;\r
+ list->us[list->num].offset = offset;\r
+ list->us[list->num].target = node;\r
+ list->num++;\r
return 0;\r
}\r
-#endif /* USE_SUBEXP_CALL */\r
+#endif /* USE_CALL */\r
\r
\r
static int\r
add_opcode(regex_t* reg, int opcode)\r
{\r
- BBUF_ADD1(reg, ((unsigned char)opcode));\r
- return 0;\r
-}\r
-\r
-#ifdef USE_COMBINATION_EXPLOSION_CHECK\r
-static int\r
-add_state_check_num(regex_t* reg, int num)\r
-{\r
- StateCheckNumType n = (StateCheckNumType )num;\r
-\r
- BBUF_ADD(reg, &n, SIZE_STATE_CHECK_NUM);\r
+ BB_ADD1(reg, opcode);\r
return 0;\r
}\r
-#endif\r
\r
static int\r
add_rel_addr(regex_t* reg, int addr)\r
{\r
RelAddrType ra = (RelAddrType )addr;\r
\r
- BBUF_ADD(reg, &ra, SIZE_RELADDR);\r
+ BB_ADD(reg, &ra, SIZE_RELADDR);\r
return 0;\r
}\r
\r
{\r
AbsAddrType ra = (AbsAddrType )addr;\r
\r
- BBUF_ADD(reg, &ra, SIZE_ABSADDR);\r
+ BB_ADD(reg, &ra, SIZE_ABSADDR);\r
return 0;\r
}\r
\r
{\r
LengthType l = (LengthType )len;\r
\r
- BBUF_ADD(reg, &l, SIZE_LENGTH);\r
+ BB_ADD(reg, &l, SIZE_LENGTH);\r
return 0;\r
}\r
\r
{\r
MemNumType n = (MemNumType )num;\r
\r
- BBUF_ADD(reg, &n, SIZE_MEMNUM);\r
+ BB_ADD(reg, &n, SIZE_MEMNUM);\r
return 0;\r
}\r
\r
+#if 0\r
static int\r
add_pointer(regex_t* reg, void* addr)\r
{\r
PointerType ptr = (PointerType )addr;\r
\r
- BBUF_ADD(reg, &ptr, SIZE_POINTER);\r
+ BB_ADD(reg, &ptr, SIZE_POINTER);\r
return 0;\r
}\r
+#endif\r
\r
static int\r
add_option(regex_t* reg, OnigOptionType option)\r
{\r
- BBUF_ADD(reg, &option, SIZE_OPTION);\r
+ BB_ADD(reg, &option, SIZE_OPTION);\r
return 0;\r
}\r
\r
static int\r
-add_opcode_rel_addr(regex_t* reg, int opcode, int addr)\r
+add_save_type(regex_t* reg, enum SaveType type)\r
{\r
- int r;\r
+ SaveType t = (SaveType )type;\r
\r
- r = add_opcode(reg, opcode);\r
- if (r) return r;\r
- r = add_rel_addr(reg, addr);\r
- return r;\r
+ BB_ADD(reg, &t, SIZE_SAVE_TYPE);\r
+ return 0;\r
}\r
\r
static int\r
-add_bytes(regex_t* reg, UChar* bytes, int len)\r
+add_update_var_type(regex_t* reg, enum UpdateVarType type)\r
{\r
- BBUF_ADD(reg, bytes, len);\r
+ UpdateVarType t = (UpdateVarType )type;\r
+\r
+ BB_ADD(reg, &t, SIZE_UPDATE_VAR_TYPE);\r
return 0;\r
}\r
\r
static int\r
-add_bitset(regex_t* reg, BitSetRef bs)\r
+add_mode(regex_t* reg, ModeType mode)\r
{\r
- BBUF_ADD(reg, bs, SIZE_BITSET);\r
+ BB_ADD(reg, &mode, SIZE_MODE);\r
return 0;\r
}\r
\r
static int\r
-add_opcode_option(regex_t* reg, int opcode, OnigOptionType option)\r
+add_opcode_rel_addr(regex_t* reg, int opcode, int addr)\r
{\r
int r;\r
\r
r = add_opcode(reg, opcode);\r
- if (r) return r;\r
- r = add_option(reg, option);\r
+ if (r != 0) return r;\r
+ r = add_rel_addr(reg, addr);\r
return r;\r
}\r
\r
+static int\r
+add_bytes(regex_t* reg, UChar* bytes, int len)\r
+{\r
+ BB_ADD(reg, bytes, len);\r
+ return 0;\r
+}\r
+\r
+static int\r
+add_bitset(regex_t* reg, BitSetRef bs)\r
+{\r
+ BB_ADD(reg, bs, SIZE_BITSET);\r
+ return 0;\r
+}\r
+\r
static int compile_length_tree(Node* node, regex_t* reg);\r
-static int compile_tree(Node* node, regex_t* reg);\r
+static int compile_tree(Node* node, regex_t* reg, ScanEnv* env);\r
\r
\r
#define IS_NEED_STR_LEN_OP_EXACT(op) \\r
}\r
\r
static int\r
-compile_tree_empty_check(Node* node, regex_t* reg, int empty_info)\r
+compile_tree_empty_check(Node* node, regex_t* reg, int empty_info, ScanEnv* env)\r
{\r
int r;\r
int saved_num_null_check = reg->num_null_check;\r
\r
- if (empty_info != 0) {\r
- r = add_opcode(reg, OP_NULL_CHECK_START);\r
- if (r) return r;\r
+ if (empty_info != QUANT_BODY_IS_NOT_EMPTY) {\r
+ r = add_opcode(reg, OP_EMPTY_CHECK_START);\r
+ if (r != 0) return r;\r
r = add_mem_num(reg, reg->num_null_check); /* NULL CHECK ID */\r
- if (r) return r;\r
+ if (r != 0) return r;\r
reg->num_null_check++;\r
}\r
\r
- r = compile_tree(node, reg);\r
- if (r) return r;\r
+ r = compile_tree(node, reg, env);\r
+ if (r != 0) return r;\r
\r
- if (empty_info != 0) {\r
- if (empty_info == NQ_TARGET_IS_EMPTY)\r
- r = add_opcode(reg, OP_NULL_CHECK_END);\r
- else if (empty_info == NQ_TARGET_IS_EMPTY_MEM)\r
- r = add_opcode(reg, OP_NULL_CHECK_END_MEMST);\r
- else if (empty_info == NQ_TARGET_IS_EMPTY_REC)\r
- r = add_opcode(reg, OP_NULL_CHECK_END_MEMST_PUSH);\r
+ if (empty_info != QUANT_BODY_IS_NOT_EMPTY) {\r
+ if (empty_info == QUANT_BODY_IS_EMPTY)\r
+ r = add_opcode(reg, OP_EMPTY_CHECK_END);\r
+ else if (empty_info == QUANT_BODY_IS_EMPTY_MEM)\r
+ r = add_opcode(reg, OP_EMPTY_CHECK_END_MEMST);\r
+ else if (empty_info == QUANT_BODY_IS_EMPTY_REC)\r
+ r = add_opcode(reg, OP_EMPTY_CHECK_END_MEMST_PUSH);\r
\r
- if (r) return r;\r
+ if (r != 0) return r;\r
r = add_mem_num(reg, saved_num_null_check); /* NULL CHECK ID */\r
}\r
return r;\r
}\r
\r
-#ifdef USE_SUBEXP_CALL\r
+#ifdef USE_CALL\r
static int\r
-compile_call(CallNode* node, regex_t* reg)\r
+compile_call(CallNode* node, regex_t* reg, ScanEnv* env)\r
{\r
int r;\r
\r
r = add_opcode(reg, OP_CALL);\r
- if (r) return r;\r
- r = unset_addr_list_add(node->unset_addr_list, BBUF_GET_OFFSET_POS(reg),\r
- node->target);\r
- if (r) return r;\r
+ if (r != 0) return r;\r
+ r = unset_addr_list_add(env->unset_addr_list, BB_GET_OFFSET_POS(reg),\r
+ NODE_CALL_BODY(node));\r
+ if (r != 0) return r;\r
r = add_abs_addr(reg, 0 /*dummy addr.*/);\r
return r;\r
}\r
#endif\r
\r
static int\r
-compile_tree_n_times(Node* node, int n, regex_t* reg)\r
+compile_tree_n_times(Node* node, int n, regex_t* reg, ScanEnv* env)\r
{\r
int i, r;\r
\r
for (i = 0; i < n; i++) {\r
- r = compile_tree(node, reg);\r
- if (r) return r;\r
+ r = compile_tree(node, reg, env);\r
+ if (r != 0) return r;\r
}\r
return 0;\r
}\r
compile_length_string_node(Node* node, regex_t* reg)\r
{\r
int rlen, r, len, prev_len, slen, ambig;\r
- OnigEncoding enc = reg->enc;\r
UChar *p, *prev;\r
StrNode* sn;\r
+ OnigEncoding enc = reg->enc;\r
\r
- sn = NSTR(node);\r
+ sn = STR_(node);\r
if (sn->end <= sn->s)\r
return 0;\r
\r
- ambig = NSTRING_IS_AMBIG(node);\r
+ ambig = NODE_STRING_IS_AMBIG(node);\r
\r
p = prev = sn->s;\r
prev_len = enclen(enc, p);\r
}\r
p += len;\r
}\r
+\r
r = add_compile_string_length(prev, prev_len, slen, reg, ambig);\r
rlen += r;\r
return rlen;\r
if (sn->end <= sn->s)\r
return 0;\r
\r
- return add_compile_string_length(sn->s, 1 /* sb */, (int)(sn->end - sn->s), reg, 0);\r
+ return add_compile_string_length(sn->s, 1 /* sb */, (int )(sn->end - sn->s),\r
+ reg, 0);\r
}\r
\r
static int\r
compile_string_node(Node* node, regex_t* reg)\r
{\r
int r, len, prev_len, slen, ambig;\r
- OnigEncoding enc = reg->enc;\r
UChar *p, *prev, *end;\r
StrNode* sn;\r
+ OnigEncoding enc = reg->enc;\r
\r
- sn = NSTR(node);\r
+ sn = STR_(node);\r
if (sn->end <= sn->s)\r
return 0;\r
\r
end = sn->end;\r
- ambig = NSTRING_IS_AMBIG(node);\r
+ ambig = NODE_STRING_IS_AMBIG(node);\r
\r
p = prev = sn->s;\r
prev_len = enclen(enc, p);\r
}\r
else {\r
r = add_compile_string(prev, prev_len, slen, reg, ambig);\r
- if (r) return r;\r
+ if (r != 0) return r;\r
\r
prev = p;\r
slen = 1;\r
\r
p += len;\r
}\r
+\r
return add_compile_string(prev, prev_len, slen, reg, ambig);\r
}\r
\r
if (sn->end <= sn->s)\r
return 0;\r
\r
- return add_compile_string(sn->s, 1 /* sb */, (int)(sn->end - sn->s), reg, 0);\r
+ return add_compile_string(sn->s, 1 /* sb */, (int )(sn->end - sn->s), reg, 0);\r
}\r
\r
static int\r
return add_bytes(reg, mbuf->p, mbuf->used);\r
#else\r
int r, pad_size;\r
- UChar* p = BBUF_GET_ADD_ADDRESS(reg) + SIZE_LENGTH;\r
+ UChar* p = BB_GET_ADD_ADDRESS(reg) + SIZE_LENGTH;\r
\r
GET_ALIGNMENT_PAD_SIZE(p, pad_size);\r
add_length(reg, mbuf->used + (WORD_ALIGNMENT_SIZE - 1));\r
{\r
int len;\r
\r
- if (IS_NCCLASS_SHARE(cc)) {\r
- len = SIZE_OPCODE + SIZE_POINTER;\r
- return len;\r
- }\r
-\r
if (IS_NULL(cc->mbuf)) {\r
len = SIZE_OPCODE + SIZE_BITSET;\r
}\r
{\r
int r;\r
\r
- if (IS_NCCLASS_SHARE(cc)) {\r
- add_opcode(reg, OP_CCLASS_NODE);\r
- r = add_pointer(reg, cc);\r
- return r;\r
- }\r
-\r
if (IS_NULL(cc->mbuf)) {\r
if (IS_NCCLASS_NOT(cc))\r
add_opcode(reg, OP_CCLASS_NOT);\r
add_opcode(reg, OP_CCLASS_MIX);\r
\r
r = add_bitset(reg, cc->bs);\r
- if (r) return r;\r
+ if (r != 0) return r;\r
r = add_multi_byte_cclass(cc->mbuf, reg);\r
}\r
}\r
}\r
\r
static int\r
-compile_range_repeat_node(QtfrNode* qn, int target_len, int empty_info,\r
- regex_t* reg)\r
+compile_range_repeat_node(QuantNode* qn, int target_len, int empty_info,\r
+ regex_t* reg, ScanEnv* env)\r
{\r
int r;\r
int num_repeat = reg->num_repeat;\r
\r
r = add_opcode(reg, qn->greedy ? OP_REPEAT : OP_REPEAT_NG);\r
- if (r) return r;\r
+ if (r != 0) return r;\r
r = add_mem_num(reg, num_repeat); /* OP_REPEAT ID */\r
reg->num_repeat++;\r
- if (r) return r;\r
+ if (r != 0) return r;\r
r = add_rel_addr(reg, target_len + SIZE_OP_REPEAT_INC);\r
- if (r) return r;\r
+ if (r != 0) return r;\r
\r
r = entry_repeat_range(reg, num_repeat, qn->lower, qn->upper);\r
- if (r) return r;\r
+ if (r != 0) return r;\r
\r
- r = compile_tree_empty_check(qn->target, reg, empty_info);\r
- if (r) return r;\r
+ r = compile_tree_empty_check(NODE_QUANT_BODY(qn), reg, empty_info, env);\r
+ if (r != 0) return r;\r
\r
if (\r
-#ifdef USE_SUBEXP_CALL\r
- reg->num_call > 0 ||\r
+#ifdef USE_CALL\r
+ NODE_IS_IN_MULTI_ENTRY(qn) ||\r
#endif\r
- IS_QUANTIFIER_IN_REPEAT(qn)) {\r
+ NODE_IS_IN_REAL_REPEAT(qn)) {\r
r = add_opcode(reg, qn->greedy ? OP_REPEAT_INC_SG : OP_REPEAT_INC_NG_SG);\r
}\r
else {\r
r = add_opcode(reg, qn->greedy ? OP_REPEAT_INC : OP_REPEAT_INC_NG);\r
}\r
- if (r) return r;\r
+ if (r != 0) return r;\r
r = add_mem_num(reg, num_repeat); /* OP_REPEAT ID */\r
return r;\r
}\r
\r
static int\r
-is_anychar_star_quantifier(QtfrNode* qn)\r
+is_anychar_infinite_greedy(QuantNode* qn)\r
{\r
if (qn->greedy && IS_REPEAT_INFINITE(qn->upper) &&\r
- NTYPE(qn->target) == NT_CANY)\r
+ NODE_IS_ANYCHAR(NODE_QUANT_BODY(qn)))\r
return 1;\r
else\r
return 0;\r
#define QUANTIFIER_EXPAND_LIMIT_SIZE 50\r
#define CKN_ON (ckn > 0)\r
\r
-#ifdef USE_COMBINATION_EXPLOSION_CHECK\r
-\r
-static int\r
-compile_length_quantifier_node(QtfrNode* qn, regex_t* reg)\r
-{\r
- int len, mod_tlen, cklen;\r
- int ckn;\r
- int infinite = IS_REPEAT_INFINITE(qn->upper);\r
- int empty_info = qn->target_empty_info;\r
- int tlen = compile_length_tree(qn->target, reg);\r
-\r
- if (tlen < 0) return tlen;\r
-\r
- ckn = ((reg->num_comb_exp_check > 0) ? qn->comb_exp_check_num : 0);\r
-\r
- cklen = (CKN_ON ? SIZE_STATE_CHECK_NUM: 0);\r
-\r
- /* anychar repeat */\r
- if (NTYPE(qn->target) == NT_CANY) {\r
- if (qn->greedy && infinite) {\r
- if (IS_NOT_NULL(qn->next_head_exact) && !CKN_ON)\r
- return SIZE_OP_ANYCHAR_STAR_PEEK_NEXT + tlen * qn->lower + cklen;\r
- else\r
- return SIZE_OP_ANYCHAR_STAR + tlen * qn->lower + cklen;\r
- }\r
- }\r
-\r
- if (empty_info != 0)\r
- mod_tlen = tlen + (SIZE_OP_NULL_CHECK_START + SIZE_OP_NULL_CHECK_END);\r
- else\r
- mod_tlen = tlen;\r
-\r
- if (infinite && qn->lower <= 1) {\r
- if (qn->greedy) {\r
- if (qn->lower == 1)\r
- len = SIZE_OP_JUMP;\r
- else\r
- len = 0;\r
-\r
- len += SIZE_OP_PUSH + cklen + mod_tlen + SIZE_OP_JUMP;\r
- }\r
- else {\r
- if (qn->lower == 0)\r
- len = SIZE_OP_JUMP;\r
- else\r
- len = 0;\r
-\r
- len += mod_tlen + SIZE_OP_PUSH + cklen;\r
- }\r
- }\r
- else if (qn->upper == 0) {\r
- if (qn->is_refered != 0) /* /(?<n>..){0}/ */\r
- len = SIZE_OP_JUMP + tlen;\r
- else\r
- len = 0;\r
- }\r
- else if (qn->upper == 1 && qn->greedy) {\r
- if (qn->lower == 0) {\r
- if (CKN_ON) {\r
- len = SIZE_OP_STATE_CHECK_PUSH + tlen;\r
- }\r
- else {\r
- len = SIZE_OP_PUSH + tlen;\r
- }\r
- }\r
- else {\r
- len = tlen;\r
- }\r
- }\r
- else if (!qn->greedy && qn->upper == 1 && qn->lower == 0) { /* '??' */\r
- len = SIZE_OP_PUSH + cklen + SIZE_OP_JUMP + tlen;\r
- }\r
- else {\r
- len = SIZE_OP_REPEAT_INC\r
- + mod_tlen + SIZE_OPCODE + SIZE_RELADDR + SIZE_MEMNUM;\r
- if (CKN_ON)\r
- len += SIZE_OP_STATE_CHECK;\r
- }\r
-\r
- return len;\r
-}\r
-\r
-static int\r
-compile_quantifier_node(QtfrNode* qn, regex_t* reg)\r
-{\r
- int r, mod_tlen;\r
- int ckn;\r
- int infinite = IS_REPEAT_INFINITE(qn->upper);\r
- int empty_info = qn->target_empty_info;\r
- int tlen = compile_length_tree(qn->target, reg);\r
-\r
- if (tlen < 0) return tlen;\r
-\r
- ckn = ((reg->num_comb_exp_check > 0) ? qn->comb_exp_check_num : 0);\r
-\r
- if (is_anychar_star_quantifier(qn)) {\r
- r = compile_tree_n_times(qn->target, qn->lower, reg);\r
- if (r) return r;\r
- if (IS_NOT_NULL(qn->next_head_exact) && !CKN_ON) {\r
- if (IS_MULTILINE(reg->options))\r
- r = add_opcode(reg, OP_ANYCHAR_ML_STAR_PEEK_NEXT);\r
- else\r
- r = add_opcode(reg, OP_ANYCHAR_STAR_PEEK_NEXT);\r
- if (r) return r;\r
- if (CKN_ON) {\r
- r = add_state_check_num(reg, ckn);\r
- if (r) return r;\r
- }\r
-\r
- return add_bytes(reg, NSTR(qn->next_head_exact)->s, 1);\r
- }\r
- else {\r
- if (IS_MULTILINE(reg->options)) {\r
- r = add_opcode(reg, (CKN_ON ?\r
- OP_STATE_CHECK_ANYCHAR_ML_STAR\r
- : OP_ANYCHAR_ML_STAR));\r
- }\r
- else {\r
- r = add_opcode(reg, (CKN_ON ?\r
- OP_STATE_CHECK_ANYCHAR_STAR\r
- : OP_ANYCHAR_STAR));\r
- }\r
- if (r) return r;\r
- if (CKN_ON)\r
- r = add_state_check_num(reg, ckn);\r
-\r
- return r;\r
- }\r
- }\r
-\r
- if (empty_info != 0)\r
- mod_tlen = tlen + (SIZE_OP_NULL_CHECK_START + SIZE_OP_NULL_CHECK_END);\r
- else\r
- mod_tlen = tlen;\r
-\r
- if (infinite && qn->lower <= 1) {\r
- if (qn->greedy) {\r
- if (qn->lower == 1) {\r
- r = add_opcode_rel_addr(reg, OP_JUMP,\r
- (CKN_ON ? SIZE_OP_STATE_CHECK_PUSH : SIZE_OP_PUSH));\r
- if (r) return r;\r
- }\r
-\r
- if (CKN_ON) {\r
- r = add_opcode(reg, OP_STATE_CHECK_PUSH);\r
- if (r) return r;\r
- r = add_state_check_num(reg, ckn);\r
- if (r) return r;\r
- r = add_rel_addr(reg, mod_tlen + SIZE_OP_JUMP);\r
- }\r
- else {\r
- r = add_opcode_rel_addr(reg, OP_PUSH, mod_tlen + SIZE_OP_JUMP);\r
- }\r
- if (r) return r;\r
- r = compile_tree_empty_check(qn->target, reg, empty_info);\r
- if (r) return r;\r
- r = add_opcode_rel_addr(reg, OP_JUMP,\r
- -(mod_tlen + (int )SIZE_OP_JUMP\r
- + (int )(CKN_ON ? SIZE_OP_STATE_CHECK_PUSH : SIZE_OP_PUSH)));\r
- }\r
- else {\r
- if (qn->lower == 0) {\r
- r = add_opcode_rel_addr(reg, OP_JUMP, mod_tlen);\r
- if (r) return r;\r
- }\r
- r = compile_tree_empty_check(qn->target, reg, empty_info);\r
- if (r) return r;\r
- if (CKN_ON) {\r
- r = add_opcode(reg, OP_STATE_CHECK_PUSH_OR_JUMP);\r
- if (r) return r;\r
- r = add_state_check_num(reg, ckn);\r
- if (r) return r;\r
- r = add_rel_addr(reg,\r
- -(mod_tlen + (int )SIZE_OP_STATE_CHECK_PUSH_OR_JUMP));\r
- }\r
- else\r
- r = add_opcode_rel_addr(reg, OP_PUSH, -(mod_tlen + (int )SIZE_OP_PUSH));\r
- }\r
- }\r
- else if (qn->upper == 0) {\r
- if (qn->is_refered != 0) { /* /(?<n>..){0}/ */\r
- r = add_opcode_rel_addr(reg, OP_JUMP, tlen);\r
- if (r) return r;\r
- r = compile_tree(qn->target, reg);\r
- }\r
- else\r
- r = 0;\r
- }\r
- else if (qn->upper == 1 && qn->greedy) {\r
- if (qn->lower == 0) {\r
- if (CKN_ON) {\r
- r = add_opcode(reg, OP_STATE_CHECK_PUSH);\r
- if (r) return r;\r
- r = add_state_check_num(reg, ckn);\r
- if (r) return r;\r
- r = add_rel_addr(reg, tlen);\r
- }\r
- else {\r
- r = add_opcode_rel_addr(reg, OP_PUSH, tlen);\r
- }\r
- if (r) return r;\r
- }\r
-\r
- r = compile_tree(qn->target, reg);\r
- }\r
- else if (!qn->greedy && qn->upper == 1 && qn->lower == 0) { /* '??' */\r
- if (CKN_ON) {\r
- r = add_opcode(reg, OP_STATE_CHECK_PUSH);\r
- if (r) return r;\r
- r = add_state_check_num(reg, ckn);\r
- if (r) return r;\r
- r = add_rel_addr(reg, SIZE_OP_JUMP);\r
- }\r
- else {\r
- r = add_opcode_rel_addr(reg, OP_PUSH, SIZE_OP_JUMP);\r
- }\r
-\r
- if (r) return r;\r
- r = add_opcode_rel_addr(reg, OP_JUMP, tlen);\r
- if (r) return r;\r
- r = compile_tree(qn->target, reg);\r
- }\r
- else {\r
- r = compile_range_repeat_node(qn, mod_tlen, empty_info, reg);\r
- if (CKN_ON) {\r
- if (r) return r;\r
- r = add_opcode(reg, OP_STATE_CHECK);\r
- if (r) return r;\r
- r = add_state_check_num(reg, ckn);\r
- }\r
- }\r
- return r;\r
-}\r
-\r
-#else /* USE_COMBINATION_EXPLOSION_CHECK */\r
-\r
static int\r
-compile_length_quantifier_node(QtfrNode* qn, regex_t* reg)\r
+compile_length_quantifier_node(QuantNode* qn, regex_t* reg)\r
{\r
int len, mod_tlen;\r
int infinite = IS_REPEAT_INFINITE(qn->upper);\r
- int empty_info = qn->target_empty_info;\r
- int tlen = compile_length_tree(qn->target, reg);\r
+ enum QuantBodyEmpty empty_info = qn->body_empty_info;\r
+ int tlen = compile_length_tree(NODE_QUANT_BODY(qn), reg);\r
\r
if (tlen < 0) return tlen;\r
+ if (tlen == 0) return 0;\r
\r
/* anychar repeat */\r
- if (NTYPE(qn->target) == NT_CANY) {\r
- if (qn->greedy && infinite) {\r
+ if (is_anychar_infinite_greedy(qn)) {\r
+ if (qn->lower <= 1 ||\r
+ int_multiply_cmp(tlen, qn->lower, QUANTIFIER_EXPAND_LIMIT_SIZE) <= 0) {\r
if (IS_NOT_NULL(qn->next_head_exact))\r
return SIZE_OP_ANYCHAR_STAR_PEEK_NEXT + tlen * qn->lower;\r
else\r
}\r
}\r
\r
- if (empty_info != 0)\r
- mod_tlen = tlen + (SIZE_OP_NULL_CHECK_START + SIZE_OP_NULL_CHECK_END);\r
- else\r
+ if (empty_info == QUANT_BODY_IS_NOT_EMPTY)\r
mod_tlen = tlen;\r
+ else\r
+ mod_tlen = tlen + (SIZE_OP_EMPTY_CHECK_START + SIZE_OP_EMPTY_CHECK_END);\r
\r
if (infinite &&\r
- (qn->lower <= 1 || tlen * qn->lower <= QUANTIFIER_EXPAND_LIMIT_SIZE)) {\r
+ (qn->lower <= 1 ||\r
+ int_multiply_cmp(tlen, qn->lower, QUANTIFIER_EXPAND_LIMIT_SIZE) <= 0)) {\r
if (qn->lower == 1 && tlen > QUANTIFIER_EXPAND_LIMIT_SIZE) {\r
len = SIZE_OP_JUMP;\r
}\r
\r
if (qn->greedy) {\r
if (IS_NOT_NULL(qn->head_exact))\r
- len += SIZE_OP_PUSH_OR_JUMP_EXACT1 + mod_tlen + SIZE_OP_JUMP;\r
+ len += SIZE_OP_PUSH_OR_JUMP_EXACT1 + mod_tlen + SIZE_OP_JUMP;\r
else if (IS_NOT_NULL(qn->next_head_exact))\r
- len += SIZE_OP_PUSH_IF_PEEK_NEXT + mod_tlen + SIZE_OP_JUMP;\r
+ len += SIZE_OP_PUSH_IF_PEEK_NEXT + mod_tlen + SIZE_OP_JUMP;\r
else\r
- len += SIZE_OP_PUSH + mod_tlen + SIZE_OP_JUMP;\r
+ len += SIZE_OP_PUSH + mod_tlen + SIZE_OP_JUMP;\r
}\r
else\r
len += SIZE_OP_JUMP + mod_tlen + SIZE_OP_PUSH;\r
len = SIZE_OP_JUMP + tlen;\r
}\r
else if (!infinite && qn->greedy &&\r
- (qn->upper == 1 || (tlen + SIZE_OP_PUSH) * qn->upper\r
- <= QUANTIFIER_EXPAND_LIMIT_SIZE)) {\r
+ (qn->upper == 1 ||\r
+ int_multiply_cmp(tlen + SIZE_OP_PUSH, qn->upper,\r
+ QUANTIFIER_EXPAND_LIMIT_SIZE) <= 0)) {\r
len = tlen * qn->lower;\r
len += (SIZE_OP_PUSH + tlen) * (qn->upper - qn->lower);\r
}\r
}\r
\r
static int\r
-compile_quantifier_node(QtfrNode* qn, regex_t* reg)\r
+compile_quantifier_node(QuantNode* qn, regex_t* reg, ScanEnv* env)\r
{\r
int i, r, mod_tlen;\r
int infinite = IS_REPEAT_INFINITE(qn->upper);\r
- int empty_info = qn->target_empty_info;\r
- int tlen = compile_length_tree(qn->target, reg);\r
+ enum QuantBodyEmpty empty_info = qn->body_empty_info;\r
+ int tlen = compile_length_tree(NODE_QUANT_BODY(qn), reg);\r
\r
if (tlen < 0) return tlen;\r
+ if (tlen == 0) return 0;\r
\r
- if (is_anychar_star_quantifier(qn)) {\r
- r = compile_tree_n_times(qn->target, qn->lower, reg);\r
- if (r) return r;\r
+ if (is_anychar_infinite_greedy(qn) &&\r
+ (qn->lower <= 1 ||\r
+ int_multiply_cmp(tlen, qn->lower, QUANTIFIER_EXPAND_LIMIT_SIZE) <= 0)) {\r
+ r = compile_tree_n_times(NODE_QUANT_BODY(qn), qn->lower, reg, env);\r
+ if (r != 0) return r;\r
if (IS_NOT_NULL(qn->next_head_exact)) {\r
- if (IS_MULTILINE(reg->options))\r
- r = add_opcode(reg, OP_ANYCHAR_ML_STAR_PEEK_NEXT);\r
+ if (IS_MULTILINE(CTYPE_OPTION(NODE_QUANT_BODY(qn), reg)))\r
+ r = add_opcode(reg, OP_ANYCHAR_ML_STAR_PEEK_NEXT);\r
else\r
- r = add_opcode(reg, OP_ANYCHAR_STAR_PEEK_NEXT);\r
- if (r) return r;\r
- return add_bytes(reg, NSTR(qn->next_head_exact)->s, 1);\r
+ r = add_opcode(reg, OP_ANYCHAR_STAR_PEEK_NEXT);\r
+ if (r != 0) return r;\r
+ return add_bytes(reg, STR_(qn->next_head_exact)->s, 1);\r
}\r
else {\r
- if (IS_MULTILINE(reg->options))\r
- return add_opcode(reg, OP_ANYCHAR_ML_STAR);\r
+ if (IS_MULTILINE(CTYPE_OPTION(NODE_QUANT_BODY(qn), reg)))\r
+ return add_opcode(reg, OP_ANYCHAR_ML_STAR);\r
else\r
- return add_opcode(reg, OP_ANYCHAR_STAR);\r
+ return add_opcode(reg, OP_ANYCHAR_STAR);\r
}\r
}\r
\r
- if (empty_info != 0)\r
- mod_tlen = tlen + (SIZE_OP_NULL_CHECK_START + SIZE_OP_NULL_CHECK_END);\r
- else\r
+ if (empty_info == QUANT_BODY_IS_NOT_EMPTY)\r
mod_tlen = tlen;\r
+ else\r
+ mod_tlen = tlen + (SIZE_OP_EMPTY_CHECK_START + SIZE_OP_EMPTY_CHECK_END);\r
\r
if (infinite &&\r
- (qn->lower <= 1 || tlen * qn->lower <= QUANTIFIER_EXPAND_LIMIT_SIZE)) {\r
+ (qn->lower <= 1 ||\r
+ int_multiply_cmp(tlen, qn->lower, QUANTIFIER_EXPAND_LIMIT_SIZE) <= 0)) {\r
if (qn->lower == 1 && tlen > QUANTIFIER_EXPAND_LIMIT_SIZE) {\r
if (qn->greedy) {\r
- if (IS_NOT_NULL(qn->head_exact))\r
- r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH_OR_JUMP_EXACT1);\r
- else if (IS_NOT_NULL(qn->next_head_exact))\r
- r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH_IF_PEEK_NEXT);\r
- else\r
- r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH);\r
+ if (IS_NOT_NULL(qn->head_exact))\r
+ r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH_OR_JUMP_EXACT1);\r
+ else if (IS_NOT_NULL(qn->next_head_exact))\r
+ r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH_IF_PEEK_NEXT);\r
+ else\r
+ r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH);\r
}\r
else {\r
- r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_JUMP);\r
+ r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_JUMP);\r
}\r
- if (r) return r;\r
+ if (r != 0) return r;\r
}\r
else {\r
- r = compile_tree_n_times(qn->target, qn->lower, reg);\r
- if (r) return r;\r
+ r = compile_tree_n_times(NODE_QUANT_BODY(qn), qn->lower, reg, env);\r
+ if (r != 0) return r;\r
}\r
\r
if (qn->greedy) {\r
if (IS_NOT_NULL(qn->head_exact)) {\r
- r = add_opcode_rel_addr(reg, OP_PUSH_OR_JUMP_EXACT1,\r
- mod_tlen + SIZE_OP_JUMP);\r
- if (r) return r;\r
- add_bytes(reg, NSTR(qn->head_exact)->s, 1);\r
- r = compile_tree_empty_check(qn->target, reg, empty_info);\r
- if (r) return r;\r
- r = add_opcode_rel_addr(reg, OP_JUMP,\r
- -(mod_tlen + (int )SIZE_OP_JUMP + (int )SIZE_OP_PUSH_OR_JUMP_EXACT1));\r
+ r = add_opcode_rel_addr(reg, OP_PUSH_OR_JUMP_EXACT1,\r
+ mod_tlen + SIZE_OP_JUMP);\r
+ if (r != 0) return r;\r
+ add_bytes(reg, STR_(qn->head_exact)->s, 1);\r
+ r = compile_tree_empty_check(NODE_QUANT_BODY(qn), reg, empty_info, env);\r
+ if (r != 0) return r;\r
+ r = add_opcode_rel_addr(reg, OP_JUMP,\r
+ -(mod_tlen + (int )SIZE_OP_JUMP + (int )SIZE_OP_PUSH_OR_JUMP_EXACT1));\r
}\r
else if (IS_NOT_NULL(qn->next_head_exact)) {\r
- r = add_opcode_rel_addr(reg, OP_PUSH_IF_PEEK_NEXT,\r
- mod_tlen + SIZE_OP_JUMP);\r
- if (r) return r;\r
- add_bytes(reg, NSTR(qn->next_head_exact)->s, 1);\r
- r = compile_tree_empty_check(qn->target, reg, empty_info);\r
- if (r) return r;\r
- r = add_opcode_rel_addr(reg, OP_JUMP,\r
- -(mod_tlen + (int )SIZE_OP_JUMP + (int )SIZE_OP_PUSH_IF_PEEK_NEXT));\r
+ r = add_opcode_rel_addr(reg, OP_PUSH_IF_PEEK_NEXT,\r
+ mod_tlen + SIZE_OP_JUMP);\r
+ if (r != 0) return r;\r
+ add_bytes(reg, STR_(qn->next_head_exact)->s, 1);\r
+ r = compile_tree_empty_check(NODE_QUANT_BODY(qn), reg, empty_info, env);\r
+ if (r != 0) return r;\r
+ r = add_opcode_rel_addr(reg, OP_JUMP,\r
+ -(mod_tlen + (int )SIZE_OP_JUMP + (int )SIZE_OP_PUSH_IF_PEEK_NEXT));\r
}\r
else {\r
- r = add_opcode_rel_addr(reg, OP_PUSH, mod_tlen + SIZE_OP_JUMP);\r
- if (r) return r;\r
- r = compile_tree_empty_check(qn->target, reg, empty_info);\r
- if (r) return r;\r
- r = add_opcode_rel_addr(reg, OP_JUMP,\r
- -(mod_tlen + (int )SIZE_OP_JUMP + (int )SIZE_OP_PUSH));\r
+ r = add_opcode_rel_addr(reg, OP_PUSH, mod_tlen + SIZE_OP_JUMP);\r
+ if (r != 0) return r;\r
+ r = compile_tree_empty_check(NODE_QUANT_BODY(qn), reg, empty_info, env);\r
+ if (r != 0) return r;\r
+ r = add_opcode_rel_addr(reg, OP_JUMP,\r
+ -(mod_tlen + (int )SIZE_OP_JUMP + (int )SIZE_OP_PUSH));\r
}\r
}\r
else {\r
r = add_opcode_rel_addr(reg, OP_JUMP, mod_tlen);\r
- if (r) return r;\r
- r = compile_tree_empty_check(qn->target, reg, empty_info);\r
- if (r) return r;\r
+ if (r != 0) return r;\r
+ r = compile_tree_empty_check(NODE_QUANT_BODY(qn), reg, empty_info, env);\r
+ if (r != 0) return r;\r
r = add_opcode_rel_addr(reg, OP_PUSH, -(mod_tlen + (int )SIZE_OP_PUSH));\r
}\r
}\r
else if (qn->upper == 0 && qn->is_refered != 0) { /* /(?<n>..){0}/ */\r
r = add_opcode_rel_addr(reg, OP_JUMP, tlen);\r
- if (r) return r;\r
- r = compile_tree(qn->target, reg);\r
+ if (r != 0) return r;\r
+ r = compile_tree(NODE_QUANT_BODY(qn), reg, env);\r
}\r
- else if (!infinite && qn->greedy &&\r
- (qn->upper == 1 || (tlen + SIZE_OP_PUSH) * qn->upper\r
- <= QUANTIFIER_EXPAND_LIMIT_SIZE)) {\r
+ else if (! infinite && qn->greedy &&\r
+ (qn->upper == 1 ||\r
+ int_multiply_cmp(tlen + SIZE_OP_PUSH, qn->upper,\r
+ QUANTIFIER_EXPAND_LIMIT_SIZE) <= 0)) {\r
int n = qn->upper - qn->lower;\r
\r
- r = compile_tree_n_times(qn->target, qn->lower, reg);\r
- if (r) return r;\r
+ r = compile_tree_n_times(NODE_QUANT_BODY(qn), qn->lower, reg, env);\r
+ if (r != 0) return r;\r
\r
for (i = 0; i < n; i++) {\r
r = add_opcode_rel_addr(reg, OP_PUSH,\r
- (n - i) * tlen + (n - i - 1) * SIZE_OP_PUSH);\r
- if (r) return r;\r
- r = compile_tree(qn->target, reg);\r
- if (r) return r;\r
+ (n - i) * tlen + (n - i - 1) * SIZE_OP_PUSH);\r
+ if (r != 0) return r;\r
+ r = compile_tree(NODE_QUANT_BODY(qn), reg, env);\r
+ if (r != 0) return r;\r
}\r
}\r
- else if (!qn->greedy && qn->upper == 1 && qn->lower == 0) { /* '??' */\r
+ else if (! qn->greedy && qn->upper == 1 && qn->lower == 0) { /* '??' */\r
r = add_opcode_rel_addr(reg, OP_PUSH, SIZE_OP_JUMP);\r
- if (r) return r;\r
+ if (r != 0) return r;\r
r = add_opcode_rel_addr(reg, OP_JUMP, tlen);\r
- if (r) return r;\r
- r = compile_tree(qn->target, reg);\r
+ if (r != 0) return r;\r
+ r = compile_tree(NODE_QUANT_BODY(qn), reg, env);\r
}\r
else {\r
- r = compile_range_repeat_node(qn, mod_tlen, empty_info, reg);\r
+ r = compile_range_repeat_node(qn, mod_tlen, empty_info, reg, env);\r
}\r
return r;\r
}\r
-#endif /* USE_COMBINATION_EXPLOSION_CHECK */\r
\r
static int\r
-compile_length_option_node(EncloseNode* node, regex_t* reg)\r
+compile_length_option_node(EnclosureNode* node, regex_t* reg)\r
{\r
int tlen;\r
OnigOptionType prev = reg->options;\r
\r
- reg->options = node->option;\r
- tlen = compile_length_tree(node->target, reg);\r
+ reg->options = node->o.options;\r
+ tlen = compile_length_tree(NODE_ENCLOSURE_BODY(node), reg);\r
reg->options = prev;\r
\r
- if (tlen < 0) return tlen;\r
-\r
- if (IS_DYNAMIC_OPTION(prev ^ node->option)) {\r
- return SIZE_OP_SET_OPTION_PUSH + SIZE_OP_SET_OPTION + SIZE_OP_FAIL\r
- + tlen + SIZE_OP_SET_OPTION;\r
- }\r
- else\r
- return tlen;\r
+ return tlen;\r
}\r
\r
static int\r
-compile_option_node(EncloseNode* node, regex_t* reg)\r
+compile_option_node(EnclosureNode* node, regex_t* reg, ScanEnv* env)\r
{\r
int r;\r
OnigOptionType prev = reg->options;\r
\r
- if (IS_DYNAMIC_OPTION(prev ^ node->option)) {\r
- r = add_opcode_option(reg, OP_SET_OPTION_PUSH, node->option);\r
- if (r) return r;\r
- r = add_opcode_option(reg, OP_SET_OPTION, prev);\r
- if (r) return r;\r
- r = add_opcode(reg, OP_FAIL);\r
- if (r) return r;\r
- }\r
-\r
- reg->options = node->option;\r
- r = compile_tree(node->target, reg);\r
+ reg->options = node->o.options;\r
+ r = compile_tree(NODE_ENCLOSURE_BODY(node), reg, env);\r
reg->options = prev;\r
\r
- if (IS_DYNAMIC_OPTION(prev ^ node->option)) {\r
- if (r) return r;\r
- r = add_opcode_option(reg, OP_SET_OPTION, prev);\r
- }\r
return r;\r
}\r
\r
static int\r
-compile_length_enclose_node(EncloseNode* node, regex_t* reg)\r
+compile_length_enclosure_node(EnclosureNode* node, regex_t* reg)\r
{\r
int len;\r
int tlen;\r
\r
- if (node->type == ENCLOSE_OPTION)\r
+ if (node->type == ENCLOSURE_OPTION)\r
return compile_length_option_node(node, reg);\r
\r
- if (node->target) {\r
- tlen = compile_length_tree(node->target, reg);\r
+ if (NODE_ENCLOSURE_BODY(node)) {\r
+ tlen = compile_length_tree(NODE_ENCLOSURE_BODY(node), reg);\r
if (tlen < 0) return tlen;\r
}\r
else\r
tlen = 0;\r
\r
switch (node->type) {\r
- case ENCLOSE_MEMORY:\r
-#ifdef USE_SUBEXP_CALL\r
- if (IS_ENCLOSE_CALLED(node)) {\r
+ case ENCLOSURE_MEMORY:\r
+#ifdef USE_CALL\r
