Leave out characters the filesystem cannot encode (Closes: #264)

devscripts/posix-locale.sh

@@ -0,0 +1,6 @@
+
+# source this file in your shell to get a POSIX locale (which will break many programs, but that's kind of the point)
+
+export LC_ALL=POSIX
+export LANG=POSIX
+export LANGUAGE=POSIX

youtube_dl/__init__.py

@@ -259,14 +259,14 @@ def sanitize_open(filename, open_mode):
 				import msvcrt
 				msvcrt.setmode(sys.stdout.fileno(), os.O_BINARY)
 			return (sys.stdout, filename)
-		stream = open(filename, open_mode)
+		stream = open(_encodeFilename(filename), open_mode)
 		return (stream, filename)
 	except (IOError, OSError), err:
 		# In case of error, try to remove win32 forbidden chars
 		filename = re.sub(ur'[/<>:"\|\?\*]', u'#', filename)
 
 		# An exception here should be caught in the caller
-		stream = open(filename, open_mode)
+		stream = open(_encodeFilename(filename), open_mode)
 		return (stream, filename)
 
 
@@ -299,6 +299,14 @@ def _unescapeHTML(s):
 	htmlParser = HTMLParser.HTMLParser()
 	return htmlParser.unescape(s)
 
+def _encodeFilename(s):
+	"""
+	@param s The name of the file (of type unicode)
+	"""
+
+	assert type(s) == type(u'')
+	return s.encode(sys.getfilesystemencoding(), 'ignore')
+
 class DownloadError(Exception):
 	"""Download Error exception.
 
@@ -563,16 +571,17 @@ class FileDownloader(object):
 		self._pps.append(pp)
 		pp.set_downloader(self)
 
-	def to_screen(self, message, skip_eol=False, ignore_encoding_errors=False):
+	def to_screen(self, message, skip_eol=False):
 		"""Print message to stdout if not in quiet mode."""
-		try:
+		assert type(message) == type(u'')
 		if not self.params.get('quiet', False):
 			terminator = [u'\n', u''][skip_eol]
-				print >>self._screen_file, (u'%s%s' % (message, terminator)).encode(preferredencoding()),
+			output = message + terminator
+
+			if 'b' not in self._screen_file.mode or sys.version_info[0] < 3: # Python 2 lies about the mode of sys.stdout/sys.stderr
+				output = output.encode(preferredencoding(), 'ignore')
+			self._screen_file.write(output)
 			self._screen_file.flush()
-		except (UnicodeEncodeError), err:
-			if not ignore_encoding_errors:
-				raise
 
 	def to_stderr(self, message):
 		"""Print message to stderr."""
@@ -622,7 +631,7 @@ class FileDownloader(object):
 	def temp_name(self, filename):
 		"""Returns a temporary filename for the given filename."""
 		if self.params.get('nopart', False) or filename == u'-' or \
-				(os.path.exists(filename) and not os.path.isfile(filename)):
+				(os.path.exists(_encodeFilename(filename)) and not os.path.isfile(_encodeFilename(filename))):
 			return filename
 		return filename + u'.part'
 
@@ -635,7 +644,7 @@ class FileDownloader(object):
 		try:
 			if old_filename == new_filename:
 				return
-			os.rename(old_filename, new_filename)
+			os.rename(_encodeFilename(old_filename), _encodeFilename(new_filename))
 		except (IOError, OSError), err:
 			self.trouble(u'ERROR: unable to rename file')
 
