Johan Wikman / SQUIRREL3

The Squirrel interpreter. See http://www.squirrel-lang.org/

Dependents:   Squirrel

sqstdlib/sqstdrex.cpp@3:7268a3ceaffc, 2014-12-16 (annotated)

Committer:
jhnwkmn
Date:
Tue Dec 16 11:39:42 2014 +0000
Revision:
3:7268a3ceaffc
Parent:
0:97a4f8cc534c 
Accepts \r as line terminator as well.

Who changed what in which revision?

UserRevisionLine numberNew contents of line
jhnwkmn 0:97a4f8cc534c 1 /* see copyright notice in squirrel.h */
jhnwkmn 0:97a4f8cc534c 2 #include <squirrel.h>
jhnwkmn 0:97a4f8cc534c 3 #include <string.h>
jhnwkmn 0:97a4f8cc534c 4 #include <ctype.h>
jhnwkmn 0:97a4f8cc534c 5 #include <setjmp.h>
jhnwkmn 0:97a4f8cc534c 6 #include <sqstdstring.h>
jhnwkmn 0:97a4f8cc534c 7
jhnwkmn 0:97a4f8cc534c 8 #ifdef _UINCODE
jhnwkmn 0:97a4f8cc534c 9 #define scisprint iswprint
jhnwkmn 0:97a4f8cc534c 10 #else
jhnwkmn 0:97a4f8cc534c 11 #define scisprint isprint
jhnwkmn 0:97a4f8cc534c 12 #endif
jhnwkmn 0:97a4f8cc534c 13
jhnwkmn 0:97a4f8cc534c 14 #ifdef _DEBUG
jhnwkmn 0:97a4f8cc534c 15 #include <stdio.h>
jhnwkmn 0:97a4f8cc534c 16
jhnwkmn 0:97a4f8cc534c 17 static const SQChar *g_nnames[] =
jhnwkmn 0:97a4f8cc534c 18 {
jhnwkmn 0:97a4f8cc534c 19 _SC("NONE"),_SC("OP_GREEDY"), _SC("OP_OR"),
jhnwkmn 0:97a4f8cc534c 20 _SC("OP_EXPR"),_SC("OP_NOCAPEXPR"),_SC("OP_DOT"), _SC("OP_CLASS"),
jhnwkmn 0:97a4f8cc534c 21 _SC("OP_CCLASS"),_SC("OP_NCLASS"),_SC("OP_RANGE"),_SC("OP_CHAR"),
jhnwkmn 0:97a4f8cc534c 22 _SC("OP_EOL"),_SC("OP_BOL"),_SC("OP_WB")
jhnwkmn 0:97a4f8cc534c 23 };
jhnwkmn 0:97a4f8cc534c 24
jhnwkmn 0:97a4f8cc534c 25 #endif
jhnwkmn 0:97a4f8cc534c 26
jhnwkmn 0:97a4f8cc534c 27 #define OP_GREEDY (MAX_CHAR+1) // * + ? {n}
jhnwkmn 0:97a4f8cc534c 28 #define OP_OR (MAX_CHAR+2)
jhnwkmn 0:97a4f8cc534c 29 #define OP_EXPR (MAX_CHAR+3) //parentesis ()
jhnwkmn 0:97a4f8cc534c 30 #define OP_NOCAPEXPR (MAX_CHAR+4) //parentesis (?:)
jhnwkmn 0:97a4f8cc534c 31 #define OP_DOT (MAX_CHAR+5)
jhnwkmn 0:97a4f8cc534c 32 #define OP_CLASS (MAX_CHAR+6)
jhnwkmn 0:97a4f8cc534c 33 #define OP_CCLASS (MAX_CHAR+7)
jhnwkmn 0:97a4f8cc534c 34 #define OP_NCLASS (MAX_CHAR+8) //negates class the [^
jhnwkmn 0:97a4f8cc534c 35 #define OP_RANGE (MAX_CHAR+9)
jhnwkmn 0:97a4f8cc534c 36 #define OP_CHAR (MAX_CHAR+10)
jhnwkmn 0:97a4f8cc534c 37 #define OP_EOL (MAX_CHAR+11)
jhnwkmn 0:97a4f8cc534c 38 #define OP_BOL (MAX_CHAR+12)
jhnwkmn 0:97a4f8cc534c 39 #define OP_WB (MAX_CHAR+13)
jhnwkmn 0:97a4f8cc534c 40
jhnwkmn 0:97a4f8cc534c 41 #define SQREX_SYMBOL_ANY_CHAR ('.')
jhnwkmn 0:97a4f8cc534c 42 #define SQREX_SYMBOL_GREEDY_ONE_OR_MORE ('+')
jhnwkmn 0:97a4f8cc534c 43 #define SQREX_SYMBOL_GREEDY_ZERO_OR_MORE ('*')
jhnwkmn 0:97a4f8cc534c 44 #define SQREX_SYMBOL_GREEDY_ZERO_OR_ONE ('?')
jhnwkmn 0:97a4f8cc534c 45 #define SQREX_SYMBOL_BRANCH ('|')
jhnwkmn 0:97a4f8cc534c 46 #define SQREX_SYMBOL_END_OF_STRING ('$')
jhnwkmn 0:97a4f8cc534c 47 #define SQREX_SYMBOL_BEGINNING_OF_STRING ('^')
jhnwkmn 0:97a4f8cc534c 48 #define SQREX_SYMBOL_ESCAPE_CHAR ('\\')
jhnwkmn 0:97a4f8cc534c 49
jhnwkmn 0:97a4f8cc534c 50
jhnwkmn 0:97a4f8cc534c 51 typedef int SQRexNodeType;
jhnwkmn 0:97a4f8cc534c 52
jhnwkmn 0:97a4f8cc534c 53 typedef struct tagSQRexNode{
jhnwkmn 0:97a4f8cc534c 54 SQRexNodeType type;
jhnwkmn 0:97a4f8cc534c 55 SQInteger left;
jhnwkmn 0:97a4f8cc534c 56 SQInteger right;
jhnwkmn 0:97a4f8cc534c 57 SQInteger next;
jhnwkmn 0:97a4f8cc534c 58 }SQRexNode;
jhnwkmn 0:97a4f8cc534c 59
jhnwkmn 0:97a4f8cc534c 60 struct SQRex{
jhnwkmn 0:97a4f8cc534c 61 const SQChar *_eol;
jhnwkmn 0:97a4f8cc534c 62 const SQChar *_bol;
jhnwkmn 0:97a4f8cc534c 63 const SQChar *_p;
jhnwkmn 0:97a4f8cc534c 64 SQInteger _first;
jhnwkmn 0:97a4f8cc534c 65 SQInteger _op;
jhnwkmn 0:97a4f8cc534c 66 SQRexNode *_nodes;
jhnwkmn 0:97a4f8cc534c 67 SQInteger _nallocated;
jhnwkmn 0:97a4f8cc534c 68 SQInteger _nsize;
jhnwkmn 0:97a4f8cc534c 69 SQInteger _nsubexpr;
jhnwkmn 0:97a4f8cc534c 70 SQRexMatch *_matches;
jhnwkmn 0:97a4f8cc534c 71 SQInteger _currsubexp;
jhnwkmn 0:97a4f8cc534c 72 void *_jmpbuf;
jhnwkmn 0:97a4f8cc534c 73 const SQChar **_error;
jhnwkmn 0:97a4f8cc534c 74 };
jhnwkmn 0:97a4f8cc534c 75
jhnwkmn 0:97a4f8cc534c 76 static SQInteger sqstd_rex_list(SQRex *exp);
jhnwkmn 0:97a4f8cc534c 77
jhnwkmn 0:97a4f8cc534c 78 static SQInteger sqstd_rex_newnode(SQRex *exp, SQRexNodeType type)
jhnwkmn 0:97a4f8cc534c 79 {
jhnwkmn 0:97a4f8cc534c 80 SQRexNode n;
jhnwkmn 0:97a4f8cc534c 81 n.type = type;
jhnwkmn 0:97a4f8cc534c 82 n.next = n.right = n.left = -1;
jhnwkmn 0:97a4f8cc534c 83 if(type == OP_EXPR)
jhnwkmn 0:97a4f8cc534c 84 n.right = exp->_nsubexpr++;
jhnwkmn 0:97a4f8cc534c 85 if(exp->_nallocated < (exp->_nsize + 1)) {
jhnwkmn 0:97a4f8cc534c 86 SQInteger oldsize = exp->_nallocated;
jhnwkmn 0:97a4f8cc534c 87 exp->_nallocated *= 2;
jhnwkmn 0:97a4f8cc534c 88 exp->_nodes = (SQRexNode *)sq_realloc(exp->_nodes, oldsize * sizeof(SQRexNode) ,exp->_nallocated * sizeof(SQRexNode));
jhnwkmn 0:97a4f8cc534c 89 }
jhnwkmn 0:97a4f8cc534c 90 exp->_nodes[exp->_nsize++] = n;
jhnwkmn 0:97a4f8cc534c 91 SQInteger newid = exp->_nsize - 1;
jhnwkmn 0:97a4f8cc534c 92 return (SQInteger)newid;
jhnwkmn 0:97a4f8cc534c 93 }
jhnwkmn 0:97a4f8cc534c 94
jhnwkmn 0:97a4f8cc534c 95 static void sqstd_rex_error(SQRex *exp,const SQChar *error)
jhnwkmn 0:97a4f8cc534c 96 {
jhnwkmn 0:97a4f8cc534c 97 if(exp->_error) *exp->_error = error;
jhnwkmn 0:97a4f8cc534c 98 longjmp(*((jmp_buf*)exp->_jmpbuf),-1);
jhnwkmn 0:97a4f8cc534c 99 }
jhnwkmn 0:97a4f8cc534c 100
jhnwkmn 0:97a4f8cc534c 101 static void sqstd_rex_expect(SQRex *exp, SQInteger n){
jhnwkmn 0:97a4f8cc534c 102 if((*exp->_p) != n)
jhnwkmn 0:97a4f8cc534c 103 sqstd_rex_error(exp, _SC("expected paren"));
jhnwkmn 0:97a4f8cc534c 104 exp->_p++;
jhnwkmn 0:97a4f8cc534c 105 }
jhnwkmn 0:97a4f8cc534c 106
jhnwkmn 0:97a4f8cc534c 107 static SQChar sqstd_rex_escapechar(SQRex *exp)
jhnwkmn 0:97a4f8cc534c 108 {
jhnwkmn 0:97a4f8cc534c 109 if(*exp->_p == SQREX_SYMBOL_ESCAPE_CHAR){
jhnwkmn 0:97a4f8cc534c 110 exp->_p++;
jhnwkmn 0:97a4f8cc534c 111 switch(*exp->_p) {
jhnwkmn 0:97a4f8cc534c 112 case 'v': exp->_p++; return '\v';
jhnwkmn 0:97a4f8cc534c 113 case 'n': exp->_p++; return '\n';
jhnwkmn 0:97a4f8cc534c 114 case 't': exp->_p++; return '\t';
jhnwkmn 0:97a4f8cc534c 115 case 'r': exp->_p++; return '\r';
jhnwkmn 0:97a4f8cc534c 116 case 'f': exp->_p++; return '\f';
jhnwkmn 0:97a4f8cc534c 117 default: return (*exp->_p++);
jhnwkmn 0:97a4f8cc534c 118 }
jhnwkmn 0:97a4f8cc534c 119 } else if(!scisprint(*exp->_p)) sqstd_rex_error(exp,_SC("letter expected"));
jhnwkmn 0:97a4f8cc534c 120 return (*exp->_p++);
jhnwkmn 0:97a4f8cc534c 121 }
jhnwkmn 0:97a4f8cc534c 122
jhnwkmn 0:97a4f8cc534c 123 static SQInteger sqstd_rex_charclass(SQRex *exp,SQInteger classid)
jhnwkmn 0:97a4f8cc534c 124 {
jhnwkmn 0:97a4f8cc534c 125 SQInteger n = sqstd_rex_newnode(exp,OP_CCLASS);
jhnwkmn 0:97a4f8cc534c 126 exp->_nodes[n].left = classid;
jhnwkmn 0:97a4f8cc534c 127 return n;
jhnwkmn 0:97a4f8cc534c 128 }
jhnwkmn 0:97a4f8cc534c 129
jhnwkmn 0:97a4f8cc534c 130 static SQInteger sqstd_rex_charnode(SQRex *exp,SQBool isclass)
jhnwkmn 0:97a4f8cc534c 131 {
jhnwkmn 0:97a4f8cc534c 132 SQChar t;
jhnwkmn 0:97a4f8cc534c 133 if(*exp->_p == SQREX_SYMBOL_ESCAPE_CHAR) {
jhnwkmn 0:97a4f8cc534c 134 exp->_p++;
jhnwkmn 0:97a4f8cc534c 135 switch(*exp->_p) {
jhnwkmn 0:97a4f8cc534c 136 case 'n': exp->_p++; return sqstd_rex_newnode(exp,'\n');
jhnwkmn 0:97a4f8cc534c 137 case 't': exp->_p++; return sqstd_rex_newnode(exp,'\t');
jhnwkmn 0:97a4f8cc534c 138 case 'r': exp->_p++; return sqstd_rex_newnode(exp,'\r');
jhnwkmn 0:97a4f8cc534c 139 case 'f': exp->_p++; return sqstd_rex_newnode(exp,'\f');
jhnwkmn 0:97a4f8cc534c 140 case 'v': exp->_p++; return sqstd_rex_newnode(exp,'\v');
jhnwkmn 0:97a4f8cc534c 141 case 'a': case 'A': case 'w': case 'W': case 's': case 'S':
jhnwkmn 0:97a4f8cc534c 142 case 'd': case 'D': case 'x': case 'X': case 'c': case 'C':
jhnwkmn 0:97a4f8cc534c 143 case 'p': case 'P': case 'l': case 'u':
jhnwkmn 0:97a4f8cc534c 144 {
jhnwkmn 0:97a4f8cc534c 145 t = *exp->_p; exp->_p++;
jhnwkmn 0:97a4f8cc534c 146 return sqstd_rex_charclass(exp,t);
jhnwkmn 0:97a4f8cc534c 147 }
jhnwkmn 0:97a4f8cc534c 148 case 'b':
jhnwkmn 0:97a4f8cc534c 149 case 'B':
jhnwkmn 0:97a4f8cc534c 150 if(!isclass) {
jhnwkmn 0:97a4f8cc534c 151 SQInteger node = sqstd_rex_newnode(exp,OP_WB);
jhnwkmn 0:97a4f8cc534c 152 exp->_nodes[node].left = *exp->_p;
jhnwkmn 0:97a4f8cc534c 153 exp->_p++;
jhnwkmn 0:97a4f8cc534c 154 return node;
jhnwkmn 0:97a4f8cc534c 155 } //else default
jhnwkmn 0:97a4f8cc534c 156 default:
jhnwkmn 0:97a4f8cc534c 157 t = *exp->_p; exp->_p++;
jhnwkmn 0:97a4f8cc534c 158 return sqstd_rex_newnode(exp,t);
jhnwkmn 0:97a4f8cc534c 159 }
jhnwkmn 0:97a4f8cc534c 160 }
jhnwkmn 0:97a4f8cc534c 161 else if(!scisprint(*exp->_p)) {
jhnwkmn 0:97a4f8cc534c 162
jhnwkmn 0:97a4f8cc534c 163 sqstd_rex_error(exp,_SC("letter expected"));
jhnwkmn 0:97a4f8cc534c 164 }
jhnwkmn 0:97a4f8cc534c 165 t = *exp->_p; exp->_p++;
jhnwkmn 0:97a4f8cc534c 166 return sqstd_rex_newnode(exp,t);
jhnwkmn 0:97a4f8cc534c 167 }
jhnwkmn 0:97a4f8cc534c 168 static SQInteger sqstd_rex_class(SQRex *exp)
jhnwkmn 0:97a4f8cc534c 169 {
jhnwkmn 0:97a4f8cc534c 170 SQInteger ret = -1;
jhnwkmn 0:97a4f8cc534c 171 SQInteger first = -1,chain;
jhnwkmn 0:97a4f8cc534c 172 if(*exp->_p == SQREX_SYMBOL_BEGINNING_OF_STRING){
jhnwkmn 0:97a4f8cc534c 173 ret = sqstd_rex_newnode(exp,OP_NCLASS);
jhnwkmn 0:97a4f8cc534c 174 exp->_p++;
jhnwkmn 0:97a4f8cc534c 175 }else ret = sqstd_rex_newnode(exp,OP_CLASS);
jhnwkmn 0:97a4f8cc534c 176
jhnwkmn 0:97a4f8cc534c 177 if(*exp->_p == ']') sqstd_rex_error(exp,_SC("empty class"));
jhnwkmn 0:97a4f8cc534c 178 chain = ret;
jhnwkmn 0:97a4f8cc534c 179 while(*exp->_p != ']' && exp->_p != exp->_eol) {
jhnwkmn 0:97a4f8cc534c 180 if(*exp->_p == '-' && first != -1){
jhnwkmn 0:97a4f8cc534c 181 SQInteger r;
jhnwkmn 0:97a4f8cc534c 182 if(*exp->_p++ == ']') sqstd_rex_error(exp,_SC("unfinished range"));
jhnwkmn 0:97a4f8cc534c 183 r = sqstd_rex_newnode(exp,OP_RANGE);
jhnwkmn 0:97a4f8cc534c 184 if(exp->_nodes[first].type>*exp->_p) sqstd_rex_error(exp,_SC("invalid range"));
jhnwkmn 0:97a4f8cc534c 185 if(exp->_nodes[first].type == OP_CCLASS) sqstd_rex_error(exp,_SC("cannot use character classes in ranges"));
jhnwkmn 0:97a4f8cc534c 186 exp->_nodes[r].left = exp->_nodes[first].type;
jhnwkmn 0:97a4f8cc534c 187 SQInteger t = sqstd_rex_escapechar(exp);
jhnwkmn 0:97a4f8cc534c 188 exp->_nodes[r].right = t;
jhnwkmn 0:97a4f8cc534c 189 exp->_nodes[chain].next = r;
jhnwkmn 0:97a4f8cc534c 190 chain = r;
jhnwkmn 0:97a4f8cc534c 191 first = -1;
jhnwkmn 0:97a4f8cc534c 192 }
jhnwkmn 0:97a4f8cc534c 193 else{
jhnwkmn 0:97a4f8cc534c 194 if(first!=-1){
jhnwkmn 0:97a4f8cc534c 195 SQInteger c = first;
jhnwkmn 0:97a4f8cc534c 196 exp->_nodes[chain].next = c;
jhnwkmn 0:97a4f8cc534c 197 chain = c;
jhnwkmn 0:97a4f8cc534c 198 first = sqstd_rex_charnode(exp,SQTrue);
jhnwkmn 0:97a4f8cc534c 199 }
jhnwkmn 0:97a4f8cc534c 200 else{
jhnwkmn 0:97a4f8cc534c 201 first = sqstd_rex_charnode(exp,SQTrue);
jhnwkmn 0:97a4f8cc534c 202 }
jhnwkmn 0:97a4f8cc534c 203 }
jhnwkmn 0:97a4f8cc534c 204 }
jhnwkmn 0:97a4f8cc534c 205 if(first!=-1){
jhnwkmn 0:97a4f8cc534c 206 SQInteger c = first;
jhnwkmn 0:97a4f8cc534c 207 exp->_nodes[chain].next = c;
jhnwkmn 0:97a4f8cc534c 208 chain = c;
jhnwkmn 0:97a4f8cc534c 209 first = -1;
jhnwkmn 0:97a4f8cc534c 210 }
jhnwkmn 0:97a4f8cc534c 211 /* hack? */
jhnwkmn 0:97a4f8cc534c 212 exp->_nodes[ret].left = exp->_nodes[ret].next;
jhnwkmn 0:97a4f8cc534c 213 exp->_nodes[ret].next = -1;
jhnwkmn 0:97a4f8cc534c 214 return ret;
jhnwkmn 0:97a4f8cc534c 215 }
jhnwkmn 0:97a4f8cc534c 216
jhnwkmn 0:97a4f8cc534c 217 static SQInteger sqstd_rex_parsenumber(SQRex *exp)
jhnwkmn 0:97a4f8cc534c 218 {
jhnwkmn 0:97a4f8cc534c 219 SQInteger ret = *exp->_p-'0';
jhnwkmn 0:97a4f8cc534c 220 SQInteger positions = 10;
jhnwkmn 0:97a4f8cc534c 221 exp->_p++;
jhnwkmn 0:97a4f8cc534c 222 while(isdigit(*exp->_p)) {
jhnwkmn 0:97a4f8cc534c 223 ret = ret*10+(*exp->_p++-'0');
jhnwkmn 0:97a4f8cc534c 224 if(positions==1000000000) sqstd_rex_error(exp,_SC("overflow in numeric constant"));
jhnwkmn 0:97a4f8cc534c 225 positions *= 10;
jhnwkmn 0:97a4f8cc534c 226 };
jhnwkmn 0:97a4f8cc534c 227 return ret;
jhnwkmn 0:97a4f8cc534c 228 }
jhnwkmn 0:97a4f8cc534c 229
jhnwkmn 0:97a4f8cc534c 230 static SQInteger sqstd_rex_element(SQRex *exp)
jhnwkmn 0:97a4f8cc534c 231 {
jhnwkmn 0:97a4f8cc534c 232 SQInteger ret = -1;
jhnwkmn 0:97a4f8cc534c 233 switch(*exp->_p)
jhnwkmn 0:97a4f8cc534c 234 {
jhnwkmn 0:97a4f8cc534c 235 case '(': {
jhnwkmn 0:97a4f8cc534c 236 SQInteger expr;
jhnwkmn 0:97a4f8cc534c 237 exp->_p++;
jhnwkmn 0:97a4f8cc534c 238
jhnwkmn 0:97a4f8cc534c 239
jhnwkmn 0:97a4f8cc534c 240 if(*exp->_p =='?') {
jhnwkmn 0:97a4f8cc534c 241 exp->_p++;
jhnwkmn 0:97a4f8cc534c 242 sqstd_rex_expect(exp,':');
jhnwkmn 0:97a4f8cc534c 243 expr = sqstd_rex_newnode(exp,OP_NOCAPEXPR);
jhnwkmn 0:97a4f8cc534c 244 }
jhnwkmn 0:97a4f8cc534c 245 else
jhnwkmn 0:97a4f8cc534c 246 expr = sqstd_rex_newnode(exp,OP_EXPR);
jhnwkmn 0:97a4f8cc534c 247 SQInteger newn = sqstd_rex_list(exp);
jhnwkmn 0:97a4f8cc534c 248 exp->_nodes[expr].left = newn;
jhnwkmn 0:97a4f8cc534c 249 ret = expr;
jhnwkmn 0:97a4f8cc534c 250 sqstd_rex_expect(exp,')');
jhnwkmn 0:97a4f8cc534c 251 }
jhnwkmn 0:97a4f8cc534c 252 break;
jhnwkmn 0:97a4f8cc534c 253 case '[':
jhnwkmn 0:97a4f8cc534c 254 exp->_p++;
jhnwkmn 0:97a4f8cc534c 255 ret = sqstd_rex_class(exp);
jhnwkmn 0:97a4f8cc534c 256 sqstd_rex_expect(exp,']');
jhnwkmn 0:97a4f8cc534c 257 break;
jhnwkmn 0:97a4f8cc534c 258 case SQREX_SYMBOL_END_OF_STRING: exp->_p++; ret = sqstd_rex_newnode(exp,OP_EOL);break;
jhnwkmn 0:97a4f8cc534c 259 case SQREX_SYMBOL_ANY_CHAR: exp->_p++; ret = sqstd_rex_newnode(exp,OP_DOT);break;
jhnwkmn 0:97a4f8cc534c 260 default:
jhnwkmn 0:97a4f8cc534c 261 ret = sqstd_rex_charnode(exp,SQFalse);
jhnwkmn 0:97a4f8cc534c 262 break;
jhnwkmn 0:97a4f8cc534c 263 }
jhnwkmn 0:97a4f8cc534c 264
jhnwkmn 0:97a4f8cc534c 265
jhnwkmn 0:97a4f8cc534c 266 SQBool isgreedy = SQFalse;
jhnwkmn 0:97a4f8cc534c 267 unsigned short p0 = 0, p1 = 0;
jhnwkmn 0:97a4f8cc534c 268 switch(*exp->_p){
jhnwkmn 0:97a4f8cc534c 269 case SQREX_SYMBOL_GREEDY_ZERO_OR_MORE: p0 = 0; p1 = 0xFFFF; exp->_p++; isgreedy = SQTrue; break;
jhnwkmn 0:97a4f8cc534c 270 case SQREX_SYMBOL_GREEDY_ONE_OR_MORE: p0 = 1; p1 = 0xFFFF; exp->_p++; isgreedy = SQTrue; break;
jhnwkmn 0:97a4f8cc534c 271 case SQREX_SYMBOL_GREEDY_ZERO_OR_ONE: p0 = 0; p1 = 1; exp->_p++; isgreedy = SQTrue; break;
jhnwkmn 0:97a4f8cc534c 272 case '{':
jhnwkmn 0:97a4f8cc534c 273 exp->_p++;
jhnwkmn 0:97a4f8cc534c 274 if(!isdigit(*exp->_p)) sqstd_rex_error(exp,_SC("number expected"));
jhnwkmn 0:97a4f8cc534c 275 p0 = (unsigned short)sqstd_rex_parsenumber(exp);
jhnwkmn 0:97a4f8cc534c 276 /*******************************/
jhnwkmn 0:97a4f8cc534c 277 switch(*exp->_p) {
jhnwkmn 0:97a4f8cc534c 278 case '}':
jhnwkmn 0:97a4f8cc534c 279 p1 = p0; exp->_p++;
jhnwkmn 0:97a4f8cc534c 280 break;
jhnwkmn 0:97a4f8cc534c 281 case ',':
jhnwkmn 0:97a4f8cc534c 282 exp->_p++;
jhnwkmn 0:97a4f8cc534c 283 p1 = 0xFFFF;
jhnwkmn 0:97a4f8cc534c 284 if(isdigit(*exp->_p)){
jhnwkmn 0:97a4f8cc534c 285 p1 = (unsigned short)sqstd_rex_parsenumber(exp);
jhnwkmn 0:97a4f8cc534c 286 }
jhnwkmn 0:97a4f8cc534c 287 sqstd_rex_expect(exp,'}');
jhnwkmn 0:97a4f8cc534c 288 break;
jhnwkmn 0:97a4f8cc534c 289 default:
jhnwkmn 0:97a4f8cc534c 290 sqstd_rex_error(exp,_SC(", or } expected"));
jhnwkmn 0:97a4f8cc534c 291 }
jhnwkmn 0:97a4f8cc534c 292 /*******************************/
jhnwkmn 0:97a4f8cc534c 293 isgreedy = SQTrue;
jhnwkmn 0:97a4f8cc534c 294 break;
jhnwkmn 0:97a4f8cc534c 295
jhnwkmn 0:97a4f8cc534c 296 }
jhnwkmn 0:97a4f8cc534c 297 if(isgreedy) {
jhnwkmn 0:97a4f8cc534c 298 SQInteger nnode = sqstd_rex_newnode(exp,OP_GREEDY);
jhnwkmn 0:97a4f8cc534c 299 exp->_nodes[nnode].left = ret;
jhnwkmn 0:97a4f8cc534c 300 exp->_nodes[nnode].right = ((p0)<<16)|p1;
jhnwkmn 0:97a4f8cc534c 301 ret = nnode;
jhnwkmn 0:97a4f8cc534c 302 }
jhnwkmn 0:97a4f8cc534c 303
jhnwkmn 0:97a4f8cc534c 304 if((*exp->_p != SQREX_SYMBOL_BRANCH) && (*exp->_p != ')') && (*exp->_p != SQREX_SYMBOL_GREEDY_ZERO_OR_MORE) && (*exp->_p != SQREX_SYMBOL_GREEDY_ONE_OR_MORE) && (*exp->_p != '\0')) {
jhnwkmn 0:97a4f8cc534c 305 SQInteger nnode = sqstd_rex_element(exp);
jhnwkmn 0:97a4f8cc534c 306 exp->_nodes[ret].next = nnode;
jhnwkmn 0:97a4f8cc534c 307 }
jhnwkmn 0:97a4f8cc534c 308
jhnwkmn 0:97a4f8cc534c 309 return ret;
jhnwkmn 0:97a4f8cc534c 310 }
jhnwkmn 0:97a4f8cc534c 311
jhnwkmn 0:97a4f8cc534c 312 static SQInteger sqstd_rex_list(SQRex *exp)
jhnwkmn 0:97a4f8cc534c 313 {
jhnwkmn 0:97a4f8cc534c 314 SQInteger ret=-1,e;
jhnwkmn 0:97a4f8cc534c 315 if(*exp->_p == SQREX_SYMBOL_BEGINNING_OF_STRING) {
jhnwkmn 0:97a4f8cc534c 316 exp->_p++;
jhnwkmn 0:97a4f8cc534c 317 ret = sqstd_rex_newnode(exp,OP_BOL);
jhnwkmn 0:97a4f8cc534c 318 }
jhnwkmn 0:97a4f8cc534c 319 e = sqstd_rex_element(exp);
jhnwkmn 0:97a4f8cc534c 320 if(ret != -1) {
jhnwkmn 0:97a4f8cc534c 321 exp->_nodes[ret].next = e;
jhnwkmn 0:97a4f8cc534c 322 }
jhnwkmn 0:97a4f8cc534c 323 else ret = e;
jhnwkmn 0:97a4f8cc534c 324
jhnwkmn 0:97a4f8cc534c 325 if(*exp->_p == SQREX_SYMBOL_BRANCH) {
jhnwkmn 0:97a4f8cc534c 326 SQInteger temp,tright;
jhnwkmn 0:97a4f8cc534c 327 exp->_p++;
jhnwkmn 0:97a4f8cc534c 328 temp = sqstd_rex_newnode(exp,OP_OR);
jhnwkmn 0:97a4f8cc534c 329 exp->_nodes[temp].left = ret;
jhnwkmn 0:97a4f8cc534c 330 tright = sqstd_rex_list(exp);
