--- parser3/src/include/pa_string.h 2001/01/26 18:34:02 1.2 +++ parser3/src/include/pa_string.h 2001/03/21 14:06:44 1.48 @@ -1,50 +1,154 @@ -/* - $Id: pa_string.h,v 1.2 2001/01/26 18:34:02 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] - link_row ........ - . . - . [ptr, size] - ...........>[link to the next chunk] + Author: Alexander Petrosyan (http://design.ru/paf) + $Id: pa_string.h,v 1.48 2001/03/21 14:06:44 paf Exp $ */ #ifndef PA_STRING_H #define PA_STRING_H +#ifdef HAVE_CONFIG_H +# include "pa_config.h" +#endif + #include -class Pool; +#include "pa_pool.h" +#include "pa_types.h" + +#define UNTAINT_TIMES_BIGGER 10 + +#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) +#else +# define STRING_APPEND_PARAMS \ + 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) +#endif +/// handy: appends const char* piece to String @see String::real_append +#define APPEND_CONST(src) APPEND(src, 0, 0, 0) + +/** + Pooled string. -class 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() }; -private: - friend Pool; + /// 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: - // the pool I'm allocated on - Pool *pool; + String(Pool& apool, const char *src=0, bool tainted=false); + String(const String& src); + size_t size() const { return fsize; } + /// convert to C string + char *cstr() const; + /** 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; } + 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; + } + bool operator != (const String& src) const { return cmp(src)!=0; } + + bool operator == (const char* b_ptr) const; + /** + 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; + +#ifndef NO_STRING_ORIGIN + /// origin of string. calculated by first row + const Origin& origin() const; +#endif + +private: - // last chank allocated count cache - int curr_chunk_rows; struct Chunk { - // the number of rows per chunk + // the number of rows in chunk int count; union Row { - // chunk item - struct { - char *ptr; // pointer to the start of string fragment - size_t size; // length of the 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]; @@ -60,34 +164,28 @@ private: // of the link to the next chunk to allocate Chunk::Row *link_row; - // new&constructors made private to enforce factory manufacturing at pool - void *operator new(size_t size, Pool *apool); +private: + // last chunk + Chunk *last_chunk; - void construct(Pool *apool); - String() { /* never */} - String(Pool *apool) { - construct(apool); - } - String(Pool *apool, char *src) { - construct(apool); - *this+=src; - } + // 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(); - int used_rows(); + void set_lang(Chunk::Row *row, Untaint_lang lang, bool forced, size_t size); -public: +private: //disabled - String(String& src); - size_t size(); - char *c_str(); - String& operator += (char *src); - bool operator == (String& src); + String& operator = (const String&) { return *this; } - unsigned int hash_code(); }; #endif