From: Mikko Rasa <tdb@tdb.fi>
Date: Sun, 28 Feb 2021 00:12:48 +0000 (+0200)
Subject: Make the GLSL parser resilient against common errors
X-Git-Url: http://git.tdb.fi/?p=libs%2Fgl.git;a=commitdiff_plain;h=7b73b63df12b3ace4231842aa291d6e1d7b3f948

Make the GLSL parser resilient against common errors

It will now attempt to continue parsing the source, with the intent of
reporting all syntax errors at once.
---

diff --git a/source/glsl/glsl_error.h b/source/glsl/glsl_error.h
index 46ca49be..1b005e0f 100644
--- a/source/glsl/glsl_error.h
+++ b/source/glsl/glsl_error.h
@@ -13,6 +13,7 @@ struct Location;
 class invalid_shader_source: public std::runtime_error
 {
 public:
+	invalid_shader_source(const std::string &w): runtime_error(w) { }
 	invalid_shader_source(const Location &, const std::string &);
 #if __cplusplus>=201103L
 	template<typename... Args>
diff --git a/source/glsl/parser.cpp b/source/glsl/parser.cpp
index 5ffc2a9b..d9fdbc4e 100644
--- a/source/glsl/parser.cpp
+++ b/source/glsl/parser.cpp
@@ -1,3 +1,4 @@
+#include <msp/core/raii.h>
 #include <msp/strings/format.h>
 #include <msp/strings/regex.h>
 #include <msp/strings/utils.h>
@@ -57,8 +58,13 @@ void Parser::parse_source(const string &name, unsigned index)
 	source_index = index;
 	source_reference(1, name);
 	tokenizer.begin(name, source);
-	while(RefPtr<Statement> statement = parse_global_declaration())
-		cur_stage->content.body.push_back(statement);
+	allow_stage_change = true;
+	while(!tokenizer.peek_token().empty())
+		if(RefPtr<Statement> statement = parse_with_recovery(&Parser::parse_global_declaration))
+			cur_stage->content.body.push_back(statement);
+
+	if(!errors.empty())
+		throw invalid_shader_source(join(errors.begin(), errors.end(), "\n"));
 }
 
 void Parser::set_required_version(const Version &ver)
@@ -181,11 +187,54 @@ bool Parser::is_identifier(const string &token)
 	return re.match(token);
 }
 
+template<typename T>
+RefPtr<T> Parser::parse_with_recovery(RefPtr<T> (Parser::*parse_func)())
+{
+	tokenizer.clear_progress_mark();
+	try
+	{
+		return (this->*parse_func)();
+	}
+	catch(const invalid_shader_source &exc)
+	{
+		errors.push_back(exc.what());
+	}
+
+	if(tokenizer.get_last_token()!=";" || !tokenizer.get_progress_mark())
+	{
+		unsigned scope_level = 0;
+		while(1)
+		{
+			if(tokenizer.peek_token()=="}" && scope_level==0)
+			{
+				if(!tokenizer.get_progress_mark())
+					tokenizer.parse_token();
+				break;
+			}
+
+			string token = tokenizer.parse_token();
+			if(token=="}")
+			{
+				--scope_level;
+				if(scope_level==0)
+					break;
+			}
+			else if(token=="{")
+				++scope_level;
+			else if(token==";" && scope_level==0)
+				break;
+			else if(token.empty())
+				break;
+		}
+	}
+
+	return RefPtr<T>();
+}
+
 RefPtr<Statement> Parser::parse_global_declaration()
 {
-	allow_stage_change = true;
 	string token = tokenizer.peek_token();
-	allow_stage_change = false;
+	SetFlag disallow(allow_stage_change, false);
 
 	if(token=="import")
 		return parse_import();
@@ -262,6 +311,11 @@ RefPtr<Statement> Parser::parse_statement()
 	}
 	else if(is_qualifier(token) || is_type(token))
 		return parse_variable_declaration();
+	else if(token==";")
+	{
+		tokenizer.parse_token();
+		throw invalid_shader_source(tokenizer.get_location(), "Empty statement not allowed");
+	}
 	else if(!token.empty())
 	{
 		RefPtr<ExpressionStatement> expr = new ExpressionStatement;
@@ -357,7 +411,8 @@ void Parser::parse_block(Block &block, bool require_braces, RefPtr<T> (Parser::*
 	if(have_braces)
 	{
 		while(tokenizer.peek_token()!="}")
-			block.body.push_back((this->*parse_content)());
+			if(RefPtr<Statement> node = parse_with_recovery(parse_content))
+				block.body.push_back(node);
 	}
 	else
 		block.body.push_back((this->*parse_content)());
diff --git a/source/glsl/parser.h b/source/glsl/parser.h
index 9ab1232a..acfb7d9e 100644
--- a/source/glsl/parser.h
+++ b/source/glsl/parser.h
@@ -24,6 +24,7 @@ private:
 	Module *module;
 	Stage *cur_stage;
 	std::set<std::string> declared_types;
+	std::vector<std::string> errors;
 
 public:
 	Parser();
@@ -60,6 +61,8 @@ private:
 	void preprocess_stage();
 
 	RefPtr<Statement> parse_global_declaration();
+	template<typename T>
+	RefPtr<T> parse_with_recovery(RefPtr<T> (Parser::*)());
 	RefPtr<Statement> parse_statement();
 	RefPtr<Import> parse_import();
 	RefPtr<Precision> parse_precision();
diff --git a/source/glsl/tokenizer.cpp b/source/glsl/tokenizer.cpp
index b7efb8f8..6ba139ff 100644
--- a/source/glsl/tokenizer.cpp
+++ b/source/glsl/tokenizer.cpp
@@ -34,11 +34,13 @@ const string &Tokenizer::peek_token(unsigned index)
 {
 	while(next_tokens.size()<=index)
 		next_tokens.push_back(parse_token_());
-	return (last_token = next_tokens[index]);
+	return next_tokens[index];
 }
 
 const string &Tokenizer::parse_token()
 {
+	progress_mark = true;
+
 	if(!next_tokens.empty())
 	{
 		last_token = next_tokens.front();
diff --git a/source/glsl/tokenizer.h b/source/glsl/tokenizer.h
index 20ea756a..46aa62fb 100644
--- a/source/glsl/tokenizer.h
+++ b/source/glsl/tokenizer.h
@@ -29,6 +29,7 @@ private:
 	std::string::const_iterator iter;
 	std::string::const_iterator source_end;
 	Location location;
+	bool progress_mark;
 	bool allow_preprocess;
 	bool suppress_line_advance;
 	std::string last_token;
@@ -40,8 +41,11 @@ public:
 	void begin(const std::string &, const std::string &);
 	const std::string &peek_token(unsigned = 0);
 	const std::string &parse_token();
+	const std::string &get_last_token() const { return last_token; }
 	void expect(const std::string &);
 	void set_location(const Location &);
+	void clear_progress_mark() { progress_mark = false; }
+	bool get_progress_mark() const { return progress_mark; }
 	const Location &get_location() const { return location; }
 private:
 	std::string parse_token_();