@@ -643,7 +652,7 @@ class FileDownloader(object):
 		"""Try to set the last-modified time of the given file."""
 		if last_modified_hdr is None:
 			return
-		if not os.path.isfile(filename):
+		if not os.path.isfile(_encodeFilename(filename)):
 			return
 		timestr = last_modified_hdr
 		if timestr is None:
@@ -659,15 +668,15 @@ class FileDownloader(object):
 
 	def report_writedescription(self, descfn):
 		""" Report that the description file is being written """
-		self.to_screen(u'[info] Writing video description to: %s' % descfn, ignore_encoding_errors=True)
+		self.to_screen(u'[info] Writing video description to: ' + descfn)
 
 	def report_writeinfojson(self, infofn):
 		""" Report that the metadata file has been written """
-		self.to_screen(u'[info] Video description metadata as JSON to: %s' % infofn, ignore_encoding_errors=True)
+		self.to_screen(u'[info] Video description metadata as JSON to: ' + infofn)
 
 	def report_destination(self, filename):
 		"""Report destination filename."""
-		self.to_screen(u'[download] Destination: %s' % filename, ignore_encoding_errors=True)
+		self.to_screen(u'[download] Destination: ' + filename)
 
 	def report_progress(self, percent_str, data_len_str, speed_str, eta_str):
 		"""Report download progress."""
@@ -769,8 +778,8 @@ class FileDownloader(object):
 			return
 
 		try:
-			dn = os.path.dirname(filename)
-			if dn != '' and not os.path.exists(dn):
+			dn = os.path.dirname(_encodeFilename(filename))
+			if dn != '' and not os.path.exists(dn): # dn is already encoded
 				os.makedirs(dn)
 		except (OSError, IOError), err:
 			self.trouble(u'ERROR: unable to create directory ' + unicode(err))
@@ -778,9 +787,9 @@ class FileDownloader(object):
 
 		if self.params.get('writedescription', False):
 			try:
-				descfn = filename + '.description'
+				descfn = filename + u'.description'
 				self.report_writedescription(descfn)
-				descfile = open(descfn, 'wb')
+				descfile = open(_encodeFilename(descfn), 'wb')
 				try:
 					descfile.write(info_dict['description'].encode('utf-8'))
 				finally:
@@ -790,7 +799,7 @@ class FileDownloader(object):
 				return
 
 		if self.params.get('writeinfojson', False):
-			infofn = filename + '.info.json'
+			infofn = filename + u'.info.json'
 			self.report_writeinfojson(infofn)
 			try:
 				json.dump
@@ -798,7 +807,7 @@ class FileDownloader(object):
 				self.trouble(u'ERROR: No JSON encoder found. Update to Python 2.6+, setup a json module, or leave out --write-info-json.')
 				return
 			try:
-				infof = open(infofn, 'wb')
+				infof = open(_encodeFilename(infofn), 'wb')
 				try:
 					json_info_dict = dict((k,v) for k,v in info_dict.iteritems() if not k in ('urlhandle',))
 					json.dump(json_info_dict, infof)
@@ -809,7 +818,7 @@ class FileDownloader(object):
 				return
 
 		if not self.params.get('skip_download', False):
-			if self.params.get('nooverwrites', False) and os.path.exists(filename):
+			if self.params.get('nooverwrites', False) and os.path.exists(_encodeFilename(filename)):
 				success = True
 			else:
 				try:
@@ -882,11 +891,11 @@ class FileDownloader(object):
 		basic_args = ['rtmpdump', '-q'] + [[], ['-W', player_url]][player_url is not None] + ['-r', url, '-o', tmpfilename]
 		retval = subprocess.call(basic_args + [[], ['-e', '-k', '1']][self.params.get('continuedl', False)])
 		while retval == 2 or retval == 1:
-			prevsize = os.path.getsize(tmpfilename)
+			prevsize = os.path.getsize(_encodeFilename(tmpfilename))
 			self.to_screen(u'\r[rtmpdump] %s bytes' % prevsize, skip_eol=True)
 			time.sleep(5.0) # This seems to be needed
 			retval = subprocess.call(basic_args + ['-e'] + [[], ['-k', '1']][retval == 1])
-			cursize = os.path.getsize(tmpfilename)
+			cursize = os.path.getsize(_encodeFilename(tmpfilename))
 			if prevsize == cursize and retval == 1:
 				break
 			 # Some rtmp streams seem abort after ~ 99.8%. Don't complain for those
@@ -895,7 +904,7 @@ class FileDownloader(object):
 				retval = 0
 				break
 		if retval == 0:
-			self.to_screen(u'\r[rtmpdump] %s bytes' % os.path.getsize(tmpfilename))
+			self.to_screen(u'\r[rtmpdump] %s bytes' % os.path.getsize(_encodeFilename(tmpfilename)))
 			self.try_rename(tmpfilename, filename)
 			return True
 		else:
@@ -907,7 +916,7 @@ class FileDownloader(object):
 		player_url = info_dict.get('player_url', None)
 
 		# Check file already present
-		if self.params.get('continuedl', False) and os.path.isfile(filename) and not self.params.get('nopart', False):
+		if self.params.get('continuedl', False) and os.path.isfile(_encodeFilename(filename)) and not self.params.get('nopart', False):
 			self.report_file_already_downloaded(filename)
 			return True
 
@@ -924,8 +933,8 @@ class FileDownloader(object):
 		request = urllib2.Request(url, None, headers)
 
 		# Establish possible resume length
-		if os.path.isfile(tmpfilename):
-			resume_len = os.path.getsize(tmpfilename)
+		if os.path.isfile(_encodeFilename(tmpfilename)):
+			resume_len = os.path.getsize(_encodeFilename(tmpfilename))
 		else:
 			resume_len = 0
 
@@ -4042,7 +4051,7 @@ class FFmpegExtractAudioPP(PostProcessor):
 	@staticmethod
 	def get_audio_codec(path):
 		try:
-			cmd = ['ffprobe', '-show_streams', '--', path]
+			cmd = ['ffprobe', '-show_streams', '--', _encodeFilename(path)]
 			handle = subprocess.Popen(cmd, stderr=file(os.path.devnull, 'w'), stdout=subprocess.PIPE)
 			output = handle.communicate()[0]
 			if handle.wait() != 0:
@@ -4063,7 +4072,7 @@ class FFmpegExtractAudioPP(PostProcessor):
 			acodec_opts = []
 		else:
 			acodec_opts = ['-acodec', codec]
-		cmd = ['ffmpeg', '-y', '-i', path, '-vn'] + acodec_opts + more_opts + ['--', out_path]
+		cmd = ['ffmpeg', '-y', '-i', _encodeFilename(path), '-vn'] + acodec_opts + more_opts + ['--', _encodeFilename(out_path)]
 		try:
 			p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
 			stdout,stderr = p.communicate()
@@ -4124,9 +4133,9 @@ class FFmpegExtractAudioPP(PostProcessor):
 				extension = 'wav'
 				more_opts += ['-f', 'wav']
 
-		(prefix, ext) = os.path.splitext(path)
-		new_path = prefix + '.' + extension
-		self._downloader.to_screen(u'[ffmpeg] Destination: %s' % new_path)
+		prefix, sep, ext = path.rpartition(u'.') # not os.path.splitext, since the latter does not work on unicode in all setups
+		new_path = prefix + sep + extension
+		self._downloader.to_screen(u'[ffmpeg] Destination: ' + new_path)
 		try:
 			self.run_ffmpeg(path, new_path, acodec, more_opts)
 		except:
@@ -4140,13 +4149,13 @@ class FFmpegExtractAudioPP(PostProcessor):
  		# Try to update the date time for extracted audio file.
 		if information.get('filetime') is not None:
 			try:
-				os.utime(new_path, (time.time(), information['filetime']))
+				os.utime(_encodeFilename(new_path), (time.time(), information['filetime']))
 			except:
 				self._downloader.to_stderr(u'WARNING: Cannot update utime of audio file')
 
 		if not self._keepvideo:
 			try:
-				os.remove(path)
+				os.remove(_encodeFilename(path))
 			except (IOError, OSError):
 				self._downloader.to_stderr(u'WARNING: Unable to remove downloaded video file')
 				return None
@@ -4194,9 +4203,9 @@ def parseOpts():
 	import optparse
 	import shlex
 
-	def _readOptions(filename):
+	def _readOptions(filename_bytes):
 		try:
-			optionf = open(filename)
+			optionf = open(filename_bytes)
 		except IOError:
 			return [] # silently skip if file is not present
 		try: