Annotation of parser3/src/include/pa_string.h, revision 1.144.2.9

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: 
1.144.2.9! paf        11: static const char* IDENT_STRING_H="$Date: 2003/01/27 17:17:06 $";
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: 
1.144.2.5  paf        17: // forwards
                     18: 
                     19: class String;
                     20: 
                     21: typedef object_ptr<String> StringPtr;
                     22: typedef object_ptr<const String> ConstStringPtr;
                     23: const ConstStringPtr ConstStringPtrZero(0);
                     24: 
1.144.2.9! paf        25: class Table;
        !            26: typedef object_ptr<Table> TablePtr;
1.144.2.5  paf        27: class SQL_Connection;
                     28: class Dictionary;
                     29: 
                     30: // helpers
                     31: 
1.9       paf        32: #ifndef NO_STRING_ORIGIN
1.33      paf        33: #      define STRING_APPEND_PARAMS \
                     34:                const char *src, size_t size,  \
1.144.2.7  paf        35:                String_UL lang, \
1.33      paf        36:                const char *file, uint line
1.54      paf        37: /// appends piece to String  @see String::real_append
                     38: #      define APPEND(src, size, lang, file, line) \
                     39:                real_append(src, size, lang, file, line)
1.9       paf        40: #else
1.33      paf        41: #      define STRING_APPEND_PARAMS \
                     42:                const char *src, \
                     43:                size_t size, \
1.144.2.7  paf        44:                String_UL lang
1.54      paf        45: /// appends piece to String  @see String::real_append
                     46: #      define APPEND(src, size, lang, file, line) \
                     47:                real_append(src, size, lang)
                     48: #endif
1.83      paf        49: /// appends clean piece to String @see String::real_append
1.54      paf        50: #define APPEND_CLEAN(src, size, file, line) \
1.62      paf        51:        APPEND(src, size, String::UL_CLEAN, file, line)
1.83      paf        52: /// appends piece to String as-is @see String::real_append
                     53: #define APPEND_AS_IS(src, size, file, line) \
                     54:        APPEND(src, size, String::UL_AS_IS, file, line)
1.48      paf        55: /// appends tainted piece to String  @see String::real_append
1.54      paf        56: #define APPEND_TAINTED(src, size, file, line) \
1.62      paf        57:        APPEND(src, size, String::UL_TAINTED, file, line)
1.48      paf        58: /// handy: appends const char* piece to String  @see String::real_append
1.83      paf        59: #define        APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
1.9       paf        60: 
1.144.2.7  paf        61: typedef uint String_UL;
                     62: 
1.144.2.2  paf        63: struct String_fragment { 
1.144.2.7  paf        64:        String_UL/*Untaint_lang*/ lang; ///< untaint flag, later untaint language
1.144.2.2  paf        65:        const char *ptr;  ///< pointer to the start
                     66:        size_t size;  ///< length
1.41      paf        67: 
1.144.2.2  paf        68: #ifndef NO_STRING_ORIGIN
                     69:        /// all String pieces hold information of where they come from
                     70:        struct Origin {
                     71:                const char *file;  ///< macros file name | load file name | sql query text
                     72:                uint line; ///< file line no | record no
                     73:        } origin;  ///< origin
                     74: #endif
                     75: };
                     76: 
                     77: /** 
                     78:        String which knows the language of all it's fragments.
1.41      paf        79: 
                     80:        All pieces remember 
                     81:        - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
                     82:        - whether they are tainted or not, 
                     83:          and the language which should be used to detaint them
                     84: */
1.48      paf        85: 
1.144.2.2  paf        86: class String: public Array<String_fragment> {
                     87:        size_t fsize;
1.144.2.5  paf        88:        int count() const; // hiding fro accidental use instead of size()
1.144.2.2  paf        89: public:
1.1       paf        90: 
1.106     parser     91:        /** piece is tainted or not. the language to use when detaint
                     92:                remember to change String_Untaint_lang_name @ untaint.C along
                     93:        */
1.27      paf        94:        enum Untaint_lang {
1.72      paf        95:                UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.122     paf        96:                // these two must go before others, there are checks for >UL_AS_IS
1.62      paf        97:                UL_CLEAN, ///< clean
1.122     paf        98:                UL_AS_IS,     ///< leave all characters intact
                     99: 
1.48      paf       100:                UL_PASS_APPENDED,
1.41      paf       101:                        /**<
                    102:                                leave language built into string being appended.
                    103:                                just a flag, that value not stored
                    104:                        */
1.122     paf       105:                UL_TAINTED,  ///< tainted, untaint language as assigned later 
                    106:                // untaint languages. assigned by ^untaint[lang]{...}
1.103     parser    107:                UL_FILE_SPEC, ///< file specification
1.74      paf       108:                UL_HTTP_HEADER,    ///< text in HTTP response header
                    109:                UL_MAIL_HEADER,    ///< text in mail header
1.48      paf       110:                UL_URI,       ///< text in uri
                    111:                UL_TABLE,     ///< ^table:set body
                    112:                UL_SQL,       ///< ^table:sql body
                    113:                UL_JS,        ///< JavaScript code
1.102     parser    114:                UL_XML,         ///< ^dom:set xml
1.48      paf       115:                UL_HTML,      ///< HTML code (for editing)
1.144.2.7  paf       116:                UL_OPTIMIZE_BIT = 0x8000  ///< flag, requiring cstr whitespace optimization
1.27      paf       117:        };
                    118: 
1.8       paf       119: public:
1.144.2.5  paf       120:        static const ConstStringPtr undefined_origin;
1.8       paf       121: 
1.144.2.1  paf       122:        //static String& OnPool(Pool& apool, const char *local_src=0, size_t src_size=0, bool tainted=false);
1.144.2.6  paf       123:        explicit String(const char *src=0, size_t src_size=0, bool tainted=false);
1.14      paf       124:        String(const String& src);
1.144.2.2  paf       125:        bool is_empty() const { return !size(); }
                    126:        size_t size() const { return fsize; }
1.69      paf       127:        /// convert to C string. if 'lang' known, forcing 'lang' to it
1.144.2.5  paf       128:        CharPtr cstr(Untaint_lang lang=UL_AS_IS, 
1.82      paf       129:                SQL_Connection *connection=0,
1.133     paf       130:                Charset *cstr_charset=0, const char *cstr_charset_name=0) const {
1.88      parser    131: 
1.144.2.5  paf       132:                CharPtr result(new char[cstr_bufsize(lang, connection, cstr_charset)]);
                    133:                char *eol=store_to(result.get(), lang, connection, cstr_charset, cstr_charset_name);
1.50      paf       134:                *eol=0;
                    135:                return result;
                    136:        }
1.144.2.9! paf       137:        char *cstr(Pool& pool, Untaint_lang lang=UL_AS_IS, 
        !           138:                SQL_Connection *connection=0,
        !           139:                Charset *cstr_charset=0, const char *cstr_charset_name=0) const {
        !           140: 
        !           141:                char *result=new(pool) char[cstr_bufsize(lang, connection, cstr_charset)];
        !           142:                char *eol=store_to(result, lang, connection, cstr_charset, cstr_charset_name);
        !           143:                *eol=0;
        !           144:                return result;
        !           145:        }
1.117     paf       146:        char *cstr_debug_origins() const;
1.108     parser    147:        /// puts pieces to buf
1.144.2.5  paf       148:        void serialize(Pool& pool, size_t prolog_size,  char *& buf, size_t& buf_size) const;
1.108     parser    149:        /// appends pieces from buf to self
1.132     paf       150:        bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file);
1.46      paf       151:        /** append fragment
1.83      paf       152:                @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46      paf       153:        */
1.9       paf       154:        String& real_append(STRING_APPEND_PARAMS);
1.44      paf       155:        /// @return <0 ==0 or >0 depending on comparison result
1.62      paf       156:        int cmp (int& partial, const String& src, 
1.72      paf       157:                size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.56      paf       158:        bool operator < (const String& src) const {     int p; return cmp(p, src)<0; }
                    159:        bool operator > (const String& src) const {     int p; return cmp(p, src)>0; }
                    160:        bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; }
                    161:        bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
1.26      paf       162:        bool operator == (const String& src) const {
                    163:                if(size()!=src.size()) // can speed up in trivial case
                    164:                        return false;
1.56      paf       165:                int p; return cmp(p, src)==0;
1.26      paf       166:        }
1.56      paf       167:        bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26      paf       168: 
1.50      paf       169:        /**
                    170:                 @param partial 
                    171:                        returns partial match status. 
1.51      paf       172:                        - -1: strings too different
                    173:                        -  0: full match
                    174:                        -  1: means @c this starts @c src
                    175:                        -  2: means @src starts @this
1.50      paf       176:        */
1.61      paf       177:        int cmp(int& partial, const char* src_ptr, size_t src_size=0, 
1.72      paf       178:                size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.139     paf       179:        /// this starts with src
                    180:        bool starts_with(const char* src_ptr, size_t src_size=0) const {
                    181:                int p; cmp(p, src_ptr, src_size);
                    182:                return p==0 || p==2;
                    183:        }
1.50      paf       184:        bool operator == (const char* src_ptr) const { 
                    185:                size_t src_size=src_ptr?strlen(src_ptr):0;
                    186:                if(size() != src_size)
                    187:                        return false;
                    188:                int partial; // unused
1.56      paf       189:                return cmp(partial, src_ptr, src_size)==0; 
1.50      paf       190:        }
1.80      paf       191:        bool operator != (const char* src_ptr) const { 
                    192:                int partial; // unused
                    193:                return cmp(partial, src_ptr, 0)!=0; 
                    194:        }
                    195: 
1.119     paf       196:        String& append(const String& src, uchar lang, bool forced=false);
1.76      paf       197:        String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
1.144.2.8  paf       198:        String& operator << (const StringPtr src) { return append(*src, UL_PASS_APPENDED); }
1.76      paf       199:        String& operator << (const char *src) { return APPEND_CONST(src); }
1.8       paf       200: 
1.144.2.7  paf       201:        /// simple hash code of string
1.14      paf       202:        uint hash_code() const;
1.100     parser    203: 
1.142     paf       204:        /// extracts first char of a string, if any
                    205:        char first_char() const {
1.144.2.2  paf       206:                return is_empty()?0:*felements[0].ptr;
1.142     paf       207:        }
1.54      paf       208: 
                    209:        /// extracts [start, finish) piece of string
1.144.2.4  paf       210:        StringPtr mid(size_t start, size_t finish) const;
1.55      paf       211: 
1.59      paf       212:        /// @return position of substr in string, -1 means "not found" [String version]
1.62      paf       213:        int pos(const String& substr, 
1.112     paf       214:                int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.59      paf       215:        /// @return position of substr in string, -1 means "not found" [const char* version]
1.97      parser    216:        int pos(const char *substr, size_t substr_size=0, 
1.112     paf       217:                int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62      paf       218: 
1.144.2.5  paf       219:        void split(Array<ConstStringPtr>& result, 
1.62      paf       220:                size_t *pos_after_ref, 
                    221:                const char *delim, size_t delim_size, 
1.86      paf       222:                Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.144.2.5  paf       223:        void split(Array<ConstStringPtr>& result, 
1.62      paf       224:                size_t *pos_after_ref, 
                    225:                const String& delim, 
1.86      paf       226:                Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62      paf       227: 
1.144.2.9! paf       228:        typedef void (*Row_action)(Table& table, Array<ConstStringPtr> *row, 
1.136     paf       229:                int prestart, int prefinish, 
                    230:                int poststart, int postfinish,
1.144.2.9! paf       231:                void *info);
1.87      parser    232:        /**
1.144.2.5  paf       233:                @return table of found items, if any.
