Ruby  2.1.10p492(2016-04-01revision54464)
regparse.h
Go to the documentation of this file.
1 #ifndef ONIGURUMA_REGPARSE_H
2 #define ONIGURUMA_REGPARSE_H
3 /**********************************************************************
4  regparse.h - Onigmo (Oniguruma-mod) (regular expression library)
5 **********************************************************************/
6 /*-
7  * Copyright (c) 2002-2007 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
8  * Copyright (c) 2011 K.Takata <kentkt AT csc DOT jp>
9  * All rights reserved.
10  *
11  * Redistribution and use in source and binary forms, with or without
12  * modification, are permitted provided that the following conditions
13  * are met:
14  * 1. Redistributions of source code must retain the above copyright
15  * notice, this list of conditions and the following disclaimer.
16  * 2. Redistributions in binary form must reproduce the above copyright
17  * notice, this list of conditions and the following disclaimer in the
18  * documentation and/or other materials provided with the distribution.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  */
32 
33 #include "regint.h"
34 
36 
37 /* node type */
38 #define NT_STR 0
39 #define NT_CCLASS 1
40 #define NT_CTYPE 2
41 #define NT_CANY 3
42 #define NT_BREF 4
43 #define NT_QTFR 5
44 #define NT_ENCLOSE 6
45 #define NT_ANCHOR 7
46 #define NT_LIST 8
47 #define NT_ALT 9
48 #define NT_CALL 10
49 
50 /* node type bit */
51 #define NTYPE2BIT(type) (1<<(type))
52 
53 #define BIT_NT_STR NTYPE2BIT(NT_STR)
54 #define BIT_NT_CCLASS NTYPE2BIT(NT_CCLASS)
55 #define BIT_NT_CTYPE NTYPE2BIT(NT_CTYPE)
56 #define BIT_NT_CANY NTYPE2BIT(NT_CANY)
57 #define BIT_NT_BREF NTYPE2BIT(NT_BREF)
58 #define BIT_NT_QTFR NTYPE2BIT(NT_QTFR)
59 #define BIT_NT_ENCLOSE NTYPE2BIT(NT_ENCLOSE)
60 #define BIT_NT_ANCHOR NTYPE2BIT(NT_ANCHOR)
61 #define BIT_NT_LIST NTYPE2BIT(NT_LIST)
62 #define BIT_NT_ALT NTYPE2BIT(NT_ALT)
63 #define BIT_NT_CALL NTYPE2BIT(NT_CALL)
64 
65 #define IS_NODE_TYPE_SIMPLE(type) \
66  ((NTYPE2BIT(type) & (BIT_NT_STR | BIT_NT_CCLASS | BIT_NT_CTYPE |\
67  BIT_NT_CANY | BIT_NT_BREF)) != 0)
68 
69 #define NTYPE(node) ((node)->u.base.type)
70 #define SET_NTYPE(node, ntype) (node)->u.base.type = (ntype)
71 
72 #define NSTR(node) (&((node)->u.str))
73 #define NCCLASS(node) (&((node)->u.cclass))
74 #define NCTYPE(node) (&((node)->u.ctype))
75 #define NBREF(node) (&((node)->u.bref))
76 #define NQTFR(node) (&((node)->u.qtfr))
77 #define NENCLOSE(node) (&((node)->u.enclose))
78 #define NANCHOR(node) (&((node)->u.anchor))
79 #define NCONS(node) (&((node)->u.cons))
80 #define NCALL(node) (&((node)->u.call))
81 
82 #define NCAR(node) (NCONS(node)->car)
83 #define NCDR(node) (NCONS(node)->cdr)
84 
85 
86 
87 #define ANCHOR_ANYCHAR_STAR_MASK (ANCHOR_ANYCHAR_STAR | ANCHOR_ANYCHAR_STAR_ML)
88 #define ANCHOR_END_BUF_MASK (ANCHOR_END_BUF | ANCHOR_SEMI_END_BUF)
89 
90 #define ENCLOSE_MEMORY (1<<0)
91 #define ENCLOSE_OPTION (1<<1)
92 #define ENCLOSE_STOP_BACKTRACK (1<<2)
93 #define ENCLOSE_CONDITION (1<<3)
94 
95 #define NODE_STR_MARGIN 16
96 #define NODE_STR_BUF_SIZE 24 /* sizeof(CClassNode) - sizeof(int)*4 */
97 #define NODE_BACKREFS_SIZE 6
98 
99 #define NSTR_RAW (1<<0) /* by backslashed number */
100 #define NSTR_AMBIG (1<<1)
101 #define NSTR_DONT_GET_OPT_INFO (1<<2)
102 
103 #define NSTRING_LEN(node) (OnigDistance )((node)->u.str.end - (node)->u.str.s)
104 #define NSTRING_SET_RAW(node) (node)->u.str.flag |= NSTR_RAW
105 #define NSTRING_CLEAR_RAW(node) (node)->u.str.flag &= ~NSTR_RAW
106 #define NSTRING_SET_AMBIG(node) (node)->u.str.flag |= NSTR_AMBIG
107 #define NSTRING_SET_DONT_GET_OPT_INFO(node) \
108  (node)->u.str.flag |= NSTR_DONT_GET_OPT_INFO
109 #define NSTRING_IS_RAW(node) (((node)->u.str.flag & NSTR_RAW) != 0)
110 #define NSTRING_IS_AMBIG(node) (((node)->u.str.flag & NSTR_AMBIG) != 0)
111 #define NSTRING_IS_DONT_GET_OPT_INFO(node) \
112  (((node)->u.str.flag & NSTR_DONT_GET_OPT_INFO) != 0)
113 
114 #define BACKREFS_P(br) \
115  (IS_NOT_NULL((br)->back_dynamic) ? (br)->back_dynamic : (br)->back_static);
116 
117 #define NQ_TARGET_ISNOT_EMPTY 0
118 #define NQ_TARGET_IS_EMPTY 1
119 #define NQ_TARGET_IS_EMPTY_MEM 2
120 #define NQ_TARGET_IS_EMPTY_REC 3
121 
122 /* status bits */
123 #define NST_MIN_FIXED (1<<0)
124 #define NST_MAX_FIXED (1<<1)
125 #define NST_CLEN_FIXED (1<<2)
126 #define NST_MARK1 (1<<3)
127 #define NST_MARK2 (1<<4)
128 #define NST_MEM_BACKREFED (1<<5)
129 #define NST_STOP_BT_SIMPLE_REPEAT (1<<6)
130 #define NST_RECURSION (1<<7)
131 #define NST_CALLED (1<<8)
132 #define NST_ADDR_FIXED (1<<9)
133 #define NST_NAMED_GROUP (1<<10)
134 #define NST_NAME_REF (1<<11)
135 #define NST_IN_REPEAT (1<<12) /* STK_REPEAT is nested in stack. */
136 #define NST_NEST_LEVEL (1<<13)
137 #define NST_BY_NUMBER (1<<14) /* {n,m} */
138 
139 #define SET_ENCLOSE_STATUS(node,f) (node)->u.enclose.state |= (f)
140 #define CLEAR_ENCLOSE_STATUS(node,f) (node)->u.enclose.state &= ~(f)
141 
142 #define IS_ENCLOSE_CALLED(en) (((en)->state & NST_CALLED) != 0)
143 #define IS_ENCLOSE_ADDR_FIXED(en) (((en)->state & NST_ADDR_FIXED) != 0)
144 #define IS_ENCLOSE_RECURSION(en) (((en)->state & NST_RECURSION) != 0)
145 #define IS_ENCLOSE_MARK1(en) (((en)->state & NST_MARK1) != 0)
146 #define IS_ENCLOSE_MARK2(en) (((en)->state & NST_MARK2) != 0)
147 #define IS_ENCLOSE_MIN_FIXED(en) (((en)->state & NST_MIN_FIXED) != 0)
148 #define IS_ENCLOSE_MAX_FIXED(en) (((en)->state & NST_MAX_FIXED) != 0)
149 #define IS_ENCLOSE_CLEN_FIXED(en) (((en)->state & NST_CLEN_FIXED) != 0)
150 #define IS_ENCLOSE_STOP_BT_SIMPLE_REPEAT(en) \
151  (((en)->state & NST_STOP_BT_SIMPLE_REPEAT) != 0)
152 #define IS_ENCLOSE_NAMED_GROUP(en) (((en)->state & NST_NAMED_GROUP) != 0)
153 #define IS_ENCLOSE_NAME_REF(en) (((en)->state & NST_NAME_REF) != 0)
154 
155 #define SET_CALL_RECURSION(node) (node)->u.call.state |= NST_RECURSION
156 #define IS_CALL_RECURSION(cn) (((cn)->state & NST_RECURSION) != 0)
157 #define IS_CALL_NAME_REF(cn) (((cn)->state & NST_NAME_REF) != 0)
158 #define IS_BACKREF_NAME_REF(bn) (((bn)->state & NST_NAME_REF) != 0)
159 #define IS_BACKREF_NEST_LEVEL(bn) (((bn)->state & NST_NEST_LEVEL) != 0)
160 #define IS_QUANTIFIER_IN_REPEAT(qn) (((qn)->state & NST_IN_REPEAT) != 0)
161 #define IS_QUANTIFIER_BY_NUMBER(qn) (((qn)->state & NST_BY_NUMBER) != 0)
162 
163 #define CALLNODE_REFNUM_UNDEF -1
164 
165 typedef struct {
169  unsigned int flag;
170  int capa; /* (allocated size - 1) or 0: use buf[] */
172 } StrNode;
173 
174 typedef struct {
176  int state;
177  struct _Node* target;
178  int lower;
179  int upper;
180  int greedy;
182  struct _Node* head_exact;
184  int is_refered; /* include called node. don't eliminate even if {0} */
185 #ifdef USE_COMBINATION_EXPLOSION_CHECK
186  int comb_exp_check_num; /* 1,2,3...: check, 0: no check */
187 #endif
188 } QtfrNode;
189 
190 typedef struct {
192  int state;
193  int type;
194  int regnum;
196  struct _Node* target;
198  /* for multiple call reference */
199  OnigDistance min_len; /* min length (byte) */
200  OnigDistance max_len; /* max length (byte) */
201  int char_len; /* character length */
202  int opt_count; /* referenced count in optimize_node_left() */
203 } EncloseNode;
204 
205 #ifdef USE_SUBEXP_CALL
206 
207 typedef struct {
208  int offset;
209  struct _Node* target;
210 } UnsetAddr;
211 
212 typedef struct {
213  int num;
214  int alloc;
216 } UnsetAddrList;
217 
218 typedef struct {
220  int state;
224  struct _Node* target; /* EncloseNode : ENCLOSE_MEMORY */
226 } CallNode;
227 
228 #endif
229 
230 typedef struct {
232  int state;
233  int back_num;
234  int back_static[NODE_BACKREFS_SIZE];
237 } BRefNode;
238 
239 typedef struct {
241  int type;
242  struct _Node* target;
243  int char_len;
245 } AnchorNode;
246 
247 typedef struct {
249  struct _Node* car;
250  struct _Node* cdr;
251 } ConsAltNode;
252 
253 typedef struct {
255  int ctype;
256  int not;
258 } CtypeNode;
259 
260 typedef struct _Node {
261  union {
271 #ifdef USE_SUBEXP_CALL
273 #endif
274  } u;
275 } Node;
276 
277 
278 #define NULL_NODE ((Node* )0)
279 
280 #define SCANENV_MEMNODES_SIZE 8
281 #define SCANENV_MEM_NODES(senv) \
282  (IS_NOT_NULL((senv)->mem_nodes_dynamic) ? \
283  (senv)->mem_nodes_dynamic : (senv)->mem_nodes_static)
284 
285 typedef struct {
298  regex_t* reg; /* for reg->names only */
299  int num_call;
300 #ifdef USE_SUBEXP_CALL
302 #endif
303  int num_mem;
304 #ifdef USE_NAMED_GROUP
306 #endif
308  Node* mem_nodes_static[SCANENV_MEMNODES_SIZE];
310 #ifdef USE_COMBINATION_EXPLOSION_CHECK
311  int num_comb_exp_check;
312  int comb_exp_max_regnum;
313  int curr_max_regnum;
314  int has_recursion;
315 #endif
317  const char* sourcefile;
319 } ScanEnv;
320 
321 
322 #define IS_SYNTAX_OP(syn, opm) (((syn)->op & (opm)) != 0)
323 #define IS_SYNTAX_OP2(syn, opm) (((syn)->op2 & (opm)) != 0)
324 #define IS_SYNTAX_BV(syn, bvm) (((syn)->behavior & (bvm)) != 0)
325 
326 #ifdef USE_NAMED_GROUP
327 typedef struct {
328  int new_val;
329 } GroupNumRemap;
330 
331 extern int onig_renumber_name_table P_((regex_t* reg, GroupNumRemap* map));
332 #endif
333 
334 extern int onig_strncmp P_((const UChar* s1, const UChar* s2, int n));
335 extern void onig_strcpy P_((UChar* dest, const UChar* src, const UChar* end));
336 extern void onig_scan_env_set_error_string P_((ScanEnv* env, int ecode, UChar* arg, UChar* arg_end));
337 extern int onig_scan_unsigned_number P_((UChar** src, const UChar* end, OnigEncoding enc));
338 extern void onig_reduce_nested_quantifier P_((Node* pnode, Node* cnode));
339 extern void onig_node_conv_to_str_node P_((Node* node, int raw));
340 extern int onig_node_str_cat P_((Node* node, const UChar* s, const UChar* end));
341 extern int onig_node_str_set P_((Node* node, const UChar* s, const UChar* end));
342 extern void onig_node_free P_((Node* node));
343 extern Node* onig_node_new_enclose P_((int type));
344 extern Node* onig_node_new_anchor P_((int type));
345 extern Node* onig_node_new_str P_((const UChar* s, const UChar* end));
346 extern Node* onig_node_new_list P_((Node* left, Node* right));
347 extern Node* onig_node_list_add P_((Node* list, Node* x));
348 extern Node* onig_node_new_alt P_((Node* left, Node* right));
349 extern void onig_node_str_clear P_((Node* node));
350 extern int onig_free_node_list P_((void));
351 extern int onig_names_free P_((regex_t* reg));
352 extern int onig_parse_make_tree P_((Node** root, const UChar* pattern, const UChar* end, regex_t* reg, ScanEnv* env));
353 extern int onig_free_shared_cclass_table P_((void));
354 
355 #ifdef ONIG_DEBUG
356 #ifdef USE_NAMED_GROUP
357 extern int onig_print_names(FILE*, regex_t*);
358 #endif
359 #endif
360 
362 
363 #endif /* ONIGURUMA_REGPARSE_H */
UChar * pattern
Definition: regparse.h:294
void onig_scan_env_set_error_string(ScanEnv *env, int ecode ARG_UNUSED, UChar *arg, UChar *arg_end)
Definition: regparse.c:6318
int is_refered
Definition: regparse.h:184
#define NODE_STR_BUF_SIZE
Definition: regparse.h:96
UChar * pattern_end
Definition: regparse.h:295
unsigned int OnigOptionType
Definition: ripper.y:348
UChar * end
Definition: regparse.h:168
int onig_free_node_list(void)
Definition: regparse.c:1112
ConsAltNode cons
Definition: regparse.h:269
int onig_node_str_cat(Node *node, const UChar *s, const UChar *end)
Definition: regparse.c:1445
int regnum
Definition: regparse.h:194
int ctype
Definition: regparse.h:255
Node * onig_node_list_add(Node *list, Node *x)
Definition: regparse.c:1261
struct _Node * target
Definition: regparse.h:224
int num_call
Definition: regparse.h:299
int sourceline
Definition: regparse.h:318
#define UChar
int onig_names_free(regex_t *reg)
Definition: regparse.c:486
int nest_level
Definition: regparse.h:236
UChar * error
Definition: regparse.h:296
int onig_node_str_set(Node *node, const UChar *s, const UChar *end)
Definition: regparse.c:1481
VALUE enc
Definition: tcltklib.c:10318
int target_empty_info
Definition: regparse.h:181
UChar * s
Definition: regparse.h:167
int opt_count
Definition: regparse.h:202
UnsetAddrList * unset_addr_list
Definition: regparse.h:301
struct _Node * next_head_exact
Definition: regparse.h:183
EncloseNode enclose
Definition: regparse.h:266
NodeBase base
Definition: regparse.h:262
int not
Definition: regparse.h:256
struct _Node * car
Definition: regparse.h:249
Node * onig_node_new_alt(Node *left, Node *right)
Definition: regparse.c:1279
#define SCANENV_MEMNODES_SIZE
Definition: regparse.h:280
UnsetAddr * us
Definition: regparse.h:215
regex_t * reg
Definition: regparse.h:298
unsigned int OnigCaseFoldType
Definition: ripper.y:120
int num_named
Definition: regparse.h:305
int char_len
Definition: regparse.h:201
unsigned int BitStatusType
Definition: regint.h:330
BRefNode bref
Definition: regparse.h:267
NodeBase base
Definition: regparse.h:166
UnsetAddrList * unset_addr_list
Definition: regparse.h:225
struct _Node * target
Definition: regparse.h:242
int group_num
Definition: regparse.h:221
struct _Node * head_exact
Definition: regparse.h:182
StrNode str
Definition: regparse.h:263
int state
Definition: regparse.h:232
const char * sourcefile
Definition: regparse.h:317
int capa
Definition: regparse.h:170
OnigDistance min_len
Definition: regparse.h:199
int AbsAddrType
Definition: regint.h:638
const OnigSyntaxType * syntax
Definition: regparse.h:289
NodeBase base
Definition: regparse.h:219
NodeBase base
Definition: regparse.h:191
int onig_free_shared_cclass_table(void)
Definition: regparse.c:5409
int * back_dynamic
Definition: regparse.h:235
int lower
Definition: regparse.h:178
Node * onig_node_new_enclose(int type)
Definition: regparse.c:1416
int mem_alloc
Definition: regparse.h:307
size_t OnigDistance
Definition: ripper.y:115
CtypeNode ctype
Definition: regparse.h:270
unsigned char buf[MIME_BUF_SIZE]
Definition: nkf.c:4308
CClassNode cclass
Definition: regparse.h:264
int upper
Definition: regparse.h:179
VALUE arg
Definition: enum.c:2427
void onig_node_conv_to_str_node(Node *node, int flag)
Definition: regparse.c:1506
int type
Definition: regparse.h:241
QtfrNode qtfr
Definition: regparse.h:265
struct _Node * target
Definition: regparse.h:209
OnigDistance max_len
Definition: regparse.h:200
unsigned int flag
Definition: regparse.h:169
register char * s
Definition: os2.c:56
NodeBase base
Definition: regparse.h:175
UChar * name_end
Definition: regparse.h:223
#define RUBY_SYMBOL_EXPORT_END
Definition: missing.h:39
NodeBase base
Definition: regparse.h:254
NodeBase base
Definition: regparse.h:248
int state
Definition: regparse.h:176
int type
Definition: tcltklib.c:112
BitStatusType backrefed_mem
Definition: regparse.h:293
#define P_(args)
Definition: oniguruma.h:71
int num_mem
Definition: regparse.h:303
CallNode call
Definition: regparse.h:272
int onig_strncmp(const UChar *s1, const UChar *s2, int n)
Definition: regparse.c:220
Node * onig_node_new_list(Node *left, Node *right)
Definition: regparse.c:1255
int ascii_range
Definition: regparse.h:257
#define RUBY_SYMBOL_EXPORT_BEGIN
Definition: missing.h:38
#define NODE_BACKREFS_SIZE
Definition: regparse.h:97
Node * onig_node_new_str(const UChar *s, const UChar *end)
Definition: regparse.c:1548
UChar * error_end
Definition: regparse.h:297
void onig_node_str_clear(Node *node)
Definition: regparse.c:1516
Node ** mem_nodes_dynamic
Definition: regparse.h:309
BitStatusType bt_mem_end
Definition: regparse.h:292
OnigEncoding enc
Definition: regparse.h:288
void onig_node_free(Node *node)
Definition: regparse.c:1029
gz end
Definition: zlib.c:2272
AnchorNode anchor
Definition: regparse.h:268
VALUE src
Definition: tcltklib.c:7943
OnigCaseFoldType case_fold_flag
Definition: regparse.h:287
void onig_reduce_nested_quantifier(Node *pnode, Node *cnode)
Definition: regparse.c:2269
struct _Node * cdr
Definition: regparse.h:250
int state
Definition: regparse.h:220
struct rb_encoding_entry * list
Definition: encoding.c:47
OnigOptionType option
Definition: regparse.h:286
struct _Node * target
Definition: regparse.h:177
int greedy
Definition: regparse.h:180
data n
Definition: enum.c:860
int onig_parse_make_tree(Node **root, const UChar *pattern, const UChar *end, regex_t *reg, ScanEnv *env)
Definition: regparse.c:6291
NodeBase base
Definition: regparse.h:231
int onig_scan_unsigned_number(UChar **src, const UChar *end, OnigEncoding enc)
Definition: regparse.c:1623
NodeBase base
Definition: regparse.h:240
int char_len
Definition: regparse.h:243
UChar * name
Definition: regparse.h:222
struct _Node * target
Definition: regparse.h:196
void onig_strcpy(UChar *dest, const UChar *src, const UChar *end)
Definition: regparse.c:232
struct _Node Node
BitStatusType bt_mem_start
Definition: regparse.h:291
int ascii_range
Definition: regparse.h:244
OnigOptionType option
Definition: regparse.h:195
AbsAddrType call_addr
Definition: regparse.h:197
#define env
int onig_renumber_name_table(regex_t *reg, GroupNumRemap *map)
Definition: regparse.c:572
int back_num
Definition: regparse.h:233
int warnings_flag
Definition: regparse.h:316
union _Node::@171 u
int offset
Definition: regparse.h:208
BitStatusType capture_history
Definition: regparse.h:290
Node * onig_node_new_anchor(int type)
Definition: regparse.c:1291