--- parser3/src/include/pa_string.h	2001/03/30 09:58:58	1.59
+++ parser3/src/include/pa_string.h	2001/04/09 14:31:40	1.78
@@ -5,7 +5,7 @@
 
 	Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
 
-	$Id: pa_string.h,v 1.59 2001/03/30 09:58:58 paf Exp $
+	$Id: pa_string.h,v 1.78 2001/04/09 14:31:40 paf Exp $
 */
 
 #ifndef PA_STRING_H
@@ -13,11 +13,14 @@
 
 #include "pa_config_includes.h"
 
+#include <string.h>
 #include <stddef.h>
 
 #include "pa_pool.h"
 #include "pa_types.h"
 
+class Table;
+
 /**
 	$MAIN:html-typo table elements must enlarge string not more that that
 	that's a tradeoff - otherwise we'd have to scan string twice:
@@ -45,13 +48,16 @@
 #endif
 /// appends clean piece to String  @see String::real_append
 #define APPEND_CLEAN(src, size, file, line) \
-	APPEND(src, size, String::UL_NO, file, line)
+	APPEND(src, size, String::UL_CLEAN, file, line)
 /// appends tainted piece to String  @see String::real_append
 #define APPEND_TAINTED(src, size, file, line) \
-	APPEND(src, size, String::UL_YES, 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)
 
+class Array;
+class SQL_Connection;
+
 /** 
 	Pooled string.
 
@@ -82,9 +88,9 @@ public:
 
 	/// 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 
+		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]{...}
 		UL_PASS_APPENDED,
 			/**<
@@ -93,7 +99,8 @@ 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
@@ -104,13 +111,16 @@ public:
 
 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, 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;
+	/// convert to C string. if 'lang' known, forcing 'lang' to it
+	char *cstr(Untaint_lang lang=UL_UNSPECIFIED, SQL_Connection *connection=0) const {
 		char *result=(char *)malloc(size()*UNTAINT_TIMES_BIGGER+1);
-		char *eol=store_to(result);
+		char *eol=store_to(result, lang, connection);
 		*eol=0;
 		return result;
 	}
@@ -119,7 +129,8 @@ public:
 	*/
 	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) const;
+	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; }
@@ -139,7 +150,8 @@ public:
 			-  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) const;
+	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)
@@ -155,17 +167,38 @@ 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) const;
+	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 String::pos(const char *substr, size_t result) const;
+	int pos(const char *substr, size_t substr_size, 
+		size_t 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, int limit=-1) const;
+	void split(Array& result, 
+		size_t *pos_after_ref, 
+		const String& delim, 
+		Untaint_lang lang, int limit=-1) const;
+
+	typedef void (*Row_action)(Table& table, Array *row, int start, int finish, 
+		void *info);
+	bool match(const String *aorigin,
+		const String& regexp, 
+		const String *options,
+		Table **table,
+		Row_action row_action, void *info) const;
 
 #ifndef NO_STRING_ORIGIN
 	/// origin of string. calculated by first row
@@ -217,8 +250,6 @@ 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