Annotation of parser3/src/include/pa_string.h, revision 1.144.2.28.2.1

1.41      paf         1: /** @file
1.43      paf         2:        Parser: string class decl.
                      3: 
1.144.2.18  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: 
1.144.2.28.2.  (paf       11:): static const char* IDENT_STRING_H="$Date: 2003/03/06 16:15:25 $";
1.144.2.10  paf        12: 
                     13: // includes
1.1       paf        14: 
1.144.2.28.2.  (paf       15:): 
1.144.2.12  paf        16: #include "pa_types.h"
                     17: #include "pa_array.h"
1.4       paf        18: 
1.144.2.12  paf        19: // forwards
1.144.2.10  paf        20: 
1.144.2.28  paf        21: class Charset;
1.144.2.22  paf        22: class Table; DECLARE_OBJECT_PTR(Table);
1.144.2.5  paf        23: class SQL_Connection;
                     24: class Dictionary;
1.144.2.15  paf        25: class Request_charsets;
1.144.2.28.2.  (paf       26:): typedef Array<String* > ArrayString; DECLARE_OBJECT_PTR(ArrayString);
1.144.2.5  paf        27: 
                     28: // helpers
                     29: 
1.9       paf        30: #ifndef NO_STRING_ORIGIN
1.33      paf        31: #      define STRING_APPEND_PARAMS \
1.144.2.18  paf        32:                const char* src, size_t size,  \
1.144.2.7  paf        33:                String_UL lang, \
1.144.2.18  paf        34:                const char* file, uint line
1.54      paf        35: /// appends piece to String  @see String::real_append
                     36: #      define APPEND(src, size, lang, file, line) \
                     37:                real_append(src, size, lang, file, line)
1.9       paf        38: #else
1.33      paf        39: #      define STRING_APPEND_PARAMS \
1.144.2.18  paf        40:                const char* src, \
1.33      paf        41:                size_t size, \
1.144.2.7  paf        42:                String_UL lang
1.54      paf        43: /// appends piece to String  @see String::real_append
                     44: #      define APPEND(src, size, lang, file, line) \
                     45:                real_append(src, size, lang)
                     46: #endif
1.83      paf        47: /// appends clean piece to String @see String::real_append
1.54      paf        48: #define APPEND_CLEAN(src, size, file, line) \
1.62      paf        49:        APPEND(src, size, String::UL_CLEAN, file, line)
1.83      paf        50: /// appends piece to String as-is @see String::real_append
                     51: #define APPEND_AS_IS(src, size, file, line) \
                     52:        APPEND(src, size, String::UL_AS_IS, file, line)
1.48      paf        53: /// appends tainted piece to String  @see String::real_append
1.54      paf        54: #define APPEND_TAINTED(src, size, file, line) \
1.62      paf        55:        APPEND(src, size, String::UL_TAINTED, file, line)
1.48      paf        56: /// handy: appends const char* piece to String  @see String::real_append
1.83      paf        57: #define        APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
1.9       paf        58: 
1.144.2.7  paf        59: typedef uint String_UL;
                     60: 
1.144.2.2  paf        61: struct String_fragment { 
1.144.2.7  paf        62:        String_UL/*Untaint_lang*/ lang; ///< untaint flag, later untaint language
1.144.2.18  paf        63:        const char* ptr;  ///< pointer to the start
1.144.2.2  paf        64:        size_t size;  ///< length
1.41      paf        65: 
1.144.2.2  paf        66: #ifndef NO_STRING_ORIGIN
                     67:        /// all String pieces hold information of where they come from
                     68:        struct Origin {
1.144.2.18  paf        69:                const char* file;  ///< macros file name | load file name | sql query text
1.144.2.2  paf        70:                uint line; ///< file line no | record no
                     71:        } origin;  ///< origin
                     72: #endif
                     73: };
                     74: 
                     75: /** 
                     76:        String which knows the language of all it's fragments.
1.41      paf        77: 
                     78:        All pieces remember 
                     79:        - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
                     80:        - whether they are tainted or not, 
                     81:          and the language which should be used to detaint them
                     82: */
1.48      paf        83: 
1.144.2.2  paf        84: class String: public Array<String_fragment> {
                     85:        size_t fsize;
1.144.2.24  paf        86:        int count() const { return 0; } // hiding fro accidental use instead of size()
1.144.2.2  paf        87: public:
1.1       paf        88: 
1.106     parser     89:        /** piece is tainted or not. the language to use when detaint
                     90:                remember to change String_Untaint_lang_name @ untaint.C along
                     91:        */
1.27      paf        92:        enum Untaint_lang {
1.72      paf        93:                UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.122     paf        94:                // these two must go before others, there are checks for >UL_AS_IS
1.62      paf        95:                UL_CLEAN, ///< clean
1.122     paf        96:                UL_AS_IS,     ///< leave all characters intact
                     97: 
1.48      paf        98:                UL_PASS_APPENDED,
1.41      paf        99:                        /**<
                    100:                                leave language built into string being appended.
                    101:                                just a flag, that value not stored
                    102:                        */
1.122     paf       103:                UL_TAINTED,  ///< tainted, untaint language as assigned later 
                    104:                // untaint languages. assigned by ^untaint[lang]{...}
1.103     parser    105:                UL_FILE_SPEC, ///< file specification
1.74      paf       106:                UL_HTTP_HEADER,    ///< text in HTTP response header
                    107:                UL_MAIL_HEADER,    ///< text in mail header
1.48      paf       108:                UL_URI,       ///< text in uri
                    109:                UL_TABLE,     ///< ^table:set body
                    110:                UL_SQL,       ///< ^table:sql body
                    111:                UL_JS,        ///< JavaScript code
1.102     parser    112:                UL_XML,         ///< ^dom:set xml
1.48      paf       113:                UL_HTML,      ///< HTML code (for editing)
1.144.2.7  paf       114:                UL_OPTIMIZE_BIT = 0x8000  ///< flag, requiring cstr whitespace optimization
1.27      paf       115:        };
                    116: 
1.8       paf       117: public:
                    118: 
1.144.2.18  paf       119:        explicit String(const char* src=0, size_t src_size=0, bool tainted=false);
1.14      paf       120:        String(const String& src);
1.144.2.2  paf       121:        bool is_empty() const { return !size(); }
                    122:        size_t size() const { return fsize; }
1.69      paf       123:        /// convert to C string. if 'lang' known, forcing 'lang' to it
1.144.2.28.2.  (paf      124:):      const char* cstr(Untaint_lang lang=UL_AS_IS, 
1.82      paf       125:                SQL_Connection *connection=0,
1.144.2.14  paf       126:                const Request_charsets *charsets=0) const {
1.88      parser    127: 
1.144.2.28.2.  (paf      128:):              char* result=new char[cstr_bufsize(lang, connection, charsets)];
1.144.2.12  paf       129:                char *eol=store_to(result, lang, connection, charsets);
1.144.2.9  paf       130:                *eol=0;
                    131:                return result;
                    132:        }
1.108     parser    133:        /// puts pieces to buf
1.144.2.28.2.  (paf      134:):      void serialize(size_t prolog_size,  char *& buf, size_t& buf_size) const;
1.108     parser    135:        /// appends pieces from buf to self
1.144.2.18  paf       136:        bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file);
1.46      paf       137:        /** append fragment
1.83      paf       138:                @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46      paf       139:        */
1.9       paf       140:        String& real_append(STRING_APPEND_PARAMS);
1.44      paf       141:        /// @return <0 ==0 or >0 depending on comparison result
1.62      paf       142:        int cmp (int& partial, const String& src, 
1.72      paf       143:                size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.56      paf       144:        bool operator < (const String& src) const {     int p; return cmp(p, src)<0; }
                    145:        bool operator > (const String& src) const {     int p; return cmp(p, src)>0; }
                    146:        bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; }
                    147:        bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
1.26      paf       148:        bool operator == (const String& src) const {
                    149:                if(size()!=src.size()) // can speed up in trivial case
                    150:                        return false;
1.56      paf       151:                int p; return cmp(p, src)==0;
1.26      paf       152:        }
1.56      paf       153:        bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26      paf       154: 
1.50      paf       155:        /**
                    156:                 @param partial 
                    157:                        returns partial match status. 
1.51      paf       158:                        - -1: strings too different
                    159:                        -  0: full match
                    160:                        -  1: means @c this starts @c src
                    161:                        -  2: means @src starts @this
1.50      paf       162:        */
1.61      paf       163:        int cmp(int& partial, const char* src_ptr, size_t src_size=0, 
1.72      paf       164:                size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.139     paf       165:        /// this starts with src
                    166:        bool starts_with(const char* src_ptr, size_t src_size=0) const {
                    167:                int p; cmp(p, src_ptr, src_size);
                    168:                return p==0 || p==2;
                    169:        }
1.50      paf       170:        bool operator == (const char* src_ptr) const { 
                    171:                size_t src_size=src_ptr?strlen(src_ptr):0;
                    172:                if(size() != src_size)
                    173:                        return false;
                    174:                int partial; // unused
1.56      paf       175:                return cmp(partial, src_ptr, src_size)==0; 
1.50      paf       176:        }
1.80      paf       177:        bool operator != (const char* src_ptr) const { 
                    178:                int partial; // unused
                    179:                return cmp(partial, src_ptr, 0)!=0; 
                    180:        }
                    181: 
1.144.2.25  paf       182:        String& append_to(String& dest, String_UL lang, bool forced) const;
                    183:        String& append(const String& src, String_UL lang, bool forced=false) { 
1.144.2.11  paf       184:                return src.append_to(*this, lang, forced);
                    185:        }
1.76      paf       186:        String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
1.144.2.28.2.  (paf      187:):      //String& operator << (const String* src) { return append(*src, UL_PASS_APPENDED); }
1.144.2.18  paf       188:        String& operator << (const char* src) { return APPEND_CONST(src); }
1.8       paf       189: 
1.144.2.7  paf       190:        /// simple hash code of string
1.14      paf       191:        uint hash_code() const;
1.100     parser    192: 
1.142     paf       193:        /// extracts first char of a string, if any
                    194:        char first_char() const {
1.144.2.2  paf       195:                return is_empty()?0:*felements[0].ptr;
1.142     paf       196:        }
1.54      paf       197: 
                    198:        /// extracts [start, finish) piece of string
1.144.2.28.2.  (paf      199:):      const String& mid(size_t start, size_t finish) const;
1.55      paf       200: 
1.59      paf       201:        /// @return position of substr in string, -1 means "not found" [String version]
1.62      paf       202:        int pos(const String& substr, 
1.112     paf       203:                int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.59      paf       204:        /// @return position of substr in string, -1 means "not found" [const char* version]
1.144.2.18  paf       205:        int pos(const char* substr, size_t substr_size=0, 
1.112     paf       206:                int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62      paf       207: 
1.144.2.22  paf       208:        void split(ArrayString& result, 
1.62      paf       209:                size_t *pos_after_ref, 
1.144.2.18  paf       210:                const char* delim, size_t delim_size, 
1.144.2.17  paf       211:                Untaint_lang lang=UL_UNSPECIFIED, int limit=-1);
1.144.2.22  paf       212:        void split(ArrayString& result, 
1.62      paf       213:                size_t *pos_after_ref, 
                    214:                const String& delim, 
1.144.2.17  paf       215:                Untaint_lang lang=UL_UNSPECIFIED, int limit=-1);
1.62      paf       216: 
1.144.2.28.2.  (paf      217:):      typedef void (*Row_action)(Table* table, ArrayString&  row, 
1.136     paf       218:                int prestart, int prefinish, 
                    219:                int poststart, int postfinish,
1.144.2.9  paf       220:                void *info);
1.87      parser    221:        /**
1.144.2.5  paf       222:                @return table of found items, if any.
1.87      parser    223:                table format is defined and fixed[can be used by others]: 
                    224:                @verbatim
                    225:                        prematch/match/postmatch/1/2/3/...
                    226:                @endverbatim
                    227:        */
1.144.2.28.2.  (paf      228:):      Table* match(Charset& source_charset,
                    229:):              const String& aorigin,          
1.64      paf       230:                const String& regexp, 
1.144.2.28.2.  (paf      231:):              const String& options,
1.99      parser    232:                Row_action row_action, void *info,
1.144.2.27  paf       233:                bool& just_matched) const;
1.87      parser    234:        enum Change_case_kind {
                    235:                CC_UPPER,
                    236:                CC_LOWER
                    237:        };
1.144.2.28.2.  (paf      238:):      const String& change_case(Charset& source_charset,
1.144.2.17  paf       239:                Change_case_kind kind);
1.144.2.28.2.  (paf      240:):      const String& replace(const Dictionary& dict) const;
1.144.2.17  paf       241:        double as_double();
                    242:        int as_int();
1.137     paf       243: 
1.144.2.28.2.  (paf      244:):      const String& join_chains() const;
1.13      paf       245: 
1.41      paf       246: #ifndef NO_STRING_ORIGIN
                    247:        /// origin of string. calculated by first row
1.144.2.2  paf       248:        const String_fragment::Origin& origin() const;
1.41      paf       249: #endif
1.8       paf       250: 
1.144.2.5  paf       251:        size_t cstr_bufsize(Untaint_lang lang=UL_UNSPECIFIED,
                    252:                SQL_Connection *connection=0,
1.144.2.14  paf       253:                const Request_charsets *charsets=0) const;
1.79      paf       254:        /// convert to C string, store to 'dest' which must be big enough for proper untaint
1.82      paf       255:        char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED, 
                    256:                SQL_Connection *connection=0,
1.144.2.14  paf       257:                const Request_charsets *charsets=0) const;
1.144.2.2  paf       258: /*
1.144.2.1  paf       259:        void join_chain(
1.130     paf       260:                                           const Chunk*& achunk, const Chunk::Row*& arow, uint& acountdown, 
1.144.2.25  paf       261:                                           String_UL& joined_lang, const char* & joined_ptr, size_t& joined_size) const;
1.144.2.2  paf       262: */
1.7       paf       263: private: //disabled
                    264: 
1.12      paf       265:        String& operator = (const String&) { return *this; }
1.7       paf       266: 
1.1       paf       267: };
1.144.2.22  paf       268: 
1.144.2.7  paf       269: 
                    270: /// simple hash code of string. used by Hash
                    271: inline uint hash_code(const String& self) {
                    272:        return self.hash_code();
                    273: }
1.144.2.5  paf       274: 
                    275: // helpers
                    276: 
                    277: #define STRING_PREPARED_FOREACH_FRAGMENT(code) \
                    278:        for(; fragment<last; fragment++) { \
                    279:                code; \
                    280:        }
                    281: 
                    282: #define STRING_FOREACH_FRAGMENT(code) \
                    283:        { \
                    284:                element_type *last = felements+fused; \
                    285:                element_type *fragment = felements; \
                    286:                STRING_PREPARED_FOREACH_FRAGMENT(code) \
                    287:        }
1.1       paf       288: 
                    289: #endif

E-mail: