--- parser3/src/main/pa_string.C	2016/09/05 22:06:13	1.255
+++ parser3/src/main/pa_string.C	2024/11/10 20:46:08	1.280
@@ -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) 2001-2024 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 1.255 2016/09/05 22:06:13 moko Exp $" IDENT_PA_STRING_H;
+volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.280 2024/11/10 20:46:08 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){
@@ -47,9 +46,10 @@ template<typename T> inline T pa_ato_any
 		}
 	}
 
-	if (base < 2 || base > 16) { /* illegal base */
+	if (base < 2 || base > 16) /* illegal base */
 		throw Exception(PARSER_RUNTIME, 0, "base to must be an integer from 2 to 16");
-	}
+	if (*pos == '-')
+		throw Exception("number.format", problem_source, problem_source ? "out of range (negative)" : "'%s' is out if range (negative)", str);
 
 	T cutoff = max / base;
 	int cutoff_digit = (int)(max - cutoff * base);
@@ -81,20 +81,26 @@ template<typename T> inline T pa_ato_any
 
 	while(char c=*pos++)
 		if(!isspace(c))
-			throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is invalid number (int)", str);
+			throw Exception("number.format", problem_source, problem_source ? "invalid number (int)" : "'%s' is an 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){
-	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 +110,31 @@ 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 an 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 an 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 +148,12 @@ 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 +171,9 @@ 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;
 }
@@ -174,7 +189,7 @@ 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;
@@ -218,21 +233,20 @@ 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(pa_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::trim(String::Trim_kind kind, const char* chars, size_t* out_start, size_t* out_length, Charset* source_charset) const {
 	size_t our_length=length();
 	if(!our_length)
@@ -601,46 +615,40 @@ size_t String::pos(Charset& charset, con
 	}
 }
 
-void String::split(ArrayString& result, size_t& pos_after, const char* delim, Language lang, int limit) const {
+void String::split(ArrayString& result, size_t pos_after, const char* delim, Language lang) const {
 	if(is_empty())
 		return;
 	size_t self_length=length();
 	if(size_t delim_length=strlen(delim)) {
 		size_t pos_before;
 		// while we have 'delim'...
-		for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
+		while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) {
 			result+=&mid(pos_after, pos_before);
 			pos_after=pos_before+delim_length;
 		}
 		// last piece
-		if(pos_after<self_length && limit) {
+		if(pos_after<self_length)
 			result+=&mid(pos_after, self_length);
-			pos_after=self_length;
-		}
 	} else { // empty delim
 		result+=this;
-		pos_after+=self_length;
 	}
 }
 
-void String::split(ArrayString& result, size_t& pos_after, const String& delim, Language lang, int limit) const {
+void String::split(ArrayString& result, size_t pos_after, const String& delim, Language lang) const {
 	if(is_empty())
 		return;
 	if(!delim.is_empty()) {
 		size_t pos_before;
 		// while we have 'delim'...
-		for(; (pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND && limit; limit--) {
+		while((pos_before=pos(delim, pos_after, lang))!=STRING_NOT_FOUND) {
 			result+=&mid(pos_after, pos_before);
 			pos_after=pos_before+delim.length();
 		}
 		// last piece
-		if(pos_after<length() && limit) {
+		if(pos_after<length())
 			result+=&mid(pos_after, length());
-			pos_after=length();
-		}
 	} else { // empty delim
 		result+=this;
-		pos_after+=length();
 	}
 }
 
@@ -653,15 +661,16 @@ 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);
 
@@ -671,8 +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;
 		}
 
@@ -693,12 +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 || (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);
@@ -950,19 +961,15 @@ bool String::deserialize(size_t prolog_s
 	return true;
 }
 
-const char* String::Body::v() const {
-	return CORD_to_const_char_star(body, length());
-}
-
 void String::Body::dump() const {
 	CORD_dump(body);
 }
 
-const char* String::Languages::v() const {
+const char* String::Languages::visualize() const {
 	if(opt.is_not_just_lang)
 		return CORD_to_const_char_star(langs, 0);
 	else
-		return (const char*)&langs;
+		return 0;
 }
 
 void String::Languages::dump() const {
@@ -972,26 +979,22 @@ void String::Languages::dump() const {
 		puts((const char*)&langs);
 }
 
-const char* String::v() const {
-	const uint LIMIT_VIEW=20;
-	char* buf=(char*)pa_malloc(MAX_STRING);
-	const char*body_view=body.v();
-	const char*langs_view=langs.v();
-	snprintf(buf, MAX_STRING, 
-		"%d:%.*s%s}   "
-		"{%d:%s",
-		langs.count(), LIMIT_VIEW, langs_view, strlen(langs_view)>LIMIT_VIEW?"...":"",
-		strlen(body_view), body_view
-	);
-
-	return buf;
-}
-
 void String::dump() const {
 	body.dump();
 	langs.dump();
 }
 
+static char *n_chars(char c, size_t length){
+	char *result=(char *)pa_malloc_atomic(length+1);
+	memset(result, c, length);
+	result[length] = '\0';
+	return result;
+}
+
+char* String::visualize_langs() const {
+	return is_not_just_lang() ? pa_strdup(langs.visualize()) : n_chars((char)just_lang(), length());
+}
+
 const String& String::trim(String::Trim_kind kind, const char* chars, Charset* source_charset) const {
 	if(is_empty())
 		return *this;