--- parser3/src/include/pa_string.h	2001/04/03 09:58:09	1.64
+++ parser3/src/include/pa_string.h	2001/04/28 13:38:12	1.85
@@ -5,7 +5,7 @@
 
 	Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
 
-	$Id: pa_string.h,v 1.64 2001/04/03 09:58:09 paf Exp $
+	$Id: pa_string.h,v 1.85 2001/04/28 13:38:12 paf Exp $
 */
 
 #ifndef PA_STRING_H
@@ -13,6 +13,7 @@
 
 #include "pa_config_includes.h"
 
+#include <string.h>
 #include <stddef.h>
 
 #include "pa_pool.h"
@@ -45,16 +46,20 @@ class Table;
 #	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 SQL_Connection;
 
 /** 
 	Pooled string.
@@ -86,7 +91,7 @@ public:
 
 	/// piece is tainted or not. the language to use when detaint 
 	enum Untaint_lang {
-		UL_UNKNOWN=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
+		UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
 		UL_CLEAN, ///< clean
 		UL_TAINTED,  ///< tainted, untaint language as assigned later 
 		// untaint languages. assigned by ^untaint[lang]{...}
@@ -97,34 +102,37 @@ public:
 			*/
 		UL_AS_IS,     ///< leave all characters intact
 		UL_FILE_NAME, ///< filename
-		UL_HEADER,    ///< text in response header
+		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_HTML,      ///< HTML code (for editing)
-		UL_HTML_TYPO  ///< HTML code with TYPOgraphic replacements (for showing)
+		UL_USER_HTML  ///< HTML code with USER chars
 	};
 
 public:
 
-	String(Pool& apool, const char *src=0, bool tainted=false);
+	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; }
-	/// convert to C string
-	char *cstr() const {
+	/// 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 *eol=store_to(result);
+		char *eol=store_to(result, lang, connection, charset);
 		*eol=0;
 		return result;
 	}
 	/** 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
 	int cmp (int& partial, const String& src, 
-		size_t this_offset=0, Untaint_lang lang=UL_UNKNOWN) const;
+		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; }
@@ -145,7 +153,7 @@ public:
 			-  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_UNKNOWN) const;
+		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)
@@ -153,6 +161,11 @@ 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.
@@ -161,19 +174,21 @@ public:
 		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& 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 [start, finish) piece of string
-	String& piece(size_t start, size_t finish) const;
+	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_UNKNOWN) const;
+		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, 
-		size_t this_offset=0, Untaint_lang lang=UL_UNKNOWN) const;
+		size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
 
 	void split(Array& result, 
 		size_t *pos_after_ref, 
@@ -184,10 +199,14 @@ public:
 		const String& delim, 
 		Untaint_lang lang, int limit=-1) const;
 
-	bool match(const String *aorigin,
+	typedef void (*Row_action)(Table& table, Array *row, int start, int finish, 
+		void *info);
+	bool match(const unsigned char *pcre_tables,
+		const String *aorigin,		
 		const String& regexp, 
-		const String& options,
-		Table **table) const;
+		const String *options,
+		Table **table,
+		Row_action row_action, void *info) const;
 
 #ifndef NO_STRING_ORIGIN
 	/// origin of string. calculated by first row
@@ -196,31 +215,32 @@ public:
 
 private:
 
+	/// several String fragments
 	struct Chunk {
-		// the number of rows in chunk
-		size_t count;
+		size_t count; ///< the number of rows in chunk
+		/// string fragment or a link to next chunk union
 		union Row {
-			// fragment
+			/// fragment
 			struct { 
-				const char *ptr;  // pointer to the start
-				size_t size;  // length
-				Untaint_lang lang; // untaint flag, later untaint language
+				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 origin;  ///< origin
 #endif
 			} item;
-			Chunk *link;  // link to the next chunk in chain
+			Chunk *link;  ///< link to the next chunk in chain
 		} rows[CR_PREALLOCATED_COUNT];
-		// next rows are here
-		Chunk *preallocated_link;
+		Chunk *preallocated_link; ///< next rows are here
 	}
-		head;  // the head chunk of the chunk chain
+		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
+	/** the address of place where lies address 
+		of the link to the next chunk to allocate
+	*/
 	Chunk::Row *link_row;
 
 private:
@@ -239,7 +259,11 @@ private:
 		return append_here == link_row;
 	}
 	void expand();
-	char *String::store_to(char *dest) 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,
+		const char *charset=0) const;
 
 private: //disabled