diff --git a/aegisub/src/subs_controller.cpp b/aegisub/src/subs_controller.cpp
index d4ca914ed..fd35d0070 100644
--- a/aegisub/src/subs_controller.cpp
+++ b/aegisub/src/subs_controller.cpp
@@ -83,9 +83,6 @@ SubsController::SubsController(agi::Context *context)
 }
 
 void SubsController::Load(agi::fs::path const& filename, std::string charset) {
-	// TextFileReader does this automatically, but relying on that results in
-	// the user being prompted twice if it can't be auto-detected, since we
-	// open the file twice below.
 	try {
 		if (charset.empty())
 			charset = CharSetDetect::GetEncoding(filename);
diff --git a/aegisub/src/subtitle_format_microdvd.cpp b/aegisub/src/subtitle_format_microdvd.cpp
index bf2756bfe..6f9c13b46 100644
--- a/aegisub/src/subtitle_format_microdvd.cpp
+++ b/aegisub/src/subtitle_format_microdvd.cpp
@@ -39,6 +39,7 @@
 #include "ass_dialogue.h"
 #include "ass_file.h"
 #include "ass_time.h"
+#include "charset_detect.h"
 #include "text_file_reader.h"
 #include "text_file_writer.h"
 #include "video_context.h"
@@ -74,7 +75,10 @@ bool MicroDVDSubtitleFormat::CanReadFile(agi::fs::path const& filename) const {
 	if (!agi::fs::HasExtension(filename, "sub")) return false;
 
 	// Since there is an infinity of .sub formats, load first line and check if it's valid
-	TextFileReader file(filename);
+	auto encoding = CharSetDetect::GetEncoding(filename);
+	if (encoding == "binary") return false;
+
+	TextFileReader file(filename, encoding);
 	if (file.HasMoreLines())
 		return regex_match(file.ReadLineFromFile(), line_regex);
 
diff --git a/aegisub/src/text_file_reader.cpp b/aegisub/src/text_file_reader.cpp
index 5b5d61e5c..a70d59e5f 100644
--- a/aegisub/src/text_file_reader.cpp
+++ b/aegisub/src/text_file_reader.cpp
@@ -23,8 +23,6 @@
 
 #include "text_file_reader.h"
 
-#include "charset_detect.h"
-
 #include <libaegisub/io.h>
 
 #include <algorithm>
@@ -32,12 +30,10 @@
 #include <boost/algorithm/string/trim.hpp>
 
 TextFileReader::TextFileReader(agi::fs::path const& filename, std::string encoding, bool trim)
-: trim(trim)
+: file(agi::io::Open(filename, true))
+, trim(trim)
+, iter(agi::line_iterator<std::string>(*file, encoding))
 {
-	if (encoding.empty())
-		encoding = CharSetDetect::GetEncoding(filename);
-	file.reset(agi::io::Open(filename, true));
-	iter = agi::line_iterator<std::string>(*file, encoding);
 }
 
 TextFileReader::~TextFileReader() {
diff --git a/aegisub/src/text_file_reader.h b/aegisub/src/text_file_reader.h
index 9b54cf3cf..2a34c86cb 100644
--- a/aegisub/src/text_file_reader.h
+++ b/aegisub/src/text_file_reader.h
@@ -43,7 +43,7 @@ public:
 	/// @param filename File to open
 	/// @param enc      Encoding to use, or empty to autodetect
 	/// @param trim     Whether to trim whitespace from lines read
-	TextFileReader(agi::fs::path const& filename, std::string encoding="", bool trim=true);
+	TextFileReader(agi::fs::path const& filename, std::string encoding, bool trim=true);
 	/// @brief Destructor
 	~TextFileReader();