From e17fa1d8c28d12e123b27761355116eff9bd53f4 Mon Sep 17 00:00:00 2001
From: Simon Rozman <simon@rozman.si>
Date: Tue, 28 Nov 2023 13:48:54 +0100
Subject: [PATCH] parser: detect spaces, characters and newline faster where
 appropriate

No need to use locale-specific character type detection when ASCII.
Locale-specific implementation on Windows is not that very fast.

Signed-off-by: Simon Rozman <simon@rozman.si>
---
 include/stdex/parser.hpp | 14 +++++++-------
 include/stdex/string.hpp | 11 +++++++++++
 2 files changed, 18 insertions(+), 7 deletions(-)

diff --git a/include/stdex/parser.hpp b/include/stdex/parser.hpp
index a71dd2d41..766a254d9 100644
--- a/include/stdex/parser.hpp
+++ b/include/stdex/parser.hpp
@@ -400,7 +400,7 @@ namespace stdex
 				_Assume_(text || start >= end);
 				if (start < end && text[start]) {
 					bool r =
-						((flags & match_multiline) || !islbreak(text[start])) &&
+						((flags & match_multiline) || !stdex::islbreak(text[start])) &&
 						std::use_facet<std::ctype<T>>(this->m_locale).is(std::ctype_base::space, text[start]);
 					if ((r && !m_invert) || (!r && m_invert)) {
 						this->interval.end = (this->interval.start = start) + 1;
@@ -445,7 +445,7 @@ namespace stdex
 					const wchar_t* chr = next_sgml_cp(text, start, end, this->interval.end, buf);
 					const wchar_t* chr_end = chr + stdex::strlen(chr);
 					bool r =
-						((flags & match_multiline) || !islbreak(chr, SIZE_MAX)) &&
+						((flags & match_multiline) || !stdex::islbreak(chr, SIZE_MAX)) &&
 						std::use_facet<std::ctype<wchar_t>>(m_locale).scan_not(std::ctype_base::space, chr, chr_end) == chr_end;
 					if ((r && !m_invert) || (!r && m_invert)) {
 						this->interval.start = start;
@@ -553,7 +553,7 @@ namespace stdex
 				_Assume_(text || start >= end);
 				if (start < end && text[start]) {
 					bool r =
-						((flags & match_multiline) || !islbreak(text[start])) &&
+						((flags & match_multiline) || !stdex::islbreak(text[start])) &&
 						std::use_facet<std::ctype<T>>(this->m_locale).is(std::ctype_base::space | std::ctype_base::punct, text[start]);
 					if ((r && !m_invert) || (!r && m_invert)) {
 						this->interval.end = (this->interval.start = start) + 1;
@@ -598,7 +598,7 @@ namespace stdex
 					const wchar_t* chr = next_sgml_cp(text, start, end, this->interval.end, buf);
 					const wchar_t* chr_end = chr + stdex::strlen(chr);
 					bool r =
-						((flags & match_multiline) || !islbreak(chr, SIZE_MAX)) &&
+						((flags & match_multiline) || !stdex::islbreak(chr, SIZE_MAX)) &&
 						std::use_facet<std::ctype<wchar_t>>(m_locale).scan_not(std::ctype_base::space | std::ctype_base::punct, chr, chr_end) == chr_end;
 					if ((r && !m_invert) || (!r && m_invert)) {
 						this->interval.start = start;
@@ -626,7 +626,7 @@ namespace stdex
 				_In_ int flags = match_default)
 			{
 				_Assume_(text || start >= end);
-				bool r = start == 0 || (start <= end && islbreak(text[start - 1]));
+				bool r = start == 0 || (start <= end && stdex::islbreak(text[start - 1]));
 				if ((r && !m_invert) || (!r && m_invert)) {
 					this->interval.end = this->interval.start = start;
 					return true;
@@ -664,7 +664,7 @@ namespace stdex
 				_In_ int flags = match_default)
 			{
 				_Assume_(text || start >= end);
-				bool r = islbreak(text[start]);
+				bool r = stdex::islbreak(text[start]);
 				if ((r && !m_invert) || (!r && m_invert)) {
 					this->interval.end = this->interval.start = start;
 					return true;
@@ -6471,7 +6471,7 @@ namespace stdex
 						k.end = this->interval.end;
 						for (;;) {
 							if (k.end < end && text[k.end]) {
-								if (isalpha(text[k.end]))
+								if (stdex::isalpha(text[k.end]))
 									k.end++;
 								else
 									break;
diff --git a/include/stdex/string.hpp b/include/stdex/string.hpp
index 184abf923..b79b581f5 100644
--- a/include/stdex/string.hpp
+++ b/include/stdex/string.hpp
@@ -177,6 +177,17 @@ namespace stdex
 		return '0' <= chr && chr <= '9';
 	}
 
+	///
+	/// Test if the given code unit is ASCII-character
+	///
+	/// \param[in] chr  Code unit
+	///
+	template <class T>
+	inline bool isalpha(_In_ T chr)
+	{
+		return islower(chr) || isupper(chr);
+	}
+
 	///
 	/// Return number of code units the glyph represents
 	///