--- parser3/src/include/pa_string.h 2001/04/08 13:11:17 1.76 +++ parser3/src/include/pa_string.h 2003/01/23 15:38:05 1.144.2.2 @@ -1,38 +1,23 @@ /** @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 1.76 2001/04/08 13:11:17 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 -#include "pa_config_includes.h" - -#include -#include +static const char* IDENT_STRING_H="$Date: 2003/01/23 15:38:05 $"; #include "pa_pool.h" #include "pa_types.h" - -class Table; - -/** - $MAIN:html-typo table elements must enlarge string not more that that - that's a tradeoff - otherwise we'd have to scan string twice: - - first for buffer length - - second for replacements themselves -*/ -#define UNTAINT_TIMES_BIGGER 10 +#include "pa_array.h" #ifndef NO_STRING_ORIGIN # define STRING_APPEND_PARAMS \ const char *src, size_t size, \ - String::Untaint_lang lang, \ + uchar lang, \ const char *file, uint line /// appends piece to String @see String::real_append # define APPEND(src, size, lang, file, line) \ @@ -41,88 +26,106 @@ class Table; # define STRING_APPEND_PARAMS \ const char *src, \ size_t size, \ - String::Untaint_lang lang + uchar 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 +/// 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_CLEAN(src, 0, 0, 0) +#define APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0) -class Array; +class Table; class SQL_Connection; +class Dictionary; -/** - Pooled string. +struct String_fragment { + const char *ptr; ///< pointer to the start + size_t size; ///< length + int/*Untaint_lang*/ lang; ///< untaint flag, later untaint language + +#ifndef NO_STRING_ORIGIN + /// all String pieces hold information of where they come from + struct Origin { + const char *file; ///< macros file name | load file name | sql query text + uint line; ///< file line no | record no + } origin; ///< origin +#endif +}; - 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: - enum { - CR_PREALLOCATED_COUNT=5, ///< default preallocated item count - CR_GROW_PERCENT=60 ///< each time the Array chunk_is_full() array expanded() - }; +class String: public Array { + size_t fsize; +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_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_TAINTED, ///< tainted, untaint language as assigned later - // untaint languages. assigned by ^untaint[lang]{...} + 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_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 = 0x80 ///< flag, requiring cstr whitespace optimization }; public: - String(Pool& apool, const char *src=0, bool tainted=false); + //static String& OnPool(Pool& apool, const char *local_src=0, size_t src_size=0, bool tainted=false); + String(const char *src=0, size_t src_size=0, bool tainted=false); String(const String& src); + bool is_empty() const { return !size(); } size_t size() const { return fsize; } /// convert to C string. if 'lang' known, forcing 'lang' to it - char *cstr(Untaint_lang lang=UL_UNSPECIFIED, SQL_Connection *connection=0) const { - char *result=(char *)malloc(size()*UNTAINT_TIMES_BIGGER+1); - char *eol=store_to(result, lang, connection); + char *cstr(Untaint_lang lang=UL_AS_IS, + SQL_Connection *connection=0, + Charset *cstr_charset=0, const char *cstr_charset_name=0) const { + + char *result=(char *)malloc(cstr_bufsize(lang, connection, cstr_charset)); + char *eol=store_to(result, lang, connection, cstr_charset, cstr_charset_name); *eol=0; return result; } + char *cstr_debug_origins() const; + /// puts pieces to buf + void serialize(size_t prolog_size, void *& buf, size_t& buf_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_CLEAN, 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 @@ -149,6 +152,11 @@ public: */ int cmp(int& partial, const char* src_ptr, size_t src_size=0, size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const; + /// this starts with src + bool starts_with(const char* src_ptr, size_t src_size=0) const { + int p; cmp(p, src_ptr, src_size); + return p==0 || p==2; + } bool operator == (const char* src_ptr) const { size_t src_size=src_ptr?strlen(src_ptr):0; if(size() != src_size) @@ -156,100 +164,94 @@ public: int partial; // unused return cmp(partial, src_ptr, src_size)==0; } + bool operator != (const char* src_ptr) const { + int partial; // unused + return cmp(partial, src_ptr, 0)!=0; + } - /** - appends other String. - - 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(const String& src, uchar lang, bool forced=false); 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 uint hash_code() const; + /// extracts first char of a string, if any + char first_char() const { + return is_empty()?0:*felements[0].ptr; + } + /// extracts [start, finish) piece of string String& mid(size_t start, size_t finish) const; /// @return position of substr in string, -1 means "not found" [String version] int pos(const String& substr, - size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const; + int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const; /// @return position of substr in string, -1 means "not found" [const char* version] - int pos(const char *substr, size_t substr_size, - size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const; + int pos(const char *substr, size_t substr_size=0, + int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const; void split(Array& result, size_t *pos_after_ref, const char *delim, size_t delim_size, - Untaint_lang lang, int limit=-1) const; + Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const; void split(Array& result, size_t *pos_after_ref, const String& delim, - Untaint_lang lang, int limit=-1) const; + Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const; - typedef void (*Row_action)(Table& table, Array *row, int start, int finish, + typedef void (*Row_action)(Table& table, Array *row, + int prestart, int prefinish, + int poststart, int postfinish, void *info); - bool match(const String *aorigin, + /** + @return true if fills table. + table format is defined and fixed[can be used by others]: + @verbatim + prematch/match/postmatch/1/2/3/... + @endverbatim + */ + bool match(Charset& source_charset, + const String *aorigin, const String& regexp, const String *options, Table **table, - Row_action row_action, void *info) const; - -#ifndef NO_STRING_ORIGIN - /// origin of string. calculated by first row - const Origin& origin() const; -#endif + Row_action row_action, void *info, + bool *was_global=0) const; + enum Change_case_kind { + CC_UPPER, + CC_LOWER + }; + String& change_case(Charset& source_charset, + Change_case_kind kind) const; + String& replace(Dictionary& dict) const; + double as_double() const; + int as_int() const; -private: + String& join_chains(char** cstr) const; - struct Chunk { - // the number of rows in chunk - size_t 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 + /// origin of string. calculated by first row + const String_fragment::Origin& origin() const; #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; private: - bool chunk_is_full() { - return append_here == link_row; - } - void expand(); - char *String::store_to(char *dest, - Untaint_lang lang=UL_UNSPECIFIED, SQL_Connection *connection=0) const; + Untaint_lang lang_of(size_t offset) const; + size_t cstr_bufsize(Untaint_lang lang, + SQL_Connection *connection, + Charset *buf_charset) const; + /// convert to C string, store to 'dest' which must be big enough for proper untaint + char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED, + SQL_Connection *connection=0, + Charset *store_to_charset=0, + const char *store_to_charset_name=0) const; +/* + void join_chain( + const Chunk*& achunk, const Chunk::Row*& arow, uint& acountdown, + uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) const; +*/ private: //disabled String& operator = (const String&) { return *this; }