--- parser3/src/main/pa_string.C	2001/04/03 09:58:10	1.62
+++ parser3/src/main/pa_string.C	2001/04/23 13:38:31	1.78
@@ -5,12 +5,12 @@
 
 	Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
 
-	$Id: pa_string.C,v 1.62 2001/04/03 09:58:10 paf Exp $
+	$Id: pa_string.C,v 1.78 2001/04/23 13:38:31 paf Exp $
 */
 
 #include "pa_config_includes.h"
 
-#include <locale.h>
+#include "pcre.h"
 
 #include "pa_pool.h"
 #include "pa_string.h"
@@ -22,17 +22,11 @@
 #include "pa_table.h"
 #include "pa_threads.h"
 
-#include "pcre.h"
-
 //#include "pa_sapi.h"
 
-// consts
-
-const int MAX_MATCH_COLUMNS=20;
-
 // String
 
-String::String(Pool& apool, const char *src, bool tasize_ted) :
+String::String(Pool& apool, const char *src, size_t src_size, bool tainted) :
 	Pooled(apool) {
 	last_chunk=&head;
 	head.count=CR_PREALLOCATED_COUNT;
@@ -42,10 +36,10 @@ String::String(Pool& apool, const char *
 	fused_rows=fsize=0;
 
 	if(src)
-		if(tasize_ted)
-			APPEND_TAINTED(src, 0, 0, 0);
+		if(tainted)
+			APPEND_TAINTED(src, src_size, 0, 0);
 		else
-			APPEND_CONST(src);
+			APPEND_CLEAN(src, src_size, 0, 0);
 }
 
 void String::expand() {
@@ -203,7 +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=
@@ -283,7 +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=
@@ -346,7 +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)
@@ -419,12 +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)
@@ -445,12 +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)
@@ -462,66 +456,133 @@ void String::split(Array& result,
 	}
 }
 
-/// @test setlocale param to auto.p  @test pcre_malloc & pcre_free substs
-bool String::match(const String *aorigin,
+static void regex_options(char *options, int *result){
+    struct Regex_option {
+		char key;
+		int clear, set;
+		int *result;
+    } regex_option[]={
+		{'i', 0, PCRE_CASELESS, result}, // a=A
+		{'s', 0, PCRE_DOTALL, result}, // \n\n$
+		{'x', 0, PCRE_EXTENDED, result}, // whitespace in regex ignored
+		{'m', PCRE_DOTALL, PCRE_MULTILINE, result}, // ^aaa\n$^bbb\n$
+		{'g', 0, true, result+1}, // many rows
+		{0},
+    };
+	result[0]=PCRE_EXTRA | PCRE_DOTALL;
+	result[1]=0;
+
+    if(options) 
+		for(Regex_option *o=regex_option; o->key; o++) 
+			if(
+				strchr(options, o->key) || 
+				strchr(options, toupper(o->key))) {
+				*(o->result)&=~o->clear;
+				*(o->result)|=o->set;
+			}
+}
+
+/**
+	returns true if fills table.
+	table format is defined and fixed[can be used by others]: 
+	@verbatim
+		pre-match/match/post-match/1/2/3/...
+	@endverbatim
+*/
+bool String::match(const unsigned char *pcre_tables,
+				   const String *aorigin,
 				   const String& regexp, 
-				   const String& options,
-				   Table **table) const { 
-	SYNCHRONIZED(true);
-	static const unsigned char *tables=0;
-	if(!tables) {
-		setlocale(LC_CTYPE, "ru");
-		tables = pcre_maketables();
-	}
-	const char *pattern=regexp.cstr();
+				   const String *options,
+				   Table **table,
+				   Row_action row_action, void *info) const { 
+
+	if(!regexp.size())
+		THROW(0, 0,
+			aorigin,
+			"regexp is empty");
+	const char *pattern=regexp.cstr(UL_AS_IS);
 	const char *errptr;
 	int erroffset;
-	pcre *code=pcre_compile(pattern, 0, 
+    int option_bits[2];  regex_options(options?options->cstr():0, option_bits);
+	pcre *code=pcre_compile(pattern, option_bits[0], 
 		&errptr, &erroffset,
-		tables);
+		pcre_tables);
 
 	if(!code)
 		THROW(0, 0,
-			&regexp.piece(erroffset, regexp.size()),
-			errptr);
+			&regexp.mid(erroffset, regexp.size()),
+			"regular expression syntax error - %s", errptr);
 	
-	int ovecsize;
-	int *ovector=(int *)malloc(sizeof(int)*(ovecsize=(1/*.match*/+MAX_MATCH_COLUMNS)*3));
-	const char *subject=cstr();
-	int length=strlen(subject);
-	int exec_result=pcre_exec(code, 0,
-          subject, length, 0/*startoffset*/,
-          0/*options*/, ovector, ovecsize);
-
-	if(exec_result==PCRE_ERROR_NOMATCH) {
-		*table=0;
-		return false;
+	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);
 	}
 
-	if(exec_result<0)
-		THROW(0, 0,
-			0,
-			"pcre_exec failed");
+	int startoffset=0;
+	const char *subject=cstr(UL_AS_IS);
+	int length=strlen(subject);
+	int ovecsize;
+	int *ovector=(int *)malloc(sizeof(int)*
+		(ovecsize=(1/*match*/+info_substrings)*3));
 
-	if(exec_result==0)
-		THROW(0, 0,
-			aorigin,
-			"produced more substrings than maximum handled by Parser, which is %d",
-				MAX_MATCH_COLUMNS);
+	{ // create table
+		Array& columns=*NEW Array(pool());
+		columns+=string_pre_match_name;
+		columns+=string_match_name;
+		columns+=string_post_match_name;
+		for(int i=1; i<=info_substrings; i++) {
+			char *column=(char *)malloc(MAX_NUMBER);
+			snprintf(column, MAX_NUMBER, "%d", i);
+			columns+=NEW String(pool(), column); // .i column name
+		}
+		*table=NEW Table(pool(), aorigin, &columns);
+	}
 
-	Array& columns=*NEW Array(pool());
-	columns+=string_match_name; // .match column name
-	Array& row=*NEW Array(pool());
-	row+=&piece(ovector[0], ovector[1]); // match column value
-	
-	for(int i=1; i<exec_result; i++) {
-		char *column=(char *)malloc(MAX_NUMBER);
-		snprintf(column, MAX_NUMBER, "%d", i);
-		columns+=NEW String(pool(), column); // .i column name
-		row+=&piece(ovector[i*2+0], ovector[i*2+1]); // .i column value
+	int exec_option_bits=0;
+	while(true) {
+		int exec_substrings=pcre_exec(code, 0,
+			subject, length, startoffset,
+			exec_option_bits, ovector, ovecsize);
+		
+		if(exec_substrings==PCRE_ERROR_NOMATCH) {
+			(*pcre_free)(code);
+			(*row_action)(**table, 0/*last time, no row*/, 0, 0, info);
+			return option_bits[1]!=0; // global=true+table, not global=false
+		}
+
+		if(exec_substrings<0) {
+			(*pcre_free)(code);
+			THROW(0, 0,
+				aorigin,
+				"regular expression execute error (%d)", 
+					exec_substrings);
+		}
+
+		Array& row=*NEW Array(pool());
+		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 mid() itself
+			row+=&mid(ovector[i*2+0], ovector[i*2+1]); // .i column value
+		}
+		
+		(*row_action)(**table, &row, startoffset, ovector[0], info);
+
+		if(!option_bits[1] || !(startoffset=ovector[1])) { // not global | going to hang
+			(*pcre_free)(code);
+			(*row_action)(**table, 0/*last time, no row*/, 0, 0, info);
+			return true;
+		}
+
+/*
+		if(option_bits[0] & PCRE_MULTILINE)
+			exec_option_bits|=PCRE_NOTBOL; // start of subject+startoffset not BOL
+*/
 	}
-	
-	*table=NEW Table(pool(), aorigin, &columns);
-	(**table)+=&row;
-	return true;
 }