+\r
+ if (node->m.regnum == 0 && NODE_IS_CALLED(node)) {\r
+ len = tlen + SIZE_OP_CALL + SIZE_OP_JUMP + SIZE_OP_RETURN;\r
+ return len;\r
+ }\r
+\r
+ if (NODE_IS_CALLED(node)) {\r
len = SIZE_OP_MEMORY_START_PUSH + tlen\r
- + SIZE_OP_CALL + SIZE_OP_JUMP + SIZE_OP_RETURN;\r
- if (BIT_STATUS_AT(reg->bt_mem_end, node->regnum))\r
- len += (IS_ENCLOSE_RECURSION(node)\r
- ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_PUSH);\r
+ + SIZE_OP_CALL + SIZE_OP_JUMP + SIZE_OP_RETURN;\r
+ if (MEM_STATUS_AT0(reg->bt_mem_end, node->m.regnum))\r
+ len += (NODE_IS_RECURSION(node)\r
+ ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_PUSH);\r
else\r
- len += (IS_ENCLOSE_RECURSION(node)\r
- ? SIZE_OP_MEMORY_END_REC : SIZE_OP_MEMORY_END);\r
+ len += (NODE_IS_RECURSION(node)\r
+ ? SIZE_OP_MEMORY_END_REC : SIZE_OP_MEMORY_END);\r
+ }\r
+ else if (NODE_IS_RECURSION(node)) {\r
+ len = SIZE_OP_MEMORY_START_PUSH;\r
+ len += tlen + (MEM_STATUS_AT0(reg->bt_mem_end, node->m.regnum)\r
+ ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_REC);\r
}\r
else\r
#endif\r
{\r
- if (BIT_STATUS_AT(reg->bt_mem_start, node->regnum))\r
- len = SIZE_OP_MEMORY_START_PUSH;\r
+ if (MEM_STATUS_AT0(reg->bt_mem_start, node->m.regnum))\r
+ len = SIZE_OP_MEMORY_START_PUSH;\r
else\r
- len = SIZE_OP_MEMORY_START;\r
+ len = SIZE_OP_MEMORY_START;\r
\r
- len += tlen + (BIT_STATUS_AT(reg->bt_mem_end, node->regnum)\r
- ? SIZE_OP_MEMORY_END_PUSH : SIZE_OP_MEMORY_END);\r
+ len += tlen + (MEM_STATUS_AT0(reg->bt_mem_end, node->m.regnum)\r
+ ? SIZE_OP_MEMORY_END_PUSH : SIZE_OP_MEMORY_END);\r
}\r
break;\r
\r
- case ENCLOSE_STOP_BACKTRACK:\r
- if (IS_ENCLOSE_STOP_BT_SIMPLE_REPEAT(node)) {\r
- if (node->target == NULL) {\r
- CHECK_NULL_RETURN_MEMERR(node->target);\r
- }\r
- QtfrNode* qn = NQTFR(node->target);\r
- tlen = compile_length_tree(qn->target, reg);\r
+ case ENCLOSURE_STOP_BACKTRACK:\r
+ if (NODE_IS_STOP_BT_SIMPLE_REPEAT(node)) {\r
+ QuantNode* qn = QUANT_(NODE_ENCLOSURE_BODY(node));\r
+ tlen = compile_length_tree(NODE_QUANT_BODY(qn), reg);\r
if (tlen < 0) return tlen;\r
\r
len = tlen * qn->lower\r
- + SIZE_OP_PUSH + tlen + SIZE_OP_POP + SIZE_OP_JUMP;\r
+ + SIZE_OP_PUSH + tlen + SIZE_OP_POP_OUT + SIZE_OP_JUMP;\r
}\r
else {\r
- len = SIZE_OP_PUSH_STOP_BT + tlen + SIZE_OP_POP_STOP_BT;\r
+ len = SIZE_OP_ATOMIC_START + tlen + SIZE_OP_ATOMIC_END;\r
}\r
break;\r
\r
- default:\r
- return ONIGERR_TYPE_BUG;\r
- break;\r
- }\r
+ case ENCLOSURE_IF_ELSE:\r
+ {\r
+ Node* cond = NODE_ENCLOSURE_BODY(node);\r
+ Node* Then = node->te.Then;\r
+ Node* Else = node->te.Else;\r
\r
- return len;\r
-}\r
+ len = compile_length_tree(cond, reg);\r
+ if (len < 0) return len;\r
+ len += SIZE_OP_PUSH;\r
+ len += SIZE_OP_ATOMIC_START + SIZE_OP_ATOMIC_END;\r
\r
-static int get_char_length_tree(Node* node, regex_t* reg, int* len);\r
+ if (IS_NOT_NULL(Then)) {\r
+ tlen = compile_length_tree(Then, reg);\r
+ if (tlen < 0) return tlen;\r
+ len += tlen;\r
+ }\r
+\r
+ if (IS_NOT_NULL(Else)) {\r
+ len += SIZE_OP_JUMP;\r
+ tlen = compile_length_tree(Else, reg);\r
+ if (tlen < 0) return tlen;\r
+ len += tlen;\r
+ }\r
+ }\r
+ break;\r
+\r
+ default:\r
+ return ONIGERR_TYPE_BUG;\r
+ break;\r
+ }\r
+\r
+ return len;\r
+}\r
+\r
+static int get_char_length_tree(Node* node, regex_t* reg, int* len);\r
\r
static int\r
-compile_enclose_node(EncloseNode* node, regex_t* reg)\r
+compile_enclosure_memory_node(EnclosureNode* node, regex_t* reg, ScanEnv* env)\r
{\r
- int r, len;\r
+ int r;\r
+ int len;\r
\r
- if (node->type == ENCLOSE_OPTION)\r
- return compile_option_node(node, reg);\r
+#ifdef USE_CALL\r
+ if (node->m.regnum == 0 && NODE_IS_CALLED(node)) {\r
+ r = add_opcode(reg, OP_CALL);\r
+ if (r != 0) return r;\r
+ node->m.called_addr = BB_GET_OFFSET_POS(reg) + SIZE_ABSADDR + SIZE_OP_JUMP;\r
+ NODE_STATUS_ADD(node, ADDR_FIXED);\r
+ r = add_abs_addr(reg, (int )node->m.called_addr);\r
+ if (r != 0) return r;\r
+ len = compile_length_tree(NODE_ENCLOSURE_BODY(node), reg);\r
+ len += SIZE_OP_RETURN;\r
+ r = add_opcode_rel_addr(reg, OP_JUMP, len);\r
+ if (r != 0) return r;\r
+\r
+ r = compile_tree(NODE_ENCLOSURE_BODY(node), reg, env);\r
+ if (r != 0) return r;\r
+ r = add_opcode(reg, OP_RETURN);\r
+ return r;\r
+ }\r
\r
- switch (node->type) {\r
- case ENCLOSE_MEMORY:\r
-#ifdef USE_SUBEXP_CALL\r
- if (IS_ENCLOSE_CALLED(node)) {\r
- r = add_opcode(reg, OP_CALL);\r
- if (r) return r;\r
- node->call_addr = BBUF_GET_OFFSET_POS(reg) + SIZE_ABSADDR + SIZE_OP_JUMP;\r
- node->state |= NST_ADDR_FIXED;\r
- r = add_abs_addr(reg, (int )node->call_addr);\r
- if (r) return r;\r
- len = compile_length_tree(node->target, reg);\r
- len += (SIZE_OP_MEMORY_START_PUSH + SIZE_OP_RETURN);\r
- if (BIT_STATUS_AT(reg->bt_mem_end, node->regnum))\r
- len += (IS_ENCLOSE_RECURSION(node)\r
- ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_PUSH);\r
- else\r
- len += (IS_ENCLOSE_RECURSION(node)\r
- ? SIZE_OP_MEMORY_END_REC : SIZE_OP_MEMORY_END);\r
+ if (NODE_IS_CALLED(node)) {\r
+ r = add_opcode(reg, OP_CALL);\r
+ if (r != 0) return r;\r
+ node->m.called_addr = BB_GET_OFFSET_POS(reg) + SIZE_ABSADDR + SIZE_OP_JUMP;\r
+ NODE_STATUS_ADD(node, ADDR_FIXED);\r
+ r = add_abs_addr(reg, (int )node->m.called_addr);\r
+ if (r != 0) return r;\r
+ len = compile_length_tree(NODE_ENCLOSURE_BODY(node), reg);\r
+ len += (SIZE_OP_MEMORY_START_PUSH + SIZE_OP_RETURN);\r
+ if (MEM_STATUS_AT0(reg->bt_mem_end, node->m.regnum))\r
+ len += (NODE_IS_RECURSION(node)\r
+ ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_PUSH);\r
+ else\r
+ len += (NODE_IS_RECURSION(node)\r
+ ? SIZE_OP_MEMORY_END_REC : SIZE_OP_MEMORY_END);\r
\r
- r = add_opcode_rel_addr(reg, OP_JUMP, len);\r
- if (r) return r;\r
- }\r
+ r = add_opcode_rel_addr(reg, OP_JUMP, len);\r
+ if (r != 0) return r;\r
+ }\r
#endif\r
- if (BIT_STATUS_AT(reg->bt_mem_start, node->regnum))\r
- r = add_opcode(reg, OP_MEMORY_START_PUSH);\r
- else\r
- r = add_opcode(reg, OP_MEMORY_START);\r
- if (r) return r;\r
- r = add_mem_num(reg, node->regnum);\r
- if (r) return r;\r
- r = compile_tree(node->target, reg);\r
- if (r) return r;\r
-#ifdef USE_SUBEXP_CALL\r
- if (IS_ENCLOSE_CALLED(node)) {\r
- if (BIT_STATUS_AT(reg->bt_mem_end, node->regnum))\r
- r = add_opcode(reg, (IS_ENCLOSE_RECURSION(node)\r
- ? OP_MEMORY_END_PUSH_REC : OP_MEMORY_END_PUSH));\r
- else\r
- r = add_opcode(reg, (IS_ENCLOSE_RECURSION(node)\r
- ? OP_MEMORY_END_REC : OP_MEMORY_END));\r
\r
- if (r) return r;\r
- r = add_mem_num(reg, node->regnum);\r
- if (r) return r;\r
- r = add_opcode(reg, OP_RETURN);\r
- }\r
- else\r
+ if (MEM_STATUS_AT0(reg->bt_mem_start, node->m.regnum))\r
+ r = add_opcode(reg, OP_MEMORY_START_PUSH);\r
+ else\r
+ r = add_opcode(reg, OP_MEMORY_START);\r
+ if (r != 0) return r;\r
+ r = add_mem_num(reg, node->m.regnum);\r
+ if (r != 0) return r;\r
+ r = compile_tree(NODE_ENCLOSURE_BODY(node), reg, env);\r
+ if (r != 0) return r;\r
+\r
+#ifdef USE_CALL\r
+ if (MEM_STATUS_AT0(reg->bt_mem_end, node->m.regnum))\r
+ r = add_opcode(reg, (NODE_IS_RECURSION(node)\r
+ ? OP_MEMORY_END_PUSH_REC : OP_MEMORY_END_PUSH));\r
+ else\r
+ r = add_opcode(reg, (NODE_IS_RECURSION(node)\r
+ ? OP_MEMORY_END_REC : OP_MEMORY_END));\r
+ if (r != 0) return r;\r
+ r = add_mem_num(reg, node->m.regnum);\r
+ if (NODE_IS_CALLED(node)) {\r
+ if (r != 0) return r;\r
+ r = add_opcode(reg, OP_RETURN);\r
+ }\r
+#else\r
+ if (MEM_STATUS_AT0(reg->bt_mem_end, node->m.regnum))\r
+ r = add_opcode(reg, OP_MEMORY_END_PUSH);\r
+ else\r
+ r = add_opcode(reg, OP_MEMORY_END);\r
+ if (r != 0) return r;\r
+ r = add_mem_num(reg, node->m.regnum);\r
#endif\r
- {\r
- if (BIT_STATUS_AT(reg->bt_mem_end, node->regnum))\r
- r = add_opcode(reg, OP_MEMORY_END_PUSH);\r
- else\r
- r = add_opcode(reg, OP_MEMORY_END);\r
- if (r) return r;\r
- r = add_mem_num(reg, node->regnum);\r
- }\r
+\r
+ return r;\r
+}\r
+\r
+static int\r
+compile_enclosure_node(EnclosureNode* node, regex_t* reg, ScanEnv* env)\r
+{\r
+ int r, len;\r
+\r
+ switch (node->type) {\r
+ case ENCLOSURE_MEMORY:\r
+ r = compile_enclosure_memory_node(node, reg, env);\r
break;\r
\r
- case ENCLOSE_STOP_BACKTRACK:\r
- if (IS_ENCLOSE_STOP_BT_SIMPLE_REPEAT(node)) {\r
- QtfrNode* qn = NQTFR(node->target);\r
- r = compile_tree_n_times(qn->target, qn->lower, reg);\r
- if (r) return r;\r
+ case ENCLOSURE_OPTION:\r
+ r = compile_option_node(node, reg, env);\r
+ break;\r
+\r
+ case ENCLOSURE_STOP_BACKTRACK:\r
+ if (NODE_IS_STOP_BT_SIMPLE_REPEAT(node)) {\r
+ QuantNode* qn = QUANT_(NODE_ENCLOSURE_BODY(node));\r
+ r = compile_tree_n_times(NODE_QUANT_BODY(qn), qn->lower, reg, env);\r
+ if (r != 0) return r;\r
\r
- len = compile_length_tree(qn->target, reg);\r
+ len = compile_length_tree(NODE_QUANT_BODY(qn), reg);\r
if (len < 0) return len;\r
\r
- r = add_opcode_rel_addr(reg, OP_PUSH, len + SIZE_OP_POP + SIZE_OP_JUMP);\r
- if (r) return r;\r
- r = compile_tree(qn->target, reg);\r
- if (r) return r;\r
- r = add_opcode(reg, OP_POP);\r
- if (r) return r;\r
+ r = add_opcode_rel_addr(reg, OP_PUSH, len + SIZE_OP_POP_OUT + SIZE_OP_JUMP);\r
+ if (r != 0) return r;\r
+ r = compile_tree(NODE_QUANT_BODY(qn), reg, env);\r
+ if (r != 0) return r;\r
+ r = add_opcode(reg, OP_POP_OUT);\r
+ if (r != 0) return r;\r
r = add_opcode_rel_addr(reg, OP_JUMP,\r
- -((int )SIZE_OP_PUSH + len + (int )SIZE_OP_POP + (int )SIZE_OP_JUMP));\r
+ -((int )SIZE_OP_PUSH + len + (int )SIZE_OP_POP_OUT + (int )SIZE_OP_JUMP));\r
}\r
else {\r
- r = add_opcode(reg, OP_PUSH_STOP_BT);\r
- if (r) return r;\r
- r = compile_tree(node->target, reg);\r
- if (r) return r;\r
- r = add_opcode(reg, OP_POP_STOP_BT);\r
+ r = add_opcode(reg, OP_ATOMIC_START);\r
+ if (r != 0) return r;\r
+ r = compile_tree(NODE_ENCLOSURE_BODY(node), reg, env);\r
+ if (r != 0) return r;\r
+ r = add_opcode(reg, OP_ATOMIC_END);\r
+ }\r
+ break;\r
+\r
+ case ENCLOSURE_IF_ELSE:\r
+ {\r
+ int cond_len, then_len, jump_len;\r
+ Node* cond = NODE_ENCLOSURE_BODY(node);\r
+ Node* Then = node->te.Then;\r
+ Node* Else = node->te.Else;\r
+\r
+ r = add_opcode(reg, OP_ATOMIC_START);\r
+ if (r != 0) return r;\r
+\r
+ cond_len = compile_length_tree(cond, reg);\r
+ if (cond_len < 0) return cond_len;\r
+ if (IS_NOT_NULL(Then)) {\r
+ then_len = compile_length_tree(Then, reg);\r
+ if (then_len < 0) return then_len;\r
+ }\r
+ else\r
+ then_len = 0;\r
+\r
+ jump_len = cond_len + then_len + SIZE_OP_ATOMIC_END;\r
+ if (IS_NOT_NULL(Else)) jump_len += SIZE_OP_JUMP;\r
+\r
+ r = add_opcode_rel_addr(reg, OP_PUSH, jump_len);\r
+ if (r != 0) return r;\r
+ r = compile_tree(cond, reg, env);\r
+ if (r != 0) return r;\r
+ r = add_opcode(reg, OP_ATOMIC_END);\r
+ if (r != 0) return r;\r
+\r
+ if (IS_NOT_NULL(Then)) {\r
+ r = compile_tree(Then, reg, env);\r
+ if (r != 0) return r;\r
+ }\r
+\r
+ if (IS_NOT_NULL(Else)) {\r
+ int else_len = compile_length_tree(Else, reg);\r
+ r = add_opcode_rel_addr(reg, OP_JUMP, else_len);\r
+ if (r != 0) return r;\r
+ r = compile_tree(Else, reg, env);\r
+ }\r
}\r
break;\r
\r
int len;\r
int tlen = 0;\r
\r
- if (node->target) {\r
- tlen = compile_length_tree(node->target, reg);\r
+ if (IS_NOT_NULL(NODE_ANCHOR_BODY(node))) {\r
+ tlen = compile_length_tree(NODE_ANCHOR_BODY(node), reg);\r
if (tlen < 0) return tlen;\r
}\r
\r
switch (node->type) {\r
case ANCHOR_PREC_READ:\r
- len = SIZE_OP_PUSH_POS + tlen + SIZE_OP_POP_POS;\r
+ len = SIZE_OP_PREC_READ_START + tlen + SIZE_OP_PREC_READ_END;\r
break;\r
case ANCHOR_PREC_READ_NOT:\r
- len = SIZE_OP_PUSH_POS_NOT + tlen + SIZE_OP_FAIL_POS;\r
+ len = SIZE_OP_PREC_READ_NOT_START + tlen + SIZE_OP_PREC_READ_NOT_END;\r
break;\r
case ANCHOR_LOOK_BEHIND:\r
len = SIZE_OP_LOOK_BEHIND + tlen;\r
break;\r
case ANCHOR_LOOK_BEHIND_NOT:\r
- len = SIZE_OP_PUSH_LOOK_BEHIND_NOT + tlen + SIZE_OP_FAIL_LOOK_BEHIND_NOT;\r
+ len = SIZE_OP_LOOK_BEHIND_NOT_START + tlen + SIZE_OP_LOOK_BEHIND_NOT_END;\r
+ break;\r
+\r
+ case ANCHOR_WORD_BOUNDARY:\r
+ case ANCHOR_NO_WORD_BOUNDARY:\r
+#ifdef USE_WORD_BEGIN_END\r
+ case ANCHOR_WORD_BEGIN:\r
+ case ANCHOR_WORD_END:\r
+#endif\r
+ len = SIZE_OP_WORD_BOUNDARY;\r
+ break;\r
+\r
+ case ANCHOR_EXTENDED_GRAPHEME_CLUSTER_BOUNDARY:\r
+ case ANCHOR_NO_EXTENDED_GRAPHEME_CLUSTER_BOUNDARY:\r
+ len = SIZE_OPCODE;\r
break;\r
\r
default:\r
}\r
\r
static int\r
-compile_anchor_node(AnchorNode* node, regex_t* reg)\r
+compile_anchor_node(AnchorNode* node, regex_t* reg, ScanEnv* env)\r
{\r
int r, len;\r
+ enum OpCode op;\r
\r
switch (node->type) {\r
case ANCHOR_BEGIN_BUF: r = add_opcode(reg, OP_BEGIN_BUF); break;\r
case ANCHOR_SEMI_END_BUF: r = add_opcode(reg, OP_SEMI_END_BUF); break;\r
case ANCHOR_BEGIN_POSITION: r = add_opcode(reg, OP_BEGIN_POSITION); break;\r
\r
- case ANCHOR_WORD_BOUND: r = add_opcode(reg, OP_WORD_BOUND); break;\r
- case ANCHOR_NOT_WORD_BOUND: r = add_opcode(reg, OP_NOT_WORD_BOUND); break;\r
+ case ANCHOR_WORD_BOUNDARY:\r
+ op = OP_WORD_BOUNDARY;\r
+ word:\r
+ r = add_opcode(reg, op);\r
+ if (r != 0) return r;\r
+ r = add_mode(reg, (ModeType )node->ascii_mode);\r
+ break;\r
+\r
+ case ANCHOR_NO_WORD_BOUNDARY:\r
+ op = OP_NO_WORD_BOUNDARY; goto word;\r
+ break;\r
#ifdef USE_WORD_BEGIN_END\r
- case ANCHOR_WORD_BEGIN: r = add_opcode(reg, OP_WORD_BEGIN); break;\r
- case ANCHOR_WORD_END: r = add_opcode(reg, OP_WORD_END); break;\r
+ case ANCHOR_WORD_BEGIN:\r
+ op = OP_WORD_BEGIN; goto word;\r
+ break;\r
+ case ANCHOR_WORD_END:\r
+ op = OP_WORD_END; goto word;\r
+ break;\r
#endif\r
\r
+ case ANCHOR_EXTENDED_GRAPHEME_CLUSTER_BOUNDARY:\r
+ r = add_opcode(reg, OP_EXTENDED_GRAPHEME_CLUSTER_BOUNDARY);\r
+ break;\r
+\r
+ case ANCHOR_NO_EXTENDED_GRAPHEME_CLUSTER_BOUNDARY:\r
+ r = add_opcode(reg, OP_NO_EXTENDED_GRAPHEME_CLUSTER_BOUNDARY);\r
+ break;\r
+\r
case ANCHOR_PREC_READ:\r
- r = add_opcode(reg, OP_PUSH_POS);\r
- if (r) return r;\r
- r = compile_tree(node->target, reg);\r
- if (r) return r;\r
- r = add_opcode(reg, OP_POP_POS);\r
+ r = add_opcode(reg, OP_PREC_READ_START);\r
+ if (r != 0) return r;\r
+ r = compile_tree(NODE_ANCHOR_BODY(node), reg, env);\r
+ if (r != 0) return r;\r
+ r = add_opcode(reg, OP_PREC_READ_END);\r
break;\r
\r
case ANCHOR_PREC_READ_NOT:\r
- len = compile_length_tree(node->target, reg);\r
+ len = compile_length_tree(NODE_ANCHOR_BODY(node), reg);\r
if (len < 0) return len;\r
- r = add_opcode_rel_addr(reg, OP_PUSH_POS_NOT, len + SIZE_OP_FAIL_POS);\r
- if (r) return r;\r
- r = compile_tree(node->target, reg);\r
- if (r) return r;\r
- r = add_opcode(reg, OP_FAIL_POS);\r
+ r = add_opcode_rel_addr(reg, OP_PREC_READ_NOT_START, len + SIZE_OP_PREC_READ_NOT_END);\r
+ if (r != 0) return r;\r
+ r = compile_tree(NODE_ANCHOR_BODY(node), reg, env);\r
+ if (r != 0) return r;\r
+ r = add_opcode(reg, OP_PREC_READ_NOT_END);\r
break;\r
\r
case ANCHOR_LOOK_BEHIND:\r
{\r
int n;\r
r = add_opcode(reg, OP_LOOK_BEHIND);\r
- if (r) return r;\r
+ if (r != 0) return r;\r
if (node->char_len < 0) {\r
- r = get_char_length_tree(node->target, reg, &n);\r
- if (r) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN;\r
+ r = get_char_length_tree(NODE_ANCHOR_BODY(node), reg, &n);\r
+ if (r != 0) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN;\r
}\r
else\r
- n = node->char_len;\r
+ n = node->char_len;\r
+\r
r = add_length(reg, n);\r
- if (r) return r;\r
- r = compile_tree(node->target, reg);\r
+ if (r != 0) return r;\r
+ r = compile_tree(NODE_ANCHOR_BODY(node), reg, env);\r
}\r
break;\r
\r
case ANCHOR_LOOK_BEHIND_NOT:\r
{\r
int n;\r
- len = compile_length_tree(node->target, reg);\r
- r = add_opcode_rel_addr(reg, OP_PUSH_LOOK_BEHIND_NOT,\r
- len + SIZE_OP_FAIL_LOOK_BEHIND_NOT);\r
- if (r) return r;\r
+\r
+ len = compile_length_tree(NODE_ANCHOR_BODY(node), reg);\r
+ r = add_opcode_rel_addr(reg, OP_LOOK_BEHIND_NOT_START,\r
+ len + SIZE_OP_LOOK_BEHIND_NOT_END);\r
+ if (r != 0) return r;\r
if (node->char_len < 0) {\r
- r = get_char_length_tree(node->target, reg, &n);\r
- if (r) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN;\r
+ r = get_char_length_tree(NODE_ANCHOR_BODY(node), reg, &n);\r
+ if (r != 0) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN;\r
}\r
else\r
- n = node->char_len;\r
+ n = node->char_len;\r
r = add_length(reg, n);\r
- if (r) return r;\r
- r = compile_tree(node->target, reg);\r
- if (r) return r;\r
- r = add_opcode(reg, OP_FAIL_LOOK_BEHIND_NOT);\r
+ if (r != 0) return r;\r
+ r = compile_tree(NODE_ANCHOR_BODY(node), reg, env);\r
+ if (r != 0) return r;\r
+ r = add_opcode(reg, OP_LOOK_BEHIND_NOT_END);\r
}\r
break;\r
\r
return r;\r
}\r
\r
+static int\r
+compile_gimmick_node(GimmickNode* node, regex_t* reg)\r
+{\r
+ int r;\r
+\r
+ switch (node->type) {\r
+ case GIMMICK_FAIL:\r
+ r = add_opcode(reg, OP_FAIL);\r
+ break;\r
+\r
+ case GIMMICK_KEEP:\r
+ r = add_opcode(reg, OP_PUSH_SAVE_VAL);\r
+ if (r != 0) return r;\r
+ r = add_save_type(reg, SAVE_KEEP);\r
+ if (r != 0) return r;\r
+ r = add_mem_num(reg, node->id);\r
+ break;\r
+\r
+ case GIMMICK_SAVE:\r
+ r = add_opcode(reg, OP_PUSH_SAVE_VAL);\r
+ if (r != 0) return r;\r
+ r = add_save_type(reg, node->detail_type);\r
+ if (r != 0) return r;\r
+ r = add_mem_num(reg, node->id);\r
+ break;\r
+\r
+ case GIMMICK_UPDATE_VAR:\r
+ r = add_opcode(reg, OP_UPDATE_VAR);\r
+ if (r != 0) return r;\r
+ r = add_update_var_type(reg, node->detail_type);\r
+ if (r != 0) return r;\r
+ r = add_mem_num(reg, node->id);\r
+ break;\r
+\r
+#ifdef USE_CALLOUT\r
+ case GIMMICK_CALLOUT:\r
+ switch (node->detail_type) {\r
+ case ONIG_CALLOUT_OF_CONTENTS:\r
+ case ONIG_CALLOUT_OF_NAME:\r
+ {\r
+ r = add_opcode(reg, (node->detail_type == ONIG_CALLOUT_OF_CONTENTS) ?\r
+ OP_CALLOUT_CONTENTS : OP_CALLOUT_NAME);\r
+ if (r != 0) return r;\r
+ if (node->detail_type == ONIG_CALLOUT_OF_NAME) {\r
+ r = add_mem_num(reg, node->id);\r
+ if (r != 0) return r;\r
+ }\r
+ r = add_mem_num(reg, node->num);\r
+ if (r != 0) return r;\r
+ }\r
+ break;\r
+\r
+ default:\r
+ r = ONIGERR_TYPE_BUG;\r
+ break;\r
+ }\r
+#endif\r
+ }\r
+\r
+ return r;\r
+}\r
+\r
+static int\r
+compile_length_gimmick_node(GimmickNode* node, regex_t* reg)\r
+{\r
+ int len;\r
+\r
+ switch (node->type) {\r
+ case GIMMICK_FAIL:\r
+ len = SIZE_OP_FAIL;\r
+ break;\r
+\r
+ case GIMMICK_KEEP:\r
+ case GIMMICK_SAVE:\r
+ len = SIZE_OP_PUSH_SAVE_VAL;\r
+ break;\r
+\r
+ case GIMMICK_UPDATE_VAR:\r
+ len = SIZE_OP_UPDATE_VAR;\r
+ break;\r
+\r
+#ifdef USE_CALLOUT\r
+ case GIMMICK_CALLOUT:\r
+ switch (node->detail_type) {\r
+ case ONIG_CALLOUT_OF_CONTENTS:\r
+ len = SIZE_OP_CALLOUT_CONTENTS;\r
+ break;\r
+ case ONIG_CALLOUT_OF_NAME:\r
+ len = SIZE_OP_CALLOUT_NAME;\r
+ break;\r
+\r
+ default:\r
+ len = ONIGERR_TYPE_BUG;\r
+ break;\r
+ }\r
+ break;\r
+#endif\r
+ }\r
+\r
+ return len;\r
+}\r
+\r
static int\r
compile_length_tree(Node* node, regex_t* reg)\r
{\r
- int len, type, r;\r
+ int len, r;\r
\r
- type = NTYPE(node);\r
- switch (type) {\r
- case NT_LIST:\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
len = 0;\r
do {\r
- r = compile_length_tree(NCAR(node), reg);\r
+ r = compile_length_tree(NODE_CAR(node), reg);\r
if (r < 0) return r;\r
len += r;\r
- } while (IS_NOT_NULL(node = NCDR(node)));\r
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));\r
r = len;\r
break;\r
\r
- case NT_ALT:\r
+ case NODE_ALT:\r
{\r
int n;\r
\r
n = r = 0;\r
do {\r
- r += compile_length_tree(NCAR(node), reg);\r
- n++;\r
- } while (IS_NOT_NULL(node = NCDR(node)));\r
+ r += compile_length_tree(NODE_CAR(node), reg);\r
+ n++;\r
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));\r
r += (SIZE_OP_PUSH + SIZE_OP_JUMP) * (n - 1);\r
}\r
break;\r
\r
- case NT_STR:\r
- if (NSTRING_IS_RAW(node))\r
- r = compile_length_string_raw_node(NSTR(node), reg);\r
+ case NODE_STRING:\r
+ if (NODE_STRING_IS_RAW(node))\r
+ r = compile_length_string_raw_node(STR_(node), reg);\r
else\r
r = compile_length_string_node(node, reg);\r
break;\r
\r
- case NT_CCLASS:\r
- r = compile_length_cclass_node(NCCLASS(node), reg);\r
+ case NODE_CCLASS:\r
+ r = compile_length_cclass_node(CCLASS_(node), reg);\r
break;\r
\r
- case NT_CTYPE:\r
- case NT_CANY:\r
+ case NODE_CTYPE:\r
r = SIZE_OPCODE;\r
break;\r
\r
- case NT_BREF:\r
+ case NODE_BACKREF:\r
{\r
- BRefNode* br = NBREF(node);\r
+ BackRefNode* br = BACKREF_(node);\r
\r
+ if (NODE_IS_CHECKER(node)) {\r
#ifdef USE_BACKREF_WITH_LEVEL\r
- if (IS_BACKREF_NEST_LEVEL(br)) {\r
- r = SIZE_OPCODE + SIZE_OPTION + SIZE_LENGTH +\r
- SIZE_LENGTH + (SIZE_MEMNUM * br->back_num);\r
- }\r
- else\r
+ if (NODE_IS_NEST_LEVEL(node)) {\r
+ r = SIZE_OPCODE + SIZE_LENGTH + SIZE_LENGTH + (SIZE_MEMNUM * br->back_num);\r
+ }\r
+ else\r
#endif\r
- if (br->back_num == 1) {\r
- r = ((!IS_IGNORECASE(reg->options) && br->back_static[0] <= 2)\r
- ? SIZE_OPCODE : (SIZE_OPCODE + SIZE_MEMNUM));\r
+ r = SIZE_OPCODE + SIZE_LENGTH + (SIZE_MEMNUM * br->back_num);\r
}\r
else {\r
- r = SIZE_OPCODE + SIZE_LENGTH + (SIZE_MEMNUM * br->back_num);\r
+#ifdef USE_BACKREF_WITH_LEVEL\r
+ if (NODE_IS_NEST_LEVEL(node)) {\r
+ r = SIZE_OPCODE + SIZE_OPTION + SIZE_LENGTH +\r
+ SIZE_LENGTH + (SIZE_MEMNUM * br->back_num);\r
+ }\r
+ else\r
+#endif\r
+ if (br->back_num == 1) {\r
+ r = ((!IS_IGNORECASE(reg->options) && br->back_static[0] <= 2)\r
+ ? SIZE_OPCODE : (SIZE_OPCODE + SIZE_MEMNUM));\r
+ }\r
+ else {\r
+ r = SIZE_OPCODE + SIZE_LENGTH + (SIZE_MEMNUM * br->back_num);\r
+ }\r
}\r
}\r
break;\r
\r
-#ifdef USE_SUBEXP_CALL\r
- case NT_CALL:\r
+#ifdef USE_CALL\r
+ case NODE_CALL:\r
r = SIZE_OP_CALL;\r
break;\r
#endif\r
\r
- case NT_QTFR:\r
- r = compile_length_quantifier_node(NQTFR(node), reg);\r
+ case NODE_QUANT:\r
+ r = compile_length_quantifier_node(QUANT_(node), reg);\r
break;\r
\r
- case NT_ENCLOSE:\r
- r = compile_length_enclose_node(NENCLOSE(node), reg);\r
+ case NODE_ENCLOSURE:\r
+ r = compile_length_enclosure_node(ENCLOSURE_(node), reg);\r
break;\r
\r
- case NT_ANCHOR:\r
- r = compile_length_anchor_node(NANCHOR(node), reg);\r
+ case NODE_ANCHOR:\r
+ r = compile_length_anchor_node(ANCHOR_(node), reg);\r
+ break;\r
+\r
+ case NODE_GIMMICK:\r
+ r = compile_length_gimmick_node(GIMMICK_(node), reg);\r
break;\r
\r
default:\r
}\r
\r
static int\r
-compile_tree(Node* node, regex_t* reg)\r
+compile_tree(Node* node, regex_t* reg, ScanEnv* env)\r
{\r
- int n, type, len, pos, r = 0;\r
+ int n, len, pos, r = 0;\r
\r
- type = NTYPE(node);\r
- switch (type) {\r
- case NT_LIST:\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
do {\r
- r = compile_tree(NCAR(node), reg);\r
- } while (r == 0 && IS_NOT_NULL(node = NCDR(node)));\r
+ r = compile_tree(NODE_CAR(node), reg, env);\r
+ } while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));\r
break;\r
\r
- case NT_ALT:\r
+ case NODE_ALT:\r
{\r
Node* x = node;\r
len = 0;\r
do {\r
- len += compile_length_tree(NCAR(x), reg);\r
- if (NCDR(x) != NULL) {\r
- len += SIZE_OP_PUSH + SIZE_OP_JUMP;\r
- }\r
- } while (IS_NOT_NULL(x = NCDR(x)));\r
+ len += compile_length_tree(NODE_CAR(x), reg);\r
+ if (IS_NOT_NULL(NODE_CDR(x))) {\r
+ len += SIZE_OP_PUSH + SIZE_OP_JUMP;\r
+ }\r
+ } while (IS_NOT_NULL(x = NODE_CDR(x)));\r
pos = reg->used + len; /* goal position */\r
\r
do {\r
- len = compile_length_tree(NCAR(node), reg);\r
- if (IS_NOT_NULL(NCDR(node))) {\r
- r = add_opcode_rel_addr(reg, OP_PUSH, len + SIZE_OP_JUMP);\r
- if (r) break;\r
- }\r
- r = compile_tree(NCAR(node), reg);\r
- if (r) break;\r
- if (IS_NOT_NULL(NCDR(node))) {\r
- len = pos - (reg->used + SIZE_OP_JUMP);\r
- r = add_opcode_rel_addr(reg, OP_JUMP, len);\r
- if (r) break;\r
- }\r
- } while (IS_NOT_NULL(node = NCDR(node)));\r
- }\r
- break;\r
-\r
- case NT_STR:\r
- if (NSTRING_IS_RAW(node))\r
- r = compile_string_raw_node(NSTR(node), reg);\r
+ len = compile_length_tree(NODE_CAR(node), reg);\r
+ if (IS_NOT_NULL(NODE_CDR(node))) {\r
+ enum OpCode push = NODE_IS_SUPER(node) ? OP_PUSH_SUPER : OP_PUSH;\r
+ r = add_opcode_rel_addr(reg, push, len + SIZE_OP_JUMP);\r
+ if (r != 0) break;\r
+ }\r
+ r = compile_tree(NODE_CAR(node), reg, env);\r
+ if (r != 0) break;\r
+ if (IS_NOT_NULL(NODE_CDR(node))) {\r
+ len = pos - (reg->used + SIZE_OP_JUMP);\r
+ r = add_opcode_rel_addr(reg, OP_JUMP, len);\r
+ if (r != 0) break;\r
+ }\r
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));\r
+ }\r
+ break;\r
+\r
+ case NODE_STRING:\r
+ if (NODE_STRING_IS_RAW(node))\r
+ r = compile_string_raw_node(STR_(node), reg);\r
else\r
r = compile_string_node(node, reg);\r
break;\r
\r
- case NT_CCLASS:\r
- r = compile_cclass_node(NCCLASS(node), reg);\r
+ case NODE_CCLASS:\r
+ r = compile_cclass_node(CCLASS_(node), reg);\r
break;\r
\r
- case NT_CTYPE:\r
+ case NODE_CTYPE:\r
{\r
int op;\r
\r
- switch (NCTYPE(node)->ctype) {\r
+ switch (CTYPE_(node)->ctype) {\r
+ case CTYPE_ANYCHAR:\r
+ if (IS_MULTILINE(CTYPE_OPTION(node, reg)))\r
+ r = add_opcode(reg, OP_ANYCHAR_ML);\r
+ else\r
+ r = add_opcode(reg, OP_ANYCHAR);\r
+ break;\r
+\r
case ONIGENC_CTYPE_WORD:\r
- if (NCTYPE(node)->not != 0) op = OP_NOT_WORD;\r
- else op = OP_WORD;\r
- break;\r
+ if (CTYPE_(node)->ascii_mode == 0) {\r
+ op = CTYPE_(node)->not != 0 ? OP_NO_WORD : OP_WORD;\r
+ }\r
+ else {\r
+ op = CTYPE_(node)->not != 0 ? OP_NO_WORD_ASCII : OP_WORD_ASCII;\r
+ }\r
+ r = add_opcode(reg, op);\r
+ break;\r
+\r
default:\r
- return ONIGERR_TYPE_BUG;\r
- break;\r
+ return ONIGERR_TYPE_BUG;\r
+ break;\r
}\r
- r = add_opcode(reg, op);\r
}\r
break;\r
\r
- case NT_CANY:\r
- if (IS_MULTILINE(reg->options))\r
- r = add_opcode(reg, OP_ANYCHAR_ML);\r
- else\r
- r = add_opcode(reg, OP_ANYCHAR);\r
- break;\r
-\r
- case NT_BREF:\r
+ case NODE_BACKREF:\r
{\r
- BRefNode* br = NBREF(node);\r
+ BackRefNode* br = BACKREF_(node);\r
\r
+ if (NODE_IS_CHECKER(node)) {\r
#ifdef USE_BACKREF_WITH_LEVEL\r
- if (IS_BACKREF_NEST_LEVEL(br)) {\r
- r = add_opcode(reg, OP_BACKREF_WITH_LEVEL);\r
- if (r) return r;\r
- r = add_option(reg, (reg->options & ONIG_OPTION_IGNORECASE));\r
- if (r) return r;\r
- r = add_length(reg, br->nest_level);\r
- if (r) return r;\r
-\r
- goto add_bacref_mems;\r
- }\r
- else\r
+ if (NODE_IS_NEST_LEVEL(node)) {\r
+ r = add_opcode(reg, OP_BACKREF_CHECK_WITH_LEVEL);\r
+ if (r != 0) return r;\r
+ r = add_length(reg, br->nest_level);\r
+ if (r != 0) return r;\r
+ }\r
+ else\r
#endif\r
- if (br->back_num == 1) {\r
- n = br->back_static[0];\r
- if (IS_IGNORECASE(reg->options)) {\r
- r = add_opcode(reg, OP_BACKREFN_IC);\r
- if (r) return r;\r
- r = add_mem_num(reg, n);\r
- }\r
- else {\r
- switch (n) {\r
- case 1: r = add_opcode(reg, OP_BACKREF1); break;\r
- case 2: r = add_opcode(reg, OP_BACKREF2); break;\r
- default:\r
- r = add_opcode(reg, OP_BACKREFN);\r
- if (r) return r;\r
- r = add_mem_num(reg, n);\r
- break;\r
- }\r
- }\r
+ {\r
+ r = add_opcode(reg, OP_BACKREF_CHECK);\r
+ if (r != 0) return r;\r
+ }\r
+\r
+ goto add_bacref_mems;\r
}\r
else {\r
- int i;\r
- int* p;\r
-\r
- if (IS_IGNORECASE(reg->options)) {\r
- r = add_opcode(reg, OP_BACKREF_MULTI_IC);\r
+#ifdef USE_BACKREF_WITH_LEVEL\r
+ if (NODE_IS_NEST_LEVEL(node)) {\r
+ r = add_opcode(reg, OP_BACKREF_WITH_LEVEL);\r
+ if (r != 0) return r;\r
+ r = add_option(reg, (reg->options & ONIG_OPTION_IGNORECASE));\r
+ if (r != 0) return r;\r
+ r = add_length(reg, br->nest_level);\r
+ if (r != 0) return r;\r
+\r
+ goto add_bacref_mems;\r
}\r
- else {\r
- r = add_opcode(reg, OP_BACKREF_MULTI);\r
+ else\r
+#endif\r
+ if (br->back_num == 1) {\r
+ n = br->back_static[0];\r
+ if (IS_IGNORECASE(reg->options)) {\r
+ r = add_opcode(reg, OP_BACKREF_N_IC);\r
+ if (r != 0) return r;\r
+ r = add_mem_num(reg, n);\r
+ }\r
+ else {\r
+ switch (n) {\r
+ case 1: r = add_opcode(reg, OP_BACKREF1); break;\r
+ case 2: r = add_opcode(reg, OP_BACKREF2); break;\r
+ default:\r
+ r = add_opcode(reg, OP_BACKREF_N);\r
+ if (r != 0) return r;\r
+ r = add_mem_num(reg, n);\r
+ break;\r
+ }\r
+ }\r
}\r
- if (r) return r;\r
+ else {\r
+ int i;\r
+ int* p;\r
\r
-#ifdef USE_BACKREF_WITH_LEVEL\r
- add_bacref_mems:\r
-#endif\r
- r = add_length(reg, br->back_num);\r
- if (r) return r;\r
- p = BACKREFS_P(br);\r
- for (i = br->back_num - 1; i >= 0; i--) {\r
- r = add_mem_num(reg, p[i]);\r
- if (r) return r;\r
- }\r
+ if (IS_IGNORECASE(reg->options)) {\r
+ r = add_opcode(reg, OP_BACKREF_MULTI_IC);\r
+ }\r
+ else {\r
+ r = add_opcode(reg, OP_BACKREF_MULTI);\r
+ }\r
+ if (r != 0) return r;\r
+\r
+ add_bacref_mems:\r
+ r = add_length(reg, br->back_num);\r
+ if (r != 0) return r;\r
+ p = BACKREFS_P(br);\r
+ for (i = br->back_num - 1; i >= 0; i--) {\r
+ r = add_mem_num(reg, p[i]);\r
+ if (r != 0) return r;\r
+ }\r
+ }\r
}\r
}\r
break;\r
\r
-#ifdef USE_SUBEXP_CALL\r
- case NT_CALL:\r
- r = compile_call(NCALL(node), reg);\r
+#ifdef USE_CALL\r
+ case NODE_CALL:\r
+ r = compile_call(CALL_(node), reg, env);\r
break;\r
#endif\r
\r
- case NT_QTFR:\r
- r = compile_quantifier_node(NQTFR(node), reg);\r
+ case NODE_QUANT:\r
+ r = compile_quantifier_node(QUANT_(node), reg, env);\r
+ break;\r
+\r
+ case NODE_ENCLOSURE:\r
+ r = compile_enclosure_node(ENCLOSURE_(node), reg, env);\r
break;\r
\r
- case NT_ENCLOSE:\r
- r = compile_enclose_node(NENCLOSE(node), reg);\r
+ case NODE_ANCHOR:\r
+ r = compile_anchor_node(ANCHOR_(node), reg, env);\r
break;\r
\r
- case NT_ANCHOR:\r
- r = compile_anchor_node(NANCHOR(node), reg);\r
+ case NODE_GIMMICK:\r
+ r = compile_gimmick_node(GIMMICK_(node), reg);\r
break;\r
\r
default:\r
#ifdef ONIG_DEBUG\r
- fprintf(stderr, "compile_tree: undefined node type %d\n", NTYPE(node));\r
+ fprintf(stderr, "compile_tree: undefined node type %d\n", NODE_TYPE(node));\r
#endif\r
break;\r
}\r
return r;\r
}\r
\r
-#ifdef USE_NAMED_GROUP\r
-\r
static int\r
noname_disable_map(Node** plink, GroupNumRemap* map, int* counter)\r
{\r
int r = 0;\r
Node* node = *plink;\r
\r
- switch (NTYPE(node)) {\r
- case NT_LIST:\r
- case NT_ALT:\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
+ case NODE_ALT:\r
do {\r
- r = noname_disable_map(&(NCAR(node)), map, counter);\r
- } while (r == 0 && IS_NOT_NULL(node = NCDR(node)));\r
+ r = noname_disable_map(&(NODE_CAR(node)), map, counter);\r
+ } while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));\r
break;\r
\r
- case NT_QTFR:\r
+ case NODE_QUANT:\r
{\r
- Node** ptarget = &(NQTFR(node)->target);\r
+ Node** ptarget = &(NODE_BODY(node));\r
Node* old = *ptarget;\r
r = noname_disable_map(ptarget, map, counter);\r
- if (*ptarget != old && NTYPE(*ptarget) == NT_QTFR) {\r
- onig_reduce_nested_quantifier(node, *ptarget);\r
+ if (*ptarget != old && NODE_TYPE(*ptarget) == NODE_QUANT) {\r
+ onig_reduce_nested_quantifier(node, *ptarget);\r
}\r
}\r
break;\r
\r
- case NT_ENCLOSE:\r
+ case NODE_ENCLOSURE:\r
{\r
- EncloseNode* en = NENCLOSE(node);\r
- if (en->type == ENCLOSE_MEMORY) {\r
- if (IS_ENCLOSE_NAMED_GROUP(en)) {\r
- (*counter)++;\r
- map[en->regnum].new_val = *counter;\r
- en->regnum = *counter;\r
- r = noname_disable_map(&(en->target), map, counter);\r
- }\r
- else {\r
- *plink = en->target;\r
- en->target = NULL_NODE;\r
- onig_node_free(node);\r
- r = noname_disable_map(plink, map, counter);\r
- }\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+ if (en->type == ENCLOSURE_MEMORY) {\r
+ if (NODE_IS_NAMED_GROUP(node)) {\r
+ (*counter)++;\r
+ map[en->m.regnum].new_val = *counter;\r
+ en->m.regnum = *counter;\r
+ r = noname_disable_map(&(NODE_BODY(node)), map, counter);\r
+ }\r
+ else {\r
+ *plink = NODE_BODY(node);\r
+ NODE_BODY(node) = NULL_NODE;\r
+ onig_node_free(node);\r
+ r = noname_disable_map(plink, map, counter);\r
+ }\r
+ }\r
+ else if (en->type == ENCLOSURE_IF_ELSE) {\r
+ r = noname_disable_map(&(NODE_ENCLOSURE_BODY(en)), map, counter);\r
+ if (r != 0) return r;\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ r = noname_disable_map(&(en->te.Then), map, counter);\r
+ if (r != 0) return r;\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else)) {\r
+ r = noname_disable_map(&(en->te.Else), map, counter);\r
+ if (r != 0) return r;\r
+ }\r
}\r
else\r
- r = noname_disable_map(&(en->target), map, counter);\r
+ r = noname_disable_map(&(NODE_BODY(node)), map, counter);\r
}\r
break;\r
\r
+ case NODE_ANCHOR:\r
+ if (IS_NOT_NULL(NODE_BODY(node)))\r
+ r = noname_disable_map(&(NODE_BODY(node)), map, counter);\r
+ break;\r
+\r
default:\r
break;\r
}\r
{\r
int i, pos, n, old_num;\r
int *backs;\r
- BRefNode* bn = NBREF(node);\r
+ BackRefNode* bn = BACKREF_(node);\r
\r
- if (! IS_BACKREF_NAME_REF(bn))\r
+ if (! NODE_IS_BY_NAME(node))\r
return ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED;\r
\r
old_num = bn->back_num;\r
{\r
int r = 0;\r
\r
- switch (NTYPE(node)) {\r
- case NT_LIST:\r
- case NT_ALT:\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
+ case NODE_ALT:\r
do {\r
- r = renumber_by_map(NCAR(node), map);\r
- } while (r == 0 && IS_NOT_NULL(node = NCDR(node)));\r
+ r = renumber_by_map(NODE_CAR(node), map);\r
+ } while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));\r
break;\r
- case NT_QTFR:\r
- r = renumber_by_map(NQTFR(node)->target, map);\r
+\r
+ case NODE_QUANT:\r
+ r = renumber_by_map(NODE_BODY(node), map);\r
break;\r
- case NT_ENCLOSE:\r
- r = renumber_by_map(NENCLOSE(node)->target, map);\r
+\r
+ case NODE_ENCLOSURE:\r
+ {\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+\r
+ r = renumber_by_map(NODE_BODY(node), map);\r
+ if (r != 0) return r;\r
+\r
+ if (en->type == ENCLOSURE_IF_ELSE) {\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ r = renumber_by_map(en->te.Then, map);\r
+ if (r != 0) return r;\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else)) {\r
+ r = renumber_by_map(en->te.Else, map);\r
+ if (r != 0) return r;\r
+ }\r
+ }\r
+ }\r
break;\r
\r
- case NT_BREF:\r
+ case NODE_BACKREF:\r
r = renumber_node_backref(node, map);\r
break;\r
\r
+ case NODE_ANCHOR:\r
+ if (IS_NOT_NULL(NODE_BODY(node)))\r
+ r = renumber_by_map(NODE_BODY(node), map);\r
+ break;\r
+\r
default:\r
break;\r
}\r
{\r
int r = 0;\r
\r
- switch (NTYPE(node)) {\r
- case NT_LIST:\r
- case NT_ALT:\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
+ case NODE_ALT:\r
do {\r
- r = numbered_ref_check(NCAR(node));\r
- } while (r == 0 && IS_NOT_NULL(node = NCDR(node)));\r
+ r = numbered_ref_check(NODE_CAR(node));\r
+ } while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));\r
break;\r
- case NT_QTFR:\r
- r = numbered_ref_check(NQTFR(node)->target);\r
+\r
+ case NODE_ANCHOR:\r
+ if (IS_NULL(NODE_BODY(node)))\r
+ break;\r
+ /* fall */\r
+ case NODE_QUANT:\r
+ r = numbered_ref_check(NODE_BODY(node));\r
break;\r
- case NT_ENCLOSE:\r
- r = numbered_ref_check(NENCLOSE(node)->target);\r
+\r
+ case NODE_ENCLOSURE:\r
+ {\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+\r
+ r = numbered_ref_check(NODE_BODY(node));\r
+ if (r != 0) return r;\r
+\r
+ if (en->type == ENCLOSURE_IF_ELSE) {\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ r = numbered_ref_check(en->te.Then);\r
+ if (r != 0) return r;\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else)) {\r
+ r = numbered_ref_check(en->te.Else);\r
+ if (r != 0) return r;\r
+ }\r
+ }\r
+ }\r
+\r
break;\r
\r
- case NT_BREF:\r
- if (! IS_BACKREF_NAME_REF(NBREF(node)))\r
+ case NODE_BACKREF:\r
+ if (! NODE_IS_BY_NAME(node))\r
return ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED;\r
break;\r
\r
disable_noname_group_capture(Node** root, regex_t* reg, ScanEnv* env)\r
{\r
int r, i, pos, counter;\r
- int Result;\r
- BitStatusType loc;\r
+ int result;\r
+ MemStatusType loc;\r
GroupNumRemap* map;\r
\r
map = (GroupNumRemap* )xmalloc(sizeof(GroupNumRemap) * (env->num_mem + 1));\r
\r
for (i = 1, pos = 1; i <= env->num_mem; i++) {\r
if (map[i].new_val > 0) {\r
- SCANENV_MEM_NODES(env)[pos] = SCANENV_MEM_NODES(env)[i];\r
+ SCANENV_MEMENV(env)[pos] = SCANENV_MEMENV(env)[i];\r
pos++;\r
}\r
}\r
\r
loc = env->capture_history;\r
- BIT_STATUS_CLEAR(env->capture_history);\r
+ MEM_STATUS_CLEAR(env->capture_history);\r
for (i = 1; i <= ONIG_MAX_CAPTURE_HISTORY_GROUP; i++) {\r
- if (BIT_STATUS_AT(loc, i)) {\r
- BIT_STATUS_ON_AT_SIMPLE(env->capture_history, map[i].new_val);\r
+ if (MEM_STATUS_AT(loc, i)) {\r
+ MEM_STATUS_ON_SIMPLE(env->capture_history, map[i].new_val);\r
}\r
}\r
\r
env->num_mem = env->num_named;\r
reg->num_mem = env->num_named;\r
-\r
- Result = onig_renumber_name_table(reg, map);\r
+ result = onig_renumber_name_table(reg, map);\r
xfree(map);\r
- return Result;\r
+ return result;\r
}\r
-#endif /* USE_NAMED_GROUP */\r
\r
-#ifdef USE_SUBEXP_CALL\r
+#ifdef USE_CALL\r
static int\r
-unset_addr_list_fix(UnsetAddrList* uslist, regex_t* reg)\r
+fix_unset_addr_list(UnsetAddrList* uslist, regex_t* reg)\r
{\r
int i, offset;\r
- EncloseNode* en;\r
+ EnclosureNode* en;\r
AbsAddrType addr;\r
\r
for (i = 0; i < uslist->num; i++) {\r
- en = NENCLOSE(uslist->us[i].target);\r
- if (! IS_ENCLOSE_ADDR_FIXED(en)) return ONIGERR_PARSER_BUG;\r
- addr = en->call_addr;\r
+ if (! NODE_IS_ADDR_FIXED(uslist->us[i].target))\r
+ return ONIGERR_PARSER_BUG;\r
+\r
+ en = ENCLOSURE_(uslist->us[i].target);\r
+ addr = en->m.called_addr;\r
offset = uslist->us[i].offset;\r
\r
- BBUF_WRITE(reg, offset, &addr, SIZE_ABSADDR);\r
+ BB_WRITE(reg, offset, &addr, SIZE_ABSADDR);\r
}\r
return 0;\r
}\r
#endif\r
\r
-#ifdef USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT\r
+\r
+#define GET_CHAR_LEN_VARLEN -1\r
+#define GET_CHAR_LEN_TOP_ALT_VARLEN -2\r
+\r
+/* fixed size pattern node only */\r
static int\r
-quantifiers_memory_node_info(Node* node)\r
+get_char_length_tree1(Node* node, regex_t* reg, int* len, int level)\r
{\r
+ int tlen;\r
int r = 0;\r
\r
- switch (NTYPE(node)) {\r
- case NT_LIST:\r
- case NT_ALT:\r
- {\r
- int v;\r
- do {\r
- v = quantifiers_memory_node_info(NCAR(node));\r
- if (v > r) r = v;\r
- } while (v >= 0 && IS_NOT_NULL(node = NCDR(node)));\r
- }\r
+ level++;\r
+ *len = 0;\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
+ do {\r
+ r = get_char_length_tree1(NODE_CAR(node), reg, &tlen, level);\r
+ if (r == 0)\r
+ *len = distance_add(*len, tlen);\r
+ } while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));\r
break;\r
\r
-#ifdef USE_SUBEXP_CALL\r
- case NT_CALL:\r
- if (IS_CALL_RECURSION(NCALL(node))) {\r
- return NQ_TARGET_IS_EMPTY_REC; /* tiny version */\r
+ case NODE_ALT:\r
+ {\r
+ int tlen2;\r
+ int varlen = 0;\r
+\r
+ r = get_char_length_tree1(NODE_CAR(node), reg, &tlen, level);\r
+ while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node))) {\r
+ r = get_char_length_tree1(NODE_CAR(node), reg, &tlen2, level);\r
+ if (r == 0) {\r
+ if (tlen != tlen2)\r
+ varlen = 1;\r
+ }\r
+ }\r
+ if (r == 0) {\r
+ if (varlen != 0) {\r
+ if (level == 1)\r
+ r = GET_CHAR_LEN_TOP_ALT_VARLEN;\r
+ else\r
+ r = GET_CHAR_LEN_VARLEN;\r
+ }\r
+ else\r
+ *len = tlen;\r
+ }\r
}\r
- else\r
- r = quantifiers_memory_node_info(NCALL(node)->target);\r
break;\r
-#endif\r
\r
- case NT_QTFR:\r
+ case NODE_STRING:\r
{\r
- QtfrNode* qn = NQTFR(node);\r
- if (qn->upper != 0) {\r
- r = quantifiers_memory_node_info(qn->target);\r
+ StrNode* sn = STR_(node);\r
+ UChar *s = sn->s;\r
+\r
+ while (s < sn->end) {\r
+ s += enclen(reg->enc, s);\r
+ (*len)++;\r
}\r
}\r
break;\r
\r
- case NT_ENCLOSE:\r
+ case NODE_QUANT:\r
{\r
- EncloseNode* en = NENCLOSE(node);\r
- switch (en->type) {\r
- case ENCLOSE_MEMORY:\r
- return NQ_TARGET_IS_EMPTY_MEM;\r
- break;\r
-\r
- case ENCLOSE_OPTION:\r
- case ENCLOSE_STOP_BACKTRACK:\r
- r = quantifiers_memory_node_info(en->target);\r
- break;\r
- default:\r
- break;\r
- }\r
- }\r
- break;\r
-\r
- case NT_BREF:\r
- case NT_STR:\r
- case NT_CTYPE:\r
- case NT_CCLASS:\r
- case NT_CANY:\r
- case NT_ANCHOR:\r
- default:\r
- break;\r
- }\r
-\r
- return r;\r
-}\r
-#endif /* USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT */\r
-\r
-static int\r
-get_min_match_length(Node* node, OnigDistance *min, ScanEnv* env)\r
-{\r
- OnigDistance tmin;\r
- int r = 0;\r
-\r
- *min = 0;\r
- switch (NTYPE(node)) {\r
- case NT_BREF:\r
- {\r
- int i;\r
- int* backs;\r
- Node** nodes = SCANENV_MEM_NODES(env);\r
- BRefNode* br = NBREF(node);\r
- if (br->state & NST_RECURSION) break;\r
-\r
- backs = BACKREFS_P(br);\r
- if (backs[0] > env->num_mem) return ONIGERR_INVALID_BACKREF;\r
- r = get_min_match_length(nodes[backs[0]], min, env);\r
- if (r != 0) break;\r
- for (i = 1; i < br->back_num; i++) {\r
- if (backs[i] > env->num_mem) return ONIGERR_INVALID_BACKREF;\r
- r = get_min_match_length(nodes[backs[i]], &tmin, env);\r
- if (r != 0) break;\r
- if (*min > tmin) *min = tmin;\r
- }\r
- }\r
- break;\r
-\r
-#ifdef USE_SUBEXP_CALL\r
- case NT_CALL:\r
- if (IS_CALL_RECURSION(NCALL(node))) {\r
- EncloseNode* en = NENCLOSE(NCALL(node)->target);\r
- if (IS_ENCLOSE_MIN_FIXED(en))\r
- *min = en->min_len;\r
- }\r
- else\r
- r = get_min_match_length(NCALL(node)->target, min, env);\r
- break;\r
-#endif\r
-\r
- case NT_LIST:\r
- do {\r
- r = get_min_match_length(NCAR(node), &tmin, env);\r
- if (r == 0) *min += tmin;\r
- } while (r == 0 && IS_NOT_NULL(node = NCDR(node)));\r
- break;\r
-\r
- case NT_ALT:\r
- {\r
- Node *x, *y;\r
- y = node;\r
- do {\r
- x = NCAR(y);\r
- r = get_min_match_length(x, &tmin, env);\r
- if (r != 0) break;\r
- if (y == node) *min = tmin;\r
- else if (*min > tmin) *min = tmin;\r
- } while (r == 0 && IS_NOT_NULL(y = NCDR(y)));\r
- }\r
- break;\r
-\r
- case NT_STR:\r
- {\r
- StrNode* sn = NSTR(node);\r
- *min = (OnigDistance)(sn->end - sn->s);\r
- }\r
- break;\r
-\r
- case NT_CTYPE:\r
- *min = 1;\r
- break;\r
-\r
- case NT_CCLASS:\r
- case NT_CANY:\r
- *min = 1;\r
- break;\r
-\r
- case NT_QTFR:\r
- {\r
- QtfrNode* qn = NQTFR(node);\r
-\r
- if (qn->lower > 0) {\r
- r = get_min_match_length(qn->target, min, env);\r
- if (r == 0)\r
- *min = distance_multiply(*min, qn->lower);\r
- }\r
- }\r
- break;\r
-\r
- case NT_ENCLOSE:\r
- {\r
- EncloseNode* en = NENCLOSE(node);\r
- switch (en->type) {\r
- case ENCLOSE_MEMORY:\r
-#ifdef USE_SUBEXP_CALL\r
- if (IS_ENCLOSE_MIN_FIXED(en))\r
- *min = en->min_len;\r
- else {\r
- r = get_min_match_length(en->target, min, env);\r
- if (r == 0) {\r
- en->min_len = *min;\r
- SET_ENCLOSE_STATUS(node, NST_MIN_FIXED);\r
- }\r
- }\r
- break;\r
-#endif\r
- case ENCLOSE_OPTION:\r
- case ENCLOSE_STOP_BACKTRACK:\r
- r = get_min_match_length(en->target, min, env);\r
- break;\r
- }\r
- }\r
- break;\r
-\r
- case NT_ANCHOR:\r
- default:\r
- break;\r
- }\r
-\r
- return r;\r
-}\r
-\r
-static int\r
-get_max_match_length(Node* node, OnigDistance *max, ScanEnv* env)\r
-{\r
- OnigDistance tmax;\r
- int r = 0;\r
-\r
- *max = 0;\r
- switch (NTYPE(node)) {\r
- case NT_LIST:\r
- do {\r
- r = get_max_match_length(NCAR(node), &tmax, env);\r
- if (r == 0)\r
- *max = distance_add(*max, tmax);\r
- } while (r == 0 && IS_NOT_NULL(node = NCDR(node)));\r
- break;\r
-\r
- case NT_ALT:\r
- do {\r
- r = get_max_match_length(NCAR(node), &tmax, env);\r
- if (r == 0 && *max < tmax) *max = tmax;\r
- } while (r == 0 && IS_NOT_NULL(node = NCDR(node)));\r
- break;\r
-\r
- case NT_STR:\r
- {\r
- StrNode* sn = NSTR(node);\r
- *max = (OnigDistance)(sn->end - sn->s);\r
- }\r
- break;\r
-\r
- case NT_CTYPE:\r
- *max = ONIGENC_MBC_MAXLEN_DIST(env->enc);\r
- break;\r
-\r
- case NT_CCLASS:\r
- case NT_CANY:\r
- *max = ONIGENC_MBC_MAXLEN_DIST(env->enc);\r
- break;\r
-\r
- case NT_BREF:\r
- {\r
- int i;\r
- int* backs;\r
- Node** nodes = SCANENV_MEM_NODES(env);\r
- BRefNode* br = NBREF(node);\r
- if (br->state & NST_RECURSION) {\r
- *max = ONIG_INFINITE_DISTANCE;\r
- break;\r
- }\r
- backs = BACKREFS_P(br);\r
- for (i = 0; i < br->back_num; i++) {\r
- if (backs[i] > env->num_mem) return ONIGERR_INVALID_BACKREF;\r
- r = get_max_match_length(nodes[backs[i]], &tmax, env);\r
- if (r != 0) break;\r
- if (*max < tmax) *max = tmax;\r
- }\r
- }\r
- break;\r
-\r
-#ifdef USE_SUBEXP_CALL\r
- case NT_CALL:\r
- if (! IS_CALL_RECURSION(NCALL(node)))\r
- r = get_max_match_length(NCALL(node)->target, max, env);\r
- else\r
- *max = ONIG_INFINITE_DISTANCE;\r
- break;\r
-#endif\r
-\r
- case NT_QTFR:\r
- {\r
- QtfrNode* qn = NQTFR(node);\r
-\r
- if (qn->upper != 0) {\r
- r = get_max_match_length(qn->target, max, env);\r
- if (r == 0 && *max != 0) {\r
- if (! IS_REPEAT_INFINITE(qn->upper))\r
- *max = distance_multiply(*max, qn->upper);\r
- else\r
- *max = ONIG_INFINITE_DISTANCE;\r
- }\r
- }\r
- }\r
- break;\r
-\r
- case NT_ENCLOSE:\r
- {\r
- EncloseNode* en = NENCLOSE(node);\r
- switch (en->type) {\r
- case ENCLOSE_MEMORY:\r
-#ifdef USE_SUBEXP_CALL\r
- if (IS_ENCLOSE_MAX_FIXED(en))\r
- *max = en->max_len;\r
- else {\r
- r = get_max_match_length(en->target, max, env);\r
- if (r == 0) {\r
- en->max_len = *max;\r
- SET_ENCLOSE_STATUS(node, NST_MAX_FIXED);\r
- }\r
- }\r
- break;\r
-#endif\r
- case ENCLOSE_OPTION:\r
- case ENCLOSE_STOP_BACKTRACK:\r
- r = get_max_match_length(en->target, max, env);\r
- break;\r
- }\r
- }\r
- break;\r
-\r
- case NT_ANCHOR:\r
- default:\r
- break;\r
- }\r
-\r
- return r;\r
-}\r
-\r
-#define GET_CHAR_LEN_VARLEN -1\r
-#define GET_CHAR_LEN_TOP_ALT_VARLEN -2\r
+ QuantNode* qn = QUANT_(node);\r
\r
-/* fixed size pattern node only */\r
-static int\r
-get_char_length_tree1(Node* node, regex_t* reg, int* len, int level)\r
-{\r
- int tlen;\r
- int r = 0;\r
-\r
- level++;\r
- *len = 0;\r
- switch (NTYPE(node)) {\r
- case NT_LIST:\r
- do {\r
- r = get_char_length_tree1(NCAR(node), reg, &tlen, level);\r
- if (r == 0)\r
- *len = distance_add(*len, tlen);\r
- } while (r == 0 && IS_NOT_NULL(node = NCDR(node)));\r
- break;\r
-\r
- case NT_ALT:\r
- {\r
- int tlen2;\r
- int varlen = 0;\r
-\r
- r = get_char_length_tree1(NCAR(node), reg, &tlen, level);\r
- while (r == 0 && IS_NOT_NULL(node = NCDR(node))) {\r
- r = get_char_length_tree1(NCAR(node), reg, &tlen2, level);\r
- if (r == 0) {\r
- if (tlen != tlen2)\r
- varlen = 1;\r
- }\r
- }\r
- if (r == 0) {\r
- if (varlen != 0) {\r
- if (level == 1)\r
- r = GET_CHAR_LEN_TOP_ALT_VARLEN;\r
- else\r
- r = GET_CHAR_LEN_VARLEN;\r
- }\r
- else\r
- *len = tlen;\r
- }\r
- }\r
- break;\r
-\r
- case NT_STR:\r
- {\r
- StrNode* sn = NSTR(node);\r
- UChar *s = sn->s;\r
- while (s < sn->end) {\r
- s += enclen(reg->enc, s);\r
- (*len)++;\r
- }\r
- }\r
- break;\r
-\r
- case NT_QTFR:\r
- {\r
- QtfrNode* qn = NQTFR(node);\r
if (qn->lower == qn->upper) {\r
- r = get_char_length_tree1(qn->target, reg, &tlen, level);\r
- if (r == 0)\r
- *len = distance_multiply(tlen, qn->lower);\r
+ if (qn->upper == 0) {\r
+ *len = 0;\r
+ }\r
+ else {\r
+ r = get_char_length_tree1(NODE_BODY(node), reg, &tlen, level);\r
+ if (r == 0)\r
+ *len = distance_multiply(tlen, qn->lower);\r
+ }\r
}\r
else\r
- r = GET_CHAR_LEN_VARLEN;\r
+ r = GET_CHAR_LEN_VARLEN;\r
}\r
break;\r
\r
-#ifdef USE_SUBEXP_CALL\r
- case NT_CALL:\r
- if (! IS_CALL_RECURSION(NCALL(node)))\r
- r = get_char_length_tree1(NCALL(node)->target, reg, len, level);\r
+#ifdef USE_CALL\r
+ case NODE_CALL:\r
+ if (! NODE_IS_RECURSION(node))\r
+ r = get_char_length_tree1(NODE_BODY(node), reg, len, level);\r
else\r
r = GET_CHAR_LEN_VARLEN;\r
break;\r
#endif\r
\r
- case NT_CTYPE:\r
- *len = 1;\r
- break;\r
-\r
- case NT_CCLASS:\r
- case NT_CANY:\r
+ case NODE_CTYPE:\r
+ case NODE_CCLASS:\r
*len = 1;\r
break;\r
\r
- case NT_ENCLOSE:\r
+ case NODE_ENCLOSURE:\r
{\r
- EncloseNode* en = NENCLOSE(node);\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+\r
switch (en->type) {\r
- case ENCLOSE_MEMORY:\r
-#ifdef USE_SUBEXP_CALL\r
- if (IS_ENCLOSE_CLEN_FIXED(en))\r
- *len = en->char_len;\r
- else {\r
- r = get_char_length_tree1(en->target, reg, len, level);\r
- if (r == 0) {\r
- en->char_len = *len;\r
- SET_ENCLOSE_STATUS(node, NST_CLEN_FIXED);\r
- }\r
- }\r
- break;\r
+ case ENCLOSURE_MEMORY:\r
+#ifdef USE_CALL\r
+ if (NODE_IS_CLEN_FIXED(node))\r
+ *len = en->char_len;\r
+ else {\r
+ r = get_char_length_tree1(NODE_BODY(node), reg, len, level);\r
+ if (r == 0) {\r
+ en->char_len = *len;\r
+ NODE_STATUS_ADD(node, CLEN_FIXED);\r
+ }\r
+ }\r
+ break;\r
#endif\r
- case ENCLOSE_OPTION:\r
- case ENCLOSE_STOP_BACKTRACK:\r
- r = get_char_length_tree1(en->target, reg, len, level);\r
- break;\r
+ case ENCLOSURE_OPTION:\r
+ case ENCLOSURE_STOP_BACKTRACK:\r
+ r = get_char_length_tree1(NODE_BODY(node), reg, len, level);\r
+ break;\r
+ case ENCLOSURE_IF_ELSE:\r
+ {\r
+ int clen, elen;\r
+\r
+ r = get_char_length_tree1(NODE_BODY(node), reg, &clen, level);\r
+ if (r == 0) {\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ r = get_char_length_tree1(en->te.Then, reg, &tlen, level);\r
+ if (r != 0) break;\r
+ }\r
+ else tlen = 0;\r
+ if (IS_NOT_NULL(en->te.Else)) {\r
+ r = get_char_length_tree1(en->te.Else, reg, &elen, level);\r
+ if (r != 0) break;\r
+ }\r
+ else elen = 0;\r
+\r
+ if (clen + tlen != elen) {\r
+ r = GET_CHAR_LEN_VARLEN;\r
+ }\r
+ else {\r
+ *len = elen;\r
+ }\r
+ }\r
+ }\r
+ break;\r
+\r
default:\r
- break;\r
+ break;\r
}\r
}\r
break;\r
\r
- case NT_ANCHOR:\r
+ case NODE_ANCHOR:\r
+ case NODE_GIMMICK:\r
break;\r
\r
+ case NODE_BACKREF:\r
+ if (NODE_IS_CHECKER(node))\r
+ break;\r
+ /* fall */\r
default:\r
r = GET_CHAR_LEN_VARLEN;\r
break;\r
\r
/* x is not included y ==> 1 : 0 */\r
static int\r
-is_not_included(Node* x, Node* y, regex_t* reg)\r
+is_exclusive(Node* x, Node* y, regex_t* reg)\r
{\r
int i, len;\r
OnigCodePoint code;\r
UChar *p;\r
- int ytype;\r
+ NodeType ytype;\r
\r
retry:\r
- ytype = NTYPE(y);\r
- switch (NTYPE(x)) {\r
- case NT_CTYPE:\r
+ ytype = NODE_TYPE(y);\r
+ switch (NODE_TYPE(x)) {\r
+ case NODE_CTYPE:\r
{\r
+ if (CTYPE_(x)->ctype == CTYPE_ANYCHAR ||\r
+ CTYPE_(y)->ctype == CTYPE_ANYCHAR)\r
+ break;\r
+\r
switch (ytype) {\r
- case NT_CTYPE:\r
- if (NCTYPE(y)->ctype == NCTYPE(x)->ctype &&\r
- NCTYPE(y)->not != NCTYPE(x)->not)\r
- return 1;\r
- else\r
- return 0;\r
- break;\r
-\r
- case NT_CCLASS:\r
+ case NODE_CTYPE:\r
+ if (CTYPE_(y)->ctype == CTYPE_(x)->ctype &&\r
+ CTYPE_(y)->not != CTYPE_(x)->not &&\r
+ CTYPE_(y)->ascii_mode == CTYPE_(x)->ascii_mode)\r
+ return 1;\r
+ else\r
+ return 0;\r
+ break;\r
+\r
+ case NODE_CCLASS:\r
swap:\r
- {\r
- Node* tmp;\r
- tmp = x; x = y; y = tmp;\r
- goto retry;\r
- }\r
- break;\r
+ {\r
+ Node* tmp;\r
+ tmp = x; x = y; y = tmp;\r
+ goto retry;\r
+ }\r
+ break;\r
\r
- case NT_STR:\r
- goto swap;\r
- break;\r
+ case NODE_STRING:\r
+ goto swap;\r
+ break;\r
\r
default:\r
- break;\r
+ break;\r
}\r
}\r
break;\r
\r
- case NT_CCLASS:\r
+ case NODE_CCLASS:\r
{\r
- CClassNode* xc = NCCLASS(x);\r
+ int range;\r
+ CClassNode* xc = CCLASS_(x);\r
+\r
switch (ytype) {\r
- case NT_CTYPE:\r
- switch (NCTYPE(y)->ctype) {\r
- case ONIGENC_CTYPE_WORD:\r
- if (NCTYPE(y)->not == 0) {\r
- if (IS_NULL(xc->mbuf) && !IS_NCCLASS_NOT(xc)) {\r
- for (i = 0; i < SINGLE_BYTE_SIZE; i++) {\r
- if (BITSET_AT(xc->bs, i)) {\r
- if (IS_CODE_SB_WORD(reg->enc, i)) return 0;\r
- }\r
- }\r
- return 1;\r
- }\r
- return 0;\r
- }\r
- else {\r
- for (i = 0; i < SINGLE_BYTE_SIZE; i++) {\r
- if (! IS_CODE_SB_WORD(reg->enc, i)) {\r
- if (!IS_NCCLASS_NOT(xc)) {\r
- if (BITSET_AT(xc->bs, i))\r
- return 0;\r
- }\r
- else {\r
- if (! BITSET_AT(xc->bs, i))\r
- return 0;\r
- }\r
- }\r
- }\r
- return 1;\r
- }\r
- break;\r
-\r
- default:\r
- break;\r
- }\r
- break;\r
-\r
- case NT_CCLASS:\r
- {\r
- int v;\r
- CClassNode* yc = NCCLASS(y);\r
-\r
- for (i = 0; i < SINGLE_BYTE_SIZE; i++) {\r
- v = BITSET_AT(xc->bs, i);\r
- if ((v != 0 && !IS_NCCLASS_NOT(xc)) ||\r
- (v == 0 && IS_NCCLASS_NOT(xc))) {\r
- v = BITSET_AT(yc->bs, i);\r
- if ((v != 0 && !IS_NCCLASS_NOT(yc)) ||\r
+ case NODE_CTYPE:\r
+ switch (CTYPE_(y)->ctype) {\r
+ case CTYPE_ANYCHAR:\r
+ return 0;\r
+ break;\r
+\r
+ case ONIGENC_CTYPE_WORD:\r
+ if (CTYPE_(y)->not == 0) {\r
+ if (IS_NULL(xc->mbuf) && !IS_NCCLASS_NOT(xc)) {\r
+ range = CTYPE_(y)->ascii_mode != 0 ? 128 : SINGLE_BYTE_SIZE;\r
+ for (i = 0; i < range; i++) {\r
+ if (BITSET_AT(xc->bs, i)) {\r
+ if (ONIGENC_IS_CODE_WORD(reg->enc, i)) return 0;\r
+ }\r
+ }\r
+ return 1;\r
+ }\r
+ return 0;\r
+ }\r
+ else {\r
+ if (IS_NOT_NULL(xc->mbuf)) return 0;\r
+ if (IS_NCCLASS_NOT(xc)) return 0;\r
+\r
+ range = CTYPE_(y)->ascii_mode != 0 ? 128 : SINGLE_BYTE_SIZE;\r
+ for (i = 0; i < range; i++) {\r
+ if (! ONIGENC_IS_CODE_WORD(reg->enc, i)) {\r
+ if (BITSET_AT(xc->bs, i))\r
+ return 0;\r
+ }\r
+ }\r
+ for (i = range; i < SINGLE_BYTE_SIZE; i++) {\r
+ if (BITSET_AT(xc->bs, i)) return 0;\r
+ }\r
+ return 1;\r
+ }\r
+ break;\r
+\r
+ default:\r
+ break;\r
+ }\r
+ break;\r
+\r
+ case NODE_CCLASS:\r
+ {\r
+ int v;\r
+ CClassNode* yc = CCLASS_(y);\r
+\r
+ for (i = 0; i < SINGLE_BYTE_SIZE; i++) {\r
+ v = BITSET_AT(xc->bs, i);\r
+ if ((v != 0 && !IS_NCCLASS_NOT(xc)) || (v == 0 && IS_NCCLASS_NOT(xc))) {\r
+ v = BITSET_AT(yc->bs, i);\r
+ if ((v != 0 && !IS_NCCLASS_NOT(yc)) ||\r
(v == 0 && IS_NCCLASS_NOT(yc)))\r
- return 0;\r
- }\r
- }\r
- if ((IS_NULL(xc->mbuf) && !IS_NCCLASS_NOT(xc)) ||\r
- (IS_NULL(yc->mbuf) && !IS_NCCLASS_NOT(yc)))\r
- return 1;\r
- return 0;\r
- }\r
- break;\r
-\r
- case NT_STR:\r
- goto swap;\r
- break;\r
+ return 0;\r
+ }\r
+ }\r
+ if ((IS_NULL(xc->mbuf) && !IS_NCCLASS_NOT(xc)) ||\r
+ (IS_NULL(yc->mbuf) && !IS_NCCLASS_NOT(yc)))\r
+ return 1;\r
+ return 0;\r
+ }\r
+ break;\r
+\r
+ case NODE_STRING:\r
+ goto swap;\r
+ break;\r
\r
default:\r
- break;\r
+ break;\r
}\r
}\r
break;\r
\r
- case NT_STR:\r
+ case NODE_STRING:\r
{\r
- StrNode* xs = NSTR(x);\r
- if (NSTRING_LEN(x) == 0)\r
- break;\r
+ StrNode* xs = STR_(x);\r
+\r
+ if (NODE_STRING_LEN(x) == 0)\r
+ break;\r
\r
- //c = *(xs->s);\r
switch (ytype) {\r
- case NT_CTYPE:\r
- switch (NCTYPE(y)->ctype) {\r
+ case NODE_CTYPE:\r
+ switch (CTYPE_(y)->ctype) {\r
+ case CTYPE_ANYCHAR:\r
+ break;\r
+\r
case ONIGENC_CTYPE_WORD:\r
- if (ONIGENC_IS_MBC_WORD(reg->enc, xs->s, xs->end))\r
- return NCTYPE(y)->not;\r
- else\r
- return !(NCTYPE(y)->not);\r
+ if (CTYPE_(y)->ascii_mode == 0) {\r
+ if (ONIGENC_IS_MBC_WORD(reg->enc, xs->s, xs->end))\r
+ return CTYPE_(y)->not;\r
+ else\r
+ return !(CTYPE_(y)->not);\r
+ }\r
+ else {\r
+ if (ONIGENC_IS_MBC_WORD_ASCII(reg->enc, xs->s, xs->end))\r
+ return CTYPE_(y)->not;\r
+ else\r
+ return !(CTYPE_(y)->not);\r
+ }\r
break;\r
default:\r
break;\r
}\r
break;\r
\r
- case NT_CCLASS:\r
+ case NODE_CCLASS:\r
{\r
- CClassNode* cc = NCCLASS(y);\r
+ CClassNode* cc = CCLASS_(y);\r
\r
code = ONIGENC_MBC_TO_CODE(reg->enc, xs->s,\r
xs->s + ONIGENC_MBC_MAXLEN(reg->enc));\r
}\r
break;\r
\r
- case NT_STR:\r
+ case NODE_STRING:\r
{\r
UChar *q;\r
- StrNode* ys = NSTR(y);\r
- len = NSTRING_LEN(x);\r
- if (len > NSTRING_LEN(y)) len = NSTRING_LEN(y);\r
- if (NSTRING_IS_AMBIG(x) || NSTRING_IS_AMBIG(y)) {\r
+ StrNode* ys = STR_(y);\r
+\r
+ len = NODE_STRING_LEN(x);\r
+ if (len > NODE_STRING_LEN(y)) len = NODE_STRING_LEN(y);\r
+ if (NODE_STRING_IS_AMBIG(x) || NODE_STRING_IS_AMBIG(y)) {\r
/* tiny version */\r
return 0;\r
}\r
}\r
}\r
break;\r
- \r
+\r
default:\r
break;\r
}\r
{\r
Node* n = NULL_NODE;\r
\r
- switch (NTYPE(node)) {\r
- case NT_BREF:\r
- case NT_ALT:\r
- case NT_CANY:\r
-#ifdef USE_SUBEXP_CALL\r
- case NT_CALL:\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_BACKREF:\r
+ case NODE_ALT:\r
+#ifdef USE_CALL\r
+ case NODE_CALL:\r
#endif\r
break;\r
\r
- case NT_CTYPE:\r
- case NT_CCLASS:\r
+ case NODE_CTYPE:\r
+ if (CTYPE_(node)->ctype == CTYPE_ANYCHAR)\r
+ break;\r
+ /* fall */\r
+ case NODE_CCLASS:\r
if (exact == 0) {\r
n = node;\r
}\r
break;\r
\r
- case NT_LIST:\r
- n = get_head_value_node(NCAR(node), exact, reg);\r
+ case NODE_LIST:\r
+ n = get_head_value_node(NODE_CAR(node), exact, reg);\r
break;\r
\r
- case NT_STR:\r
+ case NODE_STRING:\r
{\r
- StrNode* sn = NSTR(node);\r
+ StrNode* sn = STR_(node);\r
\r
if (sn->end <= sn->s)\r
- break;\r
+ break;\r
\r
if (exact != 0 &&\r
- !NSTRING_IS_RAW(node) && IS_IGNORECASE(reg->options)) {\r
+ !NODE_STRING_IS_RAW(node) && IS_IGNORECASE(reg->options)) {\r
}\r
else {\r
- n = node;\r
+ n = node;\r
}\r
}\r
break;\r
\r
- case NT_QTFR:\r
+ case NODE_QUANT:\r
{\r
- QtfrNode* qn = NQTFR(node);\r
+ QuantNode* qn = QUANT_(node);\r
if (qn->lower > 0) {\r
- if (IS_NOT_NULL(qn->head_exact))\r
- n = qn->head_exact;\r
- else\r
- n = get_head_value_node(qn->target, exact, reg);\r
+ if (IS_NOT_NULL(qn->head_exact))\r
+ n = qn->head_exact;\r
+ else\r
+ n = get_head_value_node(NODE_BODY(node), exact, reg);\r
}\r
}\r
break;\r
\r
- case NT_ENCLOSE:\r
+ case NODE_ENCLOSURE:\r
{\r
- EncloseNode* en = NENCLOSE(node);\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
switch (en->type) {\r
- case ENCLOSE_OPTION:\r
- {\r
- OnigOptionType options = reg->options;\r
-\r
- reg->options = NENCLOSE(node)->option;\r
- n = get_head_value_node(NENCLOSE(node)->target, exact, reg);\r
