--- parser3/src/include/pa_string.h 2001/05/15 14:31:58 1.89 +++ parser3/src/include/pa_string.h 2001/08/01 12:08:40 1.100 @@ -5,29 +5,26 @@ Author: Alexander Petrosyan (http://design.ru/paf) - $Id: pa_string.h,v 1.89 2001/05/15 14:31:58 parser Exp $ + $Id: pa_string.h,v 1.100 2001/08/01 12:08:40 parser Exp $ */ #ifndef PA_STRING_H #define PA_STRING_H #include "pa_config_includes.h" - -#include -#include - #include "pa_pool.h" #include "pa_types.h" class Table; /** - $MAIN:html-typo table elements must enlarge string not more that that + $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 20 +#define UNTAINT_TIMES_BIGGER 10 #ifndef NO_STRING_ORIGIN # define STRING_APPEND_PARAMS \ @@ -86,7 +83,7 @@ public: enum { CR_PREALLOCATED_COUNT=2, ///< default preallocated item count - CR_GROW_PERCENT=60 ///< each time the String chunk_is_full() string expanded() + CR_GROW_COUNT=2 ///< each time the String chunk_is_full() string expanded() }; /// piece is tainted or not. the language to use when detaint @@ -117,12 +114,14 @@ public: 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; } + /// makes store_to produce origin:text\n output + void set_origins_mode(bool aorigins_mode) const { forigins_mode=aorigins_mode; } /// convert to C string. if 'lang' known, forcing 'lang' to it char *cstr(Untaint_lang lang=UL_UNSPECIFIED, SQL_Connection *connection=0, const char *charset=0) const { - char *result=(char *)malloc(size()*UNTAINT_TIMES_BIGGER+1); + char *result=(char *)malloc(cstr_bufsize(lang)); char *eol=store_to(result, lang, connection, charset); *eol=0; return result; @@ -181,6 +180,9 @@ public: /// simple hash code of string. used by Hash uint hash_code() const; + /// 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; @@ -188,7 +190,7 @@ public: int pos(const String& substr, size_t 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, + int pos(const char *substr, size_t substr_size=0, size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const; void split(Array& result, @@ -214,13 +216,16 @@ public: const String& regexp, const String *options, Table **table, - Row_action row_action, void *info) const; + Row_action row_action, void *info, + bool *was_global=0) const; enum Change_case_kind { CC_UPPER, CC_LOWER }; String& change_case(Pool& pool, const unsigned char *pcre_tables, Change_case_kind kind) const; + double as_double() const; + int as_int() const; #ifndef NO_STRING_ORIGIN /// origin of string. calculated by first row @@ -267,14 +272,17 @@ private: // used rows in all chunks int fused_rows; + /// origins_mode changes behaviour of store_to[cstr()] + mutable bool forigins_mode; + private: bool chunk_is_full() { return append_here == link_row; } - int expand_times; void expand(); + size_t cstr_bufsize(Untaint_lang lang) 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,