--- parser3/src/include/pa_string.h 2001/03/07 09:29:53 1.26 +++ parser3/src/include/pa_string.h 2002/04/19 08:28:35 1.136 @@ -1,122 +1,322 @@ -/* - $Id: pa_string.h,v 1.26 2001/03/07 09:29:53 paf Exp $ -*/ - -/* +/** @file + Parser: string class decl. - String Chunk0 - ====== ======== - head--------------->[ptr, size, ...] - append_here-------->[ptr, size, ...] - . - . - [ptr, size, ...] - link_row----------->[link to the next chunk] + Copyright (c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) + $Id: pa_string.h,v 1.136 2002/04/19 08:28:35 paf Exp $ */ #ifndef PA_STRING_H #define PA_STRING_H -#ifdef HAVE_CONFIG_H -#include "pa_config.h" -#endif - -#include - #include "pa_pool.h" #include "pa_types.h" #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, \ + uchar 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, size_t size -# define APPEND(src, size, file, line) real_append(src, size) +# define STRING_APPEND_PARAMS \ + const char *src, \ + size_t size, \ + uchar lang +/// appends piece to String @see String::real_append +# define APPEND(src, size, lang, file, line) \ + real_append(src, size, lang) #endif -#define APPEND_CONST(src) APPEND(src, 0, 0, 0) - +/// 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_AS_IS(src, 0, 0, 0) + +class Table; +class Array; +class SQL_Connection; +class Dictionary; + +/** + 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 +*/ +#include "pa_pragma_pack_begin.h" class String : public Pooled { public: + enum { - CR_PREALLOCATED_COUNT=5, - CR_GROW_PERCENT=60 + CR_PREALLOCATED_COUNT=2, ///< default preallocated item count + CR_GROW_COUNT=1 ///< each time the String chunk_is_full() string expanded() + }; + + /** 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_AS_IS, ///< leave all characters intact + + UL_PASS_APPENDED, + /**< + leave language built into string being appended. + just a flag, that value not stored + */ + 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_OPTIMIZE_BIT = 0x80 ///< flag, requiring cstr whitespace optimization }; public: - String(Pool& apool); + String(Pool& apool, const char *src=0, size_t src_size=0, bool tainted=false); String(const String& src); - size_t size() const { return fsize; } - int used_rows() const { return fused_rows; } - char *cstr() const; + bool is_empty() const { return append_here==head.chunk.rows; } + size_t size() const; + /// convert to C string. if 'lang' known, forcing 'lang' to it + 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_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST + */ String& real_append(STRING_APPEND_PARAMS); - 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; } + /// @return <0 ==0 or >0 depending on comparison result + int cmp (int& partial, const String& src, + size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const; + bool operator < (const String& src) const { int p; return cmp(p, src)<0; } + bool operator > (const String& src) const { int p; return cmp(p, src)>0; } + bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; } + bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; } bool operator == (const String& src) const { if(size()!=src.size()) // can speed up in trivial case return false; - return cmp(src)==0; + int p; return cmp(p, src)==0; + } + bool operator != (const String& src) const { int p; return cmp(p, 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(int& partial, const char* src_ptr, size_t src_size=0, + size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) 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(partial, src_ptr, src_size)==0; + } + bool operator != (const char* src_ptr) const { + int partial; // unused + return cmp(partial, src_ptr, 0)!=0; } - bool operator != (const String& src) const { return cmp(src)!=0; } - bool operator == (char* src) const; - String& operator += (const String& src); + 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; - const Origin& origin() const { return head.rows[0].item.origin; } + /// extracts first char of a string + char first_char() const; + + /// 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, + 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=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=UL_UNSPECIFIED, int limit=-1) const; + void split(Array& result, + size_t *pos_after_ref, + const String& delim, + Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const; + + typedef void (*Row_action)(Table& table, Array *row, + int prestart, int prefinish, + int poststart, int postfinish, + void *info); + /** + @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( + const String *aorigin, + const String& regexp, + const String *options, + Table **table, + Row_action row_action, void *info, + bool *was_global=0) const; + enum Change_case_kind { + CC_UPPER, + CC_LOWER + }; + String& change_case(Pool& pool, + Change_case_kind kind) const; + String& replace(Pool& pool, Dictionary& dict) const; + double as_double() const; + int as_int() const; + +#ifndef NO_STRING_ORIGIN + /// origin of string. calculated by first row + const Origin& origin() const; +#endif private: + /** several String fragments + */ struct Chunk { - // 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 + typedef uchar count_type; + count_type count; ///< the number of rows in chunk + // here could be some padding bytes + /// string fragment or a link to next chunk union + typedef union Row { + typedef uchar item_size_type; + /// fragment + struct { + const char *ptr; ///< pointer to the start + item_size_type size; ///< length + uchar/*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 + // we are using the fact that there's no padding before this field! + Chunk *link; ///< link to the next chunk in chain + } rows_type[CR_PREALLOCATED_COUNT]; + rows_type rows; + }; + /** + 'mutable' because can write after it's end, after it was appended to somebody + @see String::append + */ + mutable struct { + Chunk chunk; + Chunk *link_storage; + } head; ///< the head chunk of the chunk chain - // next append would write to this record + /// 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; + /// last chunk + mutable Chunk *last_chunk; private: bool chunk_is_full() { - return append_here == link_row; + return append_here == last_chunk->rows+last_chunk->count; } + uint used_rows() const; void expand(); + + 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(Pool& pool, + 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; } }; +#include "pa_pragma_pack_end.h" + +#define STRING_PREPARED_FOREACH_ROW(self, body) \ + while(row!=(self).append_here) { \ + if(countdown==0) { \ + chunk=row->link; \ + row=chunk->rows; \ + countdown=chunk->count; \ + }; \ + { body } \ + row++; countdown--; \ + } + +#define STRING_PREFIX_FOREACH_ROW(self, body) { \ + const Chunk *chunk=&(self).head.chunk; \ + const Chunk::Row *row=chunk->rows; \ + uint countdown=chunk->count; \ + STRING_PREPARED_FOREACH_ROW(self, body) \ +} + +#define STRING_FOREACH_ROW(body) STRING_PREFIX_FOREACH_ROW(*this, body) +#define STRING_SRC_FOREACH_ROW(body) STRING_PREFIX_FOREACH_ROW(src, body) #endif