--- parser3/src/include/pa_string.h 2001/03/07 09:29:53 1.26 +++ parser3/src/include/pa_string.h 2001/03/25 08:52:34 1.52 @@ -1,56 +1,128 @@ -/* - $Id: pa_string.h,v 1.26 2001/03/07 09:29:53 paf Exp $ -*/ +/** @file + Parser: string class decl. -/* + Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - String Chunk0 - ====== ======== - head--------------->[ptr, size, ...] - append_here-------->[ptr, size, ...] - . - . - [ptr, size, ...] - link_row----------->[link to the next chunk] + Author: Alexander Petrosyan (http://design.ru/paf) + $Id: pa_string.h,v 1.52 2001/03/25 08:52:34 paf Exp $ */ #ifndef PA_STRING_H #define PA_STRING_H -#ifdef HAVE_CONFIG_H -#include "pa_config.h" -#endif +#include "pa_config_includes.h" #include #include "pa_pool.h" #include "pa_types.h" +/** + $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 + #ifndef NO_STRING_ORIGIN -# define STRING_APPEND_PARAMS const char *src, size_t size, char *file, uint line -# define APPEND(src, size, file, line) real_append(src, size, file, line) +# define STRING_APPEND_PARAMS \ + const char *src, size_t size, \ + String::Untaint_lang lang, \ + const char *file, uint line +/// appends clean piece to String @see String::real_append +# define APPEND(src, size, file, line) \ + real_append(src, size, String::UL_NO, file, line) +/// appends tainted piece to String @see String::real_append +# define APPEND_TAINTED(src, size, file, line) \ + real_append(src, size, String::UL_YES, file, line) +# define APPEND_SPECIFIC_TAINTED(src, size, lang, file, line) \ + real_append(src, size, lang, file, line) #else -# define STRING_APPEND_PARAMS const char *src, size_t size -# define APPEND(src, size, file, line) real_append(src, size) +# define STRING_APPEND_PARAMS \ + const char *src, \ + size_t size, \ + String::Untaint_lang lang +/// appends clean piece to String @see String::real_append +# define APPEND(src, size, file, line) \ + real_append(src, size, String::UL_NO) +/// appends tainted piece to String @see String::real_append +# define APPEND_TAINTED(src, size, file, line) \ + real_append(src, size, String::UL_YES) +# define APPEND_SPECIFIC_TAINTED(src, size, lang, file, line) \ + real_append(src, size, lang) #endif +/// handy: appends const char* piece to String @see String::real_append #define APPEND_CONST(src) APPEND(src, 0, 0, 0) +/** + Pooled string. + + Internal structure: + @verbatim + String Chunk0 + ====== ======== + head--------------->[ptr, size, ...] + append_here-------->[ptr, size, ...] + . + . + [ptr, size, ...] + link_row----------->[link to the next chunk] + @endverbatim + + 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, - CR_GROW_PERCENT=60 + CR_PREALLOCATED_COUNT=5, ///< default preallocated item count + CR_GROW_PERCENT=60 ///< each time the Array chunk_is_full() array expanded() + }; + + /// piece is tainted or not. the language to use when detaint + 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_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_URI, ///< text in uri + UL_TABLE, ///< ^table:set body + UL_SQL, ///< ^table:sql body + UL_JS, ///< JavaScript code + UL_HTML, ///< HTML code (for editing) + UL_HTML_TYPO ///< HTML code with TYPOgraphic replacements (for showing) }; public: - String(Pool& apool); + String(Pool& apool, const char *src=0, bool tainted=false); String(const String& src); size_t size() const { return fsize; } - int used_rows() const { return fused_rows; } - char *cstr() const; + /// convert to C string + char *cstr() const { + char *result=(char *)malloc(size()*UNTAINT_TIMES_BIGGER+1); + char *eol=store_to(result); + *eol=0; + return result; + } + /** append fragment + @see APPEND, 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; } @@ -63,12 +135,38 @@ public: } bool operator != (const String& src) const { return cmp(src)!=0; } - bool operator == (char* src) const; - String& operator += (const String& src); + /** + @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(const char* src_ptr, int& partial, size_t src_size=0) const; + bool operator == (const char* src_ptr) const { + size_t src_size=src_ptr?strlen(src_ptr):0; + if(size() != src_size) + return false; + int partial; // unused + return cmp(src_ptr, partial, src_size)==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); + /// simple hash code of string. used by Hash uint hash_code() const; - const Origin& origin() const { return head.rows[0].item.origin; } +#ifndef NO_STRING_ORIGIN + /// origin of string. calculated by first row + const Origin& origin() const; +#endif private: @@ -76,11 +174,14 @@ private: // the number of rows in chunk int count; union Row { - // chunk item - struct { - const char *ptr; // pointer to the start of string fragment - size_t size; // length of the fragment - Origin origin; // origin of this fragment + // 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]; @@ -112,6 +213,8 @@ private: return append_here == link_row; } void expand(); + void set_lang(Chunk::Row *row, Untaint_lang lang, bool forced, size_t size); + char *String::store_to(char *dest) const; private: //disabled