jhnwkmn 0:97a4f8cc534c 331 exp->_nodes[temp].right = tright;
jhnwkmn 0:97a4f8cc534c 332 ret = temp;
jhnwkmn 0:97a4f8cc534c 333 }
jhnwkmn 0:97a4f8cc534c 334 return ret;
jhnwkmn 0:97a4f8cc534c 335 }
jhnwkmn 0:97a4f8cc534c 336
jhnwkmn 0:97a4f8cc534c 337 static SQBool sqstd_rex_matchcclass(SQInteger cclass,SQChar c)
jhnwkmn 0:97a4f8cc534c 338 {
jhnwkmn 0:97a4f8cc534c 339 switch(cclass) {
jhnwkmn 0:97a4f8cc534c 340 case 'a': return isalpha(c)?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 341 case 'A': return !isalpha(c)?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 342 case 'w': return (isalnum(c) || c == '_')?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 343 case 'W': return (!isalnum(c) && c != '_')?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 344 case 's': return isspace(c)?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 345 case 'S': return !isspace(c)?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 346 case 'd': return isdigit(c)?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 347 case 'D': return !isdigit(c)?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 348 case 'x': return isxdigit(c)?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 349 case 'X': return !isxdigit(c)?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 350 case 'c': return iscntrl(c)?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 351 case 'C': return !iscntrl(c)?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 352 case 'p': return ispunct(c)?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 353 case 'P': return !ispunct(c)?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 354 case 'l': return islower(c)?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 355 case 'u': return isupper(c)?SQTrue:SQFalse;
jhnwkmn 0:97a4f8cc534c 356 }
jhnwkmn 0:97a4f8cc534c 357 return SQFalse; /*cannot happen*/
jhnwkmn 0:97a4f8cc534c 358 }
jhnwkmn 0:97a4f8cc534c 359
jhnwkmn 0:97a4f8cc534c 360 static SQBool sqstd_rex_matchclass(SQRex* exp,SQRexNode *node,SQChar c)
jhnwkmn 0:97a4f8cc534c 361 {
jhnwkmn 0:97a4f8cc534c 362 do {
jhnwkmn 0:97a4f8cc534c 363 switch(node->type) {
jhnwkmn 0:97a4f8cc534c 364 case OP_RANGE:
jhnwkmn 0:97a4f8cc534c 365 if(c >= node->left && c <= node->right) return SQTrue;
jhnwkmn 0:97a4f8cc534c 366 break;
jhnwkmn 0:97a4f8cc534c 367 case OP_CCLASS:
jhnwkmn 0:97a4f8cc534c 368 if(sqstd_rex_matchcclass(node->left,c)) return SQTrue;
jhnwkmn 0:97a4f8cc534c 369 break;
jhnwkmn 0:97a4f8cc534c 370 default:
jhnwkmn 0:97a4f8cc534c 371 if(c == node->type)return SQTrue;
jhnwkmn 0:97a4f8cc534c 372 }
jhnwkmn 0:97a4f8cc534c 373 } while((node->next != -1) && (node = &exp->_nodes[node->next]));
jhnwkmn 0:97a4f8cc534c 374 return SQFalse;
jhnwkmn 0:97a4f8cc534c 375 }
jhnwkmn 0:97a4f8cc534c 376
jhnwkmn 0:97a4f8cc534c 377 static const SQChar *sqstd_rex_matchnode(SQRex* exp,SQRexNode *node,const SQChar *str,SQRexNode *next)
jhnwkmn 0:97a4f8cc534c 378 {
jhnwkmn 0:97a4f8cc534c 379
jhnwkmn 0:97a4f8cc534c 380 SQRexNodeType type = node->type;
jhnwkmn 0:97a4f8cc534c 381 switch(type) {
jhnwkmn 0:97a4f8cc534c 382 case OP_GREEDY: {
jhnwkmn 0:97a4f8cc534c 383 //SQRexNode *greedystop = (node->next != -1) ? &exp->_nodes[node->next] : NULL;
jhnwkmn 0:97a4f8cc534c 384 SQRexNode *greedystop = NULL;
jhnwkmn 0:97a4f8cc534c 385 SQInteger p0 = (node->right >> 16)&0x0000FFFF, p1 = node->right&0x0000FFFF, nmaches = 0;
jhnwkmn 0:97a4f8cc534c 386 const SQChar *s=str, *good = str;
jhnwkmn 0:97a4f8cc534c 387
jhnwkmn 0:97a4f8cc534c 388 if(node->next != -1) {
jhnwkmn 0:97a4f8cc534c 389 greedystop = &exp->_nodes[node->next];
jhnwkmn 0:97a4f8cc534c 390 }
jhnwkmn 0:97a4f8cc534c 391 else {
jhnwkmn 0:97a4f8cc534c 392 greedystop = next;
jhnwkmn 0:97a4f8cc534c 393 }
jhnwkmn 0:97a4f8cc534c 394
jhnwkmn 0:97a4f8cc534c 395 while((nmaches == 0xFFFF || nmaches < p1)) {
jhnwkmn 0:97a4f8cc534c 396
jhnwkmn 0:97a4f8cc534c 397 const SQChar *stop;
jhnwkmn 0:97a4f8cc534c 398 if(!(s = sqstd_rex_matchnode(exp,&exp->_nodes[node->left],s,greedystop)))
jhnwkmn 0:97a4f8cc534c 399 break;
jhnwkmn 0:97a4f8cc534c 400 nmaches++;
jhnwkmn 0:97a4f8cc534c 401 good=s;
jhnwkmn 0:97a4f8cc534c 402 if(greedystop) {
jhnwkmn 0:97a4f8cc534c 403 //checks that 0 matches satisfy the expression(if so skips)
jhnwkmn 0:97a4f8cc534c 404 //if not would always stop(for instance if is a '?')
jhnwkmn 0:97a4f8cc534c 405 if(greedystop->type != OP_GREEDY ||
jhnwkmn 0:97a4f8cc534c 406 (greedystop->type == OP_GREEDY && ((greedystop->right >> 16)&0x0000FFFF) != 0))
jhnwkmn 0:97a4f8cc534c 407 {
jhnwkmn 0:97a4f8cc534c 408 SQRexNode *gnext = NULL;
jhnwkmn 0:97a4f8cc534c 409 if(greedystop->next != -1) {
jhnwkmn 0:97a4f8cc534c 410 gnext = &exp->_nodes[greedystop->next];
jhnwkmn 0:97a4f8cc534c 411 }else if(next && next->next != -1){
jhnwkmn 0:97a4f8cc534c 412 gnext = &exp->_nodes[next->next];
jhnwkmn 0:97a4f8cc534c 413 }
jhnwkmn 0:97a4f8cc534c 414 stop = sqstd_rex_matchnode(exp,greedystop,s,gnext);
jhnwkmn 0:97a4f8cc534c 415 if(stop) {
jhnwkmn 0:97a4f8cc534c 416 //if satisfied stop it
jhnwkmn 0:97a4f8cc534c 417 if(p0 == p1 && p0 == nmaches) break;
jhnwkmn 0:97a4f8cc534c 418 else if(nmaches >= p0 && p1 == 0xFFFF) break;
jhnwkmn 0:97a4f8cc534c 419 else if(nmaches >= p0 && nmaches <= p1) break;
jhnwkmn 0:97a4f8cc534c 420 }
jhnwkmn 0:97a4f8cc534c 421 }
jhnwkmn 0:97a4f8cc534c 422 }
jhnwkmn 0:97a4f8cc534c 423
jhnwkmn 0:97a4f8cc534c 424 if(s >= exp->_eol)
jhnwkmn 0:97a4f8cc534c 425 break;
jhnwkmn 0:97a4f8cc534c 426 }
jhnwkmn 0:97a4f8cc534c 427 if(p0 == p1 && p0 == nmaches) return good;
jhnwkmn 0:97a4f8cc534c 428 else if(nmaches >= p0 && p1 == 0xFFFF) return good;
jhnwkmn 0:97a4f8cc534c 429 else if(nmaches >= p0 && nmaches <= p1) return good;
jhnwkmn 0:97a4f8cc534c 430 return NULL;
jhnwkmn 0:97a4f8cc534c 431 }
jhnwkmn 0:97a4f8cc534c 432 case OP_OR: {
jhnwkmn 0:97a4f8cc534c 433 const SQChar *asd = str;
jhnwkmn 0:97a4f8cc534c 434 SQRexNode *temp=&exp->_nodes[node->left];
jhnwkmn 0:97a4f8cc534c 435 while( (asd = sqstd_rex_matchnode(exp,temp,asd,NULL)) ) {
jhnwkmn 0:97a4f8cc534c 436 if(temp->next != -1)
jhnwkmn 0:97a4f8cc534c 437 temp = &exp->_nodes[temp->next];
jhnwkmn 0:97a4f8cc534c 438 else
jhnwkmn 0:97a4f8cc534c 439 return asd;
jhnwkmn 0:97a4f8cc534c 440 }
jhnwkmn 0:97a4f8cc534c 441 asd = str;
jhnwkmn 0:97a4f8cc534c 442 temp = &exp->_nodes[node->right];
jhnwkmn 0:97a4f8cc534c 443 while( (asd = sqstd_rex_matchnode(exp,temp,asd,NULL)) ) {
jhnwkmn 0:97a4f8cc534c 444 if(temp->next != -1)
jhnwkmn 0:97a4f8cc534c 445 temp = &exp->_nodes[temp->next];
jhnwkmn 0:97a4f8cc534c 446 else
jhnwkmn 0:97a4f8cc534c 447 return asd;
jhnwkmn 0:97a4f8cc534c 448 }
jhnwkmn 0:97a4f8cc534c 449 return NULL;
jhnwkmn 0:97a4f8cc534c 450 break;
jhnwkmn 0:97a4f8cc534c 451 }
jhnwkmn 0:97a4f8cc534c 452 case OP_EXPR:
jhnwkmn 0:97a4f8cc534c 453 case OP_NOCAPEXPR:{
jhnwkmn 0:97a4f8cc534c 454 SQRexNode *n = &exp->_nodes[node->left];
jhnwkmn 0:97a4f8cc534c 455 const SQChar *cur = str;
jhnwkmn 0:97a4f8cc534c 456 SQInteger capture = -1;
jhnwkmn 0:97a4f8cc534c 457 if(node->type != OP_NOCAPEXPR && node->right == exp->_currsubexp) {
jhnwkmn 0:97a4f8cc534c 458 capture = exp->_currsubexp;
jhnwkmn 0:97a4f8cc534c 459 exp->_matches[capture].begin = cur;
jhnwkmn 0:97a4f8cc534c 460 exp->_currsubexp++;
jhnwkmn 0:97a4f8cc534c 461 }
jhnwkmn 0:97a4f8cc534c 462 int tempcap = exp->_currsubexp;
jhnwkmn 0:97a4f8cc534c 463 do {
jhnwkmn 0:97a4f8cc534c 464 SQRexNode *subnext = NULL;
jhnwkmn 0:97a4f8cc534c 465 if(n->next != -1) {
jhnwkmn 0:97a4f8cc534c 466 subnext = &exp->_nodes[n->next];
jhnwkmn 0:97a4f8cc534c 467 }else {
jhnwkmn 0:97a4f8cc534c 468 subnext = next;
jhnwkmn 0:97a4f8cc534c 469 }
jhnwkmn 0:97a4f8cc534c 470 if(!(cur = sqstd_rex_matchnode(exp,n,cur,subnext))) {
jhnwkmn 0:97a4f8cc534c 471 if(capture != -1){
jhnwkmn 0:97a4f8cc534c 472 exp->_matches[capture].begin = 0;
jhnwkmn 0:97a4f8cc534c 473 exp->_matches[capture].len = 0;
jhnwkmn 0:97a4f8cc534c 474 }
jhnwkmn 0:97a4f8cc534c 475 return NULL;
jhnwkmn 0:97a4f8cc534c 476 }
jhnwkmn 0:97a4f8cc534c 477 } while((n->next != -1) && (n = &exp->_nodes[n->next]));
jhnwkmn 0:97a4f8cc534c 478
jhnwkmn 0:97a4f8cc534c 479 exp->_currsubexp = tempcap;
jhnwkmn 0:97a4f8cc534c 480 if(capture != -1)
jhnwkmn 0:97a4f8cc534c 481 exp->_matches[capture].len = cur - exp->_matches[capture].begin;
jhnwkmn 0:97a4f8cc534c 482 return cur;
jhnwkmn 0:97a4f8cc534c 483 }
jhnwkmn 0:97a4f8cc534c 484 case OP_WB:
jhnwkmn 0:97a4f8cc534c 485 if((str == exp->_bol && !isspace(*str))
jhnwkmn 0:97a4f8cc534c 486 || (str == exp->_eol && !isspace(*(str-1)))
jhnwkmn 0:97a4f8cc534c 487 || (!isspace(*str) && isspace(*(str+1)))
jhnwkmn 0:97a4f8cc534c 488 || (isspace(*str) && !isspace(*(str+1))) ) {
jhnwkmn 0:97a4f8cc534c 489 return (node->left == 'b')?str:NULL;
jhnwkmn 0:97a4f8cc534c 490 }
jhnwkmn 0:97a4f8cc534c 491 return (node->left == 'b')?NULL:str;
jhnwkmn 0:97a4f8cc534c 492 case OP_BOL:
jhnwkmn 0:97a4f8cc534c 493 if(str == exp->_bol) return str;
jhnwkmn 0:97a4f8cc534c 494 return NULL;
jhnwkmn 0:97a4f8cc534c 495 case OP_EOL:
jhnwkmn 0:97a4f8cc534c 496 if(str == exp->_eol) return str;
jhnwkmn 0:97a4f8cc534c 497 return NULL;
jhnwkmn 0:97a4f8cc534c 498 case OP_DOT:{
jhnwkmn 0:97a4f8cc534c 499 str++;
jhnwkmn 0:97a4f8cc534c 500 }
jhnwkmn 0:97a4f8cc534c 501 return str;
jhnwkmn 0:97a4f8cc534c 502 case OP_NCLASS:
jhnwkmn 0:97a4f8cc534c 503 case OP_CLASS:
jhnwkmn 0:97a4f8cc534c 504 if(sqstd_rex_matchclass(exp,&exp->_nodes[node->left],*str)?(type == OP_CLASS?SQTrue:SQFalse):(type == OP_NCLASS?SQTrue:SQFalse)) {
jhnwkmn 0:97a4f8cc534c 505 str++;
jhnwkmn 0:97a4f8cc534c 506 return str;
jhnwkmn 0:97a4f8cc534c 507 }
jhnwkmn 0:97a4f8cc534c 508 return NULL;
jhnwkmn 0:97a4f8cc534c 509 case OP_CCLASS:
jhnwkmn 0:97a4f8cc534c 510 if(sqstd_rex_matchcclass(node->left,*str)) {
jhnwkmn 0:97a4f8cc534c 511 str++;
jhnwkmn 0:97a4f8cc534c 512 return str;
jhnwkmn 0:97a4f8cc534c 513 }
jhnwkmn 0:97a4f8cc534c 514 return NULL;
jhnwkmn 0:97a4f8cc534c 515 default: /* char */
jhnwkmn 0:97a4f8cc534c 516 if(*str != node->type) return NULL;
jhnwkmn 0:97a4f8cc534c 517 str++;
jhnwkmn 0:97a4f8cc534c 518 return str;
jhnwkmn 0:97a4f8cc534c 519 }
jhnwkmn 0:97a4f8cc534c 520 return NULL;
jhnwkmn 0:97a4f8cc534c 521 }
jhnwkmn 0:97a4f8cc534c 522
jhnwkmn 0:97a4f8cc534c 523 /* public api */
jhnwkmn 0:97a4f8cc534c 524 SQRex *sqstd_rex_compile(const SQChar *pattern,const SQChar **error)
jhnwkmn 0:97a4f8cc534c 525 {
jhnwkmn 0:97a4f8cc534c 526 SQRex *exp = (SQRex *)sq_malloc(sizeof(SQRex));
jhnwkmn 0:97a4f8cc534c 527 exp->_eol = exp->_bol = NULL;
jhnwkmn 0:97a4f8cc534c 528 exp->_p = pattern;
jhnwkmn 0:97a4f8cc534c 529 exp->_nallocated = (SQInteger)scstrlen(pattern) * sizeof(SQChar);
jhnwkmn 0:97a4f8cc534c 530 exp->_nodes = (SQRexNode *)sq_malloc(exp->_nallocated * sizeof(SQRexNode));
jhnwkmn 0:97a4f8cc534c 531 exp->_nsize = 0;
jhnwkmn 0:97a4f8cc534c 532 exp->_matches = 0;
jhnwkmn 0:97a4f8cc534c 533 exp->_nsubexpr = 0;
jhnwkmn 0:97a4f8cc534c 534 exp->_first = sqstd_rex_newnode(exp,OP_EXPR);
jhnwkmn 0:97a4f8cc534c 535 exp->_error = error;
jhnwkmn 0:97a4f8cc534c 536 exp->_jmpbuf = sq_malloc(sizeof(jmp_buf));
jhnwkmn 0:97a4f8cc534c 537 if(setjmp(*((jmp_buf*)exp->_jmpbuf)) == 0) {
jhnwkmn 0:97a4f8cc534c 538 SQInteger res = sqstd_rex_list(exp);
jhnwkmn 0:97a4f8cc534c 539 exp->_nodes[exp->_first].left = res;
jhnwkmn 0:97a4f8cc534c 540 if(*exp->_p!='\0')
jhnwkmn 0:97a4f8cc534c 541 sqstd_rex_error(exp,_SC("unexpected character"));
jhnwkmn 0:97a4f8cc534c 542 #ifdef _DEBUG
jhnwkmn 0:97a4f8cc534c 543 {
jhnwkmn 0:97a4f8cc534c 544 SQInteger nsize,i;
jhnwkmn 0:97a4f8cc534c 545 SQRexNode *t;
jhnwkmn 0:97a4f8cc534c 546 nsize = exp->_nsize;
jhnwkmn 0:97a4f8cc534c 547 t = &exp->_nodes[0];
jhnwkmn 0:97a4f8cc534c 548 scprintf(_SC("\n"));
jhnwkmn 0:97a4f8cc534c 549 for(i = 0;i < nsize; i++) {
jhnwkmn 0:97a4f8cc534c 550 if(exp->_nodes[i].type>MAX_CHAR)
jhnwkmn 0:97a4f8cc534c 551 scprintf(_SC("[%02d] %10s "),i,g_nnames[exp->_nodes[i].type-MAX_CHAR]);
jhnwkmn 0:97a4f8cc534c 552 else
jhnwkmn 0:97a4f8cc534c 553 scprintf(_SC("[%02d] %10c "),i,exp->_nodes[i].type);
jhnwkmn 0:97a4f8cc534c 554 scprintf(_SC("left %02d right %02d next %02d\n"),exp->_nodes[i].left,exp->_nodes[i].right,exp->_nodes[i].next);
jhnwkmn 0:97a4f8cc534c 555 }
jhnwkmn 0:97a4f8cc534c 556 scprintf(_SC("\n"));
jhnwkmn 0:97a4f8cc534c 557 }
jhnwkmn 0:97a4f8cc534c 558 #endif
jhnwkmn 0:97a4f8cc534c 559 exp->_matches = (SQRexMatch *) sq_malloc(exp->_nsubexpr * sizeof(SQRexMatch));
jhnwkmn 0:97a4f8cc534c 560 memset(exp->_matches,0,exp->_nsubexpr * sizeof(SQRexMatch));
jhnwkmn 0:97a4f8cc534c 561 }
jhnwkmn 0:97a4f8cc534c 562 else{
jhnwkmn 0:97a4f8cc534c 563 sqstd_rex_free(exp);
jhnwkmn 0:97a4f8cc534c 564 return NULL;
jhnwkmn 0:97a4f8cc534c 565 }
jhnwkmn 0:97a4f8cc534c 566 return exp;
jhnwkmn 0:97a4f8cc534c 567 }
jhnwkmn 0:97a4f8cc534c 568
jhnwkmn 0:97a4f8cc534c 569 void sqstd_rex_free(SQRex *exp)
jhnwkmn 0:97a4f8cc534c 570 {
jhnwkmn 0:97a4f8cc534c 571 if(exp) {
jhnwkmn 0:97a4f8cc534c 572 if(exp->_nodes) sq_free(exp->_nodes,exp->_nallocated * sizeof(SQRexNode));
jhnwkmn 0:97a4f8cc534c 573 if(exp->_jmpbuf) sq_free(exp->_jmpbuf,sizeof(jmp_buf));
jhnwkmn 0:97a4f8cc534c 574 if(exp->_matches) sq_free(exp->_matches,exp->_nsubexpr * sizeof(SQRexMatch));
jhnwkmn 0:97a4f8cc534c 575 sq_free(exp,sizeof(SQRex));
jhnwkmn 0:97a4f8cc534c 576 }
jhnwkmn 0:97a4f8cc534c 577 }
jhnwkmn 0:97a4f8cc534c 578
jhnwkmn 0:97a4f8cc534c 579 SQBool sqstd_rex_match(SQRex* exp,const SQChar* text)
jhnwkmn 0:97a4f8cc534c 580 {
jhnwkmn 0:97a4f8cc534c 581 const SQChar* res = NULL;
jhnwkmn 0:97a4f8cc534c 582 exp->_bol = text;
jhnwkmn 0:97a4f8cc534c 583 exp->_eol = text + scstrlen(text);
jhnwkmn 0:97a4f8cc534c 584 exp->_currsubexp = 0;
jhnwkmn 0:97a4f8cc534c 585 res = sqstd_rex_matchnode(exp,exp->_nodes,text,NULL);
jhnwkmn 0:97a4f8cc534c 586 if(res == NULL || res != exp->_eol)
jhnwkmn 0:97a4f8cc534c 587 return SQFalse;
jhnwkmn 0:97a4f8cc534c 588 return SQTrue;
jhnwkmn 0:97a4f8cc534c 589 }
jhnwkmn 0:97a4f8cc534c 590
jhnwkmn 0:97a4f8cc534c 591 SQBool sqstd_rex_searchrange(SQRex* exp,const SQChar* text_begin,const SQChar* text_end,const SQChar** out_begin, const SQChar** out_end)
jhnwkmn 0:97a4f8cc534c 592 {
jhnwkmn 0:97a4f8cc534c 593 const SQChar *cur = NULL;
jhnwkmn 0:97a4f8cc534c 594 SQInteger node = exp->_first;
jhnwkmn 0:97a4f8cc534c 595 if(text_begin >= text_end) return SQFalse;
jhnwkmn 0:97a4f8cc534c 596 exp->_bol = text_begin;
jhnwkmn 0:97a4f8cc534c 597 exp->_eol = text_end;
jhnwkmn 0:97a4f8cc534c 598 do {
jhnwkmn 0:97a4f8cc534c 599 cur = text_begin;
jhnwkmn 0:97a4f8cc534c 600 while(node != -1) {
jhnwkmn 0:97a4f8cc534c 601 exp->_currsubexp = 0;
jhnwkmn 0:97a4f8cc534c 602 cur = sqstd_rex_matchnode(exp,&exp->_nodes[node],cur,NULL);
jhnwkmn 0:97a4f8cc534c 603 if(!cur)
jhnwkmn 0:97a4f8cc534c 604 break;
jhnwkmn 0:97a4f8cc534c 605 node = exp->_nodes[node].next;
jhnwkmn 0:97a4f8cc534c 606 }
jhnwkmn 0:97a4f8cc534c 607 text_begin++;
jhnwkmn 0:97a4f8cc534c 608 } while(cur == NULL && text_begin != text_end);
jhnwkmn 0:97a4f8cc534c 609
jhnwkmn 0:97a4f8cc534c 610 if(cur == NULL)
jhnwkmn 0:97a4f8cc534c 611 return SQFalse;
jhnwkmn 0:97a4f8cc534c 612
jhnwkmn 0:97a4f8cc534c 613 --text_begin;
jhnwkmn 0:97a4f8cc534c 614
jhnwkmn 0:97a4f8cc534c 615 if(out_begin) *out_begin = text_begin;
jhnwkmn 0:97a4f8cc534c 616 if(out_end) *out_end = cur;
jhnwkmn 0:97a4f8cc534c 617 return SQTrue;
jhnwkmn 0:97a4f8cc534c 618 }
jhnwkmn 0:97a4f8cc534c 619
jhnwkmn 0:97a4f8cc534c 620 SQBool sqstd_rex_search(SQRex* exp,const SQChar* text, const SQChar** out_begin, const SQChar** out_end)
jhnwkmn 0:97a4f8cc534c 621 {
jhnwkmn 0:97a4f8cc534c 622 return sqstd_rex_searchrange(exp,text,text + scstrlen(text),out_begin,out_end);
jhnwkmn 0:97a4f8cc534c 623 }
jhnwkmn 0:97a4f8cc534c 624
jhnwkmn 0:97a4f8cc534c 625 SQInteger sqstd_rex_getsubexpcount(SQRex* exp)
jhnwkmn 0:97a4f8cc534c 626 {
jhnwkmn 0:97a4f8cc534c 627 return exp->_nsubexpr;
jhnwkmn 0:97a4f8cc534c 628 }
jhnwkmn 0:97a4f8cc534c 629
jhnwkmn 0:97a4f8cc534c 630 SQBool sqstd_rex_getsubexp(SQRex* exp, SQInteger n, SQRexMatch *subexp)
jhnwkmn 0:97a4f8cc534c 631 {
jhnwkmn 0:97a4f8cc534c 632 if( n<0 || n >= exp->_nsubexpr) return SQFalse;
jhnwkmn 0:97a4f8cc534c 633 *subexp = exp->_matches[n];
jhnwkmn 0:97a4f8cc534c 634 return SQTrue;
jhnwkmn 0:97a4f8cc534c 635 }
jhnwkmn 0:97a4f8cc534c 636