From 822be406882d8a5d29fba1d65875b7433aff21f9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Branimir=20Karad=C5=BEi=C4=87?= <branimirkaradzic@gmail.com>
Date: Sat, 8 Aug 2026 15:16:25 -0700
Subject: [PATCH] Scanner: Class::Identifier must not start with a digit.
 (#414)

---
 include/bx/string.h    |  8 ++++++++
 src/scanner.cpp        |  2 +-
 src/string.cpp         | 11 +++++++++++
 tests/scanner_test.cpp | 40 ++++++++++++++++++++++++++++++++++++++++
 tests/string_test.cpp  | 20 ++++++++++++++++++++
 5 files changed, 80 insertions(+), 1 deletion(-)

diff --git a/include/bx/string.h b/include/bx/string.h
index 99fd2ff..4575906 100644
--- a/include/bx/string.h
+++ b/include/bx/string.h
@@ -453,6 +453,14 @@ namespace bx
 	/// Returns StringView of word or empty.
 	StringView strWord(const StringView& _str);
 
+	/// Returns StringView of identifier or empty.
+	///
+	/// @remarks Unlike `strWord` the first character must be alphabetic or '_'; the rest
+	///   can additionally be numeric. `strWord` accepts a leading digit, so "123abc" is a
+	///   word but not an identifier.
+	///
+	StringView strIdentifier(const StringView& _str);
+
 	/// Returns substring in string.
 	StringView strSubstr(const StringView& _str, int32_t _start, int32_t _len = INT32_MAX);
 
diff --git a/src/scanner.cpp b/src/scanner.cpp
index 8cb923b..e1f3069 100644
--- a/src/scanner.cpp
+++ b/src/scanner.cpp
@@ -92,7 +92,7 @@ namespace bx
 
 		case Class::Identifier:
 			{
-				const StringView word = strWord(m_tail);
+				const StringView word = strIdentifier(m_tail);
 				return word.isEmpty()
 					? m_tail
 					: StringView(word.getTerm(), m_tail.getTerm() )
diff --git a/src/string.cpp b/src/string.cpp
index b27d845..4380735 100644
--- a/src/string.cpp
+++ b/src/string.cpp
@@ -538,6 +538,17 @@ namespace bx
 		return StringView(ptr, term);
 	}
 
+	StringView strIdentifier(const StringView& _str)
+	{
+		if (_str.isEmpty()
+		||  !(isAlpha(*_str.getPtr() ) || '_' == *_str.getPtr() ) )
+		{
+			return StringView(_str.getPtr(), _str.getPtr() );
+		}
+
+		return strWord(_str);
+	}
+
 	StringView strFindBlock(const StringView& _str, char _open, char _close)
 	{
 		const char* curr  = _str.getPtr();
diff --git a/tests/scanner_test.cpp b/tests/scanner_test.cpp
index 2ba5b69..ddafdc7 100644
--- a/tests/scanner_test.cpp
+++ b/tests/scanner_test.cpp
@@ -144,6 +144,46 @@ TEST_CASE("Scanner.class accept", "[scanner]")
 	}
 }
 
+TEST_CASE("Scanner.class Identifier start", "[scanner]")
+{
+	{
+		bx::Scanner sc(bx::StringView("123abc") );
+		REQUIRE(sc.accept(bx::Scanner::Class::Identifier).isEmpty() );
+		REQUIRE(sc.peek(bx::Scanner::Class::Identifier).isEmpty() );
+		REQUIRE("123abc" == sc.acceptAll() );
+	}
+
+	{
+		bx::Scanner sc(bx::StringView("5_0") );
+		REQUIRE(sc.accept(bx::Scanner::Class::Identifier).isEmpty() );
+	}
+
+	{
+		bx::Scanner sc(bx::StringView("abc123 def") );
+		REQUIRE("abc123" == sc.accept(bx::Scanner::Class::Identifier) );
+	}
+
+	{
+		bx::Scanner sc(bx::StringView("_abc9") );
+		REQUIRE("_abc9" == sc.accept(bx::Scanner::Class::Identifier) );
+	}
+
+	{
+		bx::Scanner sc(bx::StringView("ps_5_0") );
+		REQUIRE("ps_5_0" == sc.accept(bx::Scanner::Class::Identifier) );
+	}
+
+	{
+		bx::Scanner sc(bx::StringView("-abc") );
+		REQUIRE(sc.accept(bx::Scanner::Class::Identifier).isEmpty() );
+	}
+
+	{
+		bx::Scanner sc(bx::StringView("") );
+		REQUIRE(sc.accept(bx::Scanner::Class::Identifier).isEmpty() );
+	}
+}
+
 TEST_CASE("Scanner.peek", "[scanner]")
 {
 	bx::Scanner sc(bx::StringView("abc def") );
diff --git a/tests/string_test.cpp b/tests/string_test.cpp
index 3881c6d..c5356a5 100644
--- a/tests/string_test.cpp
+++ b/tests/string_test.cpp
@@ -689,6 +689,26 @@ TEST_CASE("strWord", "[string]")
 {
 	REQUIRE(bx::strWord(" abvgd-1389.0").isEmpty() );
 	REQUIRE(0 == bx::strCmp(bx::strWord("abvgd-1389.0"), "abvgd") );
+
+	REQUIRE(0 == bx::strCmp(bx::strWord("1389.0"), "1389") );
+}
+
+TEST_CASE("strIdentifier", "[string]")
+{
+	REQUIRE(bx::strIdentifier("").isEmpty() );
+	REQUIRE(bx::strIdentifier(" abvgd").isEmpty() );
+	REQUIRE(bx::strIdentifier("-abc").isEmpty() );
+
+	REQUIRE(bx::strIdentifier("123abc").isEmpty() );
+	REQUIRE(bx::strIdentifier("5_0").isEmpty() );
+
+	REQUIRE(0 == bx::strCmp(bx::strIdentifier("abvgd-1389.0"), "abvgd") );
+	REQUIRE(0 == bx::strCmp(bx::strIdentifier("abc123 def"),   "abc123") );
+	REQUIRE(0 == bx::strCmp(bx::strIdentifier("_abc9"),        "_abc9") );
+	REQUIRE(0 == bx::strCmp(bx::strIdentifier("ps_5_0"),       "ps_5_0") );
+
+	const bx::StringView test("123abc");
+	REQUIRE(test.getPtr() == bx::strIdentifier(test).getPtr() );
 }
 
 TEST_CASE("strFindEol strFindNl", "[string]")