--- parser3/src/main/pa_string.C	2001/04/03 15:35:24	1.66
+++ parser3/src/main/pa_string.C	2001/04/05 19:35:16	1.73
@@ -5,12 +5,12 @@
 
 	Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
 
-	$Id: pa_string.C,v 1.66 2001/04/03 15:35:24 paf Exp $
+	$Id: pa_string.C,v 1.73 2001/04/05 19:35:16 paf Exp $
 */
 
 #include "pa_config_includes.h"
 
-#include <locale.h>
+#include "pcre.h"
 
 #include "pa_pool.h"
 #include "pa_string.h"
@@ -22,8 +22,6 @@
 #include "pa_table.h"
 #include "pa_threads.h"
 
-#include "pcre.h"
-
 //#include "pa_sapi.h"
 
 // String
@@ -199,7 +197,7 @@ int String::cmp(int& partial, const Stri
 			break;
 
 		if(pos+a_row->item.size > this_offset) {
-			if(lang!=UL_UNKNOWN && a_row->item.lang!=lang) 
+			if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang) 
 				return -1; // wrong lang -- bail out
 
 			int size_diff=
@@ -279,7 +277,7 @@ int String::cmp(int& partial, const char
 			break;
 
 		if(pos+a_row->item.size > this_offset) {
-			if(lang!=UL_UNKNOWN && a_row->item.lang!=lang) 
+			if(lang!=UL_UNSPECIFIED && a_row->item.lang!=lang) 
 				return -1; // wrong lang -- bail out
 
 			int size_diff=
@@ -342,7 +340,7 @@ const Origin& String::origin() const {
 }
 #endif
 
-String& String::piece(size_t start, size_t finish) const {
+String& String::mid(size_t start, size_t finish) const {
 	start=max(0, start);
 	finish=min(size(), finish);
 	if(start==finish)
@@ -415,12 +413,12 @@ void String::split(Array& result,
 		int pos_before;
 		// while we have 'delim'...
 		for(; (pos_before=pos(delim, delim_size, pos_after, lang))>=0 && limit; limit--) {
-			result+=&piece(pos_after, pos_before);
+			result+=&mid(pos_after, pos_before);
 			pos_after=pos_before+delim_size;
 		}
 		// last piece
 		if(pos_after<size() && limit) {
-			result+=&piece(pos_after, size());
+			result+=&mid(pos_after, size());
 			pos_after=size();
 		}
 		if(pos_after_ref)
@@ -441,12 +439,12 @@ void String::split(Array& result,
 		int pos_before;
 		// while we have 'delim'...
 		for(; (pos_before=pos(delim, pos_after, lang))>=0 && limit; limit--) {
-			result+=&piece(pos_after, pos_before);
+			result+=&mid(pos_after, pos_before);
 			pos_after=pos_before+delim.size();
 		}
 		// last piece
 		if(pos_after<size() && limit) {
-			result+=&piece(pos_after, size());
+			result+=&mid(pos_after, size());
 			pos_after=size();
 		}
 		if(pos_after_ref)
@@ -491,7 +489,6 @@ static void regex_options(char *options,
 	@verbatim
 		pre-match/match/post-match/1/2/3/...
 	@endverbatim
-	@test setlocale param to auto.p
 */
 bool String::match(const String *aorigin,
 				   const String& regexp, 
@@ -500,12 +497,16 @@ bool String::match(const String *aorigin
 				   Row_action row_action, void *info) const { 
 
 	static const unsigned char *tables=0; {	SYNCHRONIZED(true);
-		if(!tables) {
-			setlocale(LC_CTYPE, "ru");
+		if(!tables)
 			tables=pcre_maketables();
-		}
 	}
-	const char *pattern=regexp.cstr();
+
+
+	if(!regexp.size())
+		THROW(0, 0,
+			aorigin,
+			"regexp is empty");
+	const char *pattern=regexp.cstr(UL_AS_IS);
 	const char *errptr;
 	int erroffset;
     int option_bits[2];  regex_options(options?options->cstr():0, option_bits);
@@ -513,23 +514,22 @@ bool String::match(const String *aorigin
 		&errptr, &erroffset,
 		tables);
 
-	if(!code) {
+	if(!code)
 		THROW(0, 0,
-			&regexp.piece(erroffset, regexp.size()),
-			errptr);
-	}
+			&regexp.mid(erroffset, regexp.size()),
+			"match error - %s", errptr);
 	
 	int info_substrings=pcre_info(code, 0, 0);
 	if(info_substrings<0) {
 		(*pcre_free)(code);
 		THROW(0, 0,
-		aorigin,
-		"pcre_info error #%d", 
-			info_substrings);
+			aorigin,
+			"pcre_info error #%d", 
+				info_substrings);
 	}
 
 	int startoffset=0;
-	const char *subject=cstr();
+	const char *subject=cstr(UL_AS_IS);
 	int length=strlen(subject);
 	int ovecsize;
 	int *ovector=(int *)malloc(sizeof(int)*
@@ -556,7 +556,7 @@ bool String::match(const String *aorigin
 		
 		if(exec_substrings==PCRE_ERROR_NOMATCH) {
 			(*pcre_free)(code);
-			(*row_action)(**table, 0/*last time, no row*/, info);
+			(*row_action)(**table, 0/*last time, no row*/, 0, 0, info);
 			return option_bits[1]!=0; // global=true+table, not global=false
 		}
 
@@ -569,24 +569,23 @@ bool String::match(const String *aorigin
 		}
 
 		Array& row=*NEW Array(pool());
-		row+=&piece(0, ovector[0]); // .pre-match column value
-		row+=&piece(ovector[0], ovector[1]); // .match
-		row+=&piece(ovector[1], size()); // .post-match
+		row+=&mid(0, ovector[0]); // .pre-match column value
+		row+=&mid(ovector[0], ovector[1]); // .match
+		row+=&mid(ovector[1], size()); // .post-match
 		
 		for(int i=1; i<exec_substrings; i++) {
-			// -1:-1 case handled peacefully by piece() itself
-			row+=&piece(ovector[i*2+0], ovector[i*2+1]); // .i column value
+			// -1:-1 case handled peacefully by mid() itself
+			row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value
 		}
 		
-		(*row_action)(**table, &row, info);
+		(*row_action)(**table, &row, startoffset, ovector[0], info);
 
-		if(!option_bits[1]) { // not global
+		if(!option_bits[1] || !(startoffset=ovector[1])) { // not global | going to hang
 			(*pcre_free)(code);
-			(*row_action)(**table, 0/*last time, no row*/, info);
+			(*row_action)(**table, 0/*last time, no row*/, 0, 0, info);
 			return true;
 		}
 
-		startoffset=ovector[1];
 /*
 		if(option_bits[0] & PCRE_MULTILINE)
 			exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL