From c36ffce0bdf97bff176f92b65be52adade459e4d Mon Sep 17 00:00:00 2001
From: =?utf8?q?Jean-Philippe=20Bruy=C3=A8re?= <jp_bruyere@hotmail.com>
Date: Sun, 6 Jul 2025 09:22:59 +0200
Subject: [PATCH] TokenType.UnexpectedChar to handle invalid '<' in xml
 elements

---
 CrowEditBase/src/Compiler/TokenType.cs        |  4 +-
 .../src/Parsing/IML/ImlTokenType.cs           |  1 +
 .../src/Parsing/XmlSyntaxAnalyser.cs          |  3 +-
 .../CEXmlPlugin/src/Parsing/XmlTokenType.cs   |  2 +-
 .../CEXmlPlugin/src/Parsing/XmlTokenizer.cs   | 45 +++++++++++--------
 5 files changed, 32 insertions(+), 23 deletions(-)

diff --git a/CrowEditBase/src/Compiler/TokenType.cs b/CrowEditBase/src/Compiler/TokenType.cs
index d34c355..1d25015 100644
--- a/CrowEditBase/src/Compiler/TokenType.cs
+++ b/CrowEditBase/src/Compiler/TokenType.cs
@@ -26,10 +26,10 @@ namespace CrowEditBase
 		OpenBrace				= 0x0405,
 		CloseBrace				= 0x0406,
 		DoubleQuote				= 0x0407,
-		SingleQuote				= 0x0408,		
+		SingleQuote				= 0x0408,
 
 		Operator 				= 0x0800,
 		Keyword 				= 0x1000,
-		Syntax					= 0x8000
+		UnexpectedChar			= 0x8000,
 	}
 }
\ No newline at end of file
diff --git a/plugins/CECrowPlugin/src/Parsing/IML/ImlTokenType.cs b/plugins/CECrowPlugin/src/Parsing/IML/ImlTokenType.cs
index 52849a5..ee0eaea 100644
--- a/plugins/CECrowPlugin/src/Parsing/IML/ImlTokenType.cs
+++ b/plugins/CECrowPlugin/src/Parsing/IML/ImlTokenType.cs
@@ -36,6 +36,7 @@ namespace CECrowPlugin
 		AttributeValue			= 0x2000,
 		AttributeValueOpen		= 0x2001,
 		AttributeValueClose		= 0x2002,
+		UnexpectedChar			= 0x8000,
 		Content,
 		BindingOpen 			= 0xA401,// '{'
 		BindingClose			= 0xA402,// '}'
diff --git a/plugins/CEXmlPlugin/src/Parsing/XmlSyntaxAnalyser.cs b/plugins/CEXmlPlugin/src/Parsing/XmlSyntaxAnalyser.cs
index 9832636..587fbb9 100644
--- a/plugins/CEXmlPlugin/src/Parsing/XmlSyntaxAnalyser.cs
+++ b/plugins/CEXmlPlugin/src/Parsing/XmlSyntaxAnalyser.cs
@@ -131,8 +131,7 @@ namespace CrowEdit.Xml
 						elt = processElement(new ElementSyntax(start));
 						break;
 					}
-					//wouldd be better in tokenizer, and break on unexpected tok
-					if (Peek().Is(XmlTokenType.ElementOpen) || Peek().Is(XmlTokenType.EndElementOpen)) {
+					if (Peek().Is(XmlTokenType.UnexpectedChar)) {
 						start.AddChild(new UnexpectedTokenSyntax(Read()));
 						break;
 					}
diff --git a/plugins/CEXmlPlugin/src/Parsing/XmlTokenType.cs b/plugins/CEXmlPlugin/src/Parsing/XmlTokenType.cs
index 4cbd244..a96c7d4 100644
--- a/plugins/CEXmlPlugin/src/Parsing/XmlTokenType.cs
+++ b/plugins/CEXmlPlugin/src/Parsing/XmlTokenType.cs
@@ -37,7 +37,7 @@ namespace CrowEdit.Xml
 		AttributeValueOpen		= 0x2001,
 		AttributeValueClose		= 0x2002,
 
-
+		UnexpectedChar			= 0x8000,
 		Content,
 	}
 }
\ No newline at end of file
diff --git a/plugins/CEXmlPlugin/src/Parsing/XmlTokenizer.cs b/plugins/CEXmlPlugin/src/Parsing/XmlTokenizer.cs
index 6fbf00d..400d5a8 100644
--- a/plugins/CEXmlPlugin/src/Parsing/XmlTokenizer.cs
+++ b/plugins/CEXmlPlugin/src/Parsing/XmlTokenizer.cs
@@ -113,27 +113,36 @@ namespace CrowEdit.Xml
 								curState = States.DTDObject;
 							}
 						}
-					} else if (reader.TryPeek('/')) {
-						reader.Advance ();
-						addTok (ref reader, XmlTokenType.EndElementOpen);
-						if (readName (ref reader)) {
-							addTok (ref reader, XmlTokenType.ElementName);
-							if (reader.TryPeek('>')) {
-								reader.Advance ();
-								addTok (ref reader, XmlTokenType.ClosingSign);
-
-								if (--curObjectLevel > 0)
-									curState = States.Content;
-								else
-									curState = States.Xml;
+					} else {
+						bool levelup = false;
+						if (curState == States.StartTag) {
+							addTok (ref reader, XmlTokenType.UnexpectedChar);
+							levelup = true;
+						}
+						if (reader.TryPeek('/')) {
+							reader.Advance ();
+							addTok (ref reader, XmlTokenType.EndElementOpen);
+							if (readName (ref reader)) {
+								addTok (ref reader, XmlTokenType.ElementName);
+								if (reader.TryPeek('>')) {
+									reader.Advance ();
+									addTok (ref reader, XmlTokenType.ClosingSign);
+									levelup = true;
+								}
+							}
+						} else {
+							addTok (ref reader, XmlTokenType.ElementOpen);
+							if (readName (ref reader)) {
+								addTok (ref reader, XmlTokenType.ElementName);
 							}
+							curState = States.StartTag;
 						}
-					}else{
-						addTok (ref reader, XmlTokenType.ElementOpen);
-						if (readName (ref reader)) {
-							addTok (ref reader, XmlTokenType.ElementName);
+						if (levelup) {
+							if (--curObjectLevel > 0)
+								curState = States.Content;
+							else
+								curState = States.Xml;
 						}
-						curState = States.StartTag;
 					}
 					break;
 				case '?':
-- 
2.47.3