Simplify charset detection

Originally committed to SVN as r4419.

aegisub/src/ass_file.cpp

@@ -50,6 +50,7 @@
 #include "ass_file.h"
 #include "ass_override.h"
 #include "ass_style.h"
+#include "charset_detect.h"
 #include "compat.h"
 #include "main.h"
 #include "options.h"
@@ -74,7 +75,7 @@ AssFile::~AssFile() {
 /// @param file
 /// @param charset     
 /// @param addToRecent 
-void AssFile::Load (const wxString _filename,const wxString charset,bool addToRecent) {
+void AssFile::Load (const wxString &_filename,wxString charset,bool addToRecent) {
 	bool ok = true;
 
 	try {
@@ -91,9 +92,9 @@ void AssFile::Load (const wxString _filename,const wxString charset,bool addToRe
 		fclose(file);
 
 		// Find file encoding
-		wxString enc;
-		if (charset.IsEmpty()) enc = TextFileReader::GetEncoding(_filename);
-		else enc = charset;
+		if (charset.empty()) {
+			charset = CharSetDetect::GetEncoding(_filename);
+		}
 
 		// Generic preparation
 		Clear();
@@ -104,7 +105,7 @@ void AssFile::Load (const wxString _filename,const wxString charset,bool addToRe
 		// Read file
 		if (reader) {
 			reader->SetTarget(this);
-			reader->ReadFile(_filename,enc);
+			reader->ReadFile(_filename,charset);
 		}
 
 		// Couldn't find a type

aegisub/src/ass_file.h

@@ -109,7 +109,7 @@ public:
 	AssStyle *GetStyle(wxString name);					// Gets style by its name
 
 	//wxString GetString();								// Returns the whole file as a single string
-	void Load(wxString file,wxString charset=_T(""),bool addToRecent=true);	// Load from a file
+	void Load(const wxString &file,wxString charset=_T(""),bool addToRecent=true);	// Load from a file
 	void Save(wxString file,bool setfilename=false,bool addToRecent=true,const wxString encoding=_T(""));	// Save to a file. Pass true to second argument if this isn't a copy
 	void SaveMemory(std::vector<char> &dst,const wxString encoding=_T(""));	// Save to a memory string
 	void Export(wxString file);							// Saves exported copy, with effects applied

aegisub/src/charset_detect.cpp

@@ -34,9 +34,6 @@
 /// @ingroup utility
 ///
 
-
-///////////
-// Headers
 #include "config.h"
 
 #ifndef AGI_PRE
@@ -52,16 +49,11 @@
 #include <libaegisub/log.h>
 
 #include "charset_detect.h"
-#include "text_file_reader.h"
 #include "compat.h"
 
+namespace CharSetDetect {
 
-
-/// @brief Get encoding 
-/// @param filename 
-/// @return 
-///
-wxString CharSetDetect::GetEncoding(wxString filename) {
+wxString GetEncoding(wxString const& filename) {
 	LOG_I("charset/file") << filename;
 	bool unknown = 0;
 
@@ -70,7 +62,7 @@ wxString CharSetDetect::GetEncoding(wxString filename) {
 
 	try {
 		agi::charset::DetectAll(STD_STR(filename), list);
-    } catch (const agi::charset::UnknownCharset&) {
+	} catch (const agi::charset::UnknownCharset&) {
 		unknown = 1;
 	}
 
@@ -93,3 +85,5 @@ wxString CharSetDetect::GetEncoding(wxString filename) {
 	return i_lst->second;
 }
 
+}
+

aegisub/src/charset_detect.h

@@ -34,17 +34,9 @@
 /// @ingroup utility
 ///
 
-/// DOCME
-/// @class CharSetDetect
-/// @brief Detect character set of a file
-class CharSetDetect {
-private:
-	/// Character set
-	wxString result;
-
-public:
+namespace CharSetDetect {
 	/// @brief Get character set name.
 	/// @param filename File to check
 	/// @return Character set name
-	wxString GetEncoding(wxString filename);
-};
+	wxString GetEncoding(wxString const& filename);
+}

aegisub/src/text_file_reader.cpp

@@ -48,9 +48,7 @@
 #include <libaegisub/log.h>
 
 #include "charset_conv.h"
-#ifdef WITH_UNIVCHARDET
 #include "charset_detect.h"
-#endif
 #include "text_file_reader.h"
 
 TextFileReader::TextFileReader(wxString filename, wxString enc, bool trim)
@@ -62,7 +60,7 @@ TextFileReader::TextFileReader(wxString filename, wxString enc, bool trim)
 #endif
 	if (!file.is_open()) throw L"Failed opening file for reading.";
 
-	if (encoding.IsEmpty()) encoding = GetEncoding(filename);
+	if (encoding.IsEmpty()) encoding = CharSetDetect::GetEncoding(filename);
 	if (encoding == L"binary") return;
 	encoding = AegisubCSConv::GetRealEncodingName(encoding);
 	conv = iconv_open(WCHAR_T_ENCODING, encoding.ToAscii());
@@ -75,15 +73,6 @@ TextFileReader::~TextFileReader() {
 	if (conv != (iconv_t)-1) iconv_close(conv);
 }
 
-wxString TextFileReader::GetEncoding(wxString const& filename) {
-
-	// Use universalchardet library to detect charset
-	CharSetDetect det;
-	wxString str(det.GetEncoding(filename));
-	LOG_I("file/reader/text/encoding") << str;
-	return str;
-}
-
 wchar_t TextFileReader::GetWChar() {
 	// If there's already some converted characters waiting, return the next one
 	if (++currout < outptr) {

aegisub/src/text_file_reader.h

@@ -89,9 +89,4 @@ public:
 	/// @brief Get the file encoding used by this reader
 	/// @return "unknown", "binary", or a character encoding name
 	wxString GetCurrentEncoding();
-
-	/// @brief Attempt to detect a file's encoding
-	/// @param filename The file to check
-	/// @return "unknown", "binary", or a character encoding name
-	static wxString GetEncoding(wxString const& filename);
 };