--- parser3/src/main/pa_string.C	2016/12/28 17:41:15	1.258
+++ parser3/src/main/pa_string.C	2023/12/13 20:07:11	1.271
@@ -1,8 +1,8 @@
 /** @file
 	Parser: string class. @see untalength_t.C.
 
-	Copyright (c) 2001-2015 Art. Lebedev Studio (http://www.artlebedev.com)
-	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
+	Copyright (c) 2001-2023 Art. Lebedev Studio (http://www.artlebedev.com)
+	Authors: Konstantin Morshnev <moko@design.ru>, Alexandr Petrosian <paf@design.ru>
 */
 
 #include "pa_string.h"
@@ -12,14 +12,13 @@
 #include "pa_charset.h"
 #include "pa_vregex.h"
 
-#ifndef ULLONG_MAX
-#define ULLONG_MAX 18446744073709551615ULL
-#endif
-
-volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.258 2016/12/28 17:41:15 moko Exp $" IDENT_PA_STRING_H;
+volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.271 2023/12/13 20:07:11 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){
@@ -87,14 +86,20 @@ template<typename T> inline T pa_ato_any
 }
 
 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){
-	return pa_ato_any<unsigned long long int>(str, base, problem_source, ULLONG_MAX);
+uint64_t pa_atoul(const char *str, int base, const String* problem_source){
+	if(!str)
+		return 0;
+
+	return pa_ato_any<uint64_t>(str, base, problem_source, ULLONG_MAX);
 }
 
-int pa_atoi(const char* str, const String* problem_source) {
+int pa_atoi(const char* str, int base, const String* problem_source) {
 	if(!str)
 		return 0;
 
@@ -104,26 +109,31 @@ int pa_atoi(const char* str, const Strin
 	if(!*str)
 		return 0;
 
+	const char *str_copy=str;
 	bool negative=false;
 	if(str[0]=='-') {
 		negative=true;
 		str++;
+		if(!*str || isspace(*str))
+			throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str_copy);
 	} else if(str[0]=='+') {
 		str++;
+		if(!*str || isspace(*str))
+			throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str_copy);
 	}
 
-	unsigned int result=pa_atoui(str, 0, problem_source);
+	unsigned int result=pa_atoui(str, base, 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);
+
+	throw Exception("number.format", problem_source, problem_source ? "out of range (int)" : "'%s' is out of range (int)", str_copy);
 }
 
-double pa_atod(const char* str, const String* problem_source) {
+double pa_atod(const char* str, const String* problem_source /* never null */) {
 	if(!str)
 		return 0;
 
@@ -137,8 +147,12 @@ double pa_atod(const char* str, const St
 	if(str[0]=='-') {
 		negative=true;
 		str++;
+		if(!*str || isspace(*str))
+			throw Exception("number.format", problem_source, "invalid number (double)");
 	} else if(str[0]=='+') {
 		str++;
+		if(!*str || isspace(*str))
+			throw Exception("number.format", problem_source, "invalid number (double)");
 	}
 
 	double result;
@@ -156,9 +170,9 @@ double pa_atod(const char* str, const St
 	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);
+	while(const char c=*error_pos++)
+		if(!isspace(c))
+			throw Exception("number.format", problem_source, "invalid number (double)");
 
 	return negative ? -result : result;
 }
@@ -647,7 +661,7 @@ 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;
@@ -656,6 +670,7 @@ Table* String::match(VRegex* vregex, Row
 	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);
 
@@ -665,8 +680,9 @@ 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;
 		}
 
@@ -687,12 +703,13 @@ 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 || 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);