- reg->options = options;\r
- }\r
- break;\r
-\r
- case ENCLOSE_MEMORY:\r
- case ENCLOSE_STOP_BACKTRACK:\r
- n = get_head_value_node(en->target, exact, reg);\r
- break;\r
+ case ENCLOSURE_OPTION:\r
+ {\r
+ OnigOptionType options = reg->options;\r
+\r
+ reg->options = ENCLOSURE_(node)->o.options;\r
+ n = get_head_value_node(NODE_BODY(node), exact, reg);\r
+ reg->options = options;\r
+ }\r
+ break;\r
+\r
+ case ENCLOSURE_MEMORY:\r
+ case ENCLOSURE_STOP_BACKTRACK:\r
+ case ENCLOSURE_IF_ELSE:\r
+ n = get_head_value_node(NODE_BODY(node), exact, reg);\r
+ break;\r
}\r
}\r
break;\r
\r
- case NT_ANCHOR:\r
- if (NANCHOR(node)->type == ANCHOR_PREC_READ)\r
- n = get_head_value_node(NANCHOR(node)->target, exact, reg);\r
+ case NODE_ANCHOR:\r
+ if (ANCHOR_(node)->type == ANCHOR_PREC_READ)\r
+ n = get_head_value_node(NODE_BODY(node), exact, reg);\r
break;\r
\r
+ case NODE_GIMMICK:\r
default:\r
break;\r
}\r
}\r
\r
static int\r
-check_type_tree(Node* node, int type_mask, int enclose_mask, int anchor_mask)\r
+check_type_tree(Node* node, int type_mask, int enclosure_mask, int anchor_mask)\r
{\r
- int type, r = 0;\r
+ NodeType type;\r
+ int r = 0;\r
\r
- type = NTYPE(node);\r
- if ((NTYPE2BIT(type) & type_mask) == 0)\r
+ type = NODE_TYPE(node);\r
+ if ((NODE_TYPE2BIT(type) & type_mask) == 0)\r
return 1;\r
\r
switch (type) {\r
- case NT_LIST:\r
- case NT_ALT:\r
+ case NODE_LIST:\r
+ case NODE_ALT:\r
do {\r
- r = check_type_tree(NCAR(node), type_mask, enclose_mask,\r
- anchor_mask);\r
- } while (r == 0 && IS_NOT_NULL(node = NCDR(node)));\r
+ r = check_type_tree(NODE_CAR(node), type_mask, enclosure_mask,\r
+ anchor_mask);\r
+ } while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));\r
break;\r
\r
- case NT_QTFR:\r
- r = check_type_tree(NQTFR(node)->target, type_mask, enclose_mask,\r
- anchor_mask);\r
+ case NODE_QUANT:\r
+ r = check_type_tree(NODE_BODY(node), type_mask, enclosure_mask, anchor_mask);\r
break;\r
\r
- case NT_ENCLOSE:\r
+ case NODE_ENCLOSURE:\r
{\r
- EncloseNode* en = NENCLOSE(node);\r
- if ((en->type & enclose_mask) == 0)\r
- return 1;\r
-\r
- r = check_type_tree(en->target, type_mask, enclose_mask, anchor_mask);\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+ if (((1<<en->type) & enclosure_mask) == 0)\r
+ return 1;\r
+\r
+ r = check_type_tree(NODE_BODY(node), type_mask, enclosure_mask, anchor_mask);\r
+ if (r == 0 && en->type == ENCLOSURE_IF_ELSE) {\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ r = check_type_tree(en->te.Then, type_mask, enclosure_mask, anchor_mask);\r
+ if (r != 0) break;\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else)) {\r
+ r = check_type_tree(en->te.Else, type_mask, enclosure_mask, anchor_mask);\r
+ }\r
+ }\r
}\r
break;\r
\r
- case NT_ANCHOR:\r
- type = NANCHOR(node)->type;\r
+ case NODE_ANCHOR:\r
+ type = ANCHOR_(node)->type;\r
if ((type & anchor_mask) == 0)\r
return 1;\r
\r
- if (NANCHOR(node)->target)\r
- r = check_type_tree(NANCHOR(node)->target,\r
- type_mask, enclose_mask, anchor_mask);\r
+ if (IS_NOT_NULL(NODE_BODY(node)))\r
+ r = check_type_tree(NODE_BODY(node), type_mask, enclosure_mask, anchor_mask);\r
break;\r
\r
+ case NODE_GIMMICK:\r
default:\r
break;\r
}\r
return r;\r
}\r
\r
-#ifdef USE_SUBEXP_CALL\r
+static OnigLen\r
+tree_min_len(Node* node, ScanEnv* env)\r
+{\r
+ OnigLen len;\r
+ OnigLen tmin;\r
\r
-#define RECURSION_EXIST 1\r
-#define RECURSION_INFINITE 2\r
+ len = 0;\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_BACKREF:\r
+ if (! NODE_IS_CHECKER(node)) {\r
+ int i;\r
+ int* backs;\r
+ MemEnv* mem_env = SCANENV_MEMENV(env);\r
+ BackRefNode* br = BACKREF_(node);\r
+ if (NODE_IS_RECURSION(node)) break;\r
\r
-static int\r
-subexp_inf_recursive_check(Node* node, ScanEnv* env, int head)\r
-{\r
- int type;\r
- int r = 0;\r
+ backs = BACKREFS_P(br);\r
+ len = tree_min_len(mem_env[backs[0]].node, env);\r
+ for (i = 1; i < br->back_num; i++) {\r
+ tmin = tree_min_len(mem_env[backs[i]].node, env);\r
+ if (len > tmin) len = tmin;\r
+ }\r
+ }\r
+ break;\r
\r
- type = NTYPE(node);\r
- switch (type) {\r
- case NT_LIST:\r
+#ifdef USE_CALL\r
+ case NODE_CALL:\r
{\r
- Node *x;\r
- OnigDistance min;\r
- int ret;\r
-\r
- x = node;\r
- do {\r
- ret = subexp_inf_recursive_check(NCAR(x), env, head);\r
- if (ret < 0 || ret == RECURSION_INFINITE) return ret;\r
- r |= ret;\r
- if (head) {\r
- ret = get_min_match_length(NCAR(x), &min, env);\r
- if (ret != 0) return ret;\r
- if (min != 0) head = 0;\r
- }\r
- } while (IS_NOT_NULL(x = NCDR(x)));\r
+ Node* t = NODE_BODY(node);\r
+ if (NODE_IS_RECURSION(node)) {\r
+ if (NODE_IS_MIN_FIXED(t))\r
+ len = ENCLOSURE_(t)->min_len;\r
+ }\r
+ else\r
+ len = tree_min_len(t, env);\r
}\r
break;\r
+#endif\r
+\r
+ case NODE_LIST:\r
+ do {\r
+ tmin = tree_min_len(NODE_CAR(node), env);\r
+ len = distance_add(len, tmin);\r
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));\r
+ break;\r
\r
- case NT_ALT:\r
+ case NODE_ALT:\r
{\r
- int ret;\r
- r = RECURSION_EXIST;\r
+ Node *x, *y;\r
+ y = node;\r
do {\r
- ret = subexp_inf_recursive_check(NCAR(node), env, head);\r
- if (ret < 0 || ret == RECURSION_INFINITE) return ret;\r
- r &= ret;\r
- } while (IS_NOT_NULL(node = NCDR(node)));\r
+ x = NODE_CAR(y);\r
+ tmin = tree_min_len(x, env);\r
+ if (y == node) len = tmin;\r
+ else if (len > tmin) len = tmin;\r
+ } while (IS_NOT_NULL(y = NODE_CDR(y)));\r
}\r
break;\r
\r
- case NT_QTFR:\r
- r = subexp_inf_recursive_check(NQTFR(node)->target, env, head);\r
- if (r == RECURSION_EXIST) {\r
- if (NQTFR(node)->lower == 0) r = 0;\r
+ case NODE_STRING:\r
+ {\r
+ StrNode* sn = STR_(node);\r
+ len = (int )(sn->end - sn->s);\r
}\r
break;\r
\r
- case NT_ANCHOR:\r
+ case NODE_CTYPE:\r
+ case NODE_CCLASS:\r
+ len = ONIGENC_MBC_MINLEN(env->enc);\r
+ break;\r
+\r
+ case NODE_QUANT:\r
{\r
- AnchorNode* an = NANCHOR(node);\r
- switch (an->type) {\r
- case ANCHOR_PREC_READ:\r
- case ANCHOR_PREC_READ_NOT:\r
- case ANCHOR_LOOK_BEHIND:\r
- case ANCHOR_LOOK_BEHIND_NOT:\r
- r = subexp_inf_recursive_check(an->target, env, head);\r
- break;\r
+ QuantNode* qn = QUANT_(node);\r
+\r
+ if (qn->lower > 0) {\r
+ len = tree_min_len(NODE_BODY(node), env);\r
+ len = distance_multiply(len, qn->lower);\r
}\r
}\r
break;\r
\r
- case NT_CALL:\r
- r = subexp_inf_recursive_check(NCALL(node)->target, env, head);\r
- break;\r
+ case NODE_ENCLOSURE:\r
+ {\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+ switch (en->type) {\r
+ case ENCLOSURE_MEMORY:\r
+ if (NODE_IS_MIN_FIXED(node))\r
+ len = en->min_len;\r
+ else {\r
+ if (NODE_IS_MARK1(node))\r
+ len = 0; /* recursive */\r
+ else {\r
+ NODE_STATUS_ADD(node, MARK1);\r
+ len = tree_min_len(NODE_BODY(node), env);\r
+ NODE_STATUS_REMOVE(node, MARK1);\r
\r
- case NT_ENCLOSE:\r
- if (IS_ENCLOSE_MARK2(NENCLOSE(node)))\r
- return 0;\r
- else if (IS_ENCLOSE_MARK1(NENCLOSE(node)))\r
- return (head == 0 ? RECURSION_EXIST : RECURSION_INFINITE);\r
- else {\r
- SET_ENCLOSE_STATUS(node, NST_MARK2);\r
- r = subexp_inf_recursive_check(NENCLOSE(node)->target, env, head);\r
- CLEAR_ENCLOSE_STATUS(node, NST_MARK2);\r
+ en->min_len = len;\r
+ NODE_STATUS_ADD(node, MIN_FIXED);\r
+ }\r
+ }\r
+ break;\r
+\r
+ case ENCLOSURE_OPTION:\r
+ case ENCLOSURE_STOP_BACKTRACK:\r
+ len = tree_min_len(NODE_BODY(node), env);\r
+ break;\r
+ case ENCLOSURE_IF_ELSE:\r
+ {\r
+ OnigLen elen;\r
+\r
+ len = tree_min_len(NODE_BODY(node), env);\r
+ if (IS_NOT_NULL(en->te.Then))\r
+ len += tree_min_len(en->te.Then, env);\r
+ if (IS_NOT_NULL(en->te.Else))\r
+ elen = tree_min_len(en->te.Else, env);\r
+ else elen = 0;\r
+\r
+ if (elen < len) len = elen;\r
+ }\r
+ break;\r
+ }\r
}\r
break;\r
\r
+ case NODE_GIMMICK:\r
+ {\r
+ GimmickNode* g = GIMMICK_(node);\r
+ if (g->type == GIMMICK_FAIL) {\r
+ len = INFINITE_LEN;\r
+ break;\r
+ }\r
+ }\r
+ /* fall */\r
+ case NODE_ANCHOR:\r
default:\r
break;\r
}\r
\r
- return r;\r
+ return len;\r
}\r
\r
-static int\r
-subexp_inf_recursive_check_trav(Node* node, ScanEnv* env)\r
+static OnigLen\r
+tree_max_len(Node* node, ScanEnv* env)\r
{\r
- int type;\r
- int r = 0;\r
+ OnigLen len;\r
+ OnigLen tmax;\r
\r
- type = NTYPE(node);\r
- switch (type) {\r
- case NT_LIST:\r
- case NT_ALT:\r
+ len = 0;\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
do {\r
- r = subexp_inf_recursive_check_trav(NCAR(node), env);\r
- } while (r == 0 && IS_NOT_NULL(node = NCDR(node)));\r
+ tmax = tree_max_len(NODE_CAR(node), env);\r
+ len = distance_add(len, tmax);\r
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));\r
break;\r
\r
- case NT_QTFR:\r
- r = subexp_inf_recursive_check_trav(NQTFR(node)->target, env);\r
+ case NODE_ALT:\r
+ do {\r
+ tmax = tree_max_len(NODE_CAR(node), env);\r
+ if (len < tmax) len = tmax;\r
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));\r
break;\r
\r
- case NT_ANCHOR:\r
+ case NODE_STRING:\r
{\r
- AnchorNode* an = NANCHOR(node);\r
- switch (an->type) {\r
- case ANCHOR_PREC_READ:\r
- case ANCHOR_PREC_READ_NOT:\r
- case ANCHOR_LOOK_BEHIND:\r
- case ANCHOR_LOOK_BEHIND_NOT:\r
- r = subexp_inf_recursive_check_trav(an->target, env);\r
- break;\r
+ StrNode* sn = STR_(node);\r
+ len = (OnigLen )(sn->end - sn->s);\r
+ }\r
+ break;\r
+\r
+ case NODE_CTYPE:\r
+ case NODE_CCLASS:\r
+ len = ONIGENC_MBC_MAXLEN_DIST(env->enc);\r
+ break;\r
+\r
+ case NODE_BACKREF:\r
+ if (! NODE_IS_CHECKER(node)) {\r
+ int i;\r
+ int* backs;\r
+ MemEnv* mem_env = SCANENV_MEMENV(env);\r
+ BackRefNode* br = BACKREF_(node);\r
+ if (NODE_IS_RECURSION(node)) {\r
+ len = INFINITE_LEN;\r
+ break;\r
+ }\r
+ backs = BACKREFS_P(br);\r
+ for (i = 0; i < br->back_num; i++) {\r
+ tmax = tree_max_len(mem_env[backs[i]].node, env);\r
+ if (len < tmax) len = tmax;\r
}\r
}\r
break;\r
\r
- case NT_ENCLOSE:\r
+#ifdef USE_CALL\r
+ case NODE_CALL:\r
+ if (! NODE_IS_RECURSION(node))\r
+ len = tree_max_len(NODE_BODY(node), env);\r
+ else\r
+ len = INFINITE_LEN;\r
+ break;\r
+#endif\r
+\r
+ case NODE_QUANT:\r
{\r
- EncloseNode* en = NENCLOSE(node);\r
+ QuantNode* qn = QUANT_(node);\r
\r
- if (IS_ENCLOSE_RECURSION(en)) {\r
- SET_ENCLOSE_STATUS(node, NST_MARK1);\r
- r = subexp_inf_recursive_check(en->target, env, 1);\r
- if (r > 0) return ONIGERR_NEVER_ENDING_RECURSION;\r
- CLEAR_ENCLOSE_STATUS(node, NST_MARK1);\r
+ if (qn->upper != 0) {\r
+ len = tree_max_len(NODE_BODY(node), env);\r
+ if (len != 0) {\r
+ if (! IS_REPEAT_INFINITE(qn->upper))\r
+ len = distance_multiply(len, qn->upper);\r
+ else\r
+ len = INFINITE_LEN;\r
+ }\r
}\r
- r = subexp_inf_recursive_check_trav(en->target, env);\r
}\r
+ break;\r
+\r
+ case NODE_ENCLOSURE:\r
+ {\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+ switch (en->type) {\r
+ case ENCLOSURE_MEMORY:\r
+ if (NODE_IS_MAX_FIXED(node))\r
+ len = en->max_len;\r
+ else {\r
+ if (NODE_IS_MARK1(node))\r
+ len = INFINITE_LEN;\r
+ else {\r
+ NODE_STATUS_ADD(node, MARK1);\r
+ len = tree_max_len(NODE_BODY(node), env);\r
+ NODE_STATUS_REMOVE(node, MARK1);\r
+\r
+ en->max_len = len;\r
+ NODE_STATUS_ADD(node, MAX_FIXED);\r
+ }\r
+ }\r
+ break;\r
+\r
+ case ENCLOSURE_OPTION:\r
+ case ENCLOSURE_STOP_BACKTRACK:\r
+ len = tree_max_len(NODE_BODY(node), env);\r
+ break;\r
+ case ENCLOSURE_IF_ELSE:\r
+ {\r
+ OnigLen tlen, elen;\r
+\r
+ len = tree_max_len(NODE_BODY(node), env);\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ tlen = tree_max_len(en->te.Then, env);\r
+ len = distance_add(len, tlen);\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else))\r
+ elen = tree_max_len(en->te.Else, env);\r
+ else elen = 0;\r
\r
+ if (elen > len) len = elen;\r
+ }\r
+ break;\r
+ }\r
+ }\r
break;\r
\r
+ case NODE_ANCHOR:\r
+ case NODE_GIMMICK:\r
default:\r
break;\r
}\r
\r
- return r;\r
+ return len;\r
}\r
\r
static int\r
-subexp_recursive_check(Node* node)\r
+check_backrefs(Node* node, ScanEnv* env)\r
{\r
- int r = 0;\r
+ int r;\r
\r
- switch (NTYPE(node)) {\r
- case NT_LIST:\r
- case NT_ALT:\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
+ case NODE_ALT:\r
do {\r
- r |= subexp_recursive_check(NCAR(node));\r
- } while (IS_NOT_NULL(node = NCDR(node)));\r
+ r = check_backrefs(NODE_CAR(node), env);\r
+ } while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));\r
break;\r
\r
- case NT_QTFR:\r
- r = subexp_recursive_check(NQTFR(node)->target);\r
+ case NODE_ANCHOR:\r
+ if (! ANCHOR_HAS_BODY(ANCHOR_(node))) {\r
+ r = 0;\r
+ break;\r
+ }\r
+ /* fall */\r
+ case NODE_QUANT:\r
+ r = check_backrefs(NODE_BODY(node), env);\r
break;\r
\r
- case NT_ANCHOR:\r
+ case NODE_ENCLOSURE:\r
+ r = check_backrefs(NODE_BODY(node), env);\r
{\r
- AnchorNode* an = NANCHOR(node);\r
- switch (an->type) {\r
- case ANCHOR_PREC_READ:\r
- case ANCHOR_PREC_READ_NOT:\r
- case ANCHOR_LOOK_BEHIND:\r
- case ANCHOR_LOOK_BEHIND_NOT:\r
- r = subexp_recursive_check(an->target);\r
- break;\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+\r
+ if (en->type == ENCLOSURE_IF_ELSE) {\r
+ if (r != 0) return r;\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ r = check_backrefs(en->te.Then, env);\r
+ if (r != 0) return r;\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else)) {\r
+ r = check_backrefs(en->te.Else, env);\r
+ }\r
}\r
}\r
break;\r
\r
- case NT_CALL:\r
- r = subexp_recursive_check(NCALL(node)->target);\r
- if (r != 0) SET_CALL_RECURSION(node);\r
- break;\r
+ case NODE_BACKREF:\r
+ {\r
+ int i;\r
+ BackRefNode* br = BACKREF_(node);\r
+ int* backs = BACKREFS_P(br);\r
+ MemEnv* mem_env = SCANENV_MEMENV(env);\r
\r
- case NT_ENCLOSE:\r
- if (IS_ENCLOSE_MARK2(NENCLOSE(node)))\r
- return 0;\r
- else if (IS_ENCLOSE_MARK1(NENCLOSE(node)))\r
- return 1; /* recursion */\r
- else {\r
- SET_ENCLOSE_STATUS(node, NST_MARK2);\r
- r = subexp_recursive_check(NENCLOSE(node)->target);\r
- CLEAR_ENCLOSE_STATUS(node, NST_MARK2);\r
+ for (i = 0; i < br->back_num; i++) {\r
+ if (backs[i] > env->num_mem)\r
+ return ONIGERR_INVALID_BACKREF;\r
+\r
+ NODE_STATUS_ADD(mem_env[backs[i]].node, BACKREF);\r
+ }\r
+ r = 0;\r
}\r
break;\r
\r
default:\r
+ r = 0;\r
break;\r
}\r
\r
}\r
\r
\r
+#ifdef USE_CALL\r
+\r
+#define RECURSION_EXIST (1<<0)\r
+#define RECURSION_MUST (1<<1)\r
+#define RECURSION_INFINITE (1<<2)\r
+\r
static int\r
-subexp_recursive_check_trav(Node* node, ScanEnv* env)\r
+infinite_recursive_call_check(Node* node, ScanEnv* env, int head)\r
{\r
-#define FOUND_CALLED_NODE 1\r
-\r
- int type;\r
+ int ret;\r
int r = 0;\r
\r
- type = NTYPE(node);\r
- switch (type) {\r
- case NT_LIST:\r
- case NT_ALT:\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
{\r
- int ret;\r
+ Node *x;\r
+ OnigLen min;\r
+\r
+ x = node;\r
do {\r
- ret = subexp_recursive_check_trav(NCAR(node), env);\r
- if (ret == FOUND_CALLED_NODE) r = FOUND_CALLED_NODE;\r
- else if (ret < 0) return ret;\r
- } while (IS_NOT_NULL(node = NCDR(node)));\r
+ ret = infinite_recursive_call_check(NODE_CAR(x), env, head);\r
+ if (ret < 0 || (ret & RECURSION_INFINITE) != 0) return ret;\r
+ r |= ret;\r
+ if (head != 0) {\r
+ min = tree_min_len(NODE_CAR(x), env);\r
+ if (min != 0) head = 0;\r
+ }\r
+ } while (IS_NOT_NULL(x = NODE_CDR(x)));\r
}\r
break;\r
\r
- case NT_QTFR:\r
- r = subexp_recursive_check_trav(NQTFR(node)->target, env);\r
- if (NQTFR(node)->upper == 0) {\r
- if (r == FOUND_CALLED_NODE)\r
- NQTFR(node)->is_refered = 1;\r
+ case NODE_ALT:\r
+ {\r
+ int must;\r
+\r
+ must = RECURSION_MUST;\r
+ do {\r
+ ret = infinite_recursive_call_check(NODE_CAR(node), env, head);\r
+ if (ret < 0 || (ret & RECURSION_INFINITE) != 0) return ret;\r
+\r
+ r |= (ret & RECURSION_EXIST);\r
+ must &= ret;\r
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));\r
+ r |= must;\r
}\r
break;\r
\r
- case NT_ANCHOR:\r
- {\r
- AnchorNode* an = NANCHOR(node);\r
- switch (an->type) {\r
- case ANCHOR_PREC_READ:\r
- case ANCHOR_PREC_READ_NOT:\r
- case ANCHOR_LOOK_BEHIND:\r
- case ANCHOR_LOOK_BEHIND_NOT:\r
- r = subexp_recursive_check_trav(an->target, env);\r
- break;\r
- }\r
+ case NODE_QUANT:\r
+ r = infinite_recursive_call_check(NODE_BODY(node), env, head);\r
+ if (r < 0) return r;\r
+ if ((r & RECURSION_MUST) != 0) {\r
+ if (QUANT_(node)->lower == 0)\r
+ r &= ~RECURSION_MUST;\r
}\r
break;\r
\r
- case NT_ENCLOSE:\r
+ case NODE_ANCHOR:\r
+ if (! ANCHOR_HAS_BODY(ANCHOR_(node)))\r
+ break;\r
+ /* fall */\r
+ case NODE_CALL:\r
+ r = infinite_recursive_call_check(NODE_BODY(node), env, head);\r
+ break;\r
+\r
+ case NODE_ENCLOSURE:\r
{\r
- EncloseNode* en = NENCLOSE(node);\r
-\r
- if (! IS_ENCLOSE_RECURSION(en)) {\r
- if (IS_ENCLOSE_CALLED(en)) {\r
- SET_ENCLOSE_STATUS(node, NST_MARK1);\r
- r = subexp_recursive_check(en->target);\r
- if (r != 0) SET_ENCLOSE_STATUS(node, NST_RECURSION);\r
- CLEAR_ENCLOSE_STATUS(node, NST_MARK1);\r
- }\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+\r
+ if (en->type == ENCLOSURE_MEMORY) {\r
+ if (NODE_IS_MARK2(node))\r
+ return 0;\r
+ else if (NODE_IS_MARK1(node))\r
+ return (head == 0 ? RECURSION_EXIST | RECURSION_MUST\r
+ : RECURSION_EXIST | RECURSION_MUST | RECURSION_INFINITE);\r
+ else {\r
+ NODE_STATUS_ADD(node, MARK2);\r
+ r = infinite_recursive_call_check(NODE_BODY(node), env, head);\r
+ NODE_STATUS_REMOVE(node, MARK2);\r
+ }\r
+ }\r
+ else if (en->type == ENCLOSURE_IF_ELSE) {\r
+ int eret;\r
+\r
+ ret = infinite_recursive_call_check(NODE_BODY(node), env, head);\r
+ if (ret < 0 || (ret & RECURSION_INFINITE) != 0) return ret;\r
+ r |= ret;\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ OnigLen min;\r
+ if (head != 0) {\r
+ min = tree_min_len(NODE_BODY(node), env);\r
+ }\r
+ else min = 0;\r
+\r
+ ret = infinite_recursive_call_check(en->te.Then, env, min != 0 ? 0:head);\r
+ if (ret < 0 || (ret & RECURSION_INFINITE) != 0) return ret;\r
+ r |= ret;\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else)) {\r
+ eret = infinite_recursive_call_check(en->te.Else, env, head);\r
+ if (eret < 0 || (eret & RECURSION_INFINITE) != 0) return eret;\r
+ r |= (eret & RECURSION_EXIST);\r
+ if ((eret & RECURSION_MUST) == 0)\r
+ r &= ~RECURSION_MUST;\r
+ }\r
+ }\r
+ else {\r
+ r = infinite_recursive_call_check(NODE_BODY(node), env, head);\r
}\r
- r = subexp_recursive_check_trav(en->target, env);\r
- if (IS_ENCLOSE_CALLED(en))\r
- r |= FOUND_CALLED_NODE;\r
}\r
break;\r
\r
}\r
\r
static int\r
-setup_subexp_call(Node* node, ScanEnv* env)\r
+infinite_recursive_call_check_trav(Node* node, ScanEnv* env)\r
{\r
- int type;\r
- int r = 0;\r
+ int r;\r
\r
- type = NTYPE(node);\r
- switch (type) {\r
- case NT_LIST:\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
+ case NODE_ALT:\r
do {\r
- r = setup_subexp_call(NCAR(node), env);\r
- } while (r == 0 && IS_NOT_NULL(node = NCDR(node)));\r
+ r = infinite_recursive_call_check_trav(NODE_CAR(node), env);\r
+ } while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));\r
+ break;\r
+\r
+ case NODE_ANCHOR:\r
+ if (! ANCHOR_HAS_BODY(ANCHOR_(node))) {\r
+ r = 0;\r
+ break;\r
+ }\r
+ /* fall */\r
+ case NODE_QUANT:\r
+ r = infinite_recursive_call_check_trav(NODE_BODY(node), env);\r
+ break;\r
+\r
+ case NODE_ENCLOSURE:\r
+ {\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+\r
+ if (en->type == ENCLOSURE_MEMORY) {\r
+ if (NODE_IS_RECURSION(node) && NODE_IS_CALLED(node)) {\r
+ int ret;\r
+\r
+ NODE_STATUS_ADD(node, MARK1);\r
+\r
+ ret = infinite_recursive_call_check(NODE_BODY(node), env, 1);\r
+ if (ret < 0) return ret;\r
+ else if ((ret & (RECURSION_MUST | RECURSION_INFINITE)) != 0)\r
+ return ONIGERR_NEVER_ENDING_RECURSION;\r
+\r
+ NODE_STATUS_REMOVE(node, MARK1);\r
+ }\r
+ }\r
+ else if (en->type == ENCLOSURE_IF_ELSE) {\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ r = infinite_recursive_call_check_trav(en->te.Then, env);\r
+ if (r != 0) return r;\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else)) {\r
+ r = infinite_recursive_call_check_trav(en->te.Else, env);\r
+ if (r != 0) return r;\r
+ }\r
+ }\r
+ }\r
+\r
+ r = infinite_recursive_call_check_trav(NODE_BODY(node), env);\r
+ break;\r
+\r
+ default:\r
+ r = 0;\r
break;\r
+ }\r
+\r
+ return r;\r
+}\r
+\r
+static int\r
+recursive_call_check(Node* node)\r
+{\r
+ int r;\r
\r
- case NT_ALT:\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
+ case NODE_ALT:\r
+ r = 0;\r
do {\r
- r = setup_subexp_call(NCAR(node), env);\r
- } while (r == 0 && IS_NOT_NULL(node = NCDR(node)));\r
+ r |= recursive_call_check(NODE_CAR(node));\r
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));\r
break;\r
\r
- case NT_QTFR:\r
- r = setup_subexp_call(NQTFR(node)->target, env);\r
+ case NODE_ANCHOR:\r
+ if (! ANCHOR_HAS_BODY(ANCHOR_(node))) {\r
+ r = 0;\r
+ break;\r
+ }\r
+ /* fall */\r
+ case NODE_QUANT:\r
+ r = recursive_call_check(NODE_BODY(node));\r
break;\r
- case NT_ENCLOSE:\r
- r = setup_subexp_call(NENCLOSE(node)->target, env);\r
+\r
+ case NODE_CALL:\r
+ r = recursive_call_check(NODE_BODY(node));\r
+ if (r != 0) {\r
+ if (NODE_IS_MARK1(NODE_BODY(node)))\r
+ NODE_STATUS_ADD(node, RECURSION);\r
+ }\r
break;\r
\r
- case NT_CALL:\r
+ case NODE_ENCLOSURE:\r
{\r
- CallNode* cn = NCALL(node);\r
- Node** nodes = SCANENV_MEM_NODES(env);\r
-\r
- if (cn->group_num != 0) {\r
- int gnum = cn->group_num;\r
-\r
-#ifdef USE_NAMED_GROUP\r
- if (env->num_named > 0 &&\r
- IS_SYNTAX_BV(env->syntax, ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP) &&\r
- !ONIG_IS_OPTION_ON(env->option, ONIG_OPTION_CAPTURE_GROUP)) {\r
- return ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED;\r
- }\r
-#endif\r
- if (gnum > env->num_mem) {\r
- onig_scan_env_set_error_string(env,\r
- ONIGERR_UNDEFINED_GROUP_REFERENCE, cn->name, cn->name_end);\r
- return ONIGERR_UNDEFINED_GROUP_REFERENCE;\r
- }\r
-\r
-#ifdef USE_NAMED_GROUP\r
- set_call_attr:\r
-#endif\r
- cn->target = nodes[cn->group_num];\r
- if (IS_NULL(cn->target)) {\r
- onig_scan_env_set_error_string(env,\r
- ONIGERR_UNDEFINED_NAME_REFERENCE, cn->name, cn->name_end);\r
- return ONIGERR_UNDEFINED_NAME_REFERENCE;\r
- }\r
- SET_ENCLOSE_STATUS(cn->target, NST_CALLED);\r
- BIT_STATUS_ON_AT(env->bt_mem_start, cn->group_num);\r
- cn->unset_addr_list = env->unset_addr_list;\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+\r
+ if (en->type == ENCLOSURE_MEMORY) {\r
+ if (NODE_IS_MARK2(node))\r
+ return 0;\r
+ else if (NODE_IS_MARK1(node))\r
+ return 1; /* recursion */\r
+ else {\r
+ NODE_STATUS_ADD(node, MARK2);\r
+ r = recursive_call_check(NODE_BODY(node));\r
+ NODE_STATUS_REMOVE(node, MARK2);\r
+ }\r
+ }\r
+ else if (en->type == ENCLOSURE_IF_ELSE) {\r
+ r = 0;\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ r |= recursive_call_check(en->te.Then);\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else)) {\r
+ r |= recursive_call_check(en->te.Else);\r
+ }\r
+ r |= recursive_call_check(NODE_BODY(node));\r
}\r
-#ifdef USE_NAMED_GROUP\r
else {\r
- int *refs;\r
-\r
- int n = onig_name_to_group_numbers(env->reg, cn->name, cn->name_end,\r
- &refs);\r
- if (n <= 0) {\r
- onig_scan_env_set_error_string(env,\r
- ONIGERR_UNDEFINED_NAME_REFERENCE, cn->name, cn->name_end);\r
- return ONIGERR_UNDEFINED_NAME_REFERENCE;\r
- }\r
- else if (n > 1) {\r
- onig_scan_env_set_error_string(env,\r
- ONIGERR_MULTIPLEX_DEFINITION_NAME_CALL, cn->name, cn->name_end);\r
- return ONIGERR_MULTIPLEX_DEFINITION_NAME_CALL;\r
- }\r
- else {\r
- cn->group_num = refs[0];\r
- goto set_call_attr;\r
- }\r
+ r = recursive_call_check(NODE_BODY(node));\r
}\r
-#endif\r
}\r
break;\r
\r
- case NT_ANCHOR:\r
+ default:\r
+ r = 0;\r
+ break;\r
+ }\r
+\r
+ return r;\r
+}\r
+\r
+#define IN_RECURSION (1<<0)\r
+#define FOUND_CALLED_NODE 1\r
+\r
+static int\r
+recursive_call_check_trav(Node* node, ScanEnv* env, int state)\r
+{\r
+ int r = 0;\r
+\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
+ case NODE_ALT:\r
{\r
- AnchorNode* an = NANCHOR(node);\r
+ int ret;\r
+ do {\r
+ ret = recursive_call_check_trav(NODE_CAR(node), env, state);\r
+ if (ret == FOUND_CALLED_NODE) r = FOUND_CALLED_NODE;\r
+ else if (ret < 0) return ret;\r
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));\r
+ }\r
+ break;\r
\r
- switch (an->type) {\r
- case ANCHOR_PREC_READ:\r
- case ANCHOR_PREC_READ_NOT:\r
- case ANCHOR_LOOK_BEHIND:\r
- case ANCHOR_LOOK_BEHIND_NOT:\r
- r = setup_subexp_call(an->target, env);\r
- break;\r
+ case NODE_QUANT:\r
+ r = recursive_call_check_trav(NODE_BODY(node), env, state);\r
+ if (QUANT_(node)->upper == 0) {\r
+ if (r == FOUND_CALLED_NODE)\r
+ QUANT_(node)->is_refered = 1;\r
+ }\r
+ break;\r
+\r
+ case NODE_ANCHOR:\r
+ {\r
+ AnchorNode* an = ANCHOR_(node);\r
+ if (ANCHOR_HAS_BODY(an))\r
+ r = recursive_call_check_trav(NODE_ANCHOR_BODY(an), env, state);\r
+ }\r
+ break;\r
+\r
+ case NODE_ENCLOSURE:\r
+ {\r
+ int ret;\r
+ int state1;\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+\r
+ if (en->type == ENCLOSURE_MEMORY) {\r
+ if (NODE_IS_CALLED(node) || (state & IN_RECURSION) != 0) {\r
+ if (! NODE_IS_RECURSION(node)) {\r
+ NODE_STATUS_ADD(node, MARK1);\r
+ r = recursive_call_check(NODE_BODY(node));\r
+ if (r != 0)\r
+ NODE_STATUS_ADD(node, RECURSION);\r
+ NODE_STATUS_REMOVE(node, MARK1);\r
+ }\r
+\r
+ if (NODE_IS_CALLED(node))\r
+ r = FOUND_CALLED_NODE;\r
+ }\r
+ }\r
+\r
+ state1 = state;\r
+ if (NODE_IS_RECURSION(node))\r
+ state1 |= IN_RECURSION;\r
+\r
+ ret = recursive_call_check_trav(NODE_BODY(node), env, state1);\r
+ if (ret == FOUND_CALLED_NODE)\r
+ r = FOUND_CALLED_NODE;\r
+\r
+ if (en->type == ENCLOSURE_IF_ELSE) {\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ ret = recursive_call_check_trav(en->te.Then, env, state1);\r
+ if (ret == FOUND_CALLED_NODE)\r
+ r = FOUND_CALLED_NODE;\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else)) {\r
+ ret = recursive_call_check_trav(en->te.Else, env, state1);\r
+ if (ret == FOUND_CALLED_NODE)\r
+ r = FOUND_CALLED_NODE;\r
+ }\r
}\r
}\r
break;\r
\r
return r;\r
}\r
+\r
#endif\r
\r
/* divide different length alternatives in look-behind.\r
divide_look_behind_alternatives(Node* node)\r
{\r
Node *head, *np, *insert_node;\r
- AnchorNode* an = NANCHOR(node);\r
+ AnchorNode* an = ANCHOR_(node);\r
int anc_type = an->type;\r
\r
- head = an->target;\r
- np = NCAR(head);\r
+ head = NODE_ANCHOR_BODY(an);\r
+ np = NODE_CAR(head);\r
swap_node(node, head);\r
- NCAR(node) = head;\r
- NANCHOR(head)->target = np;\r
+ NODE_CAR(node) = head;\r
+ NODE_BODY(head) = np;\r
\r
np = node;\r
- while ((np = NCDR(np)) != NULL_NODE) {\r
- insert_node = onig_node_new_anchor(anc_type);\r
+ while (IS_NOT_NULL(np = NODE_CDR(np))) {\r
+ insert_node = onig_node_new_anchor(anc_type, an->ascii_mode);\r
CHECK_NULL_RETURN_MEMERR(insert_node);\r
- NANCHOR(insert_node)->target = NCAR(np);\r
- NCAR(np) = insert_node;\r
+ NODE_BODY(insert_node) = NODE_CAR(np);\r
+ NODE_CAR(np) = insert_node;\r
}\r
\r
if (anc_type == ANCHOR_LOOK_BEHIND_NOT) {\r
np = node;\r
do {\r
- SET_NTYPE(np, NT_LIST); /* alt -> list */\r
- } while ((np = NCDR(np)) != NULL_NODE);\r
+ NODE_SET_TYPE(np, NODE_LIST); /* alt -> list */\r
+ } while (IS_NOT_NULL(np = NODE_CDR(np)));\r
}\r
return 0;\r
}\r
setup_look_behind(Node* node, regex_t* reg, ScanEnv* env)\r
{\r
int r, len;\r
- AnchorNode* an = NANCHOR(node);\r
+ AnchorNode* an = ANCHOR_(node);\r
\r
- r = get_char_length_tree(an->target, reg, &len);\r
+ r = get_char_length_tree(NODE_ANCHOR_BODY(an), reg, &len);\r
if (r == 0)\r
an->char_len = len;\r
else if (r == GET_CHAR_LEN_VARLEN)\r
static int\r
next_setup(Node* node, Node* next_node, regex_t* reg)\r
{\r
- int type;\r
+ NodeType type;\r
\r
retry:\r
- type = NTYPE(node);\r
- if (type == NT_QTFR) {\r
- QtfrNode* qn = NQTFR(node);\r
+ type = NODE_TYPE(node);\r
+ if (type == NODE_QUANT) {\r
+ QuantNode* qn = QUANT_(node);\r
if (qn->greedy && IS_REPEAT_INFINITE(qn->upper)) {\r
-#ifdef USE_QTFR_PEEK_NEXT\r
+#ifdef USE_QUANT_PEEK_NEXT\r
Node* n = get_head_value_node(next_node, 1, reg);\r
/* '\0': for UTF-16BE etc... */\r
- if (IS_NOT_NULL(n) && NSTR(n)->s[0] != '\0') {\r
- qn->next_head_exact = n;\r
+ if (IS_NOT_NULL(n) && STR_(n)->s[0] != '\0') {\r
+ qn->next_head_exact = n;\r
}\r
#endif\r
/* automatic posseivation a*b ==> (?>a*)b */\r
if (qn->lower <= 1) {\r
- int ttype = NTYPE(qn->target);\r
- if (IS_NODE_TYPE_SIMPLE(ttype)) {\r
- Node *x, *y;\r
- x = get_head_value_node(qn->target, 0, reg);\r
- if (IS_NOT_NULL(x)) {\r
- y = get_head_value_node(next_node, 0, reg);\r
- if (IS_NOT_NULL(y) && is_not_included(x, y, reg)) {\r
- Node* en = onig_node_new_enclose(ENCLOSE_STOP_BACKTRACK);\r
- CHECK_NULL_RETURN_MEMERR(en);\r
- SET_ENCLOSE_STATUS(en, NST_STOP_BT_SIMPLE_REPEAT);\r
- swap_node(node, en);\r
- NENCLOSE(node)->target = en;\r
- }\r
- }\r
- }\r
+ if (NODE_IS_SIMPLE_TYPE(NODE_BODY(node))) {\r
+ Node *x, *y;\r
+ x = get_head_value_node(NODE_BODY(node), 0, reg);\r
+ if (IS_NOT_NULL(x)) {\r
+ y = get_head_value_node(next_node, 0, reg);\r
+ if (IS_NOT_NULL(y) && is_exclusive(x, y, reg)) {\r
+ Node* en = onig_node_new_enclosure(ENCLOSURE_STOP_BACKTRACK);\r
+ CHECK_NULL_RETURN_MEMERR(en);\r
+ NODE_STATUS_ADD(en, STOP_BT_SIMPLE_REPEAT);\r
+ swap_node(node, en);\r
+ NODE_BODY(node) = en;\r
+ }\r
+ }\r
+ }\r
}\r
}\r
}\r
- else if (type == NT_ENCLOSE) {\r
- EncloseNode* en = NENCLOSE(node);\r
- if (en->type == ENCLOSE_MEMORY) {\r
- node = en->target;\r
+ else if (type == NODE_ENCLOSURE) {\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+ if (en->type == ENCLOSURE_MEMORY) {\r
+ node = NODE_BODY(node);\r
goto retry;\r
}\r
}\r
UChar *p, *end, buf[ONIGENC_MBC_CASE_FOLD_MAXLEN];\r
UChar *sbuf, *ebuf, *sp;\r
int r, i, len, sbuf_size;\r
- StrNode* sn = NSTR(node);\r
+ StrNode* sn = STR_(node);\r
\r
end = sn->end;\r
- sbuf_size = (int)(end - sn->s) * 2;\r
+ sbuf_size = (int )(end - sn->s) * 2;\r
sbuf = (UChar* )xmalloc(sbuf_size);\r
CHECK_NULL_RETURN_MEMERR(sbuf);\r
ebuf = sbuf + sbuf_size;\r
}\r
\r
static int\r
-expand_case_fold_make_rem_string(Node** rnode, UChar *s, UChar *end,\r
- regex_t* reg)\r
+expand_case_fold_make_rem_string(Node** rnode, UChar *s, UChar *end, regex_t* reg)\r
{\r
int r;\r
Node *node;\r
return r;\r
}\r
\r
- NSTRING_SET_AMBIG(node);\r
- NSTRING_SET_DONT_GET_OPT_INFO(node);\r
+ NODE_STRING_SET_AMBIG(node);\r
+ NODE_STRING_SET_DONT_GET_OPT_INFO(node);\r
*rnode = node;\r
return 0;\r
}\r
\r
static int\r
-expand_case_fold_string_alt(int item_num, OnigCaseFoldCodeItem items[],\r
- UChar *p, int slen, UChar *end,\r
- regex_t* reg, Node **rnode)\r
+expand_case_fold_string_alt(int item_num, OnigCaseFoldCodeItem items[], UChar *p,\r
+ int slen, UChar *end, regex_t* reg, Node **rnode)\r
{\r
- int r, i, j, len, varlen;\r
+ int r, i, j;\r
+ int len;\r
+ int varlen;\r
Node *anode, *var_anode, *snode, *xnode, *an;\r
UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];\r
- xnode = NULL_NODE;\r
\r
*rnode = var_anode = NULL_NODE;\r
\r
\r
xnode = onig_node_new_list(NULL, NULL);\r
if (IS_NULL(xnode)) goto mem_err;\r
- NCAR(var_anode) = xnode;\r
+ NODE_CAR(var_anode) = xnode;\r
\r
anode = onig_node_new_alt(NULL_NODE, NULL_NODE);\r
if (IS_NULL(anode)) goto mem_err;\r
- NCAR(xnode) = anode;\r
+ NODE_CAR(xnode) = anode;\r
}\r
else {\r
*rnode = anode = onig_node_new_alt(NULL_NODE, NULL_NODE);\r
snode = onig_node_new_str(p, p + slen);\r
if (IS_NULL(snode)) goto mem_err;\r
\r
- NCAR(anode) = snode;\r
+ NODE_CAR(anode) = snode;\r
\r
for (i = 0; i < item_num; i++) {\r
snode = onig_node_new_str(NULL, NULL);\r
for (j = 0; j < items[i].code_len; j++) {\r
len = ONIGENC_CODE_TO_MBC(reg->enc, items[i].code[j], buf);\r
if (len < 0) {\r
- r = len;\r
- goto mem_err2;\r
+ r = len;\r
+ goto mem_err2;\r
}\r
\r
r = onig_node_str_cat(snode, buf, buf + len);\r
}\r
\r
if (items[i].byte_len != slen) {\r
- Node *rem = NULL_NODE;\r
+ Node *rem;\r
UChar *q = p + items[i].byte_len;\r
\r
if (q < end) {\r
goto mem_err;\r
}\r
\r
- NCAR(an) = xnode;\r
+ NODE_CAR(an) = xnode;\r
}\r
else {\r
- NCAR(an) = snode;\r
+ NODE_CAR(an) = snode;\r
}\r
\r
- if (var_anode == NULL) {\r
- onig_node_free(an);\r
- onig_node_free(xnode);\r
- onig_node_free(rem);\r
- goto mem_err2;\r
- }\r
- NCDR(var_anode) = an;\r
+ NODE_CDR(var_anode) = an;\r
var_anode = an;\r
}\r
else {\r
- NCAR(an) = snode;\r
- NCDR(anode) = an;\r
+ NODE_CAR(an) = snode;\r
+ NODE_CDR(anode) = an;\r
anode = an;\r
}\r
}\r
UChar *start, *end, *p;\r
Node *top_root, *root, *snode, *prev_node;\r
OnigCaseFoldCodeItem items[ONIGENC_GET_CASE_FOLD_CODES_MAX_NUM];\r
- StrNode* sn = NSTR(node);\r
+ StrNode* sn = STR_(node);\r
\r
- if (NSTRING_IS_AMBIG(node)) return 0;\r
+ if (NODE_STRING_IS_AMBIG(node)) return 0;\r
\r
start = sn->s;\r
end = sn->end;\r
alt_num = 1;\r
p = start;\r
while (p < end) {\r
- n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(reg->enc, reg->case_fold_flag,\r
- p, end, items);\r
+ n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(reg->enc, reg->case_fold_flag, p, end,\r
+ items);\r
if (n < 0) {\r
r = n;\r
goto err;\r
\r
if (n == 0) {\r
if (IS_NULL(snode)) {\r
- if (IS_NULL(root) && IS_NOT_NULL(prev_node)) {\r
- top_root = root = onig_node_list_add(NULL_NODE, prev_node);\r
- if (IS_NULL(root)) {\r
- onig_node_free(prev_node);\r
- goto mem_err;\r
- }\r
- }\r
-\r
- prev_node = snode = onig_node_new_str(NULL, NULL);\r
- if (IS_NULL(snode)) goto mem_err;\r
- if (IS_NOT_NULL(root)) {\r
- if (IS_NULL(onig_node_list_add(root, snode))) {\r
- onig_node_free(snode);\r
- goto mem_err;\r
- }\r
- }\r
+ if (IS_NULL(root) && IS_NOT_NULL(prev_node)) {\r
+ top_root = root = onig_node_list_add(NULL_NODE, prev_node);\r
+ if (IS_NULL(root)) {\r
+ onig_node_free(prev_node);\r
+ goto mem_err;\r
+ }\r
+ }\r
+\r
+ prev_node = snode = onig_node_new_str(NULL, NULL);\r
+ if (IS_NULL(snode)) goto mem_err;\r
+ if (IS_NOT_NULL(root)) {\r
+ if (IS_NULL(onig_node_list_add(root, snode))) {\r
+ onig_node_free(snode);\r
+ goto mem_err;\r
+ }\r
+ }\r
}\r
\r
r = onig_node_str_cat(snode, p, p + len);\r
if (alt_num > THRESHOLD_CASE_FOLD_ALT_FOR_EXPANSION) break;\r
\r
if (IS_NULL(root) && IS_NOT_NULL(prev_node)) {\r
- top_root = root = onig_node_list_add(NULL_NODE, prev_node);\r
- if (IS_NULL(root)) {\r
- onig_node_free(prev_node);\r
- goto mem_err;\r
- }\r
+ top_root = root = onig_node_list_add(NULL_NODE, prev_node);\r
+ if (IS_NULL(root)) {\r
+ onig_node_free(prev_node);\r
+ goto mem_err;\r
+ }\r
}\r
\r
r = expand_case_fold_string_alt(n, items, p, len, end, reg, &prev_node);\r
if (r < 0) goto mem_err;\r
if (r == 1) {\r
- if (IS_NULL(root)) {\r
- top_root = prev_node;\r
- }\r
- else {\r
- if (IS_NULL(onig_node_list_add(root, prev_node))) {\r
- onig_node_free(prev_node);\r
- goto mem_err;\r
- }\r
- }\r
-\r
- root = NCAR(prev_node);\r
+ if (IS_NULL(root)) {\r
+ top_root = prev_node;\r
+ }\r
+ else {\r
+ if (IS_NULL(onig_node_list_add(root, prev_node))) {\r
+ onig_node_free(prev_node);\r
+ goto mem_err;\r
+ }\r
+ }\r
+\r
+ root = NODE_CAR(prev_node);\r
}\r
else { /* r == 0 */\r
- if (IS_NOT_NULL(root)) {\r
- if (IS_NULL(onig_node_list_add(root, prev_node))) {\r
- onig_node_free(prev_node);\r
- goto mem_err;\r
- }\r
- }\r
+ if (IS_NOT_NULL(root)) {\r
+ if (IS_NULL(onig_node_list_add(root, prev_node))) {\r
+ onig_node_free(prev_node);\r
+ goto mem_err;\r
+ }\r
+ }\r
}\r
\r
snode = NULL_NODE;\r
if (IS_NOT_NULL(prev_node) && IS_NULL(root)) {\r
top_root = root = onig_node_list_add(NULL_NODE, prev_node);\r
if (IS_NULL(root)) {\r
- onig_node_free(srem);\r
- onig_node_free(prev_node);\r
- goto mem_err;\r
+ onig_node_free(srem);\r
+ onig_node_free(prev_node);\r
+ goto mem_err;\r
+ }\r
+ }\r
+\r
+ if (IS_NULL(root)) {\r
+ prev_node = srem;\r
+ }\r
+ else {\r
+ if (IS_NULL(onig_node_list_add(root, srem))) {\r
+ onig_node_free(srem);\r
+ goto mem_err;\r
+ }\r
+ }\r
+ }\r
+\r
+ /* ending */\r
+ top_root = (IS_NOT_NULL(top_root) ? top_root : prev_node);\r
+ swap_node(node, top_root);\r
+ onig_node_free(top_root);\r
+ return 0;\r
+\r
+ mem_err:\r
+ r = ONIGERR_MEMORY;\r
+\r
+ err:\r
+ onig_node_free(top_root);\r
+ return r;\r
+}\r
+\r
+#ifdef USE_INSISTENT_CHECK_CAPTURES_IN_EMPTY_REPEAT\r
+static enum QuantBodyEmpty\r
+quantifiers_memory_node_info(Node* node)\r
+{\r
+ int r = QUANT_BODY_IS_EMPTY;\r
+\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
+ case NODE_ALT:\r
+ {\r
+ int v;\r
+ do {\r
+ v = quantifiers_memory_node_info(NODE_CAR(node));\r
+ if (v > r) r = v;\r
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));\r
+ }\r
+ break;\r
+\r
+#ifdef USE_CALL\r
+ case NODE_CALL:\r
+ if (NODE_IS_RECURSION(node)) {\r
+ return QUANT_BODY_IS_EMPTY_REC; /* tiny version */\r
+ }\r
+ else\r
+ r = quantifiers_memory_node_info(NODE_BODY(node));\r
+ break;\r
+#endif\r
+\r
+ case NODE_QUANT:\r
+ {\r
+ QuantNode* qn = QUANT_(node);\r
+ if (qn->upper != 0) {\r
+ r = quantifiers_memory_node_info(NODE_BODY(node));\r
+ }\r
+ }\r
+ break;\r
+\r
+ case NODE_ENCLOSURE:\r
+ {\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+ switch (en->type) {\r
+ case ENCLOSURE_MEMORY:\r
+ if (NODE_IS_RECURSION(node)) {\r
+ return QUANT_BODY_IS_EMPTY_REC;\r
+ }\r
+ return QUANT_BODY_IS_EMPTY_MEM;\r
+ break;\r
+\r
+ case ENCLOSURE_OPTION:\r
+ case ENCLOSURE_STOP_BACKTRACK:\r
+ r = quantifiers_memory_node_info(NODE_BODY(node));\r
+ break;\r
+ case ENCLOSURE_IF_ELSE:\r
+ {\r
+ int v;\r
+ r = quantifiers_memory_node_info(NODE_BODY(node));\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ v = quantifiers_memory_node_info(en->te.Then);\r
+ if (v > r) r = v;\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else)) {\r
+ v = quantifiers_memory_node_info(en->te.Else);\r
+ if (v > r) r = v;\r
+ }\r
+ }\r
+ break;\r
+ default:\r
+ break;\r
+ }\r
+ }\r
+ break;\r
+\r
+ case NODE_BACKREF:\r
+ case NODE_STRING:\r
+ case NODE_CTYPE:\r
+ case NODE_CCLASS:\r
+ case NODE_ANCHOR:\r
+ case NODE_GIMMICK:\r
+ default:\r
+ break;\r
+ }\r
+\r
+ return r;\r
+}\r
+#endif /* USE_INSISTENT_CHECK_CAPTURES_IN_EMPTY_REPEAT */\r
+\r
+\r
+#define IN_ALT (1<<0)\r
+#define IN_NOT (1<<1)\r
+#define IN_REAL_REPEAT (1<<2)\r
+#define IN_VAR_REPEAT (1<<3)\r
+#define IN_ZERO_REPEAT (1<<4)\r
+#define IN_MULTI_ENTRY (1<<5)\r
+\r
+#ifdef USE_CALL\r
+\r
+#ifdef __GNUC__\r
+__inline\r
+#endif\r
+static int\r
+setup_call_node_call(CallNode* cn, ScanEnv* env, int state)\r
+{\r
+ MemEnv* mem_env = SCANENV_MEMENV(env);\r
+\r
+ if (cn->by_number != 0) {\r
+ int gnum = cn->group_num;\r
+\r
+ if (env->num_named > 0 &&\r
+ IS_SYNTAX_BV(env->syntax, ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP) &&\r
+ ! ONIG_IS_OPTION_ON(env->options, ONIG_OPTION_CAPTURE_GROUP)) {\r
+ return ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED;\r
+ }\r
+\r
+ if (gnum > env->num_mem) {\r
+ onig_scan_env_set_error_string(env, ONIGERR_UNDEFINED_GROUP_REFERENCE,\r
+ cn->name, cn->name_end);\r
+ return ONIGERR_UNDEFINED_GROUP_REFERENCE;\r
+ }\r
+\r
+ set_call_attr:\r
+ NODE_CALL_BODY(cn) = mem_env[cn->group_num].node;\r
+ if (IS_NULL(NODE_CALL_BODY(cn))) {\r
+ onig_scan_env_set_error_string(env, ONIGERR_UNDEFINED_NAME_REFERENCE,\r
+ cn->name, cn->name_end);\r
+ return ONIGERR_UNDEFINED_NAME_REFERENCE;\r
+ }\r
+ }\r
+ else {\r
+ int *refs;\r
+\r
+ int n = onig_name_to_group_numbers(env->reg, cn->name, cn->name_end, &refs);\r
+ if (n <= 0) {\r
+ onig_scan_env_set_error_string(env, ONIGERR_UNDEFINED_NAME_REFERENCE,\r
+ cn->name, cn->name_end);\r
+ return ONIGERR_UNDEFINED_NAME_REFERENCE;\r
+ }\r
+ else if (n > 1) {\r
+ onig_scan_env_set_error_string(env, ONIGERR_MULTIPLEX_DEFINITION_NAME_CALL,\r
+ cn->name, cn->name_end);\r
+ return ONIGERR_MULTIPLEX_DEFINITION_NAME_CALL;\r
+ }\r
+ else {\r
+ cn->group_num = refs[0];\r
+ goto set_call_attr;\r
+ }\r
+ }\r
+\r
+ return 0;\r
+}\r
+\r
+static void\r
+setup_call2_call(Node* node)\r
+{\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
+ case NODE_ALT:\r
+ do {\r
+ setup_call2_call(NODE_CAR(node));\r
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));\r
+ break;\r
+\r
+ case NODE_QUANT:\r
+ setup_call2_call(NODE_BODY(node));\r
+ break;\r
+\r
+ case NODE_ANCHOR:\r
+ if (ANCHOR_HAS_BODY(ANCHOR_(node)))\r
+ setup_call2_call(NODE_BODY(node));\r
+ break;\r
+\r
+ case NODE_ENCLOSURE:\r
+ {\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+\r
+ if (en->type == ENCLOSURE_MEMORY) {\r
+ if (! NODE_IS_MARK1(node)) {\r
+ NODE_STATUS_ADD(node, MARK1);\r
+ setup_call2_call(NODE_BODY(node));\r
+ NODE_STATUS_REMOVE(node, MARK1);\r
+ }\r
+ }\r
+ else if (en->type == ENCLOSURE_IF_ELSE) {\r
+ setup_call2_call(NODE_BODY(node));\r
+ if (IS_NOT_NULL(en->te.Then))\r
+ setup_call2_call(en->te.Then);\r
+ if (IS_NOT_NULL(en->te.Else))\r
+ setup_call2_call(en->te.Else);\r
+ }\r
+ else {\r
+ setup_call2_call(NODE_BODY(node));\r
+ }\r
+ }\r
+ break;\r
+\r
+ case NODE_CALL:\r
+ if (! NODE_IS_MARK1(node)) {\r
+ NODE_STATUS_ADD(node, MARK1);\r
+ {\r
+ CallNode* cn = CALL_(node);\r
+ Node* called = NODE_CALL_BODY(cn);\r
+\r
+ cn->entry_count++;\r
+\r
+ NODE_STATUS_ADD(called, CALLED);\r
+ ENCLOSURE_(called)->m.entry_count++;\r
+ setup_call2_call(called);\r
+ }\r
+ NODE_STATUS_REMOVE(node, MARK1);\r
+ }\r
+ break;\r
+\r
+ default:\r
+ break;\r
+ }\r
+}\r
+\r
+static int\r
+setup_call(Node* node, ScanEnv* env, int state)\r
+{\r
+ int r;\r
+\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
+ case NODE_ALT:\r
+ do {\r
+ r = setup_call(NODE_CAR(node), env, state);\r
+ } while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));\r
+ break;\r
+\r
+ case NODE_QUANT:\r
+ if (QUANT_(node)->upper == 0)\r
+ state |= IN_ZERO_REPEAT;\r
+\r
+ r = setup_call(NODE_BODY(node), env, state);\r
+ break;\r
+\r
+ case NODE_ANCHOR:\r
+ if (ANCHOR_HAS_BODY(ANCHOR_(node)))\r
+ r = setup_call(NODE_BODY(node), env, state);\r
+ else\r
+ r = 0;\r
+ break;\r
+\r
+ case NODE_ENCLOSURE:\r
+ {\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+\r
+ if (en->type == ENCLOSURE_MEMORY) {\r
+ if ((state & IN_ZERO_REPEAT) != 0) {\r
+ NODE_STATUS_ADD(node, IN_ZERO_REPEAT);\r
+ ENCLOSURE_(node)->m.entry_count--;\r
+ }\r
+ r = setup_call(NODE_BODY(node), env, state);\r
+ }\r
+ else if (en->type == ENCLOSURE_IF_ELSE) {\r
+ r = setup_call(NODE_BODY(node), env, state);\r
+ if (r != 0) return r;\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ r = setup_call(en->te.Then, env, state);\r
+ if (r != 0) return r;\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else))\r
+ r = setup_call(en->te.Else, env, state);\r
+ }\r
+ else\r
+ r = setup_call(NODE_BODY(node), env, state);\r
+ }\r
+ break;\r
+\r
+ case NODE_CALL:\r
+ if ((state & IN_ZERO_REPEAT) != 0) {\r
+ NODE_STATUS_ADD(node, IN_ZERO_REPEAT);\r
+ CALL_(node)->entry_count--;\r
+ }\r
+\r
+ r = setup_call_node_call(CALL_(node), env, state);\r
+ break;\r
+\r
+ default:\r
+ r = 0;\r
+ break;\r
+ }\r
+\r
+ return r;\r
+}\r
+\r
+static int\r
+setup_call2(Node* node)\r
+{\r
+ int r = 0;\r
+\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
+ case NODE_ALT:\r
+ do {\r
+ r = setup_call2(NODE_CAR(node));\r
+ } while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));\r
+ break;\r
+\r
+ case NODE_QUANT:\r
+ if (QUANT_(node)->upper != 0)\r
+ r = setup_call2(NODE_BODY(node));\r
+ break;\r
+\r
+ case NODE_ANCHOR:\r
+ if (ANCHOR_HAS_BODY(ANCHOR_(node)))\r
+ r = setup_call2(NODE_BODY(node));\r
+ break;\r
+\r
+ case NODE_ENCLOSURE:\r
+ if (! NODE_IS_IN_ZERO_REPEAT(node))\r
+ r = setup_call2(NODE_BODY(node));\r
+\r
+ {\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+\r
+ if (r != 0) return r;\r
+ if (en->type == ENCLOSURE_IF_ELSE) {\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ r = setup_call2(en->te.Then);\r
+ if (r != 0) return r;\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else))\r
+ r = setup_call2(en->te.Else);\r
+ }\r
+ }\r
+ break;\r
+\r
+ case NODE_CALL:\r
+ if (! NODE_IS_IN_ZERO_REPEAT(node)) {\r
+ setup_call2_call(node);\r
+ }\r
+ break;\r
+\r
+ default:\r
+ break;\r
+ }\r
+\r
+ return r;\r
+}\r
+\r
+\r
+static void\r
+setup_called_state_call(Node* node, int state)\r
+{\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_ALT:\r
+ state |= IN_ALT;\r
+ /* fall */\r
+ case NODE_LIST:\r
+ do {\r
+ setup_called_state_call(NODE_CAR(node), state);\r
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));\r
+ break;\r
+\r
+ case NODE_QUANT:\r
+ {\r
+ QuantNode* qn = QUANT_(node);\r
+\r
+ if (IS_REPEAT_INFINITE(qn->upper) || qn->upper >= 2)\r
+ state |= IN_REAL_REPEAT;\r
+ if (qn->lower != qn->upper)\r
+ state |= IN_VAR_REPEAT;\r
+\r
+ setup_called_state_call(NODE_QUANT_BODY(qn), state);\r
+ }\r
+ break;\r
+\r
+ case NODE_ANCHOR:\r
+ {\r
+ AnchorNode* an = ANCHOR_(node);\r
+\r
+ switch (an->type) {\r
+ case ANCHOR_PREC_READ_NOT:\r
+ case ANCHOR_LOOK_BEHIND_NOT:\r
+ state |= IN_NOT;\r
+ /* fall */\r
+ case ANCHOR_PREC_READ:\r
+ case ANCHOR_LOOK_BEHIND:\r
+ setup_called_state_call(NODE_ANCHOR_BODY(an), state);\r
+ break;\r
+ default:\r
+ break;\r
+ }\r
+ }\r
+ break;\r
+\r
+ case NODE_ENCLOSURE:\r
+ {\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+\r
+ if (en->type == ENCLOSURE_MEMORY) {\r
+ if (NODE_IS_MARK1(node)) {\r
+ if ((~en->m.called_state & state) != 0) {\r
+ en->m.called_state |= state;\r
+ setup_called_state_call(NODE_BODY(node), state);\r
+ }\r
+ }\r
+ else {\r
+ NODE_STATUS_ADD(node, MARK1);\r
+ en->m.called_state |= state;\r
+ setup_called_state_call(NODE_BODY(node), state);\r
+ NODE_STATUS_REMOVE(node, MARK1);\r
+ }\r
+ }\r
+ else if (en->type == ENCLOSURE_IF_ELSE) {\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ setup_called_state_call(en->te.Then, state);\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else))\r
+ setup_called_state_call(en->te.Else, state);\r
+ }\r
+ else {\r
+ setup_called_state_call(NODE_BODY(node), state);\r
+ }\r
+ }\r
+ break;\r
+\r
+ case NODE_CALL:\r
+ setup_called_state_call(NODE_BODY(node), state);\r
+ break;\r
+\r
+ default:\r
+ break;\r
+ }\r
+}\r
+\r
+static void\r
+setup_called_state(Node* node, int state)\r
+{\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_ALT:\r
+ state |= IN_ALT;\r
+ /* fall */\r
+ case NODE_LIST:\r
+ do {\r
+ setup_called_state(NODE_CAR(node), state);\r
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));\r
+ break;\r
+\r
+#ifdef USE_CALL\r
+ case NODE_CALL:\r
+ setup_called_state_call(node, state);\r
+ break;\r
+#endif\r
+\r
+ case NODE_ENCLOSURE:\r
+ {\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+\r
+ switch (en->type) {\r
+ case ENCLOSURE_MEMORY:\r
+ if (en->m.entry_count > 1)\r
+ state |= IN_MULTI_ENTRY;\r
+\r
+ en->m.called_state |= state;\r
+ /* fall */\r
+ case ENCLOSURE_OPTION:\r
+ case ENCLOSURE_STOP_BACKTRACK:\r
+ setup_called_state(NODE_BODY(node), state);\r
+ break;\r
+ case ENCLOSURE_IF_ELSE:\r
+ setup_called_state(NODE_BODY(node), state);\r
+ if (IS_NOT_NULL(en->te.Then))\r
+ setup_called_state(en->te.Then, state);\r
+ if (IS_NOT_NULL(en->te.Else))\r
+ setup_called_state(en->te.Else, state);\r
+ break;\r
+ }\r
+ }\r
+ break;\r
+\r
+ case NODE_QUANT:\r
+ {\r
+ QuantNode* qn = QUANT_(node);\r
+\r
+ if (IS_REPEAT_INFINITE(qn->upper) || qn->upper >= 2)\r
+ state |= IN_REAL_REPEAT;\r
+ if (qn->lower != qn->upper)\r
+ state |= IN_VAR_REPEAT;\r
+\r
+ setup_called_state(NODE_QUANT_BODY(qn), state);\r
+ }\r
+ break;\r
+\r
+ case NODE_ANCHOR:\r
+ {\r
+ AnchorNode* an = ANCHOR_(node);\r
+\r
+ switch (an->type) {\r
+ case ANCHOR_PREC_READ_NOT:\r
+ case ANCHOR_LOOK_BEHIND_NOT:\r
+ state |= IN_NOT;\r
+ /* fall */\r
+ case ANCHOR_PREC_READ:\r
+ case ANCHOR_LOOK_BEHIND:\r
+ setup_called_state(NODE_ANCHOR_BODY(an), state);\r
+ break;\r
+ default:\r
+ break;\r
}\r
}\r
+ break;\r
\r
- if (IS_NULL(root)) {\r
- prev_node = srem;\r
- }\r
- else {\r
- if (IS_NULL(onig_node_list_add(root, srem))) {\r
- onig_node_free(srem);\r
- goto mem_err;\r
- }\r
- }\r
+ case NODE_BACKREF:\r
+ case NODE_STRING:\r
+ case NODE_CTYPE:\r
+ case NODE_CCLASS:\r
+ case NODE_GIMMICK:\r
+ default:\r
+ break;\r
}\r
+}\r
\r
- /* ending */\r
- top_root = (IS_NOT_NULL(top_root) ? top_root : prev_node);\r
- swap_node(node, top_root);\r
- onig_node_free(top_root);\r
- return 0;\r
+#endif /* USE_CALL */\r
\r
- mem_err:\r
- r = ONIGERR_MEMORY;\r
\r
- err:\r
- onig_node_free(top_root);\r
- return r;\r
-}\r
+static int setup_tree(Node* node, regex_t* reg, int state, ScanEnv* env);\r
+\r
+#ifdef __GNUC__\r
+__inline\r
+#endif\r
+static int\r
+setup_anchor(Node* node, regex_t* reg, int state, ScanEnv* env)\r
+{\r
+/* allowed node types in look-behind */\r
+#define ALLOWED_TYPE_IN_LB \\r
+ ( NODE_BIT_LIST | NODE_BIT_ALT | NODE_BIT_STRING | NODE_BIT_CCLASS \\r
+ | NODE_BIT_CTYPE | NODE_BIT_ANCHOR | NODE_BIT_ENCLOSURE | NODE_BIT_QUANT \\r
+ | NODE_BIT_CALL | NODE_BIT_GIMMICK)\r
\r
+#define ALLOWED_ENCLOSURE_IN_LB ( 1<<ENCLOSURE_MEMORY | 1<<ENCLOSURE_OPTION )\r
+#define ALLOWED_ENCLOSURE_IN_LB_NOT (1<<ENCLOSURE_OPTION)\r
\r
-#ifdef USE_COMBINATION_EXPLOSION_CHECK\r
+#define ALLOWED_ANCHOR_IN_LB \\r
+ ( ANCHOR_LOOK_BEHIND | ANCHOR_BEGIN_LINE | ANCHOR_END_LINE | ANCHOR_BEGIN_BUF \\r
+ | ANCHOR_BEGIN_POSITION | ANCHOR_WORD_BOUNDARY | ANCHOR_NO_WORD_BOUNDARY \\r
+ | ANCHOR_WORD_BEGIN | ANCHOR_WORD_END \\r
+ | ANCHOR_EXTENDED_GRAPHEME_CLUSTER_BOUNDARY \\r
+ | ANCHOR_NO_EXTENDED_GRAPHEME_CLUSTER_BOUNDARY )\r
\r
-#define CEC_THRES_NUM_BIG_REPEAT 512\r
-#define CEC_INFINITE_NUM 0x7fffffff\r
+#define ALLOWED_ANCHOR_IN_LB_NOT \\r
+ ( ANCHOR_LOOK_BEHIND | ANCHOR_LOOK_BEHIND_NOT | ANCHOR_BEGIN_LINE \\r
+ | ANCHOR_END_LINE | ANCHOR_BEGIN_BUF | ANCHOR_BEGIN_POSITION | ANCHOR_WORD_BOUNDARY \\r
+ | ANCHOR_NO_WORD_BOUNDARY | ANCHOR_WORD_BEGIN | ANCHOR_WORD_END \\r
+ | ANCHOR_EXTENDED_GRAPHEME_CLUSTER_BOUNDARY \\r
+ | ANCHOR_NO_EXTENDED_GRAPHEME_CLUSTER_BOUNDARY )\r
\r
-#define CEC_IN_INFINITE_REPEAT (1<<0)\r
-#define CEC_IN_FINITE_REPEAT (1<<1)\r
-#define CEC_CONT_BIG_REPEAT (1<<2)\r
+ int r;\r
+ AnchorNode* an = ANCHOR_(node);\r
\r
-static int\r
-setup_comb_exp_check(Node* node, int state, ScanEnv* env)\r
-{\r
- int type;\r
- int r = state;\r
+ switch (an->type) {\r
+ case ANCHOR_PREC_READ:\r
+ r = setup_tree(NODE_ANCHOR_BODY(an), reg, state, env);\r
+ break;\r
+ case ANCHOR_PREC_READ_NOT:\r
+ r = setup_tree(NODE_ANCHOR_BODY(an), reg, (state | IN_NOT), env);\r
+ break;\r
\r
- type = NTYPE(node);\r
- switch (type) {\r
- case NT_LIST:\r
+ case ANCHOR_LOOK_BEHIND:\r
{\r
- Node* prev = NULL_NODE;\r
- do {\r
- r = setup_comb_exp_check(NCAR(node), r, env);\r
- prev = NCAR(node);\r
- } while (r >= 0 && IS_NOT_NULL(node = NCDR(node)));\r
+ r = check_type_tree(NODE_ANCHOR_BODY(an), ALLOWED_TYPE_IN_LB,\r
+ ALLOWED_ENCLOSURE_IN_LB, ALLOWED_ANCHOR_IN_LB);\r
+ if (r < 0) return r;\r
+ if (r > 0) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN;\r
+ r = setup_tree(NODE_ANCHOR_BODY(an), reg, state, env);\r
+ if (r != 0) return r;\r
+ r = setup_look_behind(node, reg, env);\r
}\r
break;\r
\r
- case NT_ALT:\r
+ case ANCHOR_LOOK_BEHIND_NOT:\r
{\r
- int ret;\r
- do {\r
- ret = setup_comb_exp_check(NCAR(node), state, env);\r
- r |= ret;\r
- } while (ret >= 0 && IS_NOT_NULL(node = NCDR(node)));\r
+ r = check_type_tree(NODE_ANCHOR_BODY(an), ALLOWED_TYPE_IN_LB,\r
+ ALLOWED_ENCLOSURE_IN_LB_NOT, ALLOWED_ANCHOR_IN_LB_NOT);\r
+ if (r < 0) return r;\r
+ if (r > 0) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN;\r
+ r = setup_tree(NODE_ANCHOR_BODY(an), reg, (state | IN_NOT), env);\r
+ if (r != 0) return r;\r
+ r = setup_look_behind(node, reg, env);\r
}\r
break;\r
\r
- case NT_QTFR:\r
- {\r
- int child_state = state;\r
- int add_state = 0;\r
- QtfrNode* qn = NQTFR(node);\r
- Node* target = qn->target;\r
- int var_num;\r
-\r
- if (! IS_REPEAT_INFINITE(qn->upper)) {\r
- if (qn->upper > 1) {\r
- /* {0,1}, {1,1} are allowed */\r
- child_state |= CEC_IN_FINITE_REPEAT;\r
-\r
- /* check (a*){n,m}, (a+){n,m} => (a*){n,n}, (a+){n,n} */\r
- if (env->backrefed_mem == 0) {\r
- if (NTYPE(qn->target) == NT_ENCLOSE) {\r
- EncloseNode* en = NENCLOSE(qn->target);\r
- if (en->type == ENCLOSE_MEMORY) {\r
- if (NTYPE(en->target) == NT_QTFR) {\r
- QtfrNode* q = NQTFR(en->target);\r
- if (IS_REPEAT_INFINITE(q->upper)\r
- && q->greedy == qn->greedy) {\r
- qn->upper = (qn->lower == 0 ? 1 : qn->lower);\r
- if (qn->upper == 1)\r
- child_state = state;\r
- }\r
- }\r
- }\r
- }\r
- }\r
- }\r
- }\r
+ default:\r
+ r = 0;\r
+ break;\r
+ }\r
\r
- if (state & CEC_IN_FINITE_REPEAT) {\r
- qn->comb_exp_check_num = -1;\r
- }\r
- else {\r
- if (IS_REPEAT_INFINITE(qn->upper)) {\r
- var_num = CEC_INFINITE_NUM;\r
- child_state |= CEC_IN_INFINITE_REPEAT;\r
- }\r
- else {\r
- var_num = qn->upper - qn->lower;\r
- }\r
-\r
- if (var_num >= CEC_THRES_NUM_BIG_REPEAT)\r
- add_state |= CEC_CONT_BIG_REPEAT;\r
-\r
- if (((state & CEC_IN_INFINITE_REPEAT) != 0 && var_num != 0) ||\r
- ((state & CEC_CONT_BIG_REPEAT) != 0 &&\r
- var_num >= CEC_THRES_NUM_BIG_REPEAT)) {\r
- if (qn->comb_exp_check_num == 0) {\r
- env->num_comb_exp_check++;\r
- qn->comb_exp_check_num = env->num_comb_exp_check;\r
- if (env->curr_max_regnum > env->comb_exp_max_regnum)\r
- env->comb_exp_max_regnum = env->curr_max_regnum;\r
- }\r
- }\r
- }\r
+ return r;\r
+}\r
\r
- r = setup_comb_exp_check(target, child_state, env);\r
- r |= add_state;\r
+#ifdef __GNUC__\r
+__inline\r
+#endif\r
+static int\r
+setup_quant(Node* node, regex_t* reg, int state, ScanEnv* env)\r
+{\r
+ int r;\r
+ OnigLen d;\r
+ QuantNode* qn = QUANT_(node);\r
+ Node* body = NODE_BODY(node);\r
+\r
+ if ((state & IN_REAL_REPEAT) != 0) {\r
+ NODE_STATUS_ADD(node, IN_REAL_REPEAT);\r
+ }\r
+ if ((state & IN_MULTI_ENTRY) != 0) {\r
+ NODE_STATUS_ADD(node, IN_MULTI_ENTRY);\r
+ }\r
+\r
+ if (IS_REPEAT_INFINITE(qn->upper) || qn->upper >= 1) {\r
+ d = tree_min_len(body, env);\r
+ if (d == 0) {\r
+#ifdef USE_INSISTENT_CHECK_CAPTURES_IN_EMPTY_REPEAT\r
+ qn->body_empty_info = quantifiers_memory_node_info(body);\r
+ if (qn->body_empty_info == QUANT_BODY_IS_EMPTY_REC) {\r
+ if (NODE_TYPE(body) == NODE_ENCLOSURE &&\r
+ ENCLOSURE_(body)->type == ENCLOSURE_MEMORY) {\r
+ MEM_STATUS_ON(env->bt_mem_end, ENCLOSURE_(body)->m.regnum);\r
+ }\r
+ }\r
+#else\r
+ qn->body_empty_info = QUANT_BODY_IS_EMPTY;\r
+#endif\r
}\r
- break;\r
-\r
- case NT_ENCLOSE:\r
- {\r
- EncloseNode* en = NENCLOSE(node);\r
+ }\r
\r
- switch (en->type) {\r
- case ENCLOSE_MEMORY:\r
- {\r
- if (env->curr_max_regnum < en->regnum)\r
- env->curr_max_regnum = en->regnum;\r
+ if (IS_REPEAT_INFINITE(qn->upper) || qn->upper >= 2)\r
+ state |= IN_REAL_REPEAT;\r
+ if (qn->lower != qn->upper)\r
+ state |= IN_VAR_REPEAT;\r
\r
- r = setup_comb_exp_check(en->target, state, env);\r
- }\r
- break;\r
+ r = setup_tree(body, reg, state, env);\r
+ if (r != 0) return r;\r
\r
- default:\r
- r = setup_comb_exp_check(en->target, state, env);\r
- break;\r
+ /* expand string */\r
+#define EXPAND_STRING_MAX_LENGTH 100\r
+ if (NODE_TYPE(body) == NODE_STRING) {\r
+ if (!IS_REPEAT_INFINITE(qn->lower) && qn->lower == qn->upper &&\r
+ qn->lower > 1 && qn->lower <= EXPAND_STRING_MAX_LENGTH) {\r
+ int len = NODE_STRING_LEN(body);\r
+ StrNode* sn = STR_(body);\r
+\r
+ if (len * qn->lower <= EXPAND_STRING_MAX_LENGTH) {\r
+ int i, n = qn->lower;\r
+ onig_node_conv_to_str_node(node, STR_(body)->flag);\r
+ for (i = 0; i < n; i++) {\r
+ r = onig_node_str_cat(node, sn->s, sn->end);\r
+ if (r != 0) return r;\r
+ }\r
+ onig_node_free(body);\r
+ return r;\r
}\r
}\r
- break;\r
-\r
-#ifdef USE_SUBEXP_CALL\r
- case NT_CALL:\r
- if (IS_CALL_RECURSION(NCALL(node)))\r
- env->has_recursion = 1;\r
- else\r
- r = setup_comb_exp_check(NCALL(node)->target, state, env);\r
- break;\r
-#endif\r
+ }\r
\r
- default:\r
- break;\r
+#ifdef USE_OP_PUSH_OR_JUMP_EXACT\r
+ if (qn->greedy && (qn->body_empty_info != QUANT_BODY_IS_NOT_EMPTY)) {\r
+ if (NODE_TYPE(body) == NODE_QUANT) {\r
+ QuantNode* tqn = QUANT_(body);\r
+ if (IS_NOT_NULL(tqn->head_exact)) {\r
+ qn->head_exact = tqn->head_exact;\r
+ tqn->head_exact = NULL;\r
+ }\r
+ }\r
+ else {\r
+ qn->head_exact = get_head_value_node(NODE_BODY(node), 1, reg);\r
+ }\r
}\r
+#endif\r
\r
return r;\r
}\r
-#endif\r
-\r
-#define IN_ALT (1<<0)\r
-#define IN_NOT (1<<1)\r
-#define IN_REPEAT (1<<2)\r
-#define IN_VAR_REPEAT (1<<3)\r
\r
/* setup_tree does the following work.\r
- 1. check empty loop. (set qn->target_empty_info)\r
+ 1. check empty loop. (set qn->body_empty_info)\r
2. expand ignore-case in char class.\r
3. set memory status bit flags. (reg->mem_stats)\r
4. set qn->head_exact for [push, exact] -> [push_or_jump_exact1, exact].\r
static int\r
setup_tree(Node* node, regex_t* reg, int state, ScanEnv* env)\r
{\r
- int type;\r
int r = 0;\r
\r
- type = NTYPE(node);\r
- switch (type) {\r
- case NT_LIST:\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
{\r
Node* prev = NULL_NODE;\r
do {\r
- r = setup_tree(NCAR(node), reg, state, env);\r
- if (IS_NOT_NULL(prev) && r == 0) {\r
- r = next_setup(prev, NCAR(node), reg);\r
- }\r
- prev = NCAR(node);\r
- } while (r == 0 && IS_NOT_NULL(node = NCDR(node)));\r
+ r = setup_tree(NODE_CAR(node), reg, state, env);\r
+ if (IS_NOT_NULL(prev) && r == 0) {\r
+ r = next_setup(prev, NODE_CAR(node), reg);\r
+ }\r
+ prev = NODE_CAR(node);\r
+ } while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));\r
}\r
break;\r
\r
- case NT_ALT:\r
+ case NODE_ALT:\r
do {\r
- r = setup_tree(NCAR(node), reg, (state | IN_ALT), env);\r
- } while (r == 0 && IS_NOT_NULL(node = NCDR(node)));\r
- break;\r
-\r
- case NT_CCLASS:\r
+ r = setup_tree(NODE_CAR(node), reg, (state | IN_ALT), env);\r
+ } while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));\r
break;\r
\r
- case NT_STR:\r
- if (IS_IGNORECASE(reg->options) && !NSTRING_IS_RAW(node)) {\r
+ case NODE_STRING:\r
+ if (IS_IGNORECASE(reg->options) && !NODE_STRING_IS_RAW(node)) {\r
r = expand_case_fold_string(node, reg);\r
}\r
break;\r
\r
- case NT_CTYPE:\r
- case NT_CANY:\r
- break;\r
-\r
-#ifdef USE_SUBEXP_CALL\r
- case NT_CALL:\r
- break;\r
-#endif\r
-\r
- case NT_BREF:\r
+ case NODE_BACKREF:\r
{\r
int i;\r
int* p;\r
- Node** nodes = SCANENV_MEM_NODES(env);\r
- BRefNode* br = NBREF(node);\r
+ BackRefNode* br = BACKREF_(node);\r
p = BACKREFS_P(br);\r
for (i = 0; i < br->back_num; i++) {\r
- if (p[i] > env->num_mem) return ONIGERR_INVALID_BACKREF;\r
- BIT_STATUS_ON_AT(env->backrefed_mem, p[i]);\r
- BIT_STATUS_ON_AT(env->bt_mem_start, p[i]);\r
+ if (p[i] > env->num_mem) return ONIGERR_INVALID_BACKREF;\r
+ MEM_STATUS_ON(env->backrefed_mem, p[i]);\r
+ MEM_STATUS_ON(env->bt_mem_start, p[i]);\r
#ifdef USE_BACKREF_WITH_LEVEL\r
- if (IS_BACKREF_NEST_LEVEL(br)) {\r
- BIT_STATUS_ON_AT(env->bt_mem_end, p[i]);\r
- }\r
+ if (NODE_IS_NEST_LEVEL(node)) {\r
+ MEM_STATUS_ON(env->bt_mem_end, p[i]);\r
+ }\r
#endif\r
- SET_ENCLOSE_STATUS(nodes[p[i]], NST_MEM_BACKREFED);\r
}\r
}\r
break;\r
\r
- case NT_QTFR:\r
+ case NODE_ENCLOSURE:\r
{\r
- OnigDistance d;\r
- QtfrNode* qn = NQTFR(node);\r
- Node* target = qn->target;\r
-\r
- if ((state & IN_REPEAT) != 0) {\r
- qn->state |= NST_IN_REPEAT;\r
- }\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
\r
- if (IS_REPEAT_INFINITE(qn->upper) || qn->upper >= 1) {\r
- r = get_min_match_length(target, &d, env);\r
- if (r) break;\r
- if (d == 0) {\r
- qn->target_empty_info = NQ_TARGET_IS_EMPTY;\r
-#ifdef USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT\r
- r = quantifiers_memory_node_info(target);\r
- if (r < 0) break;\r
- if (r > 0) {\r
- qn->target_empty_info = r;\r
- }\r
-#endif\r
-#if 0\r
- r = get_max_match_length(target, &d, env);\r
- if (r == 0 && d == 0) {\r
- /* ()* ==> ()?, ()+ ==> () */\r
- qn->upper = 1;\r
- if (qn->lower > 1) qn->lower = 1;\r
- if (NTYPE(target) == NT_STR) {\r
- qn->upper = qn->lower = 0; /* /(?:)+/ ==> // */\r
- }\r
- }\r
-#endif\r
- }\r
- }\r
-\r
- state |= IN_REPEAT;\r
- if (qn->lower != qn->upper)\r
- state |= IN_VAR_REPEAT;\r
- r = setup_tree(target, reg, state, env);\r
- if (r) break;\r
-\r
- /* expand string */\r
-#define EXPAND_STRING_MAX_LENGTH 100\r
- if (NTYPE(target) == NT_STR) {\r
- if (!IS_REPEAT_INFINITE(qn->lower) && qn->lower == qn->upper &&\r
- qn->lower > 1 && qn->lower <= EXPAND_STRING_MAX_LENGTH) {\r
- int len = NSTRING_LEN(target);\r
- StrNode* sn = NSTR(target);\r
-\r
- if (len * qn->lower <= EXPAND_STRING_MAX_LENGTH) {\r
- int i, n = qn->lower;\r
- onig_node_conv_to_str_node(node, NSTR(target)->flag);\r
- for (i = 0; i < n; i++) {\r
- r = onig_node_str_cat(node, sn->s, sn->end);\r
- if (r) break;\r
- }\r
- onig_node_free(target);\r
- break; /* break case NT_QTFR: */\r
- }\r
- }\r
- }\r
+ switch (en->type) {\r
+ case ENCLOSURE_OPTION:\r
+ {\r
+ OnigOptionType options = reg->options;\r
+ reg->options = ENCLOSURE_(node)->o.options;\r
+ r = setup_tree(NODE_BODY(node), reg, state, env);\r
+ reg->options = options;\r
+ }\r
+ break;\r
\r
-#ifdef USE_OP_PUSH_OR_JUMP_EXACT\r
- if (qn->greedy && (qn->target_empty_info != 0)) {\r
- if (NTYPE(target) == NT_QTFR) {\r
- QtfrNode* tqn = NQTFR(target);\r
- if (IS_NOT_NULL(tqn->head_exact)) {\r
- qn->head_exact = tqn->head_exact;\r
- tqn->head_exact = NULL;\r
- }\r
- }\r
- else {\r
- qn->head_exact = get_head_value_node(qn->target, 1, reg);\r
- }\r
- }\r
+ case ENCLOSURE_MEMORY:\r
+#ifdef USE_CALL\r
+ state |= en->m.called_state;\r
#endif\r
- }\r
- break;\r
\r
- case NT_ENCLOSE:\r
- {\r
- EncloseNode* en = NENCLOSE(node);\r
+ if ((state & (IN_ALT | IN_NOT | IN_VAR_REPEAT | IN_MULTI_ENTRY)) != 0\r
+ || NODE_IS_RECURSION(node)) {\r
+ MEM_STATUS_ON(env->bt_mem_start, en->m.regnum);\r
+ }\r
+ r = setup_tree(NODE_BODY(node), reg, state, env);\r
+ break;\r
\r
- switch (en->type) {\r
- case ENCLOSE_OPTION:\r
- {\r
- OnigOptionType options = reg->options;\r
- reg->options = NENCLOSE(node)->option;\r
- r = setup_tree(NENCLOSE(node)->target, reg, state, env);\r
- reg->options = options;\r
- }\r
- break;\r
-\r
- case ENCLOSE_MEMORY:\r
- if ((state & (IN_ALT | IN_NOT | IN_VAR_REPEAT)) != 0) {\r
- BIT_STATUS_ON_AT(env->bt_mem_start, en->regnum);\r
- /* SET_ENCLOSE_STATUS(node, NST_MEM_IN_ALT_NOT); */\r
- }\r
- r = setup_tree(en->target, reg, state, env);\r
+ case ENCLOSURE_STOP_BACKTRACK:\r
+ {\r
+ Node* target = NODE_BODY(node);\r
+ r = setup_tree(target, reg, state, env);\r
+ if (NODE_TYPE(target) == NODE_QUANT) {\r
+ QuantNode* tqn = QUANT_(target);\r
+ if (IS_REPEAT_INFINITE(tqn->upper) && tqn->lower <= 1 &&\r
+ tqn->greedy != 0) { /* (?>a*), a*+ etc... */\r
+ if (NODE_IS_SIMPLE_TYPE(NODE_BODY(target)))\r
+ NODE_STATUS_ADD(node, STOP_BT_SIMPLE_REPEAT);\r
+ }\r
+ }\r
+ }\r
break;\r
\r
- case ENCLOSE_STOP_BACKTRACK:\r
- {\r
- Node* target = en->target;\r
- r = setup_tree(target, reg, state, env);\r
- if (NTYPE(target) == NT_QTFR) {\r
- QtfrNode* tqn = NQTFR(target);\r
- if (IS_REPEAT_INFINITE(tqn->upper) && tqn->lower <= 1 &&\r
- tqn->greedy != 0) { /* (?>a*), a*+ etc... */\r
- int qtype = NTYPE(tqn->target);\r
- if (IS_NODE_TYPE_SIMPLE(qtype))\r
- SET_ENCLOSE_STATUS(node, NST_STOP_BT_SIMPLE_REPEAT);\r
- }\r
- }\r
- }\r
- break;\r
+ case ENCLOSURE_IF_ELSE:\r
+ r = setup_tree(NODE_BODY(node), reg, (state | IN_ALT), env);\r
+ if (r != 0) return r;\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ r = setup_tree(en->te.Then, reg, (state | IN_ALT), env);\r
+ if (r != 0) return r;\r
+ }\r
+ if (IS_NOT_NULL(en->te.Else))\r
+ r = setup_tree(en->te.Else, reg, (state | IN_ALT), env);\r
+ break;\r
}\r
}\r
break;\r
\r
- case NT_ANCHOR:\r
- {\r
- AnchorNode* an = NANCHOR(node);\r
-\r
- switch (an->type) {\r
- case ANCHOR_PREC_READ:\r
- r = setup_tree(an->target, reg, state, env);\r
- break;\r
- case ANCHOR_PREC_READ_NOT:\r
- r = setup_tree(an->target, reg, (state | IN_NOT), env);\r
- break;\r
-\r
-/* allowed node types in look-behind */\r
-#define ALLOWED_TYPE_IN_LB \\r
- ( BIT_NT_LIST | BIT_NT_ALT | BIT_NT_STR | BIT_NT_CCLASS | BIT_NT_CTYPE | \\r
- BIT_NT_CANY | BIT_NT_ANCHOR | BIT_NT_ENCLOSE | BIT_NT_QTFR | BIT_NT_CALL )\r
-\r
-#define ALLOWED_ENCLOSE_IN_LB ( ENCLOSE_MEMORY )\r
-#define ALLOWED_ENCLOSE_IN_LB_NOT 0\r
-\r
-#define ALLOWED_ANCHOR_IN_LB \\r
-( ANCHOR_LOOK_BEHIND | ANCHOR_BEGIN_LINE | ANCHOR_END_LINE | ANCHOR_BEGIN_BUF | ANCHOR_BEGIN_POSITION )\r
-#define ALLOWED_ANCHOR_IN_LB_NOT \\r
-( ANCHOR_LOOK_BEHIND | ANCHOR_LOOK_BEHIND_NOT | ANCHOR_BEGIN_LINE | ANCHOR_END_LINE | ANCHOR_BEGIN_BUF | ANCHOR_BEGIN_POSITION )\r
-\r
- case ANCHOR_LOOK_BEHIND:\r
- {\r
- r = check_type_tree(an->target, ALLOWED_TYPE_IN_LB,\r
- ALLOWED_ENCLOSE_IN_LB, ALLOWED_ANCHOR_IN_LB);\r
- if (r < 0) return r;\r
- if (r > 0) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN;\r
- r = setup_look_behind(node, reg, env);\r
- if (r != 0) return r;\r
- r = setup_tree(an->target, reg, state, env);\r
- }\r
- break;\r
+ case NODE_QUANT:\r
+ r = setup_quant(node, reg, state, env);\r
+ break;\r
\r
- case ANCHOR_LOOK_BEHIND_NOT:\r
- {\r
- r = check_type_tree(an->target, ALLOWED_TYPE_IN_LB,\r
- ALLOWED_ENCLOSE_IN_LB_NOT, ALLOWED_ANCHOR_IN_LB_NOT);\r
- if (r < 0) return r;\r
- if (r > 0) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN;\r
- r = setup_look_behind(node, reg, env);\r
- if (r != 0) return r;\r
- r = setup_tree(an->target, reg, (state | IN_NOT), env);\r
- }\r
- break;\r
- }\r
- }\r
+ case NODE_ANCHOR:\r
+ r = setup_anchor(node, reg, state, env);\r
break;\r
\r
+#ifdef USE_CALL\r
+ case NODE_CALL:\r
+#endif\r
+ case NODE_CTYPE:\r
+ case NODE_CCLASS:\r
+ case NODE_GIMMICK:\r
default:\r
break;\r
}\r
return r;\r
}\r
\r
-/* set skip map for Boyer-Moor search */\r
+/* set skip map for Boyer-Moore search */\r
static int\r
set_bm_skip(UChar* s, UChar* end, OnigEncoding enc ARG_UNUSED,\r
- UChar skip[], int** int_skip)\r
+ UChar skip[], int** int_skip)\r
{\r
int i, len;\r
\r
- len = (int)(end - s);\r
+ len = (int )(end - s);\r
if (len < ONIG_CHAR_TABLE_SIZE) {\r
- for (i = 0; i < ONIG_CHAR_TABLE_SIZE; i++) skip[i] = (UChar)len;\r
+ for (i = 0; i < ONIG_CHAR_TABLE_SIZE; i++) skip[i] = len;\r
\r
for (i = 0; i < len - 1; i++)\r
- skip[s[i]] = (UChar)(len - 1 - i);\r
+ skip[s[i]] = len - 1 - i;\r
}\r
else {\r
if (IS_NULL(*int_skip)) {\r
#define OPT_EXACT_MAXLEN 24\r
\r
typedef struct {\r
- OnigDistance min; /* min byte length */\r
- OnigDistance max; /* max byte length */\r
-} MinMaxLen;\r
+ OnigLen min; /* min byte length */\r
+ OnigLen max; /* max byte length */\r
+} MinMax;\r
\r
typedef struct {\r
- MinMaxLen mmd;\r
+ MinMax mmd;\r
OnigEncoding enc;\r
OnigOptionType options;\r
OnigCaseFoldType case_fold_flag;\r
} OptEnv;\r
\r
typedef struct {\r
- int left_anchor;\r
- int right_anchor;\r
-} OptAncInfo;\r
+ int left;\r
+ int right;\r
+} OptAnc;\r
\r
typedef struct {\r
- MinMaxLen mmd; /* info position */\r
- OptAncInfo anc;\r
-\r
- int reach_end;\r
- int ignore_case;\r
- int len;\r
- UChar s[OPT_EXACT_MAXLEN];\r
-} OptExactInfo;\r
+ MinMax mmd; /* position */\r
+ OptAnc anc;\r
+ int reach_end;\r
+ int ignore_case;\r
+ int len;\r
+ UChar s[OPT_EXACT_MAXLEN];\r
+} OptExact;\r
\r
typedef struct {\r
- MinMaxLen mmd; /* info position */\r
- OptAncInfo anc;\r
-\r
- int value; /* weighted value */\r
- UChar map[ONIG_CHAR_TABLE_SIZE];\r
-} OptMapInfo;\r
+ MinMax mmd; /* position */\r
+ OptAnc anc;\r
+ int value; /* weighted value */\r
+ UChar map[ONIG_CHAR_TABLE_SIZE];\r
+} OptMap;\r
\r
typedef struct {\r
- MinMaxLen len;\r
-\r
- OptAncInfo anc;\r
- OptExactInfo exb; /* boundary */\r
- OptExactInfo exm; /* middle */\r
- OptExactInfo expr; /* prec read (?=...) */\r
-\r
- OptMapInfo map; /* boundary */\r
-} NodeOptInfo;\r
+ MinMax len;\r
+ OptAnc anc;\r
+ OptExact exb; /* boundary */\r
+ OptExact exm; /* middle */\r
+ OptExact expr; /* prec read (?=...) */\r
+ OptMap map; /* boundary */\r
+} NodeOpt;\r
\r
\r
static int\r
map_position_value(OnigEncoding enc, int i)\r
{\r
- static const short int ByteValTable[] = {\r
+ static const short int Vals[] = {\r
5, 1, 1, 1, 1, 1, 1, 1, 1, 10, 10, 1, 1, 10, 1, 1,\r
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\r
12, 4, 7, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 5,\r
6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 5, 5, 5, 5, 1\r
};\r
\r
- if (i < (int )(sizeof(ByteValTable)/sizeof(ByteValTable[0]))) {\r
+ if (i < (int )(sizeof(Vals)/sizeof(Vals[0]))) {\r
if (i == 0 && ONIGENC_MBC_MINLEN(enc) > 1)\r
return 20;\r
else\r
- return (int )ByteValTable[i];\r
+ return (int )Vals[i];\r
}\r
else\r
return 4; /* Take it easy. */\r
}\r
\r
static int\r
-distance_value(MinMaxLen* mm)\r
+distance_value(MinMax* mm)\r
{\r
/* 1000 / (min-max-dist + 1) */\r
static const short int dist_vals[] = {\r
11, 11, 11, 11, 11, 10, 10, 10, 10, 10\r
};\r
\r
- int d;\r
+ OnigLen d;\r
\r
- if (mm->max == ONIG_INFINITE_DISTANCE) return 0;\r
+ if (mm->max == INFINITE_LEN) return 0;\r
\r
d = mm->max - mm->min;\r
- if (d < (int )(sizeof(dist_vals)/sizeof(dist_vals[0])))\r
+ if (d < (OnigLen )(sizeof(dist_vals)/sizeof(dist_vals[0])))\r
/* return dist_vals[d] * 16 / (mm->min + 12); */\r
return (int )dist_vals[d];\r
else\r
}\r
\r
static int\r
-comp_distance_value(MinMaxLen* d1, MinMaxLen* d2, int v1, int v2)\r
+comp_distance_value(MinMax* d1, MinMax* d2, int v1, int v2)\r
{\r
if (v2 <= 0) return -1;\r
if (v1 <= 0) return 1;\r
}\r
\r
static int\r
-is_equal_mml(MinMaxLen* a, MinMaxLen* b)\r
+is_equal_mml(MinMax* a, MinMax* b)\r
{\r
return (a->min == b->min && a->max == b->max) ? 1 : 0;\r
}\r
\r
-\r
static void\r
-set_mml(MinMaxLen* mml, OnigDistance min, OnigDistance max)\r
+set_mml(MinMax* l, OnigLen min, OnigLen max)\r
{\r
- mml->min = min;\r
- mml->max = max;\r
+ l->min = min;\r
+ l->max = max;\r
}\r
\r
static void\r
-clear_mml(MinMaxLen* mml)\r
+clear_mml(MinMax* l)\r
{\r
- mml->min = mml->max = 0;\r
+ l->min = l->max = 0;\r
}\r
\r
static void\r
-copy_mml(MinMaxLen* to, MinMaxLen* from)\r
+copy_mml(MinMax* to, MinMax* from)\r
{\r
to->min = from->min;\r
to->max = from->max;\r
}\r
\r
static void\r
-add_mml(MinMaxLen* to, MinMaxLen* from)\r
+add_mml(MinMax* to, MinMax* from)\r
{\r
to->min = distance_add(to->min, from->min);\r
to->max = distance_add(to->max, from->max);\r
}\r
\r
-#if 0\r
-static void\r
-add_len_mml(MinMaxLen* to, OnigDistance len)\r
-{\r
- to->min = distance_add(to->min, len);\r
- to->max = distance_add(to->max, len);\r
-}\r
-#endif\r
-\r
static void\r
-alt_merge_mml(MinMaxLen* to, MinMaxLen* from)\r
+alt_merge_mml(MinMax* to, MinMax* from)\r
{\r
if (to->min > from->min) to->min = from->min;\r
if (to->max < from->max) to->max = from->max;\r
}\r
\r
static void\r
-clear_opt_anc_info(OptAncInfo* anc)\r
+clear_opt_anc_info(OptAnc* a)\r
{\r
- anc->left_anchor = 0;\r
- anc->right_anchor = 0;\r
+ a->left = 0;\r
+ a->right = 0;\r
}\r
\r
static void\r
-copy_opt_anc_info(OptAncInfo* to, OptAncInfo* from)\r
+copy_opt_anc_info(OptAnc* to, OptAnc* from)\r
{\r
*to = *from;\r
}\r
\r
static void\r
-concat_opt_anc_info(OptAncInfo* to, OptAncInfo* left, OptAncInfo* right,\r
- OnigDistance left_len, OnigDistance right_len)\r
+concat_opt_anc_info(OptAnc* to, OptAnc* left, OptAnc* right,\r
+ OnigLen left_len, OnigLen right_len)\r
{\r
clear_opt_anc_info(to);\r
\r
- to->left_anchor = left->left_anchor;\r
+ to->left = left->left;\r
if (left_len == 0) {\r
- to->left_anchor |= right->left_anchor;\r
+ to->left |= right->left;\r
}\r
\r
- to->right_anchor = right->right_anchor;\r
+ to->right = right->right;\r
if (right_len == 0) {\r
- to->right_anchor |= left->right_anchor;\r
+ to->right |= left->right;\r
+ }\r
+ else {\r
+ to->right |= (left->right & ANCHOR_PREC_READ_NOT);\r
}\r
}\r
\r
static int\r
-is_left_anchor(int anc)\r
+is_left(int a)\r
{\r
- if (anc == ANCHOR_END_BUF || anc == ANCHOR_SEMI_END_BUF ||\r
- anc == ANCHOR_END_LINE || anc == ANCHOR_PREC_READ ||\r
- anc == ANCHOR_PREC_READ_NOT)\r
+ if (a == ANCHOR_END_BUF || a == ANCHOR_SEMI_END_BUF ||\r
+ a == ANCHOR_END_LINE || a == ANCHOR_PREC_READ || a == ANCHOR_PREC_READ_NOT)\r
return 0;\r
\r
return 1;\r
}\r
\r
static int\r
-is_set_opt_anc_info(OptAncInfo* to, int anc)\r
+is_set_opt_anc_info(OptAnc* to, int anc)\r
{\r
- if ((to->left_anchor & anc) != 0) return 1;\r
+ if ((to->left & anc) != 0) return 1;\r
\r
- return ((to->right_anchor & anc) != 0 ? 1 : 0);\r
+ return ((to->right & anc) != 0 ? 1 : 0);\r
}\r
\r
static void\r
-add_opt_anc_info(OptAncInfo* to, int anc)\r
+add_opt_anc_info(OptAnc* to, int anc)\r
{\r
- if (is_left_anchor(anc))\r
- to->left_anchor |= anc;\r
+ if (is_left(anc))\r
+ to->left |= anc;\r
else\r
- to->right_anchor |= anc;\r
+ to->right |= anc;\r
}\r
\r
static void\r
-remove_opt_anc_info(OptAncInfo* to, int anc)\r
+remove_opt_anc_info(OptAnc* to, int anc)\r
{\r
- if (is_left_anchor(anc))\r
- to->left_anchor &= ~anc;\r
+ if (is_left(anc))\r
+ to->left &= ~anc;\r
else\r
- to->right_anchor &= ~anc;\r
+ to->right &= ~anc;\r
}\r
\r
static void\r
-alt_merge_opt_anc_info(OptAncInfo* to, OptAncInfo* add)\r
+alt_merge_opt_anc_info(OptAnc* to, OptAnc* add)\r
{\r
- to->left_anchor &= add->left_anchor;\r
- to->right_anchor &= add->right_anchor;\r
+ to->left &= add->left;\r
+ to->right &= add->right;\r
}\r
\r
static int\r
-is_full_opt_exact_info(OptExactInfo* ex)\r
+is_full_opt_exact(OptExact* e)\r
{\r
- return (ex->len >= OPT_EXACT_MAXLEN ? 1 : 0);\r
+ return (e->len >= OPT_EXACT_MAXLEN ? 1 : 0);\r
}\r
\r
static void\r
-clear_opt_exact_info(OptExactInfo* ex)\r
+clear_opt_exact(OptExact* e)\r
{\r
- clear_mml(&ex->mmd);\r
- clear_opt_anc_info(&ex->anc);\r
- ex->reach_end = 0;\r
- ex->ignore_case = 0;\r
- ex->len = 0;\r
- ex->s[0] = '\0';\r
+ clear_mml(&e->mmd);\r
+ clear_opt_anc_info(&e->anc);\r
+ e->reach_end = 0;\r
+ e->ignore_case = 0;\r
+ e->len = 0;\r
+ e->s[0] = '\0';\r
}\r
\r
static void\r
-copy_opt_exact_info(OptExactInfo* to, OptExactInfo* from)\r
+copy_opt_exact(OptExact* to, OptExact* from)\r
{\r
*to = *from;\r
}\r
\r
-static void\r
-concat_opt_exact_info(OptExactInfo* to, OptExactInfo* add, OnigEncoding enc)\r
+static int\r
+concat_opt_exact(OptExact* to, OptExact* add, OnigEncoding enc)\r
{\r
- int i, j, len;\r
+ int i, j, len, r;\r
UChar *p, *end;\r
- OptAncInfo tanc;\r
+ OptAnc tanc;\r
\r
if (! to->ignore_case && add->ignore_case) {\r
- if (to->len >= add->len) return ; /* avoid */\r
+ if (to->len >= add->len) return 0; /* avoid */\r
\r
to->ignore_case = 1;\r
}\r
\r
+ r = 0;\r
p = add->s;\r
end = p + add->len;\r
for (i = to->len; p < end; ) {\r
len = enclen(enc, p);\r
- if (i + len > OPT_EXACT_MAXLEN) break;\r
+ if (i + len > OPT_EXACT_MAXLEN) {\r
+ r = 1; /* 1:full */\r
+ break;\r
+ }\r
for (j = 0; j < len && p < end; j++)\r
to->s[i++] = *p++;\r
}\r
to->reach_end = (p == end ? add->reach_end : 0);\r
\r
concat_opt_anc_info(&tanc, &to->anc, &add->anc, 1, 1);\r
- if (! to->reach_end) tanc.right_anchor = 0;\r
+ if (! to->reach_end) tanc.right = 0;\r
copy_opt_anc_info(&to->anc, &tanc);\r
+\r
+ return r;\r
}\r
\r
static void\r
-concat_opt_exact_info_str(OptExactInfo* to, UChar* s, UChar* end,\r
- int raw ARG_UNUSED, OnigEncoding enc)\r
+concat_opt_exact_str(OptExact* to, UChar* s, UChar* end, OnigEncoding enc)\r
{\r
int i, j, len;\r
UChar *p;\r
}\r
\r
static void\r
-alt_merge_opt_exact_info(OptExactInfo* to, OptExactInfo* add, OptEnv* env)\r
+alt_merge_opt_exact(OptExact* to, OptExact* add, OptEnv* env)\r
{\r
int i, j, len;\r
\r
if (add->len == 0 || to->len == 0) {\r
- clear_opt_exact_info(to);\r
+ clear_opt_exact(to);\r
return ;\r
}\r
\r
if (! is_equal_mml(&to->mmd, &add->mmd)) {\r
- clear_opt_exact_info(to);\r
+ clear_opt_exact(to);\r
return ;\r
}\r
\r
to->ignore_case |= add->ignore_case;\r
\r
alt_merge_opt_anc_info(&to->anc, &add->anc);\r
- if (! to->reach_end) to->anc.right_anchor = 0;\r
+ if (! to->reach_end) to->anc.right = 0;\r
}\r
\r
static void\r
-select_opt_exact_info(OnigEncoding enc, OptExactInfo* now, OptExactInfo* alt)\r
+select_opt_exact(OnigEncoding enc, OptExact* now, OptExact* alt)\r
{\r
- int v1, v2;\r
+ int vn, va;\r
\r
- v1 = now->len;\r
- v2 = alt->len;\r
+ vn = now->len;\r
+ va = alt->len;\r
\r
- if (v2 == 0) {\r
+ if (va == 0) {\r
return ;\r
}\r
- else if (v1 == 0) {\r
- copy_opt_exact_info(now, alt);\r
+ else if (vn == 0) {\r
+ copy_opt_exact(now, alt);\r
return ;\r
}\r
- else if (v1 <= 2 && v2 <= 2) {\r
+ else if (vn <= 2 && va <= 2) {\r
/* ByteValTable[x] is big value --> low price */\r
- v2 = map_position_value(enc, now->s[0]);\r
- v1 = map_position_value(enc, alt->s[0]);\r
+ va = map_position_value(enc, now->s[0]);\r
+ vn = map_position_value(enc, alt->s[0]);\r
\r
- if (now->len > 1) v1 += 5;\r
- if (alt->len > 1) v2 += 5;\r
+ if (now->len > 1) vn += 5;\r
+ if (alt->len > 1) va += 5;\r
}\r
\r
- if (now->ignore_case == 0) v1 *= 2;\r
- if (alt->ignore_case == 0) v2 *= 2;\r
+ if (now->ignore_case == 0) vn *= 2;\r
+ if (alt->ignore_case == 0) va *= 2;\r
\r
- if (comp_distance_value(&now->mmd, &alt->mmd, v1, v2) > 0)\r
- copy_opt_exact_info(now, alt);\r
+ if (comp_distance_value(&now->mmd, &alt->mmd, vn, va) > 0)\r
+ copy_opt_exact(now, alt);\r
}\r
\r
static void\r
-clear_opt_map_info(OptMapInfo* map)\r
+clear_opt_map(OptMap* map)\r
{\r
- static const OptMapInfo clean_info = {\r
+ static const OptMap clean_info = {\r
{0, 0}, {0, 0}, 0,\r
{\r
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\r
}\r
};\r
\r
- xmemcpy(map, &clean_info, sizeof(OptMapInfo));\r
+ xmemcpy(map, &clean_info, sizeof(OptMap));\r
}\r
\r
static void\r
-copy_opt_map_info(OptMapInfo* to, OptMapInfo* from)\r
+copy_opt_map(OptMap* to, OptMap* from)\r
{\r
- *to = *from;\r
+ xmemcpy(to,from,sizeof(OptMap));\r
}\r
\r
static void\r
-add_char_opt_map_info(OptMapInfo* map, UChar c, OnigEncoding enc)\r
+add_char_opt_map(OptMap* m, UChar c, OnigEncoding enc)\r
{\r
- if (map->map[c] == 0) {\r
- map->map[c] = 1;\r
- map->value += map_position_value(enc, c);\r
+ if (m->map[c] == 0) {\r
+ m->map[c] = 1;\r
+ m->value += map_position_value(enc, c);\r
}\r
}\r
\r
static int\r
-add_char_amb_opt_map_info(OptMapInfo* map, UChar* p, UChar* end,\r
- OnigEncoding enc, OnigCaseFoldType case_fold_flag)\r
+add_char_amb_opt_map(OptMap* map, UChar* p, UChar* end,\r
+ OnigEncoding enc, OnigCaseFoldType fold_flag)\r
{\r
OnigCaseFoldCodeItem items[ONIGENC_GET_CASE_FOLD_CODES_MAX_NUM];\r
UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];\r
int i, n;\r
\r
- add_char_opt_map_info(map, p[0], enc);\r
+ add_char_opt_map(map, p[0], enc);\r
\r
- case_fold_flag = DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag);\r
- n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc, case_fold_flag, p, end, items);\r
+ fold_flag = DISABLE_CASE_FOLD_MULTI_CHAR(fold_flag);\r
+ n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc, fold_flag, p, end, items);\r
if (n < 0) return n;\r
\r
for (i = 0; i < n; i++) {\r
ONIGENC_CODE_TO_MBC(enc, items[i].code[0], buf);\r
- add_char_opt_map_info(map, buf[0], enc);\r
+ add_char_opt_map(map, buf[0], enc);\r
}\r
\r
return 0;\r
}\r
\r
static void\r
-select_opt_map_info(OptMapInfo* now, OptMapInfo* alt)\r
+select_opt_map(OptMap* now, OptMap* alt)\r
{\r
static int z = 1<<15; /* 32768: something big value */\r
\r
- int v1, v2;\r
+ int vn, va;\r
\r
if (alt->value == 0) return ;\r
if (now->value == 0) {\r
- copy_opt_map_info(now, alt);\r
+ copy_opt_map(now, alt);\r
return ;\r
}\r
\r
- v1 = z / now->value;\r
- v2 = z / alt->value;\r
- if (comp_distance_value(&now->mmd, &alt->mmd, v1, v2) > 0)\r
- copy_opt_map_info(now, alt);\r
+ vn = z / now->value;\r
+ va = z / alt->value;\r
+ if (comp_distance_value(&now->mmd, &alt->mmd, vn, va) > 0)\r
+ copy_opt_map(now, alt);\r
}\r
\r
static int\r
-comp_opt_exact_or_map_info(OptExactInfo* e, OptMapInfo* m)\r
+comp_opt_exact_or_map(OptExact* e, OptMap* m)\r
{\r
#define COMP_EM_BASE 20\r
- int ve, vm;\r
+ int ae, am;\r
\r
if (m->value <= 0) return -1;\r
\r
- ve = COMP_EM_BASE * e->len * (e->ignore_case ? 1 : 2);\r
- vm = COMP_EM_BASE * 5 * 2 / m->value;\r
- return comp_distance_value(&e->mmd, &m->mmd, ve, vm);\r
+ ae = COMP_EM_BASE * e->len * (e->ignore_case ? 1 : 2);\r
+ am = COMP_EM_BASE * 5 * 2 / m->value;\r
+ return comp_distance_value(&e->mmd, &m->mmd, ae, am);\r
}\r
\r
static void\r
-alt_merge_opt_map_info(OnigEncoding enc, OptMapInfo* to, OptMapInfo* add)\r
+alt_merge_opt_map(OnigEncoding enc, OptMap* to, OptMap* add)\r
{\r
int i, val;\r
\r
/* if (! is_equal_mml(&to->mmd, &add->mmd)) return ; */\r
if (to->value == 0) return ;\r
if (add->value == 0 || to->mmd.max < add->mmd.min) {\r
- clear_opt_map_info(to);\r
+ clear_opt_map(to);\r
return ;\r
}\r
\r
}\r
\r
static void\r
-set_bound_node_opt_info(NodeOptInfo* opt, MinMaxLen* mmd)\r
+set_bound_node_opt_info(NodeOpt* opt, MinMax* plen)\r
{\r
- copy_mml(&(opt->exb.mmd), mmd);\r
- copy_mml(&(opt->expr.mmd), mmd);\r
- copy_mml(&(opt->map.mmd), mmd);\r
+ copy_mml(&(opt->exb.mmd), plen);\r
+ copy_mml(&(opt->expr.mmd), plen);\r
+ copy_mml(&(opt->map.mmd), plen);\r
}\r
\r
static void\r
-clear_node_opt_info(NodeOptInfo* opt)\r
+clear_node_opt_info(NodeOpt* opt)\r
{\r
clear_mml(&opt->len);\r
clear_opt_anc_info(&opt->anc);\r
- clear_opt_exact_info(&opt->exb);\r
- clear_opt_exact_info(&opt->exm);\r
- clear_opt_exact_info(&opt->expr);\r
- clear_opt_map_info(&opt->map);\r
+ clear_opt_exact(&opt->exb);\r
+ clear_opt_exact(&opt->exm);\r
+ clear_opt_exact(&opt->expr);\r
+ clear_opt_map(&opt->map);\r
}\r
\r
static void\r
-copy_node_opt_info(NodeOptInfo* to, NodeOptInfo* from)\r
+copy_node_opt_info(NodeOpt* to, NodeOpt* from)\r
{\r
- *to = *from;\r
+ xmemcpy(to,from,sizeof(NodeOpt));\r
}\r
\r
static void\r
-concat_left_node_opt_info(OnigEncoding enc, NodeOptInfo* to, NodeOptInfo* add)\r
+concat_left_node_opt_info(OnigEncoding enc, NodeOpt* to, NodeOpt* add)\r
{\r
int exb_reach, exm_reach;\r
- OptAncInfo tanc;\r
+ OptAnc tanc;\r
\r
concat_opt_anc_info(&tanc, &to->anc, &add->anc, to->len.max, add->len.max);\r
copy_opt_anc_info(&to->anc, &tanc);\r
\r
if (add->exb.len > 0 && to->len.max == 0) {\r
- concat_opt_anc_info(&tanc, &to->anc, &add->exb.anc,\r
- to->len.max, add->len.max);\r
+ concat_opt_anc_info(&tanc, &to->anc, &add->exb.anc, to->len.max, add->len.max);\r
copy_opt_anc_info(&add->exb.anc, &tanc);\r
}\r
\r
if (add->map.value > 0 && to->len.max == 0) {\r
if (add->map.mmd.max == 0)\r
- add->map.anc.left_anchor |= to->anc.left_anchor;\r
+ add->map.anc.left |= to->anc.left;\r
}\r
\r
exb_reach = to->exb.reach_end;\r
\r
if (add->exb.len > 0) {\r
if (exb_reach) {\r
- concat_opt_exact_info(&to->exb, &add->exb, enc);\r
- clear_opt_exact_info(&add->exb);\r
+ concat_opt_exact(&to->exb, &add->exb, enc);\r
+ clear_opt_exact(&add->exb);\r
}\r
else if (exm_reach) {\r
- concat_opt_exact_info(&to->exm, &add->exb, enc);\r
- clear_opt_exact_info(&add->exb);\r
+ concat_opt_exact(&to->exm, &add->exb, enc);\r
+ clear_opt_exact(&add->exb);\r
}\r
}\r
- select_opt_exact_info(enc, &to->exm, &add->exb);\r
- select_opt_exact_info(enc, &to->exm, &add->exm);\r
+ select_opt_exact(enc, &to->exm, &add->exb);\r
+ select_opt_exact(enc, &to->exm, &add->exm);\r
\r
if (to->expr.len > 0) {\r
if (add->len.max > 0) {\r
if (to->expr.len > (int )add->len.max)\r
- to->expr.len = add->len.max;\r
+ to->expr.len = add->len.max;\r
\r
if (to->expr.mmd.max == 0)\r
- select_opt_exact_info(enc, &to->exb, &to->expr);\r
+ select_opt_exact(enc, &to->exb, &to->expr);\r
else\r
- select_opt_exact_info(enc, &to->exm, &to->expr);\r
+ select_opt_exact(enc, &to->exm, &to->expr);\r
}\r
}\r
else if (add->expr.len > 0) {\r
- copy_opt_exact_info(&to->expr, &add->expr);\r
+ copy_opt_exact(&to->expr, &add->expr);\r
}\r
\r
- select_opt_map_info(&to->map, &add->map);\r
-\r
+ select_opt_map(&to->map, &add->map);\r
add_mml(&to->len, &add->len);\r
}\r
\r
static void\r
-alt_merge_node_opt_info(NodeOptInfo* to, NodeOptInfo* add, OptEnv* env)\r
+alt_merge_node_opt_info(NodeOpt* to, NodeOpt* add, OptEnv* env)\r
{\r
- alt_merge_opt_anc_info (&to->anc, &add->anc);\r
- alt_merge_opt_exact_info(&to->exb, &add->exb, env);\r
- alt_merge_opt_exact_info(&to->exm, &add->exm, env);\r
- alt_merge_opt_exact_info(&to->expr, &add->expr, env);\r
- alt_merge_opt_map_info(env->enc, &to->map, &add->map);\r
+ alt_merge_opt_anc_info(&to->anc, &add->anc);\r
+ alt_merge_opt_exact(&to->exb, &add->exb, env);\r
+ alt_merge_opt_exact(&to->exm, &add->exm, env);\r
+ alt_merge_opt_exact(&to->expr, &add->expr, env);\r
+ alt_merge_opt_map(env->enc, &to->map, &add->map);\r
\r
alt_merge_mml(&to->len, &add->len);\r
}\r
#define MAX_NODE_OPT_INFO_REF_COUNT 5\r
\r
static int\r
-optimize_node_left(Node* node, NodeOptInfo* opt, OptEnv* env)\r
+optimize_nodes(Node* node, NodeOpt* opt, OptEnv* env)\r
{\r
- int type;\r
- int r = 0;\r
+ int i;\r
+ int r;\r
+ NodeOpt xo;\r
+ OnigEncoding enc;\r
\r
+ r = 0;\r
+ enc = env->enc;\r
clear_node_opt_info(opt);\r
set_bound_node_opt_info(opt, &env->mmd);\r
\r
- type = NTYPE(node);\r
- switch (type) {\r
- case NT_LIST:\r
+ switch (NODE_TYPE(node)) {\r
+ case NODE_LIST:\r
{\r
OptEnv nenv;\r
- NodeOptInfo nopt;\r
Node* nd = node;\r
\r
copy_opt_env(&nenv, env);\r
do {\r
- r = optimize_node_left(NCAR(nd), &nopt, &nenv);\r
- if (r == 0) {\r
- add_mml(&nenv.mmd, &nopt.len);\r
- concat_left_node_opt_info(env->enc, opt, &nopt);\r
- }\r
- } while (r == 0 && IS_NOT_NULL(nd = NCDR(nd)));\r
+ r = optimize_nodes(NODE_CAR(nd), &xo, &nenv);\r
+ if (r == 0) {\r
+ add_mml(&nenv.mmd, &xo.len);\r
+ concat_left_node_opt_info(enc, opt, &xo);\r
+ }\r
+ } while (r == 0 && IS_NOT_NULL(nd = NODE_CDR(nd)));\r
}\r
break;\r
\r
- case NT_ALT:\r
+ case NODE_ALT:\r
{\r
- NodeOptInfo nopt;\r
Node* nd = node;\r
\r
do {\r
- r = optimize_node_left(NCAR(nd), &nopt, env);\r
- if (r == 0) {\r
- if (nd == node) copy_node_opt_info(opt, &nopt);\r
- else alt_merge_node_opt_info(opt, &nopt, env);\r
- }\r
- } while ((r == 0) && IS_NOT_NULL(nd = NCDR(nd)));\r
+ r = optimize_nodes(NODE_CAR(nd), &xo, env);\r
+ if (r == 0) {\r
+ if (nd == node) copy_node_opt_info(opt, &xo);\r
+ else alt_merge_node_opt_info(opt, &xo, env);\r
+ }\r
+ } while ((r == 0) && IS_NOT_NULL(nd = NODE_CDR(nd)));\r
}\r
break;\r
\r
- case NT_STR:\r
+ case NODE_STRING:\r
{\r
- StrNode* sn = NSTR(node);\r
- int slen = (int)(sn->end - sn->s);\r
- int is_raw = NSTRING_IS_RAW(node);\r
-\r
- if (! NSTRING_IS_AMBIG(node)) {\r
- concat_opt_exact_info_str(&opt->exb, sn->s, sn->end,\r
- NSTRING_IS_RAW(node), env->enc);\r
- if (slen > 0) {\r
- add_char_opt_map_info(&opt->map, *(sn->s), env->enc);\r
- }\r
+ StrNode* sn = STR_(node);\r
+ int slen = (int )(sn->end - sn->s);\r
+ /* int is_raw = NODE_STRING_IS_RAW(node); */\r
+\r
+ if (! NODE_STRING_IS_AMBIG(node)) {\r
+ concat_opt_exact_str(&opt->exb, sn->s, sn->end, enc);\r
+ if (slen > 0) {\r
+ add_char_opt_map(&opt->map, *(sn->s), enc);\r
+ }\r
set_mml(&opt->len, slen, slen);\r
}\r
else {\r
int max;\r
\r
- if (NSTRING_IS_DONT_GET_OPT_INFO(node)) {\r
- int n = onigenc_strlen(env->enc, sn->s, sn->end);\r
- max = ONIGENC_MBC_MAXLEN_DIST(env->enc) * n;\r
- }\r
- else {\r
- concat_opt_exact_info_str(&opt->exb, sn->s, sn->end,\r
- is_raw, env->enc);\r
- opt->exb.ignore_case = 1;\r
+ if (NODE_STRING_IS_DONT_GET_OPT_INFO(node)) {\r
+ int n = onigenc_strlen(enc, sn->s, sn->end);\r
+ max = ONIGENC_MBC_MAXLEN_DIST(enc) * n;\r
+ }\r
+ else {\r
+ concat_opt_exact_str(&opt->exb, sn->s, sn->end, enc);\r
+ opt->exb.ignore_case = 1;\r
\r
- if (slen > 0) {\r
- r = add_char_amb_opt_map_info(&opt->map, sn->s, sn->end,\r
- env->enc, env->case_fold_flag);\r
- if (r != 0) break;\r
- }\r
+ if (slen > 0) {\r
+ r = add_char_amb_opt_map(&opt->map, sn->s, sn->end,\r
+ enc, env->case_fold_flag);\r
+ if (r != 0) break;\r
+ }\r
\r
- max = slen;\r
- }\r
+ max = slen;\r
+ }\r
\r
set_mml(&opt->len, slen, max);\r
}\r
\r
if (opt->exb.len == slen)\r
- opt->exb.reach_end = 1;\r
+ opt->exb.reach_end = 1;\r
}\r
break;\r
\r
- case NT_CCLASS:\r
+ case NODE_CCLASS:\r
{\r
- int i, z;\r
- CClassNode* cc = NCCLASS(node);\r
+ int z;\r
+ CClassNode* cc = CCLASS_(node);\r
\r
- /* no need to check ignore case. (setted in setup_tree()) */\r
+ /* no need to check ignore case. (set in setup_tree()) */\r
\r
if (IS_NOT_NULL(cc->mbuf) || IS_NCCLASS_NOT(cc)) {\r
- OnigDistance min = ONIGENC_MBC_MINLEN(env->enc);\r
- OnigDistance max = ONIGENC_MBC_MAXLEN_DIST(env->enc);\r
+ OnigLen min = ONIGENC_MBC_MINLEN(enc);\r
+ OnigLen max = ONIGENC_MBC_MAXLEN_DIST(enc);\r
\r
- set_mml(&opt->len, min, max);\r
+ set_mml(&opt->len, min, max);\r
}\r
else {\r
for (i = 0; i < SINGLE_BYTE_SIZE; i++) {\r
z = BITSET_AT(cc->bs, i);\r
- if ((z && !IS_NCCLASS_NOT(cc)) || (!z && IS_NCCLASS_NOT(cc))) {\r
- add_char_opt_map_info(&opt->map, (UChar )i, env->enc);\r
+ if ((z && ! IS_NCCLASS_NOT(cc)) || (! z && IS_NCCLASS_NOT(cc))) {\r
+ add_char_opt_map(&opt->map, (UChar )i, enc);\r
}\r
}\r
- set_mml(&opt->len, 1, 1);\r
+ set_mml(&opt->len, 1, 1);\r
}\r
}\r
break;\r
\r
- case NT_CTYPE:\r
+ case NODE_CTYPE:\r
{\r
- int i, min, max;\r
+ int min, max;\r
+ int range;\r
\r
- max = ONIGENC_MBC_MAXLEN_DIST(env->enc);\r
+ max = ONIGENC_MBC_MAXLEN_DIST(enc);\r
\r
if (max == 1) {\r
min = 1;\r
\r
- switch (NCTYPE(node)->ctype) {\r
- case ONIGENC_CTYPE_WORD:\r
- if (NCTYPE(node)->not != 0) {\r
- for (i = 0; i < SINGLE_BYTE_SIZE; i++) {\r
- if (! ONIGENC_IS_CODE_WORD(env->enc, i)) {\r
- add_char_opt_map_info(&opt->map, (UChar )i, env->enc);\r
- }\r
- }\r
- }\r
- else {\r
- for (i = 0; i < SINGLE_BYTE_SIZE; i++) {\r
- if (ONIGENC_IS_CODE_WORD(env->enc, i)) {\r
- add_char_opt_map_info(&opt->map, (UChar )i, env->enc);\r
- }\r
- }\r
- }\r
- break;\r
- }\r
+ switch (CTYPE_(node)->ctype) {\r
+ case CTYPE_ANYCHAR:\r
+ break;\r
+\r
+ case ONIGENC_CTYPE_WORD:\r
+ range = CTYPE_(node)->ascii_mode != 0 ? 128 : SINGLE_BYTE_SIZE;\r
+ if (CTYPE_(node)->not != 0) {\r
+ for (i = 0; i < range; i++) {\r
+ if (! ONIGENC_IS_CODE_WORD(enc, i)) {\r
+ add_char_opt_map(&opt->map, (UChar )i, enc);\r
+ }\r
+ }\r
+ for (i = range; i < SINGLE_BYTE_SIZE; i++) {\r
+ add_char_opt_map(&opt->map, (UChar )i, enc);\r
+ }\r
+ }\r
+ else {\r
+ for (i = 0; i < range; i++) {\r
+ if (ONIGENC_IS_CODE_WORD(enc, i)) {\r
+ add_char_opt_map(&opt->map, (UChar )i, enc);\r
+ }\r
+ }\r
+ }\r
+ break;\r
+ }\r
}\r
else {\r
- min = ONIGENC_MBC_MINLEN(env->enc);\r
+ min = ONIGENC_MBC_MINLEN(enc);\r
}\r
set_mml(&opt->len, min, max);\r
}\r
break;\r
\r
- case NT_CANY:\r
- {\r
- OnigDistance min = ONIGENC_MBC_MINLEN(env->enc);\r
- OnigDistance max = ONIGENC_MBC_MAXLEN_DIST(env->enc);\r
- set_mml(&opt->len, min, max);\r
- }\r
- break;\r
-\r
- case NT_ANCHOR:\r
- switch (NANCHOR(node)->type) {\r
+ case NODE_ANCHOR:\r
+ switch (ANCHOR_(node)->type) {\r
case ANCHOR_BEGIN_BUF:\r
case ANCHOR_BEGIN_POSITION:\r
case ANCHOR_BEGIN_LINE:\r
case ANCHOR_END_BUF:\r
case ANCHOR_SEMI_END_BUF:\r
case ANCHOR_END_LINE:\r
- add_opt_anc_info(&opt->anc, NANCHOR(node)->type);\r
+ case ANCHOR_PREC_READ_NOT:\r
+ case ANCHOR_LOOK_BEHIND:\r
+ add_opt_anc_info(&opt->anc, ANCHOR_(node)->type);\r
break;\r
\r
case ANCHOR_PREC_READ:\r
{\r
- NodeOptInfo nopt;\r
+ r = optimize_nodes(NODE_BODY(node), &xo, env);\r
+ if (r == 0) {\r
+ if (xo.exb.len > 0)\r
+ copy_opt_exact(&opt->expr, &xo.exb);\r
+ else if (xo.exm.len > 0)\r
+ copy_opt_exact(&opt->expr, &xo.exm);\r
\r
- r = optimize_node_left(NANCHOR(node)->target, &nopt, env);\r
- if (r == 0) {\r
- if (nopt.exb.len > 0)\r
- copy_opt_exact_info(&opt->expr, &nopt.exb);\r
- else if (nopt.exm.len > 0)\r
- copy_opt_exact_info(&opt->expr, &nopt.exm);\r
+ opt->expr.reach_end = 0;\r
\r
- opt->expr.reach_end = 0;\r
-\r
- if (nopt.map.value > 0)\r
- copy_opt_map_info(&opt->map, &nopt.map);\r
- }\r
+ if (xo.map.value > 0)\r
+ copy_opt_map(&opt->map, &xo.map);\r
+ }\r
}\r
break;\r
\r
- case ANCHOR_PREC_READ_NOT:\r
- case ANCHOR_LOOK_BEHIND: /* Sorry, I can't make use of it. */\r
case ANCHOR_LOOK_BEHIND_NOT:\r
break;\r
}\r
break;\r
\r
- case NT_BREF:\r
- {\r
- int i;\r
+ case NODE_BACKREF:\r
+ if (! NODE_IS_CHECKER(node)) {\r
int* backs;\r
- OnigDistance min, max, tmin, tmax;\r
- Node** nodes = SCANENV_MEM_NODES(env->scan_env);\r
- BRefNode* br = NBREF(node);\r
+ OnigLen min, max, tmin, tmax;\r
+ MemEnv* mem_env = SCANENV_MEMENV(env->scan_env);\r
+ BackRefNode* br = BACKREF_(node);\r
\r
- if (br->state & NST_RECURSION) {\r
- set_mml(&opt->len, 0, ONIG_INFINITE_DISTANCE);\r
- break;\r
+ if (NODE_IS_RECURSION(node)) {\r
+ set_mml(&opt->len, 0, INFINITE_LEN);\r
+ break;\r
}\r
backs = BACKREFS_P(br);\r
- r = get_min_match_length(nodes[backs[0]], &min, env->scan_env);\r
- if (r != 0) break;\r
- r = get_max_match_length(nodes[backs[0]], &max, env->scan_env);\r
- if (r != 0) break;\r
+ min = tree_min_len(mem_env[backs[0]].node, env->scan_env);\r
+ max = tree_max_len(mem_env[backs[0]].node, env->scan_env);\r
for (i = 1; i < br->back_num; i++) {\r
- r = get_min_match_length(nodes[backs[i]], &tmin, env->scan_env);\r
- if (r != 0) break;\r
- r = get_max_match_length(nodes[backs[i]], &tmax, env->scan_env);\r
- if (r != 0) break;\r
- if (min > tmin) min = tmin;\r
- if (max < tmax) max = tmax;\r
+ tmin = tree_min_len(mem_env[backs[i]].node, env->scan_env);\r
+ tmax = tree_max_len(mem_env[backs[i]].node, env->scan_env);\r
+ if (min > tmin) min = tmin;\r
+ if (max < tmax) max = tmax;\r
}\r
- if (r == 0) set_mml(&opt->len, min, max);\r
+ set_mml(&opt->len, min, max);\r
}\r
break;\r
\r
-#ifdef USE_SUBEXP_CALL\r
- case NT_CALL:\r
- if (IS_CALL_RECURSION(NCALL(node)))\r
- set_mml(&opt->len, 0, ONIG_INFINITE_DISTANCE);\r
+#ifdef USE_CALL\r
+ case NODE_CALL:\r
+ if (NODE_IS_RECURSION(node))\r
+ set_mml(&opt->len, 0, INFINITE_LEN);\r
else {\r
OnigOptionType save = env->options;\r
- env->options = NENCLOSE(NCALL(node)->target)->option;\r
- r = optimize_node_left(NCALL(node)->target, opt, env);\r
+ env->options = ENCLOSURE_(NODE_BODY(node))->o.options;\r
+ r = optimize_nodes(NODE_BODY(node), opt, env);\r
env->options = save;\r
}\r
break;\r
#endif\r
\r
- case NT_QTFR:\r
+ case NODE_QUANT:\r
{\r
- int i;\r
- OnigDistance min, max;\r
- NodeOptInfo nopt;\r
- QtfrNode* qn = NQTFR(node);\r
-\r
- r = optimize_node_left(qn->target, &nopt, env);\r
- if (r) break;\r
-\r
- if (qn->lower == 0 && IS_REPEAT_INFINITE(qn->upper)) {\r
- if (env->mmd.max == 0 &&\r
- NTYPE(qn->target) == NT_CANY && qn->greedy) {\r
- if (IS_MULTILINE(env->options))\r
- add_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_STAR_ML);\r
- else\r
- add_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_STAR);\r
- }\r
+ OnigLen min, max;\r
+ QuantNode* qn = QUANT_(node);\r
+\r
+ r = optimize_nodes(NODE_BODY(node), &xo, env);\r
+ if (r != 0) break;\r
+\r
+ if (qn->lower > 0) {\r
+ copy_node_opt_info(opt, &xo);\r
+ if (xo.exb.len > 0) {\r
+ if (xo.exb.reach_end) {\r
+ for (i = 2; i <= qn->lower && ! is_full_opt_exact(&opt->exb); i++) {\r
+ int rc = concat_opt_exact(&opt->exb, &xo.exb, enc);\r
+ if (rc > 0) break;\r
+ }\r
+ if (i < qn->lower) opt->exb.reach_end = 0;\r
+ }\r
+ }\r
+\r
+ if (qn->lower != qn->upper) {\r
+ opt->exb.reach_end = 0;\r
+ opt->exm.reach_end = 0;\r
+ }\r
+ if (qn->lower > 1)\r
+ opt->exm.reach_end = 0;\r
+ }\r
+\r
+ if (IS_REPEAT_INFINITE(qn->upper)) {\r
+ if (env->mmd.max == 0 &&\r
+ NODE_IS_ANYCHAR(NODE_BODY(node)) && qn->greedy != 0) {\r
+ if (IS_MULTILINE(CTYPE_OPTION(NODE_QUANT_BODY(qn), env)))\r
+ add_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_INF_ML);\r
+ else\r
+ add_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_INF);\r
+ }\r
+\r
+ max = (xo.len.max > 0 ? INFINITE_LEN : 0);\r
}\r
else {\r
- if (qn->lower > 0) {\r
- copy_node_opt_info(opt, &nopt);\r
- if (nopt.exb.len > 0) {\r
- if (nopt.exb.reach_end) {\r
- for (i = 2; i <= qn->lower &&\r
- ! is_full_opt_exact_info(&opt->exb); i++) {\r
- concat_opt_exact_info(&opt->exb, &nopt.exb, env->enc);\r
- }\r
- if (i < qn->lower) {\r
- opt->exb.reach_end = 0;\r
- }\r
- }\r
- }\r
-\r
- if (qn->lower != qn->upper) {\r
- opt->exb.reach_end = 0;\r
- opt->exm.reach_end = 0;\r
- }\r
- if (qn->lower > 1)\r
- opt->exm.reach_end = 0;\r
- }\r
+ max = distance_multiply(xo.len.max, qn->upper);\r
}\r
\r
- min = distance_multiply(nopt.len.min, qn->lower);\r
- if (IS_REPEAT_INFINITE(qn->upper))\r
- max = (nopt.len.max > 0 ? ONIG_INFINITE_DISTANCE : 0);\r
- else\r
- max = distance_multiply(nopt.len.max, qn->upper);\r
-\r
+ min = distance_multiply(xo.len.min, qn->lower);\r
set_mml(&opt->len, min, max);\r
}\r
break;\r
\r
- case NT_ENCLOSE:\r
- {\r
- EncloseNode* en = NENCLOSE(node);\r
-\r
- switch (en->type) {\r
- case ENCLOSE_OPTION:\r
- {\r
- OnigOptionType save = env->options;\r
-\r
- env->options = en->option;\r
- r = optimize_node_left(en->target, opt, env);\r
- env->options = save;\r
- }\r
- break;\r
-\r
- case ENCLOSE_MEMORY:\r
-#ifdef USE_SUBEXP_CALL\r
- en->opt_count++;\r
- if (en->opt_count > MAX_NODE_OPT_INFO_REF_COUNT) {\r
- OnigDistance min, max;\r
-\r
- min = 0;\r
- max = ONIG_INFINITE_DISTANCE;\r
- if (IS_ENCLOSE_MIN_FIXED(en)) min = en->min_len;\r
- if (IS_ENCLOSE_MAX_FIXED(en)) max = en->max_len;\r
- set_mml(&opt->len, min, max);\r
- }\r
- else\r
-#endif\r
- {\r
- r = optimize_node_left(en->target, opt, env);\r
-\r
- if (is_set_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_STAR_MASK)) {\r
- if (BIT_STATUS_AT(env->scan_env->backrefed_mem, en->regnum))\r
- remove_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_STAR_MASK);\r
- }\r
- }\r
- break;\r
-\r
- case ENCLOSE_STOP_BACKTRACK:\r
- r = optimize_node_left(en->target, opt, env);\r
- break;\r
- }\r
- }\r
+ case NODE_ENCLOSURE:\r
+ {\r
+ EnclosureNode* en = ENCLOSURE_(node);\r
+\r
+ switch (en->type) {\r
+ case ENCLOSURE_OPTION:\r
+ {\r
+ OnigOptionType save = env->options;\r
+\r
+ env->options = en->o.options;\r
+ r = optimize_nodes(NODE_BODY(node), opt, env);\r
+ env->options = save;\r
+ }\r
+ break;\r
+\r
+ case ENCLOSURE_MEMORY:\r
+#ifdef USE_CALL\r
+ en->opt_count++;\r
+ if (en->opt_count > MAX_NODE_OPT_INFO_REF_COUNT) {\r
+ OnigLen min, max;\r
+\r
+ min = 0;\r
+ max = INFINITE_LEN;\r
+ if (NODE_IS_MIN_FIXED(node)) min = en->min_len;\r
+ if (NODE_IS_MAX_FIXED(node)) max = en->max_len;\r
+ set_mml(&opt->len, min, max);\r
+ }\r
+ else\r
+#endif\r
+ {\r
+ r = optimize_nodes(NODE_BODY(node), opt, env);\r
+ if (is_set_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_INF_MASK)) {\r
+ if (MEM_STATUS_AT0(env->scan_env->backrefed_mem, en->m.regnum))\r
+ remove_opt_anc_info(&opt->anc, ANCHOR_ANYCHAR_INF_MASK);\r
+ }\r
+ }\r
+ break;\r
+\r
+ case ENCLOSURE_STOP_BACKTRACK:\r
+ r = optimize_nodes(NODE_BODY(node), opt, env);\r
+ break;\r
+\r
+ case ENCLOSURE_IF_ELSE:\r
+ {\r
+ OptEnv nenv;\r
+\r
+ copy_opt_env(&nenv, env);\r
+ r = optimize_nodes(NODE_ENCLOSURE_BODY(en), &xo, &nenv);\r
+ if (r == 0) {\r
+ add_mml(&nenv.mmd, &xo.len);\r
+ concat_left_node_opt_info(enc, opt, &xo);\r
+ if (IS_NOT_NULL(en->te.Then)) {\r
+ r = optimize_nodes(en->te.Then, &xo, &nenv);\r
+ if (r == 0) {\r
+ concat_left_node_opt_info(enc, opt, &xo);\r
+ }\r
+ }\r
+\r
+ if (IS_NOT_NULL(en->te.Else)) {\r
+ r = optimize_nodes(en->te.Else, &xo, env);\r
+ if (r == 0)\r
+ alt_merge_node_opt_info(opt, &xo, env);\r
+ }\r
+ }\r
+ }\r
+ break;\r
+ }\r
+ }\r
+ break;\r
+\r
+ case NODE_GIMMICK:\r
break;\r
\r
default:\r
#ifdef ONIG_DEBUG\r
- fprintf(stderr, "optimize_node_left: undefined node type %d\n",\r
- NTYPE(node));\r
+ fprintf(stderr, "optimize_nodes: undefined node type %d\n", NODE_TYPE(node));\r
#endif\r
r = ONIGERR_TYPE_BUG;\r
break;\r
}\r
\r
static int\r
-set_optimize_exact_info(regex_t* reg, OptExactInfo* e)\r
+set_optimize_exact(regex_t* reg, OptExact* e)\r
{\r
int r;\r
\r
CHECK_NULL_RETURN_MEMERR(reg->exact);\r
xmemcpy(reg->exact, e->s, e->len);\r
reg->exact_end = reg->exact + e->len;\r
- reg->optimize = ONIG_OPTIMIZE_EXACT_IC;\r
+ reg->optimize = OPTIMIZE_EXACT_IC;\r
}\r
else {\r
int allow_reverse;\r
\r
- reg->exact = str_dup(e->s, e->s + e->len);\r
+ reg->exact = onigenc_strdup(reg->enc, e->s, e->s + e->len);\r
CHECK_NULL_RETURN_MEMERR(reg->exact);\r
reg->exact_end = reg->exact + e->len;\r
\r
allow_reverse =\r
- ONIGENC_IS_ALLOWED_REVERSE_MATCH(reg->enc, reg->exact, reg->exact_end);\r
+ ONIGENC_IS_ALLOWED_REVERSE_MATCH(reg->enc, reg->exact, reg->exact_end);\r
\r
if (e->len >= 3 || (e->len >= 2 && allow_reverse)) {\r
r = set_bm_skip(reg->exact, reg->exact_end, reg->enc,\r
- reg->map, &(reg->int_map));\r
- if (r) return r;\r
+ reg->map, &(reg->int_map));\r
+ if (r != 0) return r;\r
\r
reg->optimize = (allow_reverse != 0\r
- ? ONIG_OPTIMIZE_EXACT_BM : ONIG_OPTIMIZE_EXACT_BM_NOT_REV);\r
+ ? OPTIMIZE_EXACT_BM : OPTIMIZE_EXACT_BM_NO_REV);\r
}\r
else {\r
- reg->optimize = ONIG_OPTIMIZE_EXACT;\r
+ reg->optimize = OPTIMIZE_EXACT;\r
}\r
}\r
\r
reg->dmin = e->mmd.min;\r
reg->dmax = e->mmd.max;\r
\r
- if (reg->dmin != ONIG_INFINITE_DISTANCE) {\r
- reg->threshold_len = reg->dmin + (int)(reg->exact_end - reg->exact);\r
+ if (reg->dmin != INFINITE_LEN) {\r
+ reg->threshold_len = reg->dmin + (int )(reg->exact_end - reg->exact);\r
}\r
\r
return 0;\r
}\r
\r
static void\r
-set_optimize_map_info(regex_t* reg, OptMapInfo* m)\r
+set_optimize_map(regex_t* reg, OptMap* m)\r
{\r
int i;\r
\r
for (i = 0; i < ONIG_CHAR_TABLE_SIZE; i++)\r
reg->map[i] = m->map[i];\r
\r
- reg->optimize = ONIG_OPTIMIZE_MAP;\r
+ reg->optimize = OPTIMIZE_MAP;\r
reg->dmin = m->mmd.min;\r
reg->dmax = m->mmd.max;\r
\r
- if (reg->dmin != ONIG_INFINITE_DISTANCE) {\r
+ if (reg->dmin != INFINITE_LEN) {\r
reg->threshold_len = reg->dmin + 1;\r
}\r
}\r
\r
static void\r
-set_sub_anchor(regex_t* reg, OptAncInfo* anc)\r
+set_sub_anchor(regex_t* reg, OptAnc* anc)\r
{\r
- reg->sub_anchor |= anc->left_anchor & ANCHOR_BEGIN_LINE;\r
- reg->sub_anchor |= anc->right_anchor & ANCHOR_END_LINE;\r
+ reg->sub_anchor |= anc->left & ANCHOR_BEGIN_LINE;\r
+ reg->sub_anchor |= anc->right & ANCHOR_END_LINE;\r
}\r
\r
-#ifdef ONIG_DEBUG\r
+#if defined(ONIG_DEBUG_COMPILE) || defined(ONIG_DEBUG_MATCH)\r
static void print_optimize_info(FILE* f, regex_t* reg);\r
#endif\r
\r
static int\r
set_optimize_info_from_tree(Node* node, regex_t* reg, ScanEnv* scan_env)\r
{\r
-\r
int r;\r
- NodeOptInfo opt;\r
+ NodeOpt opt;\r
OptEnv env;\r
\r
env.enc = reg->enc;\r
env.options = reg->options;\r
env.case_fold_flag = reg->case_fold_flag;\r
- env.scan_env = scan_env;\r
+ env.scan_env = scan_env;\r
clear_mml(&env.mmd);\r
\r
- r = optimize_node_left(node, &opt, &env);\r
- if (r) return r;\r
+ r = optimize_nodes(node, &opt, &env);\r
+ if (r != 0) return r;\r
+\r
+ reg->anchor = opt.anc.left & (ANCHOR_BEGIN_BUF |\r
+ ANCHOR_BEGIN_POSITION | ANCHOR_ANYCHAR_INF | ANCHOR_ANYCHAR_INF_ML |\r
+ ANCHOR_LOOK_BEHIND);\r
\r
- reg->anchor = opt.anc.left_anchor & (ANCHOR_BEGIN_BUF |\r
- ANCHOR_BEGIN_POSITION | ANCHOR_ANYCHAR_STAR | ANCHOR_ANYCHAR_STAR_ML);\r
+ if ((opt.anc.left & (ANCHOR_LOOK_BEHIND | ANCHOR_PREC_READ_NOT)) != 0)\r
+ reg->anchor &= ~ANCHOR_ANYCHAR_INF_ML;\r
\r
- reg->anchor |= opt.anc.right_anchor & (ANCHOR_END_BUF | ANCHOR_SEMI_END_BUF);\r
+ reg->anchor |= opt.anc.right & (ANCHOR_END_BUF | ANCHOR_SEMI_END_BUF |\r
+ ANCHOR_PREC_READ_NOT);\r
\r
if (reg->anchor & (ANCHOR_END_BUF | ANCHOR_SEMI_END_BUF)) {\r
reg->anchor_dmin = opt.len.min;\r
}\r
\r
if (opt.exb.len > 0 || opt.exm.len > 0) {\r
- select_opt_exact_info(reg->enc, &opt.exb, &opt.exm);\r
- if (opt.map.value > 0 &&\r
- comp_opt_exact_or_map_info(&opt.exb, &opt.map) > 0) {\r
+ select_opt_exact(reg->enc, &opt.exb, &opt.exm);\r
+ if (opt.map.value > 0 && comp_opt_exact_or_map(&opt.exb, &opt.map) > 0) {\r
goto set_map;\r
}\r
else {\r
- r = set_optimize_exact_info(reg, &opt.exb);\r
+ r = set_optimize_exact(reg, &opt.exb);\r
set_sub_anchor(reg, &opt.exb.anc);\r
}\r
}\r
else if (opt.map.value > 0) {\r
set_map:\r
- set_optimize_map_info(reg, &opt.map);\r
+ set_optimize_map(reg, &opt.map);\r
set_sub_anchor(reg, &opt.map.anc);\r
}\r
else {\r
- reg->sub_anchor |= opt.anc.left_anchor & ANCHOR_BEGIN_LINE;\r
+ reg->sub_anchor |= opt.anc.left & ANCHOR_BEGIN_LINE;\r
if (opt.len.max == 0)\r
- reg->sub_anchor |= opt.anc.right_anchor & ANCHOR_END_LINE;\r
+ reg->sub_anchor |= opt.anc.right & ANCHOR_END_LINE;\r
}\r
\r
#if defined(ONIG_DEBUG_COMPILE) || defined(ONIG_DEBUG_MATCH)\r
static void\r
clear_optimize_info(regex_t* reg)\r
{\r
- reg->optimize = ONIG_OPTIMIZE_NONE;\r
+ reg->optimize = OPTIMIZE_NONE;\r
reg->anchor = 0;\r
reg->anchor_dmin = 0;\r
reg->anchor_dmax = 0;\r
#ifdef ONIG_DEBUG\r
\r
static void print_enc_string(FILE* fp, OnigEncoding enc,\r
- const UChar *s, const UChar *end)\r
+ const UChar *s, const UChar *end)\r
{\r
fprintf(fp, "\nPATTERN: /");\r
\r
while (p < end) {\r
code = ONIGENC_MBC_TO_CODE(enc, p, end);\r
if (code >= 0x80) {\r
- fprintf(fp, " 0x%04x ", (int )code);\r
+ fprintf(fp, " 0x%04x ", (int )code);\r
}\r
else {\r
- fputc((int )code, fp);\r
+ fputc((int )code, fp);\r
}\r
\r
p += enclen(enc, p);\r
fprintf(fp, "/\n");\r
}\r
\r
+#endif /* ONIG_DEBUG */\r
+\r
+#if defined(ONIG_DEBUG_COMPILE) || defined(ONIG_DEBUG_MATCH)\r
+\r
static void\r
-print_distance_range(FILE* f, OnigDistance a, OnigDistance b)\r
+print_distance_range(FILE* f, OnigLen a, OnigLen b)\r
{\r
- if (a == ONIG_INFINITE_DISTANCE)\r
+ if (a == INFINITE_LEN)\r
fputs("inf", f);\r
else\r
fprintf(f, "(%u)", a);\r
\r
fputs("-", f);\r
\r
- if (b == ONIG_INFINITE_DISTANCE)\r
+ if (b == INFINITE_LEN)\r
fputs("inf", f);\r
else\r
fprintf(f, "(%u)", b);\r
q = 1;\r
fprintf(f, "end-line");\r
}\r
- if (anchor & ANCHOR_ANYCHAR_STAR) {\r
+ if (anchor & ANCHOR_ANYCHAR_INF) {\r
if (q) fprintf(f, ", ");\r
q = 1;\r
- fprintf(f, "anychar-star");\r
+ fprintf(f, "anychar-inf");\r
}\r
- if (anchor & ANCHOR_ANYCHAR_STAR_ML) {\r
+ if (anchor & ANCHOR_ANYCHAR_INF_ML) {\r
if (q) fprintf(f, ", ");\r
- fprintf(f, "anychar-star-pl");\r
+ fprintf(f, "anychar-inf-ml");\r
}\r
\r
fprintf(f, "]");\r
for (p = reg->exact; p < reg->exact_end; p++) {\r
fputc(*p, f);\r
}\r
- fprintf(f, "]: length: %d\n", (reg->exact_end - reg->exact));\r
+ fprintf(f, "]: length: %ld\n", (reg->exact_end - reg->exact));\r
}\r
- else if (reg->optimize & ONIG_OPTIMIZE_MAP) {\r
+ else if (reg->optimize & OPTIMIZE_MAP) {\r
int c, i, n = 0;\r
\r
for (i = 0; i < ONIG_CHAR_TABLE_SIZE; i++)\r
c = 0;\r
fputc('[', f);\r
for (i = 0; i < ONIG_CHAR_TABLE_SIZE; i++) {\r
- if (reg->map[i] != 0) {\r
+ if (reg->map[i] != 0) {\r
if (c > 0) fputs(", ", f);\r
c++;\r
if (ONIGENC_MBC_MAXLEN(reg->enc) == 1 &&\r
}\r
}\r
}\r
-#endif /* ONIG_DEBUG */\r
+#endif\r
+\r
+\r
+extern RegexExt*\r
+onig_get_regex_ext(regex_t* reg)\r
+{\r
+ if (IS_NULL(REG_EXTP(reg))) {\r
+ RegexExt* ext = (RegexExt* )xmalloc(sizeof(*ext));\r
+ if (IS_NULL(ext)) return 0;\r
+\r
+ ext->pattern = 0;\r
+ ext->pattern_end = 0;\r
+#ifdef USE_CALLOUT\r
+ ext->tag_table = 0;\r
+ ext->callout_num = 0;\r
+ ext->callout_list_alloc = 0;\r
+ ext->callout_list = 0;\r
+#endif\r
+\r
+ REG_EXTPL(reg) = (void* )ext;\r
+ }\r
+\r
+ return REG_EXTP(reg);\r
+}\r
+\r
+static void\r
+free_regex_ext(RegexExt* ext)\r
+{\r
+ if (IS_NOT_NULL(ext)) {\r
+ if (IS_NOT_NULL(ext->pattern))\r
+ xfree((void* )ext->pattern);\r
+\r
+#ifdef USE_CALLOUT\r
+ if (IS_NOT_NULL(ext->tag_table))\r
+ onig_callout_tag_table_free(ext->tag_table);\r
+\r
+ if (IS_NOT_NULL(ext->callout_list))\r
+ onig_free_reg_callout_list(ext->callout_num, ext->callout_list);\r
+#endif\r
+\r
+ xfree(ext);\r
+ }\r
+}\r
+\r
+extern int\r
+onig_ext_set_pattern(regex_t* reg, const UChar* pattern, const UChar* pattern_end)\r
+{\r
+ RegexExt* ext;\r
+ UChar* s;\r
+\r
+ ext = onig_get_regex_ext(reg);\r
+ CHECK_NULL_RETURN_MEMERR(ext);\r
+\r
+ s = onigenc_strdup(reg->enc, pattern, pattern_end);\r
+ CHECK_NULL_RETURN_MEMERR(s);\r
+\r
+ ext->pattern = s;\r
+ ext->pattern_end = s + (pattern_end - pattern);\r
+\r
+ return ONIG_NORMAL;\r
+}\r
\r
\r
extern void\r
if (IS_NOT_NULL(reg->int_map)) xfree(reg->int_map);\r
if (IS_NOT_NULL(reg->int_map_backward)) xfree(reg->int_map_backward);\r
if (IS_NOT_NULL(reg->repeat_range)) xfree(reg->repeat_range);\r
- if (IS_NOT_NULL(reg->chain)) onig_free(reg->chain);\r
+ if (IS_NOT_NULL(REG_EXTP(reg))) {\r
+ free_regex_ext(REG_EXTP(reg));\r
+ REG_EXTPL(reg) = 0;\r
+ }\r
\r
-#ifdef USE_NAMED_GROUP\r
onig_names_free(reg);\r
-#endif\r
}\r
}\r
\r
}\r
\r
#define REGEX_TRANSFER(to,from) do {\\r
- (to)->state = ONIG_STATE_MODIFY;\\r
onig_free_body(to);\\r
xmemcpy(to, from, sizeof(regex_t));\\r
xfree(from);\\r
extern void\r
onig_transfer(regex_t* to, regex_t* from)\r
{\r
- THREAD_ATOMIC_START;\r
REGEX_TRANSFER(to, from);\r
- THREAD_ATOMIC_END;\r
-}\r
-\r
-#define REGEX_CHAIN_HEAD(reg) do {\\r
- while (IS_NOT_NULL((reg)->chain)) {\\r
- (reg) = (reg)->chain;\\r
- }\\r
-} while (0)\r
-\r
-extern void\r
-onig_chain_link_add(regex_t* to, regex_t* add)\r
-{\r
- THREAD_ATOMIC_START;\r
- REGEX_CHAIN_HEAD(to);\r
- to->chain = add;\r
- THREAD_ATOMIC_END;\r
}\r
\r
-extern void\r
-onig_chain_reduce(regex_t* reg)\r
-{\r
- regex_t *head, *prev;\r
-\r
- prev = reg;\r
- head = prev->chain;\r
- if (IS_NOT_NULL(head)) {\r
- reg->state = ONIG_STATE_MODIFY;\r
- while (IS_NOT_NULL(head->chain)) {\r
- prev = head;\r
- head = head->chain;\r
- }\r
- prev->chain = (regex_t* )NULL;\r
- REGEX_TRANSFER(reg, head);\r
- }\r
-}\r
\r
-#ifdef ONIG_DEBUG\r
-static void print_compiled_byte_code_list P_((FILE* f, regex_t* reg));\r
-#endif\r
-#ifdef ONIG_DEBUG_PARSE_TREE\r
+#ifdef ONIG_DEBUG_PARSE\r
static void print_tree P_((FILE* f, Node* node));\r
#endif\r
\r
extern int\r
onig_compile(regex_t* reg, const UChar* pattern, const UChar* pattern_end,\r
- OnigErrorInfo* einfo)\r
+ OnigErrorInfo* einfo)\r
{\r
#define COMPILE_INIT_SIZE 20\r
\r
int r, init_size;\r
Node* root;\r
ScanEnv scan_env;\r
-#ifdef USE_SUBEXP_CALL\r
+#ifdef USE_CALL\r
UnsetAddrList uslist;\r
#endif\r
\r
- if (IS_NOT_NULL(einfo)) einfo->par = (UChar* )NULL;\r
-\r
- reg->state = ONIG_STATE_COMPILING;\r
+ root = 0;\r
+ if (IS_NOT_NULL(einfo)) {\r
+ einfo->enc = reg->enc;\r
+ einfo->par = (UChar* )NULL;\r
+ }\r
\r
#ifdef ONIG_DEBUG\r
print_enc_string(stderr, reg->enc, pattern, pattern_end);\r
#endif\r
\r
if (reg->alloc == 0) {\r
- init_size = ((int)(pattern_end - pattern)) * 2;\r
+ init_size = (int )(pattern_end - pattern) * 2;\r
if (init_size <= 0) init_size = COMPILE_INIT_SIZE;\r
- r = BBUF_INIT(reg, init_size);\r
+ r = BB_INIT(reg, init_size);\r
if (r != 0) goto end;\r
}\r
else\r
reg->num_null_check = 0;\r
reg->repeat_range_alloc = 0;\r
reg->repeat_range = (OnigRepeatRange* )NULL;\r
-#ifdef USE_COMBINATION_EXPLOSION_CHECK\r
- reg->num_comb_exp_check = 0;\r
-#endif\r
\r
- r = onig_parse_make_tree(&root, pattern, pattern_end, reg, &scan_env);\r
- if (r != 0 || root == NULL) goto err;\r
+ r = onig_parse_tree(&root, pattern, pattern_end, reg, &scan_env);\r
+ if (r != 0) goto err;\r
\r
-#ifdef USE_NAMED_GROUP\r
/* mixed use named group and no-named group */\r
if (scan_env.num_named > 0 &&\r
IS_SYNTAX_BV(scan_env.syntax, ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP) &&\r
- !ONIG_IS_OPTION_ON(reg->options, ONIG_OPTION_CAPTURE_GROUP)) {\r
+ ! ONIG_IS_OPTION_ON(reg->options, ONIG_OPTION_CAPTURE_GROUP)) {\r
if (scan_env.num_named != scan_env.num_mem)\r
r = disable_noname_group_capture(&root, reg, &scan_env);\r
else\r
\r
if (r != 0) goto err;\r
}\r
-#endif\r
\r
-#ifdef USE_SUBEXP_CALL\r
+ r = check_backrefs(root, &scan_env);\r
+ if (r != 0) goto err;\r
+\r
+#ifdef USE_CALL\r
if (scan_env.num_call > 0) {\r
r = unset_addr_list_init(&uslist, scan_env.num_call);\r
if (r != 0) goto err;\r
scan_env.unset_addr_list = &uslist;\r
- r = setup_subexp_call(root, &scan_env);\r
+ r = setup_call(root, &scan_env, 0);\r
+ if (r != 0) goto err_unset;\r
+ r = setup_call2(root);\r
if (r != 0) goto err_unset;\r
- r = subexp_recursive_check_trav(root, &scan_env);\r
+ r = recursive_call_check_trav(root, &scan_env, 0);\r
if (r < 0) goto err_unset;\r
- r = subexp_inf_recursive_check_trav(root, &scan_env);\r
+ r = infinite_recursive_call_check_trav(root, &scan_env);\r
if (r != 0) goto err_unset;\r
\r
- reg->num_call = scan_env.num_call;\r
+ setup_called_state(root, 0);\r
}\r
- else\r
- reg->num_call = 0;\r
+\r
+ reg->num_call = scan_env.num_call;\r
#endif\r
\r
r = setup_tree(root, reg, 0, &scan_env);\r
if (r != 0) goto err_unset;\r
\r
-#ifdef ONIG_DEBUG_PARSE_TREE\r
+#ifdef ONIG_DEBUG_PARSE\r
print_tree(stderr, root);\r
#endif\r
\r
reg->bt_mem_start = scan_env.bt_mem_start;\r
reg->bt_mem_start |= reg->capture_history;\r
if (IS_FIND_CONDITION(reg->options))\r
- BIT_STATUS_ON_ALL(reg->bt_mem_end);\r
+ MEM_STATUS_ON_ALL(reg->bt_mem_end);\r
else {\r
reg->bt_mem_end = scan_env.bt_mem_end;\r
reg->bt_mem_end |= reg->capture_history;\r
}\r
-\r
-#ifdef USE_COMBINATION_EXPLOSION_CHECK\r
- if (scan_env.backrefed_mem == 0\r
-#ifdef USE_SUBEXP_CALL\r
- || scan_env.num_call == 0\r
-#endif\r
- ) {\r
- setup_comb_exp_check(root, 0, &scan_env);\r
-#ifdef USE_SUBEXP_CALL\r
- if (scan_env.has_recursion != 0) {\r
- scan_env.num_comb_exp_check = 0;\r
- }\r
- else\r
-#endif\r
- if (scan_env.comb_exp_max_regnum > 0) {\r
- int i;\r
- for (i = 1; i <= scan_env.comb_exp_max_regnum; i++) {\r
- if (BIT_STATUS_AT(scan_env.backrefed_mem, i) != 0) {\r
- scan_env.num_comb_exp_check = 0;\r
- break;\r
- }\r
- }\r
- }\r
- }\r
-\r
- reg->num_comb_exp_check = scan_env.num_comb_exp_check;\r
-#endif\r
+ reg->bt_mem_start |= reg->bt_mem_end;\r
\r
clear_optimize_info(reg);\r
#ifndef ONIG_DONT_OPTIMIZE\r
if (r != 0) goto err_unset;\r
#endif\r
\r
- if (IS_NOT_NULL(scan_env.mem_nodes_dynamic)) {\r
- xfree(scan_env.mem_nodes_dynamic);\r
- scan_env.mem_nodes_dynamic = (Node** )NULL;\r
+ if (IS_NOT_NULL(scan_env.mem_env_dynamic)) {\r
+ xfree(scan_env.mem_env_dynamic);\r
+ scan_env.mem_env_dynamic = (MemEnv* )NULL;\r
}\r
\r
- r = compile_tree(root, reg);\r
+ r = compile_tree(root, reg, &scan_env);\r
if (r == 0) {\r
+ if (scan_env.keep_num > 0) {\r
+ r = add_opcode(reg, OP_UPDATE_VAR);\r
+ if (r != 0) goto err;\r
+ r = add_update_var_type(reg, UPDATE_VAR_KEEP_FROM_STACK_LAST);\r
+ if (r != 0) goto err;\r
+ r = add_mem_num(reg, 0 /* not used */);\r
+ if (r != 0) goto err;\r
+ }\r
+\r
r = add_opcode(reg, OP_END);\r
-#ifdef USE_SUBEXP_CALL\r
+#ifdef USE_CALL\r
if (scan_env.num_call > 0) {\r
- r = unset_addr_list_fix(&uslist, reg);\r
+ r = fix_unset_addr_list(&uslist, reg);\r
unset_addr_list_end(&uslist);\r
- if (r) goto err;\r
+ if (r != 0) goto err;\r
}\r
#endif\r
\r
- if ((reg->num_repeat != 0) || (reg->bt_mem_end != 0))\r
+ if ((reg->num_repeat != 0) || (reg->bt_mem_end != 0)\r
+#ifdef USE_CALLOUT\r
+ || (IS_NOT_NULL(REG_EXTP(reg)) && REG_EXTP(reg)->callout_num != 0)\r
+#endif\r
+ )\r
reg->stack_pop_level = STACK_POP_LEVEL_ALL;\r
else {\r
if (reg->bt_mem_start != 0)\r
- reg->stack_pop_level = STACK_POP_LEVEL_MEM_START;\r
+ reg->stack_pop_level = STACK_POP_LEVEL_MEM_START;\r
else\r
- reg->stack_pop_level = STACK_POP_LEVEL_FREE;\r
+ reg->stack_pop_level = STACK_POP_LEVEL_FREE;\r
}\r
}\r
-#ifdef USE_SUBEXP_CALL\r
+#ifdef USE_CALL\r
else if (scan_env.num_call > 0) {\r
unset_addr_list_end(&uslist);\r
}\r
onig_node_free(root);\r
\r
#ifdef ONIG_DEBUG_COMPILE\r
-#ifdef USE_NAMED_GROUP\r
onig_print_names(stderr, reg);\r
-#endif\r
- print_compiled_byte_code_list(stderr, reg);\r
+ onig_print_compiled_byte_code_list(stderr, reg);\r
#endif\r
\r
end:\r
- reg->state = ONIG_STATE_NORMAL;\r
return r;\r
\r
err_unset:\r
-#ifdef USE_SUBEXP_CALL\r
+#ifdef USE_CALL\r
if (scan_env.num_call > 0) {\r
unset_addr_list_end(&uslist);\r
}\r
err:\r
if (IS_NOT_NULL(scan_env.error)) {\r
if (IS_NOT_NULL(einfo)) {\r
- einfo->enc = scan_env.enc;\r
einfo->par = scan_env.error;\r
einfo->par_end = scan_env.error_end;\r
}\r
}\r
\r
onig_node_free(root);\r
- if (IS_NOT_NULL(scan_env.mem_nodes_dynamic))\r
- xfree(scan_env.mem_nodes_dynamic);\r
+ if (IS_NOT_NULL(scan_env.mem_env_dynamic))\r
+ xfree(scan_env.mem_env_dynamic);\r
return r;\r
}\r
\r
-#ifdef USE_RECOMPILE_API\r
+\r
+static int onig_inited = 0;\r
+\r
extern int\r
-onig_recompile(regex_t* reg, const UChar* pattern, const UChar* pattern_end,\r
- OnigOptionType option, OnigEncoding enc, OnigSyntaxType* syntax,\r
- OnigErrorInfo* einfo)\r
+onig_reg_init(regex_t* reg, OnigOptionType option, OnigCaseFoldType case_fold_flag,\r
+ OnigEncoding enc, OnigSyntaxType* syntax)\r
{\r
int r;\r
- regex_t *new_reg;\r
\r
- r = onig_new(&new_reg, pattern, pattern_end, option, enc, syntax, einfo);\r
- if (r) return r;\r
- if (ONIG_STATE(reg) == ONIG_STATE_NORMAL) {\r
- onig_transfer(reg, new_reg);\r
- }\r
- else {\r
- onig_chain_link_add(reg, new_reg);\r
- }\r
- return 0;\r
-}\r
-#endif\r
+ xmemset(reg, 0, sizeof(*reg));\r
\r
-static int onig_inited = 0;\r
+ if (onig_inited == 0) {\r
+#if 0\r
+ return ONIGERR_LIBRARY_IS_NOT_INITIALIZED;\r
+#else\r
+ r = onig_initialize(&enc, 1);\r
+ if (r != 0)\r
+ return ONIGERR_FAIL_TO_INITIALIZE;\r
\r
-extern int\r
-onig_reg_init(regex_t* reg, OnigOptionType option,\r
- OnigCaseFoldType case_fold_flag,\r
- OnigEncoding enc, OnigSyntaxType* syntax)\r
-{\r
- if (! onig_inited)\r
- onig_init();\r
+ onig_warning("You didn't call onig_initialize() explicitly");\r
+#endif\r
+ }\r
\r
if (IS_NULL(reg))\r
return ONIGERR_INVALID_ARGUMENT;\r
return ONIGERR_INVALID_COMBINATION_OF_OPTIONS;\r
}\r
\r
- (reg)->state = ONIG_STATE_MODIFY;\r
-\r
if ((option & ONIG_OPTION_NEGATE_SINGLELINE) != 0) {\r
option |= syntax->options;\r
option &= ~ONIG_OPTION_SINGLELINE;\r
(reg)->exact = (UChar* )NULL;\r
(reg)->int_map = (int* )NULL;\r
(reg)->int_map_backward = (int* )NULL;\r
- (reg)->chain = (regex_t* )NULL;\r
+ REG_EXTPL(reg) = NULL;\r
\r
(reg)->p = (UChar* )NULL;\r
(reg)->alloc = 0;\r
}\r
\r
extern int\r
-onig_new_without_alloc(regex_t* reg, const UChar* pattern,\r
- const UChar* pattern_end, OnigOptionType option, OnigEncoding enc,\r
- OnigSyntaxType* syntax, OnigErrorInfo* einfo)\r
+onig_new_without_alloc(regex_t* reg,\r
+ const UChar* pattern, const UChar* pattern_end,\r
+ OnigOptionType option, OnigEncoding enc,\r
+ OnigSyntaxType* syntax, OnigErrorInfo* einfo)\r
{\r
int r;\r
\r
r = onig_reg_init(reg, option, ONIGENC_CASE_FOLD_DEFAULT, enc, syntax);\r
- if (r) return r;\r
+ if (r != 0) return r;\r
\r
r = onig_compile(reg, pattern, pattern_end, einfo);\r
return r;\r
\r
extern int\r
onig_new(regex_t** reg, const UChar* pattern, const UChar* pattern_end,\r
- OnigOptionType option, OnigEncoding enc, OnigSyntaxType* syntax,\r
- OnigErrorInfo* einfo)\r
+ OnigOptionType option, OnigEncoding enc, OnigSyntaxType* syntax,\r
+ OnigErrorInfo* einfo)\r
{\r
int r;\r
\r
if (IS_NULL(*reg)) return ONIGERR_MEMORY;\r
\r
r = onig_reg_init(*reg, option, ONIGENC_CASE_FOLD_DEFAULT, enc, syntax);\r
- if (r) goto err;\r
+ if (r != 0) goto err;\r
\r
r = onig_compile(*reg, pattern, pattern_end, einfo);\r
- if (r) {\r
+ if (r != 0) {\r
err:\r
onig_free(*reg);\r
*reg = NULL;\r
return r;\r
}\r
\r
-\r
extern int\r
-onig_init(void)\r
+onig_initialize(OnigEncoding encodings[], int n)\r
{\r
+ int i;\r
+ int r;\r
+\r
if (onig_inited != 0)\r
return 0;\r
\r
- THREAD_SYSTEM_INIT;\r
- THREAD_ATOMIC_START;\r
+ onigenc_init();\r
\r
onig_inited = 1;\r
\r
- onigenc_init();\r
- /* onigenc_set_default_caseconv_table((UChar* )0); */\r
-\r
-#ifdef ONIG_DEBUG_STATISTICS\r
- onig_statistics_init();\r
-#endif\r
+ for (i = 0; i < n; i++) {\r
+ OnigEncoding enc = encodings[i];\r
+ r = onig_initialize_encoding(enc);\r
+ if (r != 0)\r
+ return r;\r
+ }\r
\r
- THREAD_ATOMIC_END;\r
- return 0;\r
+ return ONIG_NORMAL;\r
}\r
\r
+typedef struct EndCallListItem {\r
+ struct EndCallListItem* next;\r
+ void (*func)(void);\r
+} EndCallListItemType;\r
\r
-static OnigEndCallListItemType* EndCallTop;\r
+static EndCallListItemType* EndCallTop;\r
\r
extern void onig_add_end_call(void (*func)(void))\r
{\r
- OnigEndCallListItemType* item;\r
+ EndCallListItemType* item;\r
\r
- item = (OnigEndCallListItemType* )xmalloc(sizeof(*item));\r
+ item = (EndCallListItemType* )xmalloc(sizeof(*item));\r
if (item == 0) return ;\r
\r
item->next = EndCallTop;\r
static void\r
exec_end_call_list(void)\r
{\r
- OnigEndCallListItemType* prev;\r
+ EndCallListItemType* prev;\r
void (*func)(void);\r
\r
while (EndCallTop != 0) {\r
extern int\r
onig_end(void)\r
{\r
- THREAD_ATOMIC_START;\r
-\r
exec_end_call_list();\r
\r
-#ifdef ONIG_DEBUG_STATISTICS\r
- onig_print_statistics(stderr);\r
+#ifdef USE_CALLOUT\r
+ onig_global_callout_names_free();\r
#endif\r
\r
-#ifdef USE_SHARED_CCLASS_TABLE\r
- onig_free_shared_cclass_table();\r
-#endif\r
-\r
-#ifdef USE_PARSE_TREE_NODE_RECYCLE\r
- onig_free_node_list();\r
-#endif\r
+ onigenc_end();\r
\r
onig_inited = 0;\r
\r
- THREAD_ATOMIC_END;\r
- THREAD_SYSTEM_END;\r
return 0;\r
}\r
\r
}\r
\r
extern int\r
-onig_is_code_in_cc_len(int elen, OnigCodePoint code, CClassNode* cc)\r
+onig_is_code_in_cc_len(int elen, OnigCodePoint code, /* CClassNode* */ void* cc_arg)\r
{\r
int found;\r
+ CClassNode* cc = (CClassNode* )cc_arg;\r
\r
if (elen > 1 || (code >= SINGLE_BYTE_SIZE)) {\r
if (IS_NULL(cc->mbuf)) {\r
}\r
\r
\r
-#ifdef ONIG_DEBUG\r
-\r
-/* arguments type */\r
-#define ARG_SPECIAL -1\r
-#define ARG_NON 0\r
-#define ARG_RELADDR 1\r
-#define ARG_ABSADDR 2\r
-#define ARG_LENGTH 3\r
-#define ARG_MEMNUM 4\r
-#define ARG_OPTION 5\r
-#define ARG_STATE_CHECK 6\r
-\r
-OnigOpInfoType OnigOpInfo[] = {\r
- { OP_FINISH, "finish", ARG_NON },\r
- { OP_END, "end", ARG_NON },\r
- { OP_EXACT1, "exact1", ARG_SPECIAL },\r
- { OP_EXACT2, "exact2", ARG_SPECIAL },\r
- { OP_EXACT3, "exact3", ARG_SPECIAL },\r
- { OP_EXACT4, "exact4", ARG_SPECIAL },\r
- { OP_EXACT5, "exact5", ARG_SPECIAL },\r
- { OP_EXACTN, "exactn", ARG_SPECIAL },\r
- { OP_EXACTMB2N1, "exactmb2-n1", ARG_SPECIAL },\r
- { OP_EXACTMB2N2, "exactmb2-n2", ARG_SPECIAL },\r
- { OP_EXACTMB2N3, "exactmb2-n3", ARG_SPECIAL },\r
- { OP_EXACTMB2N, "exactmb2-n", ARG_SPECIAL },\r
- { OP_EXACTMB3N, "exactmb3n" , ARG_SPECIAL },\r
- { OP_EXACTMBN, "exactmbn", ARG_SPECIAL },\r
- { OP_EXACT1_IC, "exact1-ic", ARG_SPECIAL },\r
- { OP_EXACTN_IC, "exactn-ic", ARG_SPECIAL },\r
- { OP_CCLASS, "cclass", ARG_SPECIAL },\r
- { OP_CCLASS_MB, "cclass-mb", ARG_SPECIAL },\r
- { OP_CCLASS_MIX, "cclass-mix", ARG_SPECIAL },\r
- { OP_CCLASS_NOT, "cclass-not", ARG_SPECIAL },\r
- { OP_CCLASS_MB_NOT, "cclass-mb-not", ARG_SPECIAL },\r
- { OP_CCLASS_MIX_NOT, "cclass-mix-not", ARG_SPECIAL },\r
- { OP_CCLASS_NODE, "cclass-node", ARG_SPECIAL },\r
- { OP_ANYCHAR, "anychar", ARG_NON },\r
- { OP_ANYCHAR_ML, "anychar-ml", ARG_NON },\r
- { OP_ANYCHAR_STAR, "anychar*", ARG_NON },\r
- { OP_ANYCHAR_ML_STAR, "anychar-ml*", ARG_NON },\r
- { OP_ANYCHAR_STAR_PEEK_NEXT, "anychar*-peek-next", ARG_SPECIAL },\r
- { OP_ANYCHAR_ML_STAR_PEEK_NEXT, "anychar-ml*-peek-next", ARG_SPECIAL },\r
- { OP_WORD, "word", ARG_NON },\r
- { OP_NOT_WORD, "not-word", ARG_NON },\r
- { OP_WORD_BOUND, "word-bound", ARG_NON },\r
- { OP_NOT_WORD_BOUND, "not-word-bound", ARG_NON },\r
- { OP_WORD_BEGIN, "word-begin", ARG_NON },\r
- { OP_WORD_END, "word-end", ARG_NON },\r
- { OP_BEGIN_BUF, "begin-buf", ARG_NON },\r
- { OP_END_BUF, "end-buf", ARG_NON },\r
- { OP_BEGIN_LINE, "begin-line", ARG_NON },\r
- { OP_END_LINE, "end-line", ARG_NON },\r
- { OP_SEMI_END_BUF, "semi-end-buf", ARG_NON },\r
- { OP_BEGIN_POSITION, "begin-position", ARG_NON },\r
- { OP_BACKREF1, "backref1", ARG_NON },\r
- { OP_BACKREF2, "backref2", ARG_NON },\r
- { OP_BACKREFN, "backrefn", ARG_MEMNUM },\r
- { OP_BACKREFN_IC, "backrefn-ic", ARG_SPECIAL },\r
- { OP_BACKREF_MULTI, "backref_multi", ARG_SPECIAL },\r
- { OP_BACKREF_MULTI_IC, "backref_multi-ic", ARG_SPECIAL },\r
- { OP_BACKREF_WITH_LEVEL, "backref_at_level", ARG_SPECIAL },\r
- { OP_MEMORY_START_PUSH, "mem-start-push", ARG_MEMNUM },\r
- { OP_MEMORY_START, "mem-start", ARG_MEMNUM },\r
- { OP_MEMORY_END_PUSH, "mem-end-push", ARG_MEMNUM },\r
- { OP_MEMORY_END_PUSH_REC, "mem-end-push-rec", ARG_MEMNUM },\r
- { OP_MEMORY_END, "mem-end", ARG_MEMNUM },\r
- { OP_MEMORY_END_REC, "mem-end-rec", ARG_MEMNUM },\r
- { OP_SET_OPTION_PUSH, "set-option-push", ARG_OPTION },\r
- { OP_SET_OPTION, "set-option", ARG_OPTION },\r
- { OP_FAIL, "fail", ARG_NON },\r
- { OP_JUMP, "jump", ARG_RELADDR },\r
- { OP_PUSH, "push", ARG_RELADDR },\r
- { OP_POP, "pop", ARG_NON },\r
- { OP_PUSH_OR_JUMP_EXACT1, "push-or-jump-e1", ARG_SPECIAL },\r
- { OP_PUSH_IF_PEEK_NEXT, "push-if-peek-next", ARG_SPECIAL },\r
- { OP_REPEAT, "repeat", ARG_SPECIAL },\r
- { OP_REPEAT_NG, "repeat-ng", ARG_SPECIAL },\r
- { OP_REPEAT_INC, "repeat-inc", ARG_MEMNUM },\r
- { OP_REPEAT_INC_NG, "repeat-inc-ng", ARG_MEMNUM },\r
- { OP_REPEAT_INC_SG, "repeat-inc-sg", ARG_MEMNUM },\r
- { OP_REPEAT_INC_NG_SG, "repeat-inc-ng-sg", ARG_MEMNUM },\r
- { OP_NULL_CHECK_START, "null-check-start", ARG_MEMNUM },\r
- { OP_NULL_CHECK_END, "null-check-end", ARG_MEMNUM },\r
- { OP_NULL_CHECK_END_MEMST,"null-check-end-memst", ARG_MEMNUM },\r
- { OP_NULL_CHECK_END_MEMST_PUSH,"null-check-end-memst-push", ARG_MEMNUM },\r
- { OP_PUSH_POS, "push-pos", ARG_NON },\r
- { OP_POP_POS, "pop-pos", ARG_NON },\r
- { OP_PUSH_POS_NOT, "push-pos-not", ARG_RELADDR },\r
- { OP_FAIL_POS, "fail-pos", ARG_NON },\r
- { OP_PUSH_STOP_BT, "push-stop-bt", ARG_NON },\r
- { OP_POP_STOP_BT, "pop-stop-bt", ARG_NON },\r
- { OP_LOOK_BEHIND, "look-behind", ARG_SPECIAL },\r
- { OP_PUSH_LOOK_BEHIND_NOT, "push-look-behind-not", ARG_SPECIAL },\r
- { OP_FAIL_LOOK_BEHIND_NOT, "fail-look-behind-not", ARG_NON },\r
- { OP_CALL, "call", ARG_ABSADDR },\r
- { OP_RETURN, "return", ARG_NON },\r
- { OP_STATE_CHECK_PUSH, "state-check-push", ARG_SPECIAL },\r
- { OP_STATE_CHECK_PUSH_OR_JUMP, "state-check-push-or-jump", ARG_SPECIAL },\r
- { OP_STATE_CHECK, "state-check", ARG_STATE_CHECK },\r
- { OP_STATE_CHECK_ANYCHAR_STAR, "state-check-anychar*", ARG_STATE_CHECK },\r
- { OP_STATE_CHECK_ANYCHAR_ML_STAR,\r
- "state-check-anychar-ml*", ARG_STATE_CHECK },\r
- { -1, "", ARG_NON }\r
-};\r
-\r
-static char*\r
-op2name(int opcode)\r
-{\r
- int i;\r
-\r
- for (i = 0; OnigOpInfo[i].opcode >= 0; i++) {\r
- if (opcode == OnigOpInfo[i].opcode)\r
- return OnigOpInfo[i].name;\r
- }\r
- return "";\r
-}\r
-\r
-static int\r
-op2arg_type(int opcode)\r
-{\r
- int i;\r
-\r
- for (i = 0; OnigOpInfo[i].opcode >= 0; i++) {\r
- if (opcode == OnigOpInfo[i].opcode)\r
- return OnigOpInfo[i].arg_type;\r
- }\r
- return ARG_SPECIAL;\r
-}\r
-\r
-static void\r
-Indent(FILE* f, int indent)\r
-{\r
- int i;\r
- for (i = 0; i < indent; i++) putc(' ', f);\r
-}\r
+#ifdef ONIG_DEBUG_PARSE\r
\r
static void\r
p_string(FILE* f, int len, UChar* s)\r
}\r
\r
static void\r
-p_len_string(FILE* f, LengthType len, int mb_len, UChar* s)\r
-{\r
- int x = len * mb_len;\r
-\r
- fprintf(f, ":%d:", len);\r
- while (x-- > 0) { fputc(*s++, f); }\r
-}\r
-\r
-extern void\r
-onig_print_compiled_byte_code(FILE* f, UChar* bp, UChar** nextp,\r
- OnigEncoding enc)\r
-{\r
- int i, n, arg_type;\r
- RelAddrType addr;\r
- LengthType len;\r
- MemNumType mem;\r
- StateCheckNumType scn;\r
- OnigCodePoint code;\r
- UChar *q;\r
-\r
- fprintf(f, "[%s", op2name(*bp));\r
- arg_type = op2arg_type(*bp);\r
- if (arg_type != ARG_SPECIAL) {\r
- bp++;\r
- switch (arg_type) {\r
- case ARG_NON:\r
- break;\r
- case ARG_RELADDR:\r
- GET_RELADDR_INC(addr, bp);\r
- fprintf(f, ":(%d)", addr);\r
- break;\r
- case ARG_ABSADDR:\r
- GET_ABSADDR_INC(addr, bp);\r
- fprintf(f, ":(%d)", addr);\r
- break;\r
- case ARG_LENGTH:\r
- GET_LENGTH_INC(len, bp);\r
- fprintf(f, ":%d", len);\r
- break;\r
- case ARG_MEMNUM:\r
- mem = *((MemNumType* )bp);\r
- bp += SIZE_MEMNUM;\r
- fprintf(f, ":%d", mem);\r
- break;\r
- case ARG_OPTION:\r
- {\r
- OnigOptionType option = *((OnigOptionType* )bp);\r
- bp += SIZE_OPTION;\r
- fprintf(f, ":%d", option);\r
- }\r
- break;\r
-\r
- case ARG_STATE_CHECK:\r
- scn = *((StateCheckNumType* )bp);\r
- bp += SIZE_STATE_CHECK_NUM;\r
- fprintf(f, ":%d", scn);\r
- break;\r
- }\r
- }\r
- else {\r
- switch (*bp++) {\r
- case OP_EXACT1:\r
- case OP_ANYCHAR_STAR_PEEK_NEXT:\r
- case OP_ANYCHAR_ML_STAR_PEEK_NEXT:\r
- p_string(f, 1, bp++); break;\r
- case OP_EXACT2:\r
- p_string(f, 2, bp); bp += 2; break;\r
- case OP_EXACT3:\r
- p_string(f, 3, bp); bp += 3; break;\r
- case OP_EXACT4:\r
- p_string(f, 4, bp); bp += 4; break;\r
- case OP_EXACT5:\r
- p_string(f, 5, bp); bp += 5; break;\r
- case OP_EXACTN:\r
- GET_LENGTH_INC(len, bp);\r
- p_len_string(f, len, 1, bp);\r
- bp += len;\r
- break;\r
- \r
- case OP_EXACTMB2N1:\r
- p_string(f, 2, bp); bp += 2; break;\r
- case OP_EXACTMB2N2:\r
- p_string(f, 4, bp); bp += 4; break;\r
- case OP_EXACTMB2N3:\r
- p_string(f, 6, bp); bp += 6; break;\r
- case OP_EXACTMB2N:\r
- GET_LENGTH_INC(len, bp);\r
- p_len_string(f, len, 2, bp);\r
- bp += len * 2;\r
- break;\r
- case OP_EXACTMB3N:\r
- GET_LENGTH_INC(len, bp);\r
- p_len_string(f, len, 3, bp);\r
- bp += len * 3;\r
- break;\r
- case OP_EXACTMBN:\r
- {\r
- int mb_len;\r
- \r
- GET_LENGTH_INC(mb_len, bp);\r
- GET_LENGTH_INC(len, bp);\r
- fprintf(f, ":%d:%d:", mb_len, len);\r
- n = len * mb_len;\r
- while (n-- > 0) { fputc(*bp++, f); }\r
- }\r
- break;\r
-\r
- case OP_EXACT1_IC:\r
- len = enclen(enc, bp);\r
- p_string(f, len, bp);\r
- bp += len;\r
- break;\r
- case OP_EXACTN_IC:\r
- GET_LENGTH_INC(len, bp);\r
- p_len_string(f, len, 1, bp);\r
- bp += len;\r
- break;\r
-\r
- case OP_CCLASS:\r
- n = bitset_on_num((BitSetRef )bp);\r
- bp += SIZE_BITSET;\r
- fprintf(f, ":%d", n);\r
- break;\r
-\r
- case OP_CCLASS_NOT:\r
- n = bitset_on_num((BitSetRef )bp);\r
- bp += SIZE_BITSET;\r
- fprintf(f, ":%d", n);\r
- break;\r
-\r
- case OP_CCLASS_MB:\r
- case OP_CCLASS_MB_NOT:\r
- GET_LENGTH_INC(len, bp);\r
- q = bp;\r
-#ifndef PLATFORM_UNALIGNED_WORD_ACCESS\r
- ALIGNMENT_RIGHT(q);\r
-#endif\r
- GET_CODE_POINT(code, q);\r
- bp += len;\r
- fprintf(f, ":%d:%d", (int )code, len);\r
- break;\r
-\r
- case OP_CCLASS_MIX:\r
- case OP_CCLASS_MIX_NOT:\r
- n = bitset_on_num((BitSetRef )bp);\r
- bp += SIZE_BITSET;\r
- GET_LENGTH_INC(len, bp);\r
- q = bp;\r
-#ifndef PLATFORM_UNALIGNED_WORD_ACCESS\r
- ALIGNMENT_RIGHT(q);\r
-#endif\r
- GET_CODE_POINT(code, q);\r
- bp += len;\r
- fprintf(f, ":%d:%d:%d", n, (int )code, len);\r
- break;\r
-\r
- case OP_CCLASS_NODE:\r
- {\r
- CClassNode *cc;\r
-\r
- GET_POINTER_INC(cc, bp);\r
- n = bitset_on_num(cc->bs);\r
- fprintf(f, ":%u:%d", (unsigned int )cc, n);\r
- }\r
- break;\r
-\r
- case OP_BACKREFN_IC:\r
- mem = *((MemNumType* )bp);\r
- bp += SIZE_MEMNUM;\r
- fprintf(f, ":%d", mem);\r
- break;\r
-\r
- case OP_BACKREF_MULTI_IC:\r
- case OP_BACKREF_MULTI:\r
- fputs(" ", f);\r
- GET_LENGTH_INC(len, bp);\r
- for (i = 0; i < len; i++) {\r
- GET_MEMNUM_INC(mem, bp);\r
- if (i > 0) fputs(", ", f);\r
- fprintf(f, "%d", mem);\r
- }\r
- break;\r
-\r
- case OP_BACKREF_WITH_LEVEL:\r
- {\r
- OnigOptionType option;\r
- LengthType level;\r
-\r
- GET_OPTION_INC(option, bp);\r
- fprintf(f, ":%d", option);\r
- GET_LENGTH_INC(level, bp);\r
- fprintf(f, ":%d", level);\r
-\r
- fputs(" ", f);\r
- GET_LENGTH_INC(len, bp);\r
- for (i = 0; i < len; i++) {\r
- GET_MEMNUM_INC(mem, bp);\r
- if (i > 0) fputs(", ", f);\r
- fprintf(f, "%d", mem);\r
- }\r
- }\r
- break;\r
-\r
- case OP_REPEAT:\r
- case OP_REPEAT_NG:\r
- {\r
- mem = *((MemNumType* )bp);\r
- bp += SIZE_MEMNUM;\r
- addr = *((RelAddrType* )bp);\r
- bp += SIZE_RELADDR;\r
- fprintf(f, ":%d:%d", mem, addr);\r
- }\r
- break;\r
-\r
- case OP_PUSH_OR_JUMP_EXACT1:\r
- case OP_PUSH_IF_PEEK_NEXT:\r
- addr = *((RelAddrType* )bp);\r
- bp += SIZE_RELADDR;\r
- fprintf(f, ":(%d)", addr);\r
- p_string(f, 1, bp);\r
- bp += 1;\r
- break;\r
-\r
- case OP_LOOK_BEHIND:\r
- GET_LENGTH_INC(len, bp);\r
- fprintf(f, ":%d", len);\r
- break;\r
-\r
- case OP_PUSH_LOOK_BEHIND_NOT:\r
- GET_RELADDR_INC(addr, bp);\r
- GET_LENGTH_INC(len, bp);\r
- fprintf(f, ":%d:(%d)", len, addr);\r
- break;\r
-\r
- case OP_STATE_CHECK_PUSH:\r
- case OP_STATE_CHECK_PUSH_OR_JUMP:\r
- scn = *((StateCheckNumType* )bp);\r
- bp += SIZE_STATE_CHECK_NUM;\r
- addr = *((RelAddrType* )bp);\r
- bp += SIZE_RELADDR;\r
- fprintf(f, ":%d:(%d)", scn, addr);\r
- break;\r
-\r
- default:\r
- fprintf(stderr, "onig_print_compiled_byte_code: undefined code %d\n",\r
- *--bp);\r
- }\r
- }\r
- fputs("]", f);\r
- if (nextp) *nextp = bp;\r
-}\r
-\r
-static void\r
-print_compiled_byte_code_list(FILE* f, regex_t* reg)\r
+Indent(FILE* f, int indent)\r
{\r
- int ncode;\r
- UChar* bp = reg->p;\r
- UChar* end = reg->p + reg->used;\r
-\r
- fprintf(f, "code length: %d\n", reg->used);\r
-\r
- ncode = 0;\r
- while (bp < end) {\r
- ncode++;\r
- if (bp > reg->p) {\r
- if (ncode % 5 == 0)\r
- fprintf(f, "\n");\r
- else\r
- fputs(" ", f);\r
- }\r
- onig_print_compiled_byte_code(f, bp, &bp, reg->enc);\r
- }\r
-\r
- fprintf(f, "\n");\r
+ int i;\r
+ for (i = 0; i < indent; i++) putc(' ', f);\r
}\r
\r
static void\r
print_indent_tree(FILE* f, Node* node, int indent)\r
{\r
- int i, type;\r
- int add = 3;\r
+ int i;\r
+ NodeType type;\r
UChar* p;\r
+ int add = 3;\r
\r
Indent(f, indent);\r
if (IS_NULL(node)) {\r
exit (0);\r
}\r
\r
- type = NTYPE(node);\r
+ type = NODE_TYPE(node);\r
switch (type) {\r
- case NT_LIST:\r
- case NT_ALT:\r
- if (NTYPE(node) == NT_LIST)\r
- fprintf(f, "<list:%x>\n", (int )node);\r
+ case NODE_LIST:\r
+ case NODE_ALT:\r
+ if (type == NODE_LIST)\r
+ fprintf(f, "<list:%p>\n", node);\r
else\r
- fprintf(f, "<alt:%x>\n", (int )node);\r
+ fprintf(f, "<alt:%p>\n", node);\r
\r
- print_indent_tree(f, NCAR(node), indent + add);\r
- while (IS_NOT_NULL(node = NCDR(node))) {\r
- if (NTYPE(node) != type) {\r
- fprintf(f, "ERROR: list/alt right is not a cons. %d\n", NTYPE(node));\r
- exit(0);\r
+ print_indent_tree(f, NODE_CAR(node), indent + add);\r
+ while (IS_NOT_NULL(node = NODE_CDR(node))) {\r
+ if (NODE_TYPE(node) != type) {\r
+ fprintf(f, "ERROR: list/alt right is not a cons. %d\n", NODE_TYPE(node));\r
+ exit(0);\r
}\r
- print_indent_tree(f, NCAR(node), indent + add);\r
+ print_indent_tree(f, NODE_CAR(node), indent + add);\r
}\r
break;\r
\r
- case NT_STR:\r
- fprintf(f, "<string%s:%x>",\r
- (NSTRING_IS_RAW(node) ? "-raw" : ""), (int )node);\r
- for (p = NSTR(node)->s; p < NSTR(node)->end; p++) {\r
+ case NODE_STRING:\r
+ fprintf(f, "<string%s:%p>", (NODE_STRING_IS_RAW(node) ? "-raw" : ""), node);\r
+ for (p = STR_(node)->s; p < STR_(node)->end; p++) {\r
if (*p >= 0x20 && *p < 0x7f)\r
- fputc(*p, f);\r
+ fputc(*p, f);\r
else {\r
- fprintf(f, " 0x%02x", *p);\r
+ fprintf(f, " 0x%02x", *p);\r
}\r
}\r
break;\r
\r
- case NT_CCLASS:\r
- fprintf(f, "<cclass:%x>", (int )node);\r
- if (IS_NCCLASS_NOT(NCCLASS(node))) fputs(" not", f);\r
- if (NCCLASS(node)->mbuf) {\r
- BBuf* bbuf = NCCLASS(node)->mbuf;\r
+ case NODE_CCLASS:\r
+ fprintf(f, "<cclass:%p>", node);\r
+ if (IS_NCCLASS_NOT(CCLASS_(node))) fputs(" not", f);\r
+ if (CCLASS_(node)->mbuf) {\r
+ BBuf* bbuf = CCLASS_(node)->mbuf;\r
for (i = 0; i < bbuf->used; i++) {\r
- if (i > 0) fprintf(f, ",");\r
- fprintf(f, "%0x", bbuf->p[i]);\r
+ if (i > 0) fprintf(f, ",");\r
+ fprintf(f, "%0x", bbuf->p[i]);\r
}\r
}\r
break;\r
\r
- case NT_CTYPE:\r
- fprintf(f, "<ctype:%x> ", (int )node);\r
- switch (NCTYPE(node)->ctype) {\r
+ case NODE_CTYPE:\r
+ fprintf(f, "<ctype:%p> ", node);\r
+ switch (CTYPE_(node)->ctype) {\r
+ case CTYPE_ANYCHAR:\r
+ fprintf(f, "<anychar:%p>", node);\r
+ break;\r
+\r
case ONIGENC_CTYPE_WORD:\r
- if (NCTYPE(node)->not != 0)\r
- fputs("not word", f);\r
+ if (CTYPE_(node)->not != 0)\r
+ fputs("not word", f);\r
else\r
- fputs("word", f);\r
+ fputs("word", f);\r
+\r
+ if (CTYPE_(node)->ascii_mode != 0)\r
+ fputs(" (ascii)", f);\r
+\r
break;\r
\r
default:\r
}\r
break;\r
\r
- case NT_CANY:\r
- fprintf(f, "<anychar:%x>", (int )node);\r
- break;\r
-\r
- case NT_ANCHOR:\r
- fprintf(f, "<anchor:%x> ", (int )node);\r
- switch (NANCHOR(node)->type) {\r
- case ANCHOR_BEGIN_BUF: fputs("begin buf", f); break;\r
- case ANCHOR_END_BUF: fputs("end buf", f); break;\r
- case ANCHOR_BEGIN_LINE: fputs("begin line", f); break;\r
- case ANCHOR_END_LINE: fputs("end line", f); break;\r
- case ANCHOR_SEMI_END_BUF: fputs("semi end buf", f); break;\r
- case ANCHOR_BEGIN_POSITION: fputs("begin position", f); break;\r
+ case NODE_ANCHOR:\r
+ fprintf(f, "<anchor:%p> ", node);\r
+ switch (ANCHOR_(node)->type) {\r
+ case ANCHOR_BEGIN_BUF: fputs("begin buf", f); break;\r
+ case ANCHOR_END_BUF: fputs("end buf", f); break;\r
+ case ANCHOR_BEGIN_LINE: fputs("begin line", f); break;\r
+ case ANCHOR_END_LINE: fputs("end line", f); break;\r
+ case ANCHOR_SEMI_END_BUF: fputs("semi end buf", f); break;\r
+ case ANCHOR_BEGIN_POSITION: fputs("begin position", f); break;\r
\r
- case ANCHOR_WORD_BOUND: fputs("word bound", f); break;\r
- case ANCHOR_NOT_WORD_BOUND: fputs("not word bound", f); break;\r
+ case ANCHOR_WORD_BOUNDARY: fputs("word boundary", f); break;\r
+ case ANCHOR_NO_WORD_BOUNDARY: fputs("not word boundary", f); break;\r
#ifdef USE_WORD_BEGIN_END\r
- case ANCHOR_WORD_BEGIN: fputs("word begin", f); break;\r
- case ANCHOR_WORD_END: fputs("word end", f); break;\r
+ case ANCHOR_WORD_BEGIN: fputs("word begin", f); break;\r
+ case ANCHOR_WORD_END: fputs("word end", f); break;\r
#endif\r
- case ANCHOR_PREC_READ: fputs("prec read", f); break;\r
- case ANCHOR_PREC_READ_NOT: fputs("prec read not", f); break;\r
- case ANCHOR_LOOK_BEHIND: fputs("look_behind", f); break;\r
- case ANCHOR_LOOK_BEHIND_NOT: fputs("look_behind_not",f); break;\r
+ case ANCHOR_EXTENDED_GRAPHEME_CLUSTER_BOUNDARY:\r
+ fputs("extended-grapheme-cluster boundary", f); break;\r
+ case ANCHOR_NO_EXTENDED_GRAPHEME_CLUSTER_BOUNDARY:\r
+ fputs("no-extended-grapheme-cluster boundary", f); break;\r
+ case ANCHOR_PREC_READ:\r
+ fprintf(f, "prec read\n");\r
+ print_indent_tree(f, NODE_BODY(node), indent + add);\r
+ break;\r
+ case ANCHOR_PREC_READ_NOT:\r
+ fprintf(f, "prec read not\n");\r
+ print_indent_tree(f, NODE_BODY(node), indent + add);\r
+ break;\r
+ case ANCHOR_LOOK_BEHIND:\r
+ fprintf(f, "look behind\n");\r
+ print_indent_tree(f, NODE_BODY(node), indent + add);\r
+ break;\r
+ case ANCHOR_LOOK_BEHIND_NOT:\r
+ fprintf(f, "look behind not\n");\r
+ print_indent_tree(f, NODE_BODY(node), indent + add);\r
+ break;\r
\r
default:\r
fprintf(f, "ERROR: undefined anchor type.\n");\r
}\r
break;\r
\r
- case NT_BREF:\r
+ case NODE_BACKREF:\r
{\r
int* p;\r
- BRefNode* br = NBREF(node);\r
+ BackRefNode* br = BACKREF_(node);\r
p = BACKREFS_P(br);\r
- fprintf(f, "<backref:%x>", (int )node);\r
+ fprintf(f, "<backref%s:%p>", NODE_IS_CHECKER(node) ? "-checker" : "", node);\r
for (i = 0; i < br->back_num; i++) {\r
- if (i > 0) fputs(", ", f);\r
- fprintf(f, "%d", p[i]);\r
+ if (i > 0) fputs(", ", f);\r
+ fprintf(f, "%d", p[i]);\r
}\r
}\r
break;\r
\r
-#ifdef USE_SUBEXP_CALL\r
- case NT_CALL:\r
+#ifdef USE_CALL\r
+ case NODE_CALL:\r
{\r
- CallNode* cn = NCALL(node);\r
- fprintf(f, "<call:%x>", (int )node);\r
+ CallNode* cn = CALL_(node);\r
+ fprintf(f, "<call:%p>", node);\r
p_string(f, cn->name_end - cn->name, cn->name);\r
}\r
break;\r
#endif\r
\r
- case NT_QTFR:\r
- fprintf(f, "<quantifier:%x>{%d,%d}%s\n", (int )node,\r
- NQTFR(node)->lower, NQTFR(node)->upper,\r
- (NQTFR(node)->greedy ? "" : "?"));\r
- print_indent_tree(f, NQTFR(node)->target, indent + add);\r
+ case NODE_QUANT:\r
+ fprintf(f, "<quantifier:%p>{%d,%d}%s\n", node,\r
+ QUANT_(node)->lower, QUANT_(node)->upper,\r
+ (QUANT_(node)->greedy ? "" : "?"));\r
+ print_indent_tree(f, NODE_BODY(node), indent + add);\r
break;\r
\r
- case NT_ENCLOSE:\r
- fprintf(f, "<enclose:%x> ", (int )node);\r
- switch (NENCLOSE(node)->type) {\r
- case ENCLOSE_OPTION:\r
- fprintf(f, "option:%d", NENCLOSE(node)->option);\r
+ case NODE_ENCLOSURE:\r
+ fprintf(f, "<enclosure:%p> ", node);\r
+ switch (ENCLOSURE_(node)->type) {\r
+ case ENCLOSURE_OPTION:\r
+ fprintf(f, "option:%d", ENCLOSURE_(node)->o.options);\r
break;\r
- case ENCLOSE_MEMORY:\r
- fprintf(f, "memory:%d", NENCLOSE(node)->regnum);\r
+ case ENCLOSURE_MEMORY:\r
+ fprintf(f, "memory:%d", ENCLOSURE_(node)->m.regnum);\r
break;\r
- case ENCLOSE_STOP_BACKTRACK:\r
+ case ENCLOSURE_STOP_BACKTRACK:\r
fprintf(f, "stop-bt");\r
break;\r
\r
break;\r
}\r
fprintf(f, "\n");\r
- print_indent_tree(f, NENCLOSE(node)->target, indent + add);\r
+ print_indent_tree(f, NODE_BODY(node), indent + add);\r
+ break;\r
+\r
+ case NODE_GIMMICK:\r
+ fprintf(f, "<gimmick:%p> ", node);\r
+ switch (GIMMICK_(node)->type) {\r
+ case GIMMICK_FAIL:\r
+ fprintf(f, "fail");\r
+ break;\r
+ case GIMMICK_KEEP:\r
+ fprintf(f, "keep:%d", GIMMICK_(node)->id);\r
+ break;\r
+ case GIMMICK_SAVE:\r
+ fprintf(f, "save:%d:%d", GIMMICK_(node)->detail_type, GIMMICK_(node)->id);\r
+ break;\r
+ case GIMMICK_UPDATE_VAR:\r
+ fprintf(f, "update_var:%d:%d", GIMMICK_(node)->detail_type, GIMMICK_(node)->id);\r
+ break;\r
+#ifdef USE_CALLOUT\r
+ case GIMMICK_CALLOUT:\r
+ switch (GIMMICK_(node)->detail_type) {\r
+ case ONIG_CALLOUT_OF_CONTENTS:\r
+ fprintf(f, "callout:contents:%d", GIMMICK_(node)->num);\r
+ break;\r
+ case ONIG_CALLOUT_OF_NAME:\r
+ fprintf(f, "callout:name:%d:%d", GIMMICK_(node)->id, GIMMICK_(node)->num);\r
+ break;\r
+ }\r
+#endif\r
+ }\r
break;\r
\r
default:\r
- fprintf(f, "print_indent_tree: undefined node type %d\n", NTYPE(node));\r
+ fprintf(f, "print_indent_tree: undefined node type %d\n", NODE_TYPE(node));\r
break;\r
}\r
\r
- if (type != NT_LIST && type != NT_ALT && type != NT_QTFR &&\r
- type != NT_ENCLOSE)\r
+ if (type != NODE_LIST && type != NODE_ALT && type != NODE_QUANT &&\r
+ type != NODE_ENCLOSURE)\r
fprintf(f, "\n");\r
fflush(f);\r
}\r
-#endif /* ONIG_DEBUG */\r
\r
-#ifdef ONIG_DEBUG_PARSE_TREE\r
static void\r
print_tree(FILE* f, Node* node)\r
{\r