Introduce a feature in TextFileReader that checks whether a file seems to be binary rather than text.

Use this to make the MicroDVD (.sub) format reader check that it isn't being fed something that might actually be VobSub (.sub+.idx) which makes Aegisub crash.
Hopefully this doesn't break anything else.
Fixes #1351.

Originally committed to SVN as r5773.

aegisub/src/subtitle_format_microdvd.cpp

@@ -78,7 +78,7 @@ bool MicroDVDSubtitleFormat::CanReadFile(wxString filename) {
 
 	// Since there is an infinity of .sub formats, load first line and check if it's valid
 	TextFileReader file(filename);
-	if (file.HasMoreLines()) {
+	if (file.HasMoreLines() && !file.SmellsBinary()) {
 		wxRegEx exp(_T("^[\\{\\[]([0-9]+)[\\}\\]][\\{\\[]([0-9]+)[\\}\\]](.*)$"),wxRE_ADVANCED);
 		return exp.Matches(file.ReadLineFromFile());
 	}
@@ -94,13 +94,18 @@ bool MicroDVDSubtitleFormat::CanWriteFile(wxString filename) {
 }
 
 
+DEFINE_SIMPLE_EXCEPTION(MicrodvdFormatParseError, SubtitleFormatParseError, "subtitle_io/parse/microdvd")
+
 ///////////////
 // Read a file
 void MicroDVDSubtitleFormat::ReadFile(wxString filename,wxString forceEncoding) {
 	// Load and prepare regexp
-	TextFileReader file(filename);
+	TextFileReader file(filename, forceEncoding);
 	wxRegEx exp(_T("^[\\{\\[]([0-9]+)[\\}\\]][\\{\\[]([0-9]+)[\\}\\]](.*)$"),wxRE_ADVANCED);
 
+	if (file.SmellsBinary())
+		throw new MicrodvdFormatParseError(_T("File seems to be binary. Maybe it's a VobSub format file? (Aegisub does not support VobSub format.)"), 0);
+
 	// Load default
 	LoadDefault(false);
 

aegisub/src/text_file_reader.cpp

@@ -307,6 +307,21 @@ void TextFileReader::Open() {
 	}
 #endif
 	open = true;
+
+	// Check if file seems binary
+	size_t binaryFactor = 0;
+	const size_t bufsize = 512;
+	char buf[bufsize];
+	file.get(buf, bufsize);
+	size_t bytes_read = file.gcount();
+	file.seekg(0, std::ios_base::beg);
+	for (size_t i = 0; i < bytes_read; ++i) {
+		if (((unsigned char)buf[i]) < 32) {
+			if (buf[i] != '\r' && buf[i] != '\n' && buf[i] != '\t')
+				binaryFactor++;
+		}
+	}
+	isBinary = (binaryFactor > 8) || (binaryFactor > bytes_read/8);
 }
 
 
@@ -352,3 +367,4 @@ void TextFileReader::EnsureValid(wxString enc) {
 wxString TextFileReader::GetCurrentEncoding() {
 	return encoding;
 }
+

aegisub/src/text_file_reader.h

@@ -66,6 +66,7 @@ private:
 	bool open;
 	bool customConv;
 	bool trim;
+	bool isBinary;
 
 	void Open();
 	void Close();
@@ -81,6 +82,7 @@ public:
 	static void EnsureValid(const wxString encoding);
 	wxString GetCurrentEncoding();
 	static wxString GetEncoding(const wxString filename);
+	bool SmellsBinary() { return !Is16 && isBinary; }
 };