1.87      parser    234:                table format is defined and fixed[can be used by others]: 
                    235:                @verbatim
                    236:                        prematch/match/postmatch/1/2/3/...
                    237:                @endverbatim
                    238:        */
1.144.2.9! paf       239:        TablePtr match(Charset& source_charset,
1.144.2.5  paf       240:                ConstStringPtr aorigin,         
1.64      paf       241:                const String& regexp, 
1.144.2.5  paf       242:                ConstStringPtr options,
1.99      parser    243:                Row_action row_action, void *info,
1.144.2.9! paf       244:                bool *was_global=0) const;
1.87      parser    245:        enum Change_case_kind {
                    246:                CC_UPPER,
                    247:                CC_LOWER
                    248:        };
1.144.2.5  paf       249:        StringPtr change_case(Pool& pool, Charset& source_charset,
1.87      parser    250:                Change_case_kind kind) const;
1.144.2.5  paf       251:        StringPtr replace(Pool& pool, Dictionary& dict) const;
1.96      parser    252:        double as_double() const;
                    253:        int as_int() const;
1.137     paf       254: 
1.144.2.5  paf       255:        StringPtr join_chains(Pool& pool) const;
1.13      paf       256: 
1.41      paf       257: #ifndef NO_STRING_ORIGIN
                    258:        /// origin of string. calculated by first row
1.144.2.2  paf       259:        const String_fragment::Origin& origin() const;
1.41      paf       260: #endif
1.8       paf       261: 
1.144.2.5  paf       262:        size_t cstr_bufsize(Untaint_lang lang=UL_UNSPECIFIED,
                    263:                SQL_Connection *connection=0,
                    264:                Charset *buf_charset=0) const;
1.79      paf       265:        /// convert to C string, store to 'dest' which must be big enough for proper untaint
1.82      paf       266:        char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED, 
                    267:                SQL_Connection *connection=0,
1.133     paf       268:                Charset *store_to_charset=0,
                    269:                const char *store_to_charset_name=0) const;
1.144.2.2  paf       270: /*
1.144.2.1  paf       271:        void join_chain(
1.130     paf       272:                                           const Chunk*& achunk, const Chunk::Row*& arow, uint& acountdown, 
1.119     paf       273:                                           uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) const;
1.144.2.2  paf       274: */
1.7       paf       275: private: //disabled
                    276: 
1.12      paf       277:        String& operator = (const String&) { return *this; }
1.7       paf       278: 
1.1       paf       279: };
1.144.2.7  paf       280: 
                    281: /// simple hash code of string. used by Hash
                    282: inline uint hash_code(const String& self) {
                    283:        return self.hash_code();
                    284: }
                    285: /// simple hash code of string. used by Hash
                    286: inline uint hash_code(ConstStringPtr self) {
                    287:        return self->hash_code();
                    288: }
1.144.2.5  paf       289: 
                    290: // statics
                    291: 
                    292: const ConstStringPtr String::undefined_origin=ConstStringPtr(0);
                    293: 
                    294: // helpers
                    295: 
                    296: #define STRING_PREPARED_FOREACH_FRAGMENT(code) \
                    297:        for(; fragment<last; fragment++) { \
                    298:                code; \
                    299:        }
                    300: 
                    301: #define STRING_FOREACH_FRAGMENT(code) \
                    302:        { \
                    303:                element_type *last = felements+fused; \
                    304:                element_type *fragment = felements; \
                    305:                STRING_PREPARED_FOREACH_FRAGMENT(code) \
                    306:        }
1.1       paf       307: 
                    308: #endif

E-mail: