Ruby: regparse.h Source File

00001 #ifndef ONIGURUMA_REGPARSE_H
00002 #define ONIGURUMA_REGPARSE_H
00003 /**********************************************************************
00004   regparse.h -  Onigmo (Oniguruma-mod) (regular expression library)
00005 **********************************************************************/
00006 /*-
00007  * Copyright (c) 2002-2007  K.Kosako  <sndgk393 AT ybb DOT ne DOT jp>
00008  * Copyright (c) 2011       K.Takata  <kentkt AT csc DOT jp>
00009  * All rights reserved.
00010  *
00011  * Redistribution and use in source and binary forms, with or without
00012  * modification, are permitted provided that the following conditions
00013  * are met:
00014  * 1. Redistributions of source code must retain the above copyright
00015  *    notice, this list of conditions and the following disclaimer.
00016  * 2. Redistributions in binary form must reproduce the above copyright
00017  *    notice, this list of conditions and the following disclaimer in the
00018  *    documentation and/or other materials provided with the distribution.
00019  *
00020  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
00021  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
00022  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
00023  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
00024  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
00025  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
00026  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
00027  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
00028  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
00029  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
00030  * SUCH DAMAGE.
00031  */
00032 
00033 #include "regint.h"
00034 
00035 RUBY_SYMBOL_EXPORT_BEGIN
00036 
00037 /* node type */
00038 #define NT_STR         0
00039 #define NT_CCLASS      1
00040 #define NT_CTYPE       2
00041 #define NT_CANY        3
00042 #define NT_BREF        4
00043 #define NT_QTFR        5
00044 #define NT_ENCLOSE     6
00045 #define NT_ANCHOR      7
00046 #define NT_LIST        8
00047 #define NT_ALT         9
00048 #define NT_CALL       10
00049 
00050 /* node type bit */
00051 #define NTYPE2BIT(type)      (1<<(type))
00052 
00053 #define BIT_NT_STR        NTYPE2BIT(NT_STR)
00054 #define BIT_NT_CCLASS     NTYPE2BIT(NT_CCLASS)
00055 #define BIT_NT_CTYPE      NTYPE2BIT(NT_CTYPE)
00056 #define BIT_NT_CANY       NTYPE2BIT(NT_CANY)
00057 #define BIT_NT_BREF       NTYPE2BIT(NT_BREF)
00058 #define BIT_NT_QTFR       NTYPE2BIT(NT_QTFR)
00059 #define BIT_NT_ENCLOSE    NTYPE2BIT(NT_ENCLOSE)
00060 #define BIT_NT_ANCHOR     NTYPE2BIT(NT_ANCHOR)
00061 #define BIT_NT_LIST       NTYPE2BIT(NT_LIST)
00062 #define BIT_NT_ALT        NTYPE2BIT(NT_ALT)
00063 #define BIT_NT_CALL       NTYPE2BIT(NT_CALL)
00064 
00065 #define IS_NODE_TYPE_SIMPLE(type) \
00066   ((NTYPE2BIT(type) & (BIT_NT_STR | BIT_NT_CCLASS | BIT_NT_CTYPE |\
00067                        BIT_NT_CANY | BIT_NT_BREF)) != 0)
00068 
00069 #define NTYPE(node)             ((node)->u.base.type)
00070 #define SET_NTYPE(node, ntype)   (node)->u.base.type = (ntype)
00071 
00072 #define NSTR(node)         (&((node)->u.str))
00073 #define NCCLASS(node)      (&((node)->u.cclass))
00074 #define NCTYPE(node)       (&((node)->u.ctype))
00075 #define NBREF(node)        (&((node)->u.bref))
00076 #define NQTFR(node)        (&((node)->u.qtfr))
00077 #define NENCLOSE(node)     (&((node)->u.enclose))
00078 #define NANCHOR(node)      (&((node)->u.anchor))
00079 #define NCONS(node)        (&((node)->u.cons))
00080 #define NCALL(node)        (&((node)->u.call))
00081 
00082 #define NCAR(node)         (NCONS(node)->car)
00083 #define NCDR(node)         (NCONS(node)->cdr)
00084 
00085 
00086 
00087 #define ANCHOR_ANYCHAR_STAR_MASK (ANCHOR_ANYCHAR_STAR | ANCHOR_ANYCHAR_STAR_ML)
00088 #define ANCHOR_END_BUF_MASK      (ANCHOR_END_BUF | ANCHOR_SEMI_END_BUF)
00089 
00090 #define ENCLOSE_MEMORY           (1<<0)
00091 #define ENCLOSE_OPTION           (1<<1)
00092 #define ENCLOSE_STOP_BACKTRACK   (1<<2)
00093 #define ENCLOSE_CONDITION        (1<<3)
00094 
00095 #define NODE_STR_MARGIN         16
00096 #define NODE_STR_BUF_SIZE       24  /* sizeof(CClassNode) - sizeof(int)*4 */
00097 #define NODE_BACKREFS_SIZE       6
00098 
00099 #define NSTR_RAW                (1<<0) /* by backslashed number */
00100 #define NSTR_AMBIG              (1<<1)
00101 #define NSTR_DONT_GET_OPT_INFO  (1<<2)
00102 
00103 #define NSTRING_LEN(node) (OnigDistance )((node)->u.str.end - (node)->u.str.s)
00104 #define NSTRING_SET_RAW(node)          (node)->u.str.flag |= NSTR_RAW
00105 #define NSTRING_CLEAR_RAW(node)        (node)->u.str.flag &= ~NSTR_RAW
00106 #define NSTRING_SET_AMBIG(node)        (node)->u.str.flag |= NSTR_AMBIG
00107 #define NSTRING_SET_DONT_GET_OPT_INFO(node) \
00108   (node)->u.str.flag |= NSTR_DONT_GET_OPT_INFO
00109 #define NSTRING_IS_RAW(node)          (((node)->u.str.flag & NSTR_RAW)   != 0)
00110 #define NSTRING_IS_AMBIG(node)        (((node)->u.str.flag & NSTR_AMBIG) != 0)
00111 #define NSTRING_IS_DONT_GET_OPT_INFO(node) \
00112   (((node)->u.str.flag & NSTR_DONT_GET_OPT_INFO) != 0)
00113 
00114 #define BACKREFS_P(br) \
00115   (IS_NOT_NULL((br)->back_dynamic) ? (br)->back_dynamic : (br)->back_static);
00116 
00117 #define NQ_TARGET_ISNOT_EMPTY     0
00118 #define NQ_TARGET_IS_EMPTY        1
00119 #define NQ_TARGET_IS_EMPTY_MEM    2
00120 #define NQ_TARGET_IS_EMPTY_REC    3
00121 
00122 /* status bits */
00123 #define NST_MIN_FIXED             (1<<0)
00124 #define NST_MAX_FIXED             (1<<1)
00125 #define NST_CLEN_FIXED            (1<<2)
00126 #define NST_MARK1                 (1<<3)
00127 #define NST_MARK2                 (1<<4)
00128 #define NST_MEM_BACKREFED         (1<<5)
00129 #define NST_STOP_BT_SIMPLE_REPEAT (1<<6)
00130 #define NST_RECURSION             (1<<7)
00131 #define NST_CALLED                (1<<8)
00132 #define NST_ADDR_FIXED            (1<<9)
00133 #define NST_NAMED_GROUP           (1<<10)
00134 #define NST_NAME_REF              (1<<11)
00135 #define NST_IN_REPEAT             (1<<12) /* STK_REPEAT is nested in stack. */
00136 #define NST_NEST_LEVEL            (1<<13)
00137 #define NST_BY_NUMBER             (1<<14) /* {n,m} */
00138 
00139 #define SET_ENCLOSE_STATUS(node,f)      (node)->u.enclose.state |=  (f)
00140 #define CLEAR_ENCLOSE_STATUS(node,f)    (node)->u.enclose.state &= ~(f)
00141 
00142 #define IS_ENCLOSE_CALLED(en)          (((en)->state & NST_CALLED)        != 0)
00143 #define IS_ENCLOSE_ADDR_FIXED(en)      (((en)->state & NST_ADDR_FIXED)    != 0)
00144 #define IS_ENCLOSE_RECURSION(en)       (((en)->state & NST_RECURSION)     != 0)
00145 #define IS_ENCLOSE_MARK1(en)           (((en)->state & NST_MARK1)         != 0)
00146 #define IS_ENCLOSE_MARK2(en)           (((en)->state & NST_MARK2)         != 0)
00147 #define IS_ENCLOSE_MIN_FIXED(en)       (((en)->state & NST_MIN_FIXED)     != 0)
00148 #define IS_ENCLOSE_MAX_FIXED(en)       (((en)->state & NST_MAX_FIXED)     != 0)
00149 #define IS_ENCLOSE_CLEN_FIXED(en)      (((en)->state & NST_CLEN_FIXED)    != 0)
00150 #define IS_ENCLOSE_STOP_BT_SIMPLE_REPEAT(en) \
00151     (((en)->state & NST_STOP_BT_SIMPLE_REPEAT) != 0)
00152 #define IS_ENCLOSE_NAMED_GROUP(en)     (((en)->state & NST_NAMED_GROUP)   != 0)
00153 #define IS_ENCLOSE_NAME_REF(en)        (((en)->state & NST_NAME_REF)      != 0)
00154 
00155 #define SET_CALL_RECURSION(node)       (node)->u.call.state |= NST_RECURSION
00156 #define IS_CALL_RECURSION(cn)          (((cn)->state & NST_RECURSION)  != 0)
00157 #define IS_CALL_NAME_REF(cn)           (((cn)->state & NST_NAME_REF)   != 0)
00158 #define IS_BACKREF_NAME_REF(bn)        (((bn)->state & NST_NAME_REF)   != 0)
00159 #define IS_BACKREF_NEST_LEVEL(bn)      (((bn)->state & NST_NEST_LEVEL) != 0)
00160 #define IS_QUANTIFIER_IN_REPEAT(qn)    (((qn)->state & NST_IN_REPEAT)  != 0)
00161 #define IS_QUANTIFIER_BY_NUMBER(qn)    (((qn)->state & NST_BY_NUMBER)  != 0)
00162 
00163 #define CALLNODE_REFNUM_UNDEF  -1
00164 
00165 typedef struct {
00166   NodeBase base;
00167   UChar* s;
00168   UChar* end;
00169   unsigned int flag;
00170   int    capa;    /* (allocated size - 1) or 0: use buf[] */
00171   UChar  buf[NODE_STR_BUF_SIZE];
00172 } StrNode;
00173 
00174 typedef struct {
00175   NodeBase base;
00176   int state;
00177   struct _Node* target;
00178   int lower;
00179   int upper;
00180   int greedy;
00181   int target_empty_info;
00182   struct _Node* head_exact;
00183   struct _Node* next_head_exact;
00184   int is_refered;     /* include called node. don't eliminate even if {0} */
00185 #ifdef USE_COMBINATION_EXPLOSION_CHECK
00186   int comb_exp_check_num;  /* 1,2,3...: check,  0: no check  */
00187 #endif
00188 } QtfrNode;
00189 
00190 typedef struct {
00191   NodeBase base;
00192   int state;
00193   int type;
00194   int regnum;
00195   OnigOptionType option;
00196   struct _Node*  target;
00197   AbsAddrType    call_addr;
00198   /* for multiple call reference */
00199   OnigDistance min_len; /* min length (byte) */
00200   OnigDistance max_len; /* max length (byte) */
00201   int char_len;         /* character length  */
00202   int opt_count;        /* referenced count in optimize_node_left() */
00203 } EncloseNode;
00204 
00205 #ifdef USE_SUBEXP_CALL
00206 
00207 typedef struct {
00208   int           offset;
00209   struct _Node* target;
00210 } UnsetAddr;
00211 
00212 typedef struct {
00213   int        num;
00214   int        alloc;
00215   UnsetAddr* us;
00216 } UnsetAddrList;
00217 
00218 typedef struct {
00219   NodeBase base;
00220   int     state;
00221   int     group_num;
00222   UChar*  name;
00223   UChar*  name_end;
00224   struct _Node*  target;  /* EncloseNode : ENCLOSE_MEMORY */
00225   UnsetAddrList* unset_addr_list;
00226 } CallNode;
00227 
00228 #endif
00229 
00230 typedef struct {
00231   NodeBase base;
00232   int  state;
00233   int  back_num;
00234   int  back_static[NODE_BACKREFS_SIZE];
00235   int* back_dynamic;
00236   int  nest_level;
00237 } BRefNode;
00238 
00239 typedef struct {
00240   NodeBase base;
00241   int type;
00242   struct _Node* target;
00243   int char_len;
00244   int ascii_range;
00245 } AnchorNode;
00246 
00247 typedef struct {
00248   NodeBase base;
00249   struct _Node* car;
00250   struct _Node* cdr;
00251 } ConsAltNode;
00252 
00253 typedef struct {
00254   NodeBase base;
00255   int ctype;
00256   int not;
00257   int ascii_range;
00258 } CtypeNode;
00259 
00260 typedef struct _Node {
00261   union {
00262     NodeBase     base;
00263     StrNode      str;
00264     CClassNode   cclass;
00265     QtfrNode     qtfr;
00266     EncloseNode  enclose;
00267     BRefNode     bref;
00268     AnchorNode   anchor;
00269     ConsAltNode  cons;
00270     CtypeNode    ctype;
00271 #ifdef USE_SUBEXP_CALL
00272     CallNode     call;
00273 #endif
00274   } u;
00275 } Node;
00276 
00277 
00278 #define NULL_NODE  ((Node* )0)
00279 
00280 #define SCANENV_MEMNODES_SIZE               8
00281 #define SCANENV_MEM_NODES(senv)   \
00282  (IS_NOT_NULL((senv)->mem_nodes_dynamic) ? \
00283     (senv)->mem_nodes_dynamic : (senv)->mem_nodes_static)
00284 
00285 typedef struct {
00286   OnigOptionType   option;
00287   OnigCaseFoldType case_fold_flag;
00288   OnigEncoding     enc;
00289   const OnigSyntaxType* syntax;
00290   BitStatusType    capture_history;
00291   BitStatusType    bt_mem_start;
00292   BitStatusType    bt_mem_end;
00293   BitStatusType    backrefed_mem;
00294   UChar*           pattern;
00295   UChar*           pattern_end;
00296   UChar*           error;
00297   UChar*           error_end;
00298   regex_t*         reg;       /* for reg->names only */
00299   int              num_call;
00300 #ifdef USE_SUBEXP_CALL
00301   UnsetAddrList*   unset_addr_list;
00302 #endif
00303   int              num_mem;
00304 #ifdef USE_NAMED_GROUP
00305   int              num_named;
00306 #endif
00307   int              mem_alloc;
00308   Node*            mem_nodes_static[SCANENV_MEMNODES_SIZE];
00309   Node**           mem_nodes_dynamic;
00310 #ifdef USE_COMBINATION_EXPLOSION_CHECK
00311   int num_comb_exp_check;
00312   int comb_exp_max_regnum;
00313   int curr_max_regnum;
00314   int has_recursion;
00315 #endif
00316   int warnings_flag;
00317   const char* sourcefile;
00318   int sourceline;
00319 } ScanEnv;
00320 
00321 
00322 #define IS_SYNTAX_OP(syn, opm)    (((syn)->op  & (opm)) != 0)
00323 #define IS_SYNTAX_OP2(syn, opm)   (((syn)->op2 & (opm)) != 0)
00324 #define IS_SYNTAX_BV(syn, bvm)    (((syn)->behavior & (bvm)) != 0)
00325 
00326 #ifdef USE_NAMED_GROUP
00327 typedef struct {
00328   int new_val;
00329 } GroupNumRemap;
00330 
00331 extern int    onig_renumber_name_table P_((regex_t* reg, GroupNumRemap* map));
00332 #endif
00333 
00334 extern int    onig_strncmp P_((const UChar* s1, const UChar* s2, int n));
00335 extern void   onig_strcpy P_((UChar* dest, const UChar* src, const UChar* end));
00336 extern void   onig_scan_env_set_error_string P_((ScanEnv* env, int ecode, UChar* arg, UChar* arg_end));
00337 extern int    onig_scan_unsigned_number P_((UChar** src, const UChar* end, OnigEncoding enc));
00338 extern void   onig_reduce_nested_quantifier P_((Node* pnode, Node* cnode));
00339 extern void   onig_node_conv_to_str_node P_((Node* node, int raw));
00340 extern int    onig_node_str_cat P_((Node* node, const UChar* s, const UChar* end));
00341 extern int    onig_node_str_set P_((Node* node, const UChar* s, const UChar* end));
00342 extern void   onig_node_free P_((Node* node));
00343 extern Node*  onig_node_new_enclose P_((int type));
00344 extern Node*  onig_node_new_anchor P_((int type));
00345 extern Node*  onig_node_new_str P_((const UChar* s, const UChar* end));
00346 extern Node*  onig_node_new_list P_((Node* left, Node* right));
00347 extern Node*  onig_node_list_add P_((Node* list, Node* x));
00348 extern Node*  onig_node_new_alt P_((Node* left, Node* right));
00349 extern void   onig_node_str_clear P_((Node* node));
00350 extern int    onig_free_node_list P_((void));
00351 extern int    onig_names_free P_((regex_t* reg));
00352 extern int    onig_parse_make_tree P_((Node** root, const UChar* pattern, const UChar* end, regex_t* reg, ScanEnv* env));
00353 extern int    onig_free_shared_cclass_table P_((void));
00354 
00355 #ifdef ONIG_DEBUG
00356 #ifdef USE_NAMED_GROUP
00357 extern int onig_print_names(FILE*, regex_t*);
00358 #endif
00359 #endif
00360 
00361 RUBY_SYMBOL_EXPORT_END
00362 
00363 #endif /* ONIGURUMA_REGPARSE_H */
00364