--- parser3/src/main/pa_string.C	2020/10/10 06:08:37	1.263
+++ parser3/src/main/pa_string.C	2026/04/25 13:38:46	1.285
@@ -1,8 +1,8 @@
 /** @file
 	Parser: string class. @see untalength_t.C.
 
-	Copyright (c) 2001-2017 Art. Lebedev Studio (http://www.artlebedev.com)
-	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
+	Copyright (c) 2001-2026 Art. Lebedev Studio (https://www.artlebedev.com)
+	Authors: Konstantin Morshnev <moko@design.ru>, Alexandr Petrosian <paf@design.ru>
 */
 
 #include "pa_string.h"
@@ -12,162 +12,13 @@
 #include "pa_charset.h"
 #include "pa_vregex.h"
 
-volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.263 2020/10/10 06:08:37 moko Exp $" IDENT_PA_STRING_H;
+volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.285 2026/04/25 13:38:46 moko Exp $" IDENT_PA_STRING_H;
 
 const String String::Empty;
 
 #define COMPILE_ASSERT(x) extern int assert_checker[(x) ? 1 : -1]
 COMPILE_ASSERT(sizeof(String::Languages) == sizeof(CORD));
 
-// pa_atoui is based on Manuel Novoa III _strto_l for uClibc
-
-template<typename T> inline T pa_ato_any(const char *str, int base, const String* problem_source,const T max){
-	T result = 0;
-	const char *pos = str;
-
-	while (isspace(*pos)) /* skip leading whitespace */
-		++pos;
-
-	if (base == 16 && *pos == '0') { /* handle option prefix */
-		++pos;
-		if (*pos == 'x' || *pos == 'X') {
-			++pos;
-		}
-	}
-
-	if (base == 0) { /* dynamic base */
-		base = 10; /* default is 10 */
-		if (*pos == '0') {
-			++pos;
-			if (*pos == 'x' || *pos == 'X'){
-				++pos;
-				base=16;
-			}
-		}
-	}
-
-	if (base < 2 || base > 16) { /* illegal base */
-		throw Exception(PARSER_RUNTIME, 0, "base to must be an integer from 2 to 16");
-	}
-
-	T cutoff = max / base;
-	int cutoff_digit = (int)(max - cutoff * base);
-
-	while(true) {
-		int digit;
-		
-		if ((*pos >= '0') && (*pos <= '9')) {
-			digit = (*pos - '0');
-		} else if (*pos >= 'a') {
-			digit = (*pos - 'a' + 10);
-		} else if (*pos >= 'A') {
-			digit = (*pos - 'A' + 10);
-		} else break;
-
-		if (digit >= base) {
-			break;
-		}
-		
-		++pos;
-		
-		/* adjust number, with overflow check */
-		if ((result > cutoff) || ((result == cutoff) && (digit > cutoff_digit))) {
-			throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
-		} else {
-			result  = result * base + digit;
-		}
-	}
-
-	while(char c=*pos++)
-		if(!isspace(c))
-			throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str);
-
-	return result;
-}
-
-unsigned int pa_atoui(const char *str, int base, const String* problem_source){
-	if(!str)
-		return 0;
-
-	return pa_ato_any<unsigned int>(str, base, problem_source, UINT_MAX);
-}
-
-unsigned long long pa_atoul(const char *str, int base, const String* problem_source){
-	if(!str)
-		return 0;
-
-	return pa_ato_any<unsigned long long int>(str, base, problem_source, ULLONG_MAX);
-}
-
-int pa_atoi(const char* str, const String* problem_source) {
-	if(!str)
-		return 0;
-
-	while(isspace(*str))
-		str++;
-
-	if(!*str)
-		return 0;
-
-	bool negative=false;
-	if(str[0]=='-') {
-		negative=true;
-		str++;
-	} else if(str[0]=='+') {
-		str++;
-	}
-
-	unsigned int result=pa_atoui(str, 0, problem_source);
-
-	if(negative && result <= ((unsigned int)(-(1+INT_MIN)))+1)
-		return -(int)result;
-	
-	if(result<=INT_MAX)
-		return (int)result;
-	
-	throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str);
-}
-
-double pa_atod(const char* str, const String* problem_source) {
-	if(!str)
-		return 0;
-
-	while(isspace(*str))
-		str++;
-
-	if(!*str)
-		return 0;
-
-	bool negative=false;
-	if(str[0]=='-') {
-		negative=true;
-		str++;
-	} else if(str[0]=='+') {
-		str++;
-	}
-
-	double result;
-	if(str[0]=='0') {
-		if(str[1]=='x' || str[1]=='X') {
-			// 0xABC
-			result=(double)pa_atoul(str, 0, problem_source);
-			return negative ? -result : result;
-		} else {
-			 // skip leading 0000, to disable octal interpretation
-			do str++; while(*str=='0');
-		}
-	}
-
-	char *error_pos;
-	result=strtod(str, &error_pos);
-
-	while(char c=*error_pos++)
-		if(!isspace((unsigned char)c))
-			throw Exception("number.format", problem_source, problem_source ? "invalid number (double)" : "'%s' is invalid number (double)", str);
-
-	return negative ? -result : result;
-}
-
 // cord lib extension
 
 #ifndef DOXYGEN
@@ -179,7 +30,7 @@ typedef struct {
 
 static int CORD_range_contains_chr_greater_then_proc(char c, size_t size, void* client_data)
 {
-	register chr_data * d = (chr_data *)client_data;
+	chr_data * d = (chr_data *)client_data;
 
 	if (d -> countdown<=0) return(2);
 	d -> countdown -= size;
@@ -223,19 +74,27 @@ public:
 		*this+=new String("match");
 		*this+=new String("postmatch");
 		for(int i=0; i<MAX_MATCH_GROUPS; i++) {
-			*this+=new String(String::Body::Format(1+i), String::L_CLEAN);
+			*this+=new String(String::Body::uitoa(1+i), String::L_CLEAN);
 		}
 	}
 };
 
-Table string_match_table_template(new String_match_table_template_columns);
+static Table &string_match_table_template(){
+	static Table *singleton=NULL;
+	if(!singleton)
+		singleton=new Table(new String_match_table_template_columns);
+	return *singleton;
+}
 
 // String::Body methods
 
-String::Body String::Body::Format(int value) {
-	char local[MAX_NUMBER];
-	size_t length=snprintf(local, MAX_NUMBER, "%d", value);
-	return String::Body(pa_strdup(local, length));
+String::Body String::Body::uitoa(size_t aindex) {
+	static const size_t CACHE_SIZE=256;
+	static String::Body cache[CACHE_SIZE];
+	if (aindex < CACHE_SIZE)
+		return !cache[aindex] ? (cache[aindex]=Body(pa_uitoa(aindex))) : cache[aindex];
+	else
+		return Body(pa_uitoa(aindex));
 }
 
 String::Body String::Body::trim(String::Trim_kind kind, const char* chars, size_t* out_start, size_t* out_length, Charset* source_charset) const {
@@ -257,7 +116,7 @@ String::Body String::Body::trim(String::
 	size_t start=0;
 	size_t end=our_length;
 	if(!chars)
-		chars=" \t\n"; // white space
+		chars=" \t\r\n"; // white space
 
 	if(fast){
 		// from left...
@@ -267,9 +126,9 @@ String::Body String::Body::trim(String::
 				char c=CORD_pos_fetch(pos);
 				if(strchr(chars, c)) {
 					if(++start==our_length)
-						return 0; // all chars are empty, just return empty string
+						return String::Body(); // all chars are empty, just return empty string
 				} else
-					break;			
+					break;
 
 				CORD_next(pos);
 			}
@@ -282,9 +141,9 @@ String::Body String::Body::trim(String::
 				char c=CORD_pos_fetch(pos);
 				if(strchr(chars, c)) {
 					if(--end==0) // optimization: NO need to check for 'end>=start', that's(<) impossible
-						return 0; // all chars are empty, just return empty string
+						return String::Body(); // all chars are empty, just return empty string
 				} else
-					break;			
+					break;
 
 				CORD_prev(pos);
 			}
@@ -312,7 +171,7 @@ String::Body String::Body::trim(String::
 				if(found){
 					start+=char_length;
 					if(start==our_length)
-						return 0; // all chars are empty, just return empty string
+						return String::Body(); // all chars are empty, just return empty string
 				} else
 					break;
 			}
@@ -336,7 +195,7 @@ String::Body String::Body::trim(String::
 				if(found){
 					end-=char_length;
 					if(end==0)
-						return 0; // all chars are empty, just return empty string
+						return String::Body(); // all chars are empty, just return empty string
 				} else
 					break;
 			}
@@ -613,7 +472,7 @@ void String::split(ArrayString& result,
 	if(size_t delim_length=strlen(delim)) {
 		size_t pos_before;
 		// while we have 'delim'...
-		while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) {
+		while((pos_before=pos(String::Body(delim), pos_after, lang)) != STRING_NOT_FOUND) {
 			result+=&mid(pos_after, pos_before);
 			pos_after=pos_before+delim_length;
 		}
@@ -652,15 +511,16 @@ Table* String::match(VRegex* vregex, Row
 
 	const char* subject=cstr();
 	size_t subject_length=length();
-	const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3;
+	const int ovector_size=(1/*match*/+MAX_MATCH_GROUPS)*3; /* 1/3 is used as workspace by pcre_exec() */
 	int ovector[ovector_size];
 
 	Table::Action_options table_options;
-	Table& table=*new Table(string_match_table_template, table_options);
+	Table& table=*new Table(string_match_table_template(), table_options);
 
 	int prestart=0;
 	int poststart=0;
 	int postfinish=length();
+	int action_was_executed=-1;
 	while(true) {
 		int exec_result=vregex->exec(subject, subject_length, ovector, ovector_size, prestart);
 
@@ -670,8 +530,9 @@ Table* String::match(VRegex* vregex, Row
 		int prefinish=ovector[0];
 		poststart=ovector[1];
 
-		if (prestart==poststart && subject[poststart]=='\n'){
+		if (prestart==poststart && action_was_executed==1){
 			prestart++;
+			action_was_executed=0;
 			continue;
 		}
 
@@ -692,12 +553,13 @@ Table* String::match(VRegex* vregex, Row
 		}
 		
 		matches_count++;
-		row_action(table, row, prestart, prefinish, poststart, postfinish, info);
+		row_action(table, row, prestart - !action_was_executed, prefinish, poststart, postfinish, info);
 
-		if(!global || prestart==poststart) // last step
+		if(!global || (size_t)poststart>=subject_length) // last step, avoid prestart++ after last char
 			break;
 
 		prestart=poststart;
+		action_was_executed=1;
 	}
 
 	row_action(table, 0/*last time, no raw*/, 0, 0, poststart, postfinish, info);
@@ -756,7 +618,7 @@ String& String::change_case(Charset& sou
 		}
 	}
 	result.langs=langs;
-	result.body=new_cstr;
+	result.body=String::Body(new_cstr);
 
 	return result;
 }