From: Mikko Rasa <tdb@tdb.fi>
Date: Sun, 14 Feb 2021 17:56:15 +0000 (+0200)
Subject: Follow the GLSL syntax rules more closely
X-Git-Url: http://git.tdb.fi/?p=libs%2Fgl.git;a=commitdiff_plain;h=c1ec86d6b9ceb6c71f787fed3b2ea6c75457a474

Follow the GLSL syntax rules more closely

Fix a bug which allowed preprocessor directives anywhere on a line,
reject characters which are not allowed in GLSL, and parse numeric
literals more robustly.
---

diff --git a/source/glsl/glsl_error.cpp b/source/glsl/glsl_error.cpp
index ea8e9306..aff897ca 100644
--- a/source/glsl/glsl_error.cpp
+++ b/source/glsl/glsl_error.cpp
@@ -11,6 +11,10 @@ invalid_shader_source::invalid_shader_source(const Location &loc, const string &
 	runtime_error(format("%s:%d: %s", loc.name, loc.line, message))
 { }
 
+syntax_error::syntax_error(const Location &loc, const string &token, const string &message):
+	invalid_shader_source(loc, "Syntax error near '%s': %s", token, message)
+{ }
+
 parse_error::parse_error(const Location &loc, const string &token, const string &expected):
 	invalid_shader_source(loc, "Parse error at '%s': expected %s", token, expected)
 { }
diff --git a/source/glsl/glsl_error.h b/source/glsl/glsl_error.h
index 63c49bbd..1ae521ca 100644
--- a/source/glsl/glsl_error.h
+++ b/source/glsl/glsl_error.h
@@ -23,6 +23,13 @@ public:
 	virtual ~invalid_shader_source() throw() { }
 };
 
+class syntax_error: public invalid_shader_source
+{
+public:
+	syntax_error(const Location &, const std::string &, const std::string &);
+	virtual ~syntax_error() throw() { }
+};
+
 class parse_error: public invalid_shader_source
 {
 public:
diff --git a/source/glsl/preprocessor.cpp b/source/glsl/preprocessor.cpp
index 3c190221..bf2d4b4d 100644
--- a/source/glsl/preprocessor.cpp
+++ b/source/glsl/preprocessor.cpp
@@ -14,8 +14,6 @@ Preprocessor::Preprocessor(Tokenizer &t):
 
 void Preprocessor::preprocess()
 {
-	tokenizer.expect("#");
-
 	string token = tokenizer.peek_token();
 	if(token=="pragma")
 		preprocess_pragma();
diff --git a/source/glsl/tokenizer.cpp b/source/glsl/tokenizer.cpp
index be432143..e73afd52 100644
--- a/source/glsl/tokenizer.cpp
+++ b/source/glsl/tokenizer.cpp
@@ -61,17 +61,21 @@ string Tokenizer::parse_token_()
 	while(1)
 	{
 		skip_comment_and_whitespace();
+		bool allow_preproc = allow_preprocess;
+		allow_preprocess = false;
 		if(iter==source_end)
 			return string();
-		else if(allow_preprocess && *iter=='#')
+		else if(allow_preproc && *iter=='#')
 		{
-			allow_preprocess = false;
+			++iter;
 			preprocess();
 		}
 		else if(isalpha(*iter) || *iter=='_')
 			return parse_identifier();
 		else if(isdigit(*iter))
 			return parse_number();
+		else if(*iter=='"' || *iter=='#' || *iter=='$' || *iter=='\'' || *iter=='@' || *iter=='\\' || *iter=='`')
+			throw syntax_error(location, string(1, *iter), "Invalid character in source");
 		else
 			return parse_other();
 	}
@@ -106,23 +110,45 @@ string Tokenizer::parse_identifier()
 
 string Tokenizer::parse_number()
 {
-	bool accept_sign = false;
+	bool got_fract = false;
 	string number;
 	while(iter!=source_end)
 	{
-		if(isdigit(*iter) || *iter=='.')
+		if(isdigit(*iter))
 			number += *iter++;
-		else if(*iter=='e' || *iter=='E')
+		else if(!got_fract && *iter=='.')
 		{
 			number += *iter++;
-			accept_sign = true;
+			got_fract = true;
 		}
-		else if(accept_sign && (*iter=='+' || *iter=='-'))
-			number += *iter++;
 		else
 			break;
 	}
 
+	bool require_digit = false;
+	if(iter!=source_end && (*iter=='e' || *iter=='E'))
+	{
+		number += *iter++;
+		if(iter!=source_end && (*iter=='-' || *iter=='+'))
+			number += *iter++;
+		require_digit = true;
+		while(iter!=source_end)
+		{
+			if(isdigit(*iter))
+			{
+				number += *iter++;
+				require_digit = false;
+			}
+			else
+				break;
+		}
+	}
+
+	if(require_digit)
+		throw syntax_error(location, number, "Incomplete numeric literal");
+	else if(isalnum(*iter) || *iter=='_')
+		throw syntax_error(location, number, "Garbage at end of numeric literal");
+
 	return number;
 }