Fixed bug reported in "parseSignals should fail if frame data length is too small"

Fixes #29

Author: SimonCahill

include/libdbc/message.hpp

@@ -19,6 +19,7 @@ struct Message {
 		ErrorBigEndian,
 		ErrorUnknownID,
 		ErrorInvalidConversion,
+		ErrorInvalidSignalSize,
 	};
 
 	ParseSignalsStatus parse_signals(const std::vector<uint8_t>& data, std::vector<double>& values) const;

src/message.cpp

@@ -1,3 +1,5 @@
+#include <arpa/inet.h>
+
 #include <cstddef>
 #include <cstdint>
 #include <libdbc/message.hpp>
@@ -35,15 +37,20 @@ Message::ParseSignalsStatus Message::parse_signals(const std::vector<uint8_t>& d
 	uint64_t data_little_endian = 0;
 	uint64_t data_big_endian = 0;
 	for (std::size_t i = 0; i < size; i++) {
-		data_little_endian |= ((uint64_t)data[i]) << i * ONE_BYTE;
 		data_big_endian = (data_big_endian << ONE_BYTE) | (uint64_t)data[i];
 	}
+	data_little_endian = ntohl(data_little_endian);
 
 	// TODO: does this also work on a big endian machine?
 
 	const auto len = size * 8;
 	uint64_t value = 0;
 	for (const auto& signal : m_signals) {
+
+		if (signal.size > len) {
+			return ParseSignalsStatus::ErrorInvalidSignalSize;
+		}
+
 		if (signal.is_bigendian) {
 			uint32_t start_bit = ONE_BYTE * (signal.start_bit / ONE_BYTE) + (SEVEN_BITS - (signal.start_bit % ONE_BYTE)); // Calculation taken from python CAN
 			value = data_big_endian << start_bit;