Annotation of parser3/src/include/pa_string.h, revision 1.144.2.4

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: 
1.144.2.4! paf        11: static const char* IDENT_STRING_H="$Date: 2003/01/23 17:05:34 $";
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: 
1.144.2.4! paf        46: class String;
        !            47: 
        !            48: typedef object_ptr<String> StringPtr;
        !            49: typedef object_ptr<const String> ConstStringPtr;
        !            50: 
1.135     paf        51: class Table;
1.71      paf        52: class SQL_Connection;
1.101     parser     53: class Dictionary;
1.62      paf        54: 
1.144.2.2  paf        55: struct String_fragment { 
                     56:        const char *ptr;  ///< pointer to the start
                     57:        size_t size;  ///< length
                     58:        int/*Untaint_lang*/ lang; ///< untaint flag, later untaint language
1.41      paf        59: 
1.144.2.2  paf        60: #ifndef NO_STRING_ORIGIN
                     61:        /// all String pieces hold information of where they come from
                     62:        struct Origin {
                     63:                const char *file;  ///< macros file name | load file name | sql query text
                     64:                uint line; ///< file line no | record no
                     65:        } origin;  ///< origin
                     66: #endif
                     67: };
                     68: 
                     69: /** 
                     70:        String which knows the language of all it's fragments.
1.41      paf        71: 
                     72:        All pieces remember 
                     73:        - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
                     74:        - whether they are tainted or not, 
                     75:          and the language which should be used to detaint them
                     76: */
1.48      paf        77: 
1.144.2.2  paf        78: class String: public Array<String_fragment> {
                     79:        size_t fsize;
                     80: public:
1.1       paf        81: 
1.106     parser     82:        /** piece is tainted or not. the language to use when detaint
                     83:                remember to change String_Untaint_lang_name @ untaint.C along
                     84:        */
1.27      paf        85:        enum Untaint_lang {
1.72      paf        86:                UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.122     paf        87:                // these two must go before others, there are checks for >UL_AS_IS
1.62      paf        88:                UL_CLEAN, ///< clean
1.122     paf        89:                UL_AS_IS,     ///< leave all characters intact
                     90: 
1.48      paf        91:                UL_PASS_APPENDED,
1.41      paf        92:                        /**<
                     93:                                leave language built into string being appended.
                     94:                                just a flag, that value not stored
                     95:                        */
1.122     paf        96:                UL_TAINTED,  ///< tainted, untaint language as assigned later 
                     97:                // untaint languages. assigned by ^untaint[lang]{...}
1.103     parser     98:                UL_FILE_SPEC, ///< file specification
1.74      paf        99:                UL_HTTP_HEADER,    ///< text in HTTP response header
                    100:                UL_MAIL_HEADER,    ///< text in mail header
1.48      paf       101:                UL_URI,       ///< text in uri
                    102:                UL_TABLE,     ///< ^table:set body
                    103:                UL_SQL,       ///< ^table:sql body
                    104:                UL_JS,        ///< JavaScript code
1.102     parser    105:                UL_XML,         ///< ^dom:set xml
1.48      paf       106:                UL_HTML,      ///< HTML code (for editing)
1.119     paf       107:                UL_OPTIMIZE_BIT = 0x80  ///< flag, requiring cstr whitespace optimization
1.27      paf       108:        };
                    109: 
1.8       paf       110: public:
                    111: 
1.144.2.1  paf       112:        //static String& OnPool(Pool& apool, const char *local_src=0, size_t src_size=0, bool tainted=false);
                    113:        String(const char *src=0, size_t src_size=0, bool tainted=false);
1.14      paf       114:        String(const String& src);
1.144.2.2  paf       115:        bool is_empty() const { return !size(); }
                    116:        size_t size() const { return fsize; }
1.69      paf       117:        /// convert to C string. if 'lang' known, forcing 'lang' to it
1.109     paf       118:        char *cstr(Untaint_lang lang=UL_AS_IS, 
1.82      paf       119:                SQL_Connection *connection=0,
1.133     paf       120:                Charset *cstr_charset=0, const char *cstr_charset_name=0) const {
1.88      parser    121: 
1.121     paf       122:                char *result=(char *)malloc(cstr_bufsize(lang, connection, cstr_charset));
1.133     paf       123:                char *eol=store_to(result, lang, connection, cstr_charset, cstr_charset_name);
1.50      paf       124:                *eol=0;
                    125:                return result;
                    126:        }
1.117     paf       127:        char *cstr_debug_origins() const;
1.108     parser    128:        /// puts pieces to buf
                    129:        void serialize(size_t prolog_size,  void *& buf, size_t& buf_size) const;
                    130:        /// appends pieces from buf to self
1.132     paf       131:        bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file);
1.46      paf       132:        /** append fragment
1.83      paf       133:                @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46      paf       134:        */
1.9       paf       135:        String& real_append(STRING_APPEND_PARAMS);
1.44      paf       136:        /// @return <0 ==0 or >0 depending on comparison result
1.62      paf       137:        int cmp (int& partial, const String& src, 
1.72      paf       138:                size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.56      paf       139:        bool operator < (const String& src) const {     int p; return cmp(p, src)<0; }
                    140:        bool operator > (const String& src) const {     int p; return cmp(p, src)>0; }
                    141:        bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; }
                    142:        bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
1.26      paf       143:        bool operator == (const String& src) const {
                    144:                if(size()!=src.size()) // can speed up in trivial case
                    145:                        return false;
1.56      paf       146:                int p; return cmp(p, src)==0;
1.26      paf       147:        }
1.56      paf       148:        bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26      paf       149: 
1.50      paf       150:        /**
                    151:                 @param partial 
                    152:                        returns partial match status. 
1.51      paf       153:                        - -1: strings too different
                    154:                        -  0: full match
                    155:                        -  1: means @c this starts @c src
                    156:                        -  2: means @src starts @this
1.50      paf       157:        */
1.61      paf       158:        int cmp(int& partial, const char* src_ptr, size_t src_size=0, 
1.72      paf       159:                size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.139     paf       160:        /// this starts with src
                    161:        bool starts_with(const char* src_ptr, size_t src_size=0) const {
                    162:                int p; cmp(p, src_ptr, src_size);
                    163:                return p==0 || p==2;
                    164:        }
1.50      paf       165:        bool operator == (const char* src_ptr) const { 
                    166:                size_t src_size=src_ptr?strlen(src_ptr):0;
                    167:                if(size() != src_size)
                    168:                        return false;
                    169:                int partial; // unused
1.56      paf       170:                return cmp(partial, src_ptr, src_size)==0; 
1.50      paf       171:        }
1.80      paf       172:        bool operator != (const char* src_ptr) const { 
                    173:                int partial; // unused
                    174:                return cmp(partial, src_ptr, 0)!=0; 
                    175:        }
                    176: 
1.119     paf       177:        String& append(const String& src, uchar lang, bool forced=false);
1.76      paf       178:        String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
                    179:        String& operator << (const char *src) { return APPEND_CONST(src); }
1.8       paf       180: 
1.41      paf       181:        /// simple hash code of string. used by Hash
1.14      paf       182:        uint hash_code() const;
1.100     parser    183: 
1.142     paf       184:        /// extracts first char of a string, if any
                    185:        char first_char() const {
1.144.2.2  paf       186:                return is_empty()?0:*felements[0].ptr;
1.142     paf       187:        }
1.54      paf       188: 
                    189:        /// extracts [start, finish) piece of string
1.144.2.4! paf       190:        StringPtr mid(size_t start, size_t finish) const;
1.55      paf       191: 
1.59      paf       192:        /// @return position of substr in string, -1 means "not found" [String version]
1.62      paf       193:        int pos(const String& substr, 
1.112     paf       194:                int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.59      paf       195:        /// @return position of substr in string, -1 means "not found" [const char* version]
1.97      parser    196:        int pos(const char *substr, size_t substr_size=0, 
1.112     paf       197:                int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62      paf       198: 
                    199:        void split(Array& result, 
                    200:                size_t *pos_after_ref, 
                    201:                const char *delim, size_t delim_size, 
1.86      paf       202:                Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62      paf       203:        void split(Array& result, 
                    204:                size_t *pos_after_ref, 
                    205:                const String& delim, 
1.86      paf       206:                Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62      paf       207: 
1.136     paf       208:        typedef void (*Row_action)(Table& table, Array *row, 
                    209:                int prestart, int prefinish, 
                    210:                int poststart, int postfinish,
1.68      paf       211:                void *info);
1.87      parser    212:        /**
                    213:                @return true if fills table.
                    214:                table format is defined and fixed[can be used by others]: 
                    215:                @verbatim
                    216:                        prematch/match/postmatch/1/2/3/...
                    217:                @endverbatim
                    218:        */
1.144.2.1  paf       219:        bool match(Charset& source_charset,
1.81      paf       220:                const String *aorigin,          
1.64      paf       221:                const String& regexp, 
1.65      paf       222:                const String *options,
1.66      paf       223:                Table **table,
1.99      parser    224:                Row_action row_action, void *info,
                    225:                bool *was_global=0) const;
1.87      parser    226:        enum Change_case_kind {
                    227:                CC_UPPER,
                    228:                CC_LOWER
                    229:        };
1.144.2.1  paf       230:        String& change_case(Charset& source_charset,
1.87      parser    231:                Change_case_kind kind) const;
1.144.2.1  paf       232:        String& replace(Dictionary& dict) const;
1.96      parser    233:        double as_double() const;
                    234:        int as_int() const;
1.137     paf       235: 
1.144.2.1  paf       236:        String& join_chains(char** cstr) const;
1.13      paf       237: 
1.41      paf       238: #ifndef NO_STRING_ORIGIN
                    239:        /// origin of string. calculated by first row
1.144.2.2  paf       240:        const String_fragment::Origin& origin() const;
1.41      paf       241: #endif
1.8       paf       242: 
1.129     paf       243: private:
1.10      paf       244: 
1.125     paf       245:        Untaint_lang lang_of(size_t offset) const;
1.79      paf       246: 
1.116     paf       247:        size_t cstr_bufsize(Untaint_lang lang,
                    248:                SQL_Connection *connection,
1.121     paf       249:                Charset *buf_charset) const;
1.79      paf       250:        /// convert to C string, store to 'dest' which must be big enough for proper untaint
1.82      paf       251:        char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED, 
                    252:                SQL_Connection *connection=0,
1.133     paf       253:                Charset *store_to_charset=0,
                    254:                const char *store_to_charset_name=0) const;
1.144.2.2  paf       255: /*
1.144.2.1  paf       256:        void join_chain(
1.130     paf       257:                                           const Chunk*& achunk, const Chunk::Row*& arow, uint& acountdown, 
1.119     paf       258:                                           uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) const;
1.144.2.2  paf       259: */
1.7       paf       260: private: //disabled
                    261: 
1.12      paf       262:        String& operator = (const String&) { return *this; }
1.7       paf       263: 
1.1       paf       264: };
                    265: 
                    266: #endif

E-mail: