Annotation of parser3/src/include/pa_string.h, revision 1.144.2.2
1.41 paf 1: /** @file
1.43 paf 2: Parser: string class decl.
3:
1.144 paf 4: Copyright (c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com)
1.124 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.1 paf 6: */
7:
8: #ifndef PA_STRING_H
9: #define PA_STRING_H
1.140 paf 10:
1.144.2.2! paf 11: static const char* IDENT_STRING_H="$Date: 2003/01/22 15:39:07 $";
1.1 paf 12:
1.15 paf 13: #include "pa_pool.h"
1.4 paf 14: #include "pa_types.h"
1.144.2.2! paf 15: #include "pa_array.h"
1.4 paf 16:
1.9 paf 17: #ifndef NO_STRING_ORIGIN
1.33 paf 18: # define STRING_APPEND_PARAMS \
19: const char *src, size_t size, \
1.119 paf 20: uchar lang, \
1.33 paf 21: const char *file, uint line
1.54 paf 22: /// appends piece to String @see String::real_append
23: # define APPEND(src, size, lang, file, line) \
24: real_append(src, size, lang, file, line)
1.9 paf 25: #else
1.33 paf 26: # define STRING_APPEND_PARAMS \
27: const char *src, \
28: size_t size, \
1.119 paf 29: uchar lang
1.54 paf 30: /// appends piece to String @see String::real_append
31: # define APPEND(src, size, lang, file, line) \
32: real_append(src, size, lang)
33: #endif
1.83 paf 34: /// appends clean piece to String @see String::real_append
1.54 paf 35: #define APPEND_CLEAN(src, size, file, line) \
1.62 paf 36: APPEND(src, size, String::UL_CLEAN, file, line)
1.83 paf 37: /// appends piece to String as-is @see String::real_append
38: #define APPEND_AS_IS(src, size, file, line) \
39: APPEND(src, size, String::UL_AS_IS, file, line)
1.48 paf 40: /// appends tainted piece to String @see String::real_append
1.54 paf 41: #define APPEND_TAINTED(src, size, file, line) \
1.62 paf 42: APPEND(src, size, String::UL_TAINTED, file, line)
1.48 paf 43: /// handy: appends const char* piece to String @see String::real_append
1.83 paf 44: #define APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
1.9 paf 45:
1.135 paf 46: class Table;
1.71 paf 47: class SQL_Connection;
1.101 parser 48: class Dictionary;
1.62 paf 49:
1.144.2.2! paf 50: struct String_fragment {
! 51: const char *ptr; ///< pointer to the start
! 52: size_t size; ///< length
! 53: int/*Untaint_lang*/ lang; ///< untaint flag, later untaint language
1.41 paf 54:
1.144.2.2! paf 55: #ifndef NO_STRING_ORIGIN
! 56: /// all String pieces hold information of where they come from
! 57: struct Origin {
! 58: const char *file; ///< macros file name | load file name | sql query text
! 59: uint line; ///< file line no | record no
! 60: } origin; ///< origin
! 61: #endif
! 62: };
! 63:
! 64: /**
! 65: String which knows the language of all it's fragments.
1.41 paf 66:
67: All pieces remember
68: - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
69: - whether they are tainted or not,
70: and the language which should be used to detaint them
71: */
1.48 paf 72:
1.144.2.2! paf 73: class String: public Array<String_fragment> {
! 74: size_t fsize;
! 75: public:
1.1 paf 76:
1.106 parser 77: /** piece is tainted or not. the language to use when detaint
78: remember to change String_Untaint_lang_name @ untaint.C along
79: */
1.27 paf 80: enum Untaint_lang {
1.72 paf 81: UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.122 paf 82: // these two must go before others, there are checks for >UL_AS_IS
1.62 paf 83: UL_CLEAN, ///< clean
1.122 paf 84: UL_AS_IS, ///< leave all characters intact
85:
1.48 paf 86: UL_PASS_APPENDED,
1.41 paf 87: /**<
88: leave language built into string being appended.
89: just a flag, that value not stored
90: */
1.122 paf 91: UL_TAINTED, ///< tainted, untaint language as assigned later
92: // untaint languages. assigned by ^untaint[lang]{...}
1.103 parser 93: UL_FILE_SPEC, ///< file specification
1.74 paf 94: UL_HTTP_HEADER, ///< text in HTTP response header
95: UL_MAIL_HEADER, ///< text in mail header
1.48 paf 96: UL_URI, ///< text in uri
97: UL_TABLE, ///< ^table:set body
98: UL_SQL, ///< ^table:sql body
99: UL_JS, ///< JavaScript code
1.102 parser 100: UL_XML, ///< ^dom:set xml
1.48 paf 101: UL_HTML, ///< HTML code (for editing)
1.119 paf 102: UL_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization
1.27 paf 103: };
104:
1.8 paf 105: public:
106:
1.144.2.1 paf 107: //static String& OnPool(Pool& apool, const char *local_src=0, size_t src_size=0, bool tainted=false);
108: String(const char *src=0, size_t src_size=0, bool tainted=false);
1.14 paf 109: String(const String& src);
1.144.2.2! paf 110: bool is_empty() const { return !size(); }
! 111: size_t size() const { return fsize; }
1.69 paf 112: /// convert to C string. if 'lang' known, forcing 'lang' to it
1.109 paf 113: char *cstr(Untaint_lang lang=UL_AS_IS,
1.82 paf 114: SQL_Connection *connection=0,
1.133 paf 115: Charset *cstr_charset=0, const char *cstr_charset_name=0) const {
1.88 parser 116:
1.121 paf 117: char *result=(char *)malloc(cstr_bufsize(lang, connection, cstr_charset));
1.133 paf 118: char *eol=store_to(result, lang, connection, cstr_charset, cstr_charset_name);
1.50 paf 119: *eol=0;
120: return result;
121: }
1.117 paf 122: char *cstr_debug_origins() const;
1.108 parser 123: /// puts pieces to buf
124: void serialize(size_t prolog_size, void *& buf, size_t& buf_size) const;
125: /// appends pieces from buf to self
1.132 paf 126: bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file);
1.46 paf 127: /** append fragment
1.83 paf 128: @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46 paf 129: */
1.9 paf 130: String& real_append(STRING_APPEND_PARAMS);
1.44 paf 131: /// @return <0 ==0 or >0 depending on comparison result
1.62 paf 132: int cmp (int& partial, const String& src,
1.72 paf 133: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.56 paf 134: bool operator < (const String& src) const { int p; return cmp(p, src)<0; }
135: bool operator > (const String& src) const { int p; return cmp(p, src)>0; }
136: bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; }
137: bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
1.26 paf 138: bool operator == (const String& src) const {
139: if(size()!=src.size()) // can speed up in trivial case
140: return false;
1.56 paf 141: int p; return cmp(p, src)==0;
1.26 paf 142: }
1.56 paf 143: bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26 paf 144:
1.50 paf 145: /**
146: @param partial
147: returns partial match status.
