--- parser3/src/include/pa_string.h 2001/03/21 14:06:44 1.48 +++ parser3/src/include/pa_string.h 2003/03/20 11:11:55 1.144.2.28.2.8 @@ -1,186 +1,272 @@ /** @file Parser: string class decl. - Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - - Author: Alexander Petrosyan (http://design.ru/paf) - - $Id: pa_string.h,v 1.48 2001/03/21 14:06:44 paf Exp $ + Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) */ #ifndef PA_STRING_H #define PA_STRING_H -#ifdef HAVE_CONFIG_H -# include "pa_config.h" -#endif +static const char* IDENT_STRING_H="$Date: 2003/03/20 11:11:55 $"; -#include +// includes -#include "pa_pool.h" #include "pa_types.h" +#include "pa_array.h" + +extern "C" { // cord's author forgot to do that +#include "cord.h" +}; + +// forwards -#define UNTAINT_TIMES_BIGGER 10 +class Charset; +class Table; +class SQL_Connection; +class Dictionary; +class Request_charsets; +class String; +typedef Array ArrayString; + +// helpers #ifndef NO_STRING_ORIGIN # define STRING_APPEND_PARAMS \ - const char *src, size_t size, \ - bool tainted, \ - const char *file, uint line -/// appends clean piece to String @see String::real_append -# define APPEND(src, size, file, line) real_append(src, size, false, file, line) -/// appends tainted piece to String @see String::real_append -# define APPEND_TAINTED(src, size, file, line) real_append(src, size, true, file, line) + const char* src, size_t size, \ + Untaint_lang lang, \ + const char* file, uint line +/// appends piece to String @see String::real_append +# define APPEND(src, size, lang, file, line) \ + real_append(src, size, lang, file, line) #else # define STRING_APPEND_PARAMS \ - const char *src, \ + const char* src, \ size_t size, \ - bool tainted -/// appends clean piece to String @see String::real_append -# define APPEND(src, size, file, line) real_append(src, size, false) -/// appends tainted piece to String @see String::real_append -# define APPEND_TAINTED(src, size, file, line) real_append(src, size, true) + Untaint_lang lang +/// appends piece to String @see String::real_append +# define APPEND(src, size, lang, file, line) \ + real_append(src, size, lang) #endif +/// appends clean piece to String @see String::real_append +#define APPEND_CLEAN(src, size, file, line) \ + APPEND(src, size, String::UL_CLEAN, file, line) +/// appends piece to String as-is @see String::real_append +#define APPEND_AS_IS(src, size, file, line) \ + APPEND(src, size, String::UL_AS_IS, file, line) +/// appends tainted piece to String @see String::real_append +#define APPEND_TAINTED(src, size, file, line) \ + APPEND(src, size, String::UL_TAINTED, file, line) /// handy: appends const char* piece to String @see String::real_append -#define APPEND_CONST(src) APPEND(src, 0, 0, 0) +#define APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0) -/** - Pooled string. +/// this is result of pos functions which mean that substr were not found +#define STRING_NOT_FOUND ((size_t)-1) - Internal structure: - @verbatim - String Chunk0 - ====== ======== - head--------------->[ptr, size, ...] - append_here-------->[ptr, size, ...] - . - . - [ptr, size, ...] - link_row----------->[link to the next chunk] - @endverbatim + +/** + String which knows the language of all it's fragments. All pieces remember - - the file and its line they are from [can be turned off by NO_STRING_ORIGIN] - whether they are tainted or not, and the language which should be used to detaint them */ -class String : public Pooled { -public: +class String: public PA_Object { - enum { - CR_PREALLOCATED_COUNT=5, ///< default preallocated item count - CR_GROW_PERCENT=60 ///< each time the Array chunk_is_full() array expanded() - }; +public: - /// piece is tainted or not. the language to use when detaint + /** piece is tainted or not. the language to use when detaint + remember to change String_Untaint_lang_name @ untaint.C along + */ enum Untaint_lang { - UL_UNKNOWN=0, ///< when get by name fails - UL_NO, ///< clean - UL_YES, ///< tainted, untaint language as assigned later - // untaint languages. assigned by ^untaint[lang]{...} + UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum + // these two must go before others, there are checks for >UL_AS_IS + UL_CLEAN, ///< clean + UL_AS_IS, ///< leave all characters intact + UL_PASS_APPENDED, /**< leave language built into string being appended. just a flag, that value not stored */ - UL_AS_IS, ///< leave all characters intact - UL_FILE_NAME, ///< filename - UL_HEADER, ///< text in response header + UL_TAINTED, ///< tainted, untaint language as assigned later + // untaint languages. assigned by ^untaint[lang]{...} + UL_FILE_SPEC, ///< file specification + UL_HTTP_HEADER, ///< text in HTTP response header + UL_MAIL_HEADER, ///< text in mail header UL_URI, ///< text in uri UL_TABLE, ///< ^table:set body UL_SQL, ///< ^table:sql body UL_JS, ///< JavaScript code + UL_XML, ///< ^dom:set xml UL_HTML, ///< HTML code (for editing) - UL_HTML_TYPO ///< HTML code with TYPOgraphic replacements (for showing) + UL_OPTIMIZE_BIT = 0x8000 ///< flag, requiring cstr whitespace optimization }; + struct Fragment { + Untaint_lang lang; ///< untaint flag, later untaint language + size_t size; ///< size + Fragment(Untaint_lang alang, size_t asize): lang(alang), size(asize) {} + }; + + class ArrayFragment: public Array { + void simple_append(element_type src) { + *static_cast *>(this)+=src; + } + public: + ArrayFragment& operator += (element_type src) { + if(count()) { // not empty? + // try to join with last + Fragment& last=get(count()-1); + if(last.lang==src.lang) { + last.size+=src.size; + return *this; + } + } + simple_append(src); + return *this; + } + ArrayFragment& append(const ArrayFragment& src, size_t substr_start, size_t substr_end); + }; + + struct C { + char *str; + size_t size; + operator char *() { return str; } + C(char *astr, size_t asize): str(astr), size(asize) {} + }; + +private: + + CORD body; ///< all characters of string + ArrayFragment fragments; ///< fragment language+size info + public: - String(Pool& apool, const char *src=0, bool tainted=false); + explicit String(const char* src=0, size_t src_size=0, bool tainted=false); String(const String& src); - size_t size() const { return fsize; } - /// convert to C string - char *cstr() const; + operator bool() const { return body!=CORD_EMPTY; } + size_t size() const { return CORD_len(body); } + + /// convert to C string. if 'lang' known, forcing 'lang' to it + CORD cstr_to_cord(Untaint_lang lang=UL_AS_IS, + SQL_Connection *connection=0, + const Request_charsets *charsets=0) const; + + /// convert to C string. if 'lang' known, forcing 'lang' to it + const char* cstr(Untaint_lang lang=UL_AS_IS, + SQL_Connection *connection=0, + const Request_charsets *charsets=0) const { + return CORD_to_const_char_star(cstr_to_cord(lang, connection, charsets)); + } + /// convert to C string. if 'lang' known, forcing 'lang' to it + char *cstrm(Untaint_lang lang=UL_AS_IS, + SQL_Connection *connection=0, + const Request_charsets *charsets=0) const { + return CORD_to_char_star(cstr_to_cord(lang, connection, charsets)); + } + /// puts pieces to buf + C serialize(size_t prolog_size) const; + /// appends pieces from buf to self + bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char* file); /** append fragment - @see APPEND, APPEND_TAINTED, APPEND_CONST + @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST */ String& real_append(STRING_APPEND_PARAMS); - /// @return <0 ==0 or >0 depending on comparison result - int cmp (const String& src) const; - bool operator < (const String& src) const { return cmp(src)<0; } - bool operator > (const String& src) const { return cmp(src)>0; } - bool operator <= (const String& src) const { return cmp(src)<=0; } - bool operator >= (const String& src) const { return cmp(src)>=0; } - bool operator == (const String& src) const { - if(size()!=src.size()) // can speed up in trivial case - return false; - return cmp(src)==0; + /** + @param partial + returns partial match status. + - -1: strings too different + - 0: full match + - 1: means @c this starts @c src + - 2: means @src starts @this + */ + //int cmp(CORD substr, size_t this_offset=0) const; + /// String version of @see cmp(CORD, size_t) + //int cmp(const String& src, size_t this_offset=0) const; + + bool operator < (const String& src) const { return CORD_cmp(body, src.body)<0; } + bool operator > (const String& src) const { return CORD_cmp(body, src.body)>0; } + bool operator <= (const String& src) const { return CORD_cmp(body, src.body)<=0; } + bool operator >= (const String& src) const { return CORD_cmp(body, src.body)>=0; } + bool operator == (const String& src) const { return CORD_cmp(body, src.body)==0; } + bool operator != (const String& src) const { return CORD_cmp(body, src.body)!=0; } + + /// this starts with y + bool starts_with(CORD y) const { + return CORD_ncmp(body, 0/*x_start*/, y, 0/*y_start*/, CORD_len(y))==0; } - bool operator != (const String& src) const { return cmp(src)!=0; } - - bool operator == (const char* b_ptr) const; - /** - appends other String. + /// y starts with this + bool this_starts(CORD y) const { + return CORD_ncmp(body, 0/*x_start*/, y, 0/*y_start*/, CORD_len(y))==0; + } + bool operator == (CORD y) const { return CORD_cmp(body, y)==0; } + bool operator != (CORD y) const { return CORD_cmp(body, y)!=0; } - marking all tainted pieces of it with @a lang. - or marking ALL pieces of it with a @a lang when @a forced to. - */ - String& append(const String& src, Untaint_lang lang, bool forced=false); + String& append_to(String& dest, Untaint_lang lang, bool forced) const; + String& append(const String& src, Untaint_lang lang, bool forced=false) { + return src.append_to(*this, lang, forced); + } + String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); } + //String& operator << (const String* src) { return append(*src, UL_PASS_APPENDED); } + String& operator << (const char* src) { return APPEND_CONST(src); } - /// simple hash code of string. used by Hash + /// simple hash code of string uint hash_code() const; -#ifndef NO_STRING_ORIGIN - /// origin of string. calculated by first row - const Origin& origin() const; -#endif - -private: - - struct Chunk { - // the number of rows in chunk - int count; - union Row { - // fragment - struct { - const char *ptr; // pointer to the start - size_t size; // length - Untaint_lang lang; // untaint flag, later untaint language -#ifndef NO_STRING_ORIGIN - Origin origin; // origin -#endif - } item; - Chunk *link; // link to the next chunk in chain - } rows[CR_PREALLOCATED_COUNT]; - // next rows are here - Chunk *preallocated_link; - } - head; // the head chunk of the chunk chain - - // next append would write to this record - Chunk::Row *append_here; - - // the address of place where lies address - // of the link to the next chunk to allocate - Chunk::Row *link_row; - -private: - // last chunk - Chunk *last_chunk; - - // string size - size_t fsize; - - // used rows in all chunks - int fused_rows; + /// extracts first char of a string, if any + char first_char() const { + return operator bool()?CORD_fetch(body, 0):0; + } -private: + /// extracts [start, finish) piece of string + const String& mid(size_t substr_start, size_t substr_end) const; - bool chunk_is_full() { - return append_here == link_row; - } - void expand(); - void set_lang(Chunk::Row *row, Untaint_lang lang, bool forced, size_t size); + /** + ignore lang if it's UL_UNSPECIFIED + but when specified: look for substring that lies in ONE fragment in THAT lang + @return position of substr in string, -1 means "not found" [const char* version] + */ + size_t pos(CORD substr, + size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const; + /// String version of @see pos(const char*, int, Untaint_lang) + size_t pos(const String& substr, + size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const; + + void split(ArrayString& result, + size_t& pos_after_ref, + const char* delim, + Untaint_lang lang=UL_UNSPECIFIED, int limit=-1); + void split(ArrayString& result, + size_t& pos_after_ref, + const String& delim, + Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const; + + typedef void (*Row_action)(Table* table, ArrayString* row, + int prestart, int prefinish, + int poststart, int postfinish, + void *info); + /** + @return table of found items, if any. + table format is defined and fixed[can be used by others]: + @verbatim + prematch/match/postmatch/1/2/3/... + @endverbatim + */ + Table* match(Charset& source_charset, + const String& regexp, + const String& options, + Row_action row_action, void *info, + bool& just_matched) const; + enum Change_case_kind { + CC_UPPER, + CC_LOWER + }; + String& change_case(Charset& source_charset, + Change_case_kind kind); + const String& replace(const Dictionary& dict) const; + double as_double() const; + int as_int() const; private: //disabled @@ -188,4 +274,9 @@ private: //disabled }; +/// simple hash code of string. used by Hash +inline uint hash_code(const String& self) { + return self.hash_code(); +} + #endif