Annotation of parser3/src/types/pa_vregex.C, revision 1.10
1.1 misha 1: /** @file
2: Parser: @b regex class.
3:
4: Copyright(c) 2001-2009 ArtLebedev Group (http://www.artlebedev.com)
5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
6: */
7:
1.10 ! misha 8: static const char * const IDENT_VREGEX_C="$Date: 2009-09-17 02:17:25 $";
1.1 misha 9:
10: #include "pa_vregex.h"
1.2 misha 11: #include "pa_vint.h"
1.8 misha 12: #include "pa_vstring.h"
13:
14:
15: // defines
16:
17: #define REGEX_PATTERN_NAME "pattern"
18: #define REGEX_OPTIONS_NAME "options"
1.1 misha 19:
20:
21: char* get_pcre_exec_error_text(int exec_result){
22: switch(exec_result){
23: case PCRE_ERROR_BADUTF8:
24: case PCRE_ERROR_BADUTF8_OFFSET:
25: return "UTF-8 validation failed during pcre_exec (%d).";
26: break;
27: default:
28: return "execution error (%d)";
29: }
30: }
31:
32:
33: Value& VRegex::as_expr_result(bool/*return_string_as_is=false*/) {
34: return *new VInt(as_int());
35: }
36:
37: void VRegex::regex_options(const String* options, int* result){
38: struct Regex_option {
39: const char* key;
40: const char* keyAlt;
41: int clear;
42: int set;
43: int *result;
44: } regex_option[]={
45: {"i", "I", 0, PCRE_CASELESS, result}, // a=A
46: {"s", "S", 0, PCRE_DOTALL, result}, // ^\n\n$ [default]
47: {"m", "M", PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$
48: {"x", 0, 0, PCRE_EXTENDED, result}, // whitespace in regex ignored
49: {"U", 0, 0, PCRE_UNGREEDY, result}, // ungreedy patterns (greedy by default)
50: {"g", "G", 0, MF_GLOBAL_SEARCH, result+1}, // many rows
51: {"'", 0, 0, MF_NEED_PRE_POST_MATCH, result+1},
52: {"n", 0, 0, MF_JUST_COUNT_MATCHES, result+1},
53: {0, 0, 0, 0, 0}
54: };
55: result[0]=PCRE_EXTRA /* backslash+non-special char causes error */
56: | PCRE_DOTALL /* dot matches all chars including newline char */
57: | PCRE_DOLLAR_ENDONLY /* dollar matches only end of string, but not newline chars */;
58: result[1]=0;
59:
60: if(options && !options->is_empty())
61: for(Regex_option *o=regex_option; o->key; o++)
62: if(
63: options->pos(o->key)!=STRING_NOT_FOUND
64: || (o->keyAlt && options->pos(o->keyAlt)!=STRING_NOT_FOUND)
65: ){
66: *o->result &= ~o->clear;
67: *o->result |= o->set;
68: }
69: }
70:
71:
72: void VRegex::set(Charset& acharset, const String* aregex, const String* aoptions){
73: if(aregex->is_empty())
74: throw Exception(PARSER_RUNTIME,
75: 0,
76: "regexp is empty");
77:
78: fcharset=&acharset;
1.5 misha 79:
1.7 misha 80: fpattern=aregex->untaint_cstr(String::L_REGEX);
1.1 misha 81:
1.9 misha 82: foptions_cstr=aoptions?aoptions->cstr():0;
1.8 misha 83:
1.1 misha 84: regex_options(aoptions, foptions);
85: }
86:
87:
88: void VRegex::compile(){
89: const char* err_ptr;
90: int err_offset;
91: int options=foptions[0];
92:
93: // @todo (for UTF-8): check string & pattern and use PCRE_NO_UTF8_CHECK option
94: if(fcharset->isUTF8())
95: options|=PCRE_UTF8;
96:
97: fcode=pcre_compile(fpattern, options,
98: &err_ptr, &err_offset,
99: fcharset->pcre_tables);
100:
101: if(!fcode){
102: throw Exception(PCRE_EXCEPTION_TYPE,
1.4 misha 103: new String(fpattern+err_offset, String::L_TAINTED),
1.1 misha 104: "regular expression syntax error - %s", err_ptr);
105: }
106:
107: }
108:
109:
110: size_t VRegex::full_info(int type){
111: size_t result;
112: int fullinfo_result=pcre_fullinfo(fcode, fextra, type, &result);
113: if(fullinfo_result<0){
114: throw Exception(PCRE_EXCEPTION_TYPE,
1.4 misha 115: new String(fpattern, String::L_TAINTED),
1.1 misha 116: "pcre_full_info error (%d)", fullinfo_result);
117: }
118:
119: return result;
120: };
121:
122:
123: size_t VRegex::get_info_size(){
124: return full_info(PCRE_INFO_SIZE);
125: }
126:
127:
128: size_t VRegex::get_study_size(){
129: return full_info(PCRE_INFO_STUDYSIZE);
130: }
131:
132: void VRegex::study(){
133: if(fstudied)
134: return;
135:
136: const char* err_ptr;
137: fextra=pcre_study(fcode, 0/*options*/, &err_ptr);
138:
139: if(err_ptr){
140: throw Exception(PCRE_EXCEPTION_TYPE,
1.4 misha 141: new String(fpattern, String::L_TAINTED),
1.1 misha 142: "pcre_study error: %s", err_ptr);
143: }
144:
145: fstudied=true;
146: }
147:
148:
149: int VRegex::exec(const char* string, size_t string_len, int* ovector, int ovector_size, int prestart){
150: int result=pcre_exec(fcode, fextra,
151: string, string_len, prestart,
1.10 ! misha 152: prestart>0 ? PCRE_NO_UTF8_CHECK : 0, ovector, ovector_size);
1.1 misha 153:
154: if(result<0 && result!=PCRE_ERROR_NOMATCH){
155: throw Exception(PCRE_EXCEPTION_TYPE,
1.4 misha 156: new String(fpattern, String::L_TAINTED),
1.1 misha 157: get_pcre_exec_error_text(result), result);
158: }
159:
160: return result;
161: }
162:
163:
1.8 misha 164: Value* VRegex::get_element(const String& aname) {
165: if(aname == REGEX_PATTERN_NAME)
166: return new VString(*new String(fpattern, String::L_TAINTED));
167:
168: if(aname == REGEX_OPTIONS_NAME)
169: return new VString(*new String(foptions_cstr, String::L_TAINTED));
170:
171: // .CLASS, .CLASS_NAME
172: if(Value* result=VStateless_object::get_element(aname))
173: return result;
174:
175: throw Exception(PARSER_RUNTIME,
176: &aname,
177: "reading of invalid field");
178: }
E-mail:
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to pa_vregex.C CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [parser3project] / parser3 / src / types