1.51 paf 148: - -1: strings too different
149: - 0: full match
150: - 1: means @c this starts @c src
151: - 2: means @src starts @this
1.50 paf 152: */
1.61 paf 153: int cmp(int& partial, const char* src_ptr, size_t src_size=0,
1.72 paf 154: size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.139 paf 155: /// this starts with src
156: bool starts_with(const char* src_ptr, size_t src_size=0) const {
157: int p; cmp(p, src_ptr, src_size);
158: return p==0 || p==2;
159: }
1.50 paf 160: bool operator == (const char* src_ptr) const {
161: size_t src_size=src_ptr?strlen(src_ptr):0;
162: if(size() != src_size)
163: return false;
164: int partial; // unused
1.56 paf 165: return cmp(partial, src_ptr, src_size)==0;
1.50 paf 166: }
1.80 paf 167: bool operator != (const char* src_ptr) const {
168: int partial; // unused
169: return cmp(partial, src_ptr, 0)!=0;
170: }
171:
1.119 paf 172: String& append(const String& src, uchar lang, bool forced=false);
1.76 paf 173: String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
174: String& operator << (const char *src) { return APPEND_CONST(src); }
1.8 paf 175:
1.41 paf 176: /// simple hash code of string. used by Hash
1.14 paf 177: uint hash_code() const;
1.100 parser 178:
1.142 paf 179: /// extracts first char of a string, if any
180: char first_char() const {
1.144.2.2! paf 181: return is_empty()?0:*felements[0].ptr;
1.142 paf 182: }
1.54 paf 183:
184: /// extracts [start, finish) piece of string
1.70 paf 185: String& mid(size_t start, size_t finish) const;
1.55 paf 186:
1.59 paf 187: /// @return position of substr in string, -1 means "not found" [String version]
1.62 paf 188: int pos(const String& substr,
1.112 paf 189: int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.59 paf 190: /// @return position of substr in string, -1 means "not found" [const char* version]
1.97 parser 191: int pos(const char *substr, size_t substr_size=0,
1.112 paf 192: int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62 paf 193:
194: void split(Array& result,
195: size_t *pos_after_ref,
196: const char *delim, size_t delim_size,
1.86 paf 197: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 198: void split(Array& result,
199: size_t *pos_after_ref,
200: const String& delim,
1.86 paf 201: Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62 paf 202:
1.136 paf 203: typedef void (*Row_action)(Table& table, Array *row,
204: int prestart, int prefinish,
205: int poststart, int postfinish,
1.68 paf 206: void *info);
1.87 parser 207: /**
208: @return true if fills table.
209: table format is defined and fixed[can be used by others]:
210: @verbatim
211: prematch/match/postmatch/1/2/3/...
212: @endverbatim
213: */
1.144.2.1 paf 214: bool match(Charset& source_charset,
1.81 paf 215: const String *aorigin,
1.64 paf 216: const String& regexp,
1.65 paf 217: const String *options,
1.66 paf 218: Table **table,
1.99 parser 219: Row_action row_action, void *info,
220: bool *was_global=0) const;
1.87 parser 221: enum Change_case_kind {
222: CC_UPPER,
223: CC_LOWER
224: };
1.144.2.1 paf 225: String& change_case(Charset& source_charset,
1.87 parser 226: Change_case_kind kind) const;
1.144.2.1 paf 227: String& replace(Dictionary& dict) const;
1.96 parser 228: double as_double() const;
229: int as_int() const;
1.137 paf 230:
1.144.2.1 paf 231: String& join_chains(char** cstr) const;
1.13 paf 232:
1.41 paf 233: #ifndef NO_STRING_ORIGIN
234: /// origin of string. calculated by first row
1.144.2.2! paf 235: const String_fragment::Origin& origin() const;
1.41 paf 236: #endif
1.8 paf 237:
1.129 paf 238: private:
1.10 paf 239:
1.125 paf 240: Untaint_lang lang_of(size_t offset) const;
1.79 paf 241:
1.116 paf 242: size_t cstr_bufsize(Untaint_lang lang,
243: SQL_Connection *connection,
1.121 paf 244: Charset *buf_charset) const;
1.79 paf 245: /// convert to C string, store to 'dest' which must be big enough for proper untaint
1.82 paf 246: char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED,
247: SQL_Connection *connection=0,
1.133 paf 248: Charset *store_to_charset=0,
249: const char *store_to_charset_name=0) const;
1.144.2.2! paf 250: /*
1.144.2.1 paf 251: void join_chain(
1.130 paf 252: const Chunk*& achunk, const Chunk::Row*& arow, uint& acountdown,
1.119 paf 253: uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) const;
1.144.2.2! paf 254: */
1.7 paf 255: private: //disabled
256:
1.12 paf 257: String& operator = (const String&) { return *this; }
1.7 paf 258:
1.1 paf 259: };
260:
261: #endif
E-mail: