Merge pull request #5033 from jix/liberty-fixes

liberty: More robust parsing

Author: jix

passes/techmap/libparse.cc

@@ -75,7 +75,7 @@ bool LibertyInputStream::extend_buffer_once()
 		buffer.resize(buf_end + chunk_size);
 	}
 
-	size_t read_size = f.rdbuf()->sgetn(buffer.data() + buf_end, chunk_size);
+	size_t read_size = f.rdbuf()->sgetn((char *)buffer.data() + buf_end, chunk_size);
 	buf_end += read_size;
 	if (read_size < chunk_size)
 		eof = true;
@@ -436,6 +436,9 @@ void LibertyParser::report_unexpected_token(int tok)
 		eReport += "'.";
 		error(eReport);
 		break;
+	case EOF:
+		error("Unexpected end of file");
+		break;
 	default:
 		eReport = "Unexpected token: ";
 		eReport += static_cast<char>(tok);
@@ -484,7 +487,7 @@ void LibertyParser::parse_vector_range(int tok)
 	}
 }
 
-LibertyAst *LibertyParser::parse()
+LibertyAst *LibertyParser::parse(bool top_level)
 {
 	std::string str;
 
@@ -498,7 +501,13 @@ LibertyAst *LibertyParser::parse()
 	while ((tok == 'n') || (tok == ';'))
 		tok = lexer(str);
 
-	if (tok == '}' || tok < 0)
+	if (tok == EOF) {
+		if (top_level)
+			return NULL;
+		report_unexpected_token(tok);
+	}
+
+	if (tok == '}')
 		return NULL;
 
 	if (tok != 'v') {
@@ -571,12 +580,18 @@ LibertyAst *LibertyParser::parse()
 		}
 
 		if (tok == '{') {
+			bool terminated = false;
 			while (1) {
-				LibertyAst *child = parse();
-				if (child == NULL)
+				LibertyAst *child = parse(false);
+				if (child == NULL) {
+					terminated = true;
 					break;
+				}
 				ast->children.push_back(child);
 			}
+			if (!terminated) {
+				report_unexpected_token(EOF);
+			}
 			break;
 		}
 

passes/techmap/libparse.h

@@ -92,7 +92,7 @@ namespace Yosys
 
 	class LibertyInputStream {
 		std::istream &f;
-		std::vector<char> buffer;
+		std::vector<unsigned char> buffer;
 		size_t buf_pos = 0;
 		size_t buf_end = 0;
 		bool eof = false;
@@ -107,7 +107,7 @@ namespace Yosys
 		LibertyInputStream(std::istream &f) : f(f) {}
 
 		size_t buffered_size() { return buf_end - buf_pos; }
-		const char *buffered_data() { return buffer.data() + buf_pos; }
+		const unsigned char *buffered_data() { return buffer.data() + buf_pos; }
 
 		int get() {
 			if (buf_pos == buf_end)
@@ -165,7 +165,7 @@ namespace Yosys
 
 		void report_unexpected_token(int tok);
 		void parse_vector_range(int tok);
-		LibertyAst *parse();
+		LibertyAst *parse(bool top_level);
 		void error() const;
 		void error(const std::string &str) const;
 
@@ -174,18 +174,29 @@ namespace Yosys
 		const LibertyAst *ast = nullptr;
 
 		LibertyParser(std::istream &f) : f(f), line(1) {
-			shared_ast.reset(parse());
+			shared_ast.reset(parse(true));
 			ast = shared_ast.get();
+			if (!ast) {
+#ifdef FILTERLIB
+				fprintf(stderr, "No entries found in liberty file.\n");
+				exit(1);
+#else
+				log_error("No entries found in liberty file.\n");
+#endif
+			}
 		}
 
 #ifndef FILTERLIB
 		LibertyParser(std::istream &f, const std::string &fname) : f(f), line(1) {
 			shared_ast = LibertyAstCache::instance.cached_ast(fname);
 			if (!shared_ast) {
-				shared_ast.reset(parse());
+				shared_ast.reset(parse(true));
 				LibertyAstCache::instance.parsed_ast(fname, shared_ast);
 			}
 			ast = shared_ast.get();
+			if (!ast) {
+				log_error("No entries found in liberty file `%s'.\n", fname.c_str());
+			}
 		}
 #endif
 	};

tests/liberty/non-ascii.lib

@@ -0,0 +1,14 @@
+// The parser used to choke on the copyright symbol even in comments
+// © ® ø Φ
+library(dummy) {
+  cell(buffer) {
+    area : 1 ;
+    pin(A) {
+      direction : input ;
+    }
+    pin(Y) {
+      direction : output ;
+      function : "A" ;
+    }
+  }
+}

tests/liberty/non-ascii.lib.filtered.ok

@@ -0,0 +1,12 @@
+library(dummy) {
+  cell(buffer) {
+    area : 1 ;
+    pin(A) {
+      direction : input ;
+    }
+    pin(Y) {
+      direction : output ;
+      function : "A" ;
+    }
+  }
+}

tests/liberty/non-ascii.lib.verilogsim.ok

@@ -0,0 +1,5 @@
+module buffer (A, Y);
+  input A;
+  output Y;
+  assign Y = A; // "A"
+endmodule