--- parser3/src/main/pa_string.C	2016/09/05 21:59:22	1.254
+++ parser3/src/main/pa_string.C	2020/10/10 06:08:37	1.263
@@ -1,7 +1,7 @@
 /** @file
 	Parser: string class. @see untalength_t.C.
 
-	Copyright (c) 2001-2015 Art. Lebedev Studio (http://www.artlebedev.com)
+	Copyright (c) 2001-2017 Art. Lebedev Studio (http://www.artlebedev.com)
 	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
 */
 
@@ -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 1.254 2016/09/05 21:59:22 moko Exp $" IDENT_PA_STRING_H;
+volatile const char * IDENT_PA_STRING_C="$Id: pa_string.C,v 1.263 2020/10/10 06:08:37 moko Exp $" IDENT_PA_STRING_H;
 
 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){
@@ -87,10 +86,16 @@ 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){
+	if(!str)
+		return 0;
+
 	return pa_ato_any<unsigned long long int>(str, base, problem_source, ULLONG_MAX);
 }
 
@@ -480,7 +485,8 @@ struct CORD_length_info {
 };
 
 int CORD_batched_len(const char* s, CORD_length_info* info){
-	info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s));	return 0;
+	info->len += lengthUTF8( (const XMLByte *)s, (const XMLByte *)s+strlen(s));
+	return 0;
 }
 
 // can be called only for IS_FUNCTION(CORD) which are used in large String::Body::mid
@@ -527,7 +533,7 @@ String& String::mid(size_t substr_begin,
 String& String::mid(Charset& charset, size_t from, size_t to, size_t helper_length) const {
 	String& result=*new String;
 
-	size_t self_length=(helper_length)?helper_length:length(charset);
+	size_t self_length=helper_length ? helper_length : length(charset);
 
 	if(!self_length)
 		return result;
@@ -600,46 +606,40 @@ 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();
 	}
 }
 
@@ -949,19 +949,15 @@ 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 {
@@ -971,26 +967,22 @@ 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;