Switch to bytes because lots of the old things are not valid UTF-8

efnet-irclogger-convert.py

@@ -7,47 +7,48 @@ filename = sys.argv[1]
 
 date = filename.rsplit('/', 1)[-1][:10]
 
-with open(filename, 'r') as fp:
+with open(filename, 'rb') as fp:
 	for line in fp:
 		origLine = line
-		if not (line[0] == '[' and line[3] == ':' and line[6:8] == '] ' and line[-1] == '\n'):
+		if not (line[0:1] == b'[' and line[3:4] == b':' and line[6:8] == b'] ' and line[-1:] == b'\n'):
 			print(f'MALFORMED LINE: {line!r}', file = sys.stderr)
 			continue
-		time = line[1:6]
+		time = line[1:6].decode('ascii')
 		line = line[8:-1]
 		ts = datetime.datetime(int(date[:4]), int(date[5:7]), int(date[8:]), int(time[:2]), int(time[3:]), 0).replace(tzinfo = datetime.timezone.utc).timestamp()
-		if line.startswith('<'): #PRIVMSG
-			print(f'{ts} PRIVMSG {line}')
-		elif line.startswith('* '): #ACTION
-			print(f'{ts} ACTION {line[2:]}')
-		elif line.startswith('*** '):
-			words = line.split(' ')[1:]
-			if words[1:3] == ['has', 'joined']: # JOIN
-				print(f'{ts} JOIN {words[0]} joins')
-			elif words[0] == 'Joins:': # JOIN
-				print(f'{ts} JOIN {words[1]} joins')
-			elif words[1:3] == ['has', 'left']: # PART
-				reason = f' [{" ".join(words[3:])}]' if len(words) > 4 or words[3] != '' else ''
-				print(f'{ts} PART {words[0]} leaves{reason}')
-			elif words[0] == 'Parts:': # PART
-				reason = f' [{" ".join(words[3:])[1:-1]}]' if len(words) > 4 or words[3] != '()' else ''
-				print(f'{ts} PART {words[1]} leaves{reason}')
-			elif words[1:4] == ['has', 'quit', 'IRC']: # QUIT
-				print(f'{ts} QUIT {words[0]} quits [{" ".join(words[4:])[1:-1]}]')
-			elif words[0] == 'Quits:': # QUIT
-				reason = f' [{" ".join(words[3:])[1:-1]}]' if len(words) > 4 or words[3] != '()' else ''
-				print(f'{ts} QUIT {words[0]} quits{reason}')
-			elif words[1:4] == ['was', 'kicked', 'by']: # KICK
-				print(f'{ts} KICK {words[0]} is kicked by {words[4]} [{" ".join(words[5:])[1:-1]}]')
-			elif words[1:3] == ['sets', 'mode:']: # MODE
-				print(f'{ts} MODE {line[4:]}')
-			elif words[1:4] == ['changes', 'topic', 'to:']: # TOPIC
-				print(f'{ts} TOPIC {words[0]} sets the topic to: {" ".join(words[4:])}')
-			elif words[1:4] == ['changes', 'topic', 'to']: # TOPIC
-				print(f'{ts} TOPIC {words[0]} sets the topic to: {" ".join(words[4:])[1:-1]}')
-			elif words[1:5] == ['is', 'now', 'known', 'as']: # NICK
-				print(f'{ts} NICK {line[4:]}')
-			elif words[1:3] == ['starts', 'logging']: # Silently ignore (there's already a JOIN)
+		if line.startswith(b'<'): #PRIVMSG
+			sys.stdout.buffer.write(f'{ts} PRIVMSG '.encode('ascii') + line)
+		elif line.startswith(b'* '): #ACTION
+			sys.stdout.buffer.write(f'{ts} ACTION '.encode('ascii') + line[2:])
+		elif line.startswith(b'*** '):
+			words = line.split(b' ')[1:]
+			if words[1:3] == [b'has', b'joined']: # JOIN
+				sys.stdout.buffer.write(f'{ts} JOIN '.encode('ascii') + words[0] + b' joins')
+			elif words[0] == b'Joins:': # JOIN
+				sys.stdout.buffer.write(f'{ts} JOIN '.encode('ascii') + words[1] + b' joins')
+			elif words[1:3] == [b'has', b'left']: # PART
+				reason = (b' [' + b' '.join(words[3:]) + b']') if len(words) > 4 or words[3] != b'' else b''
+				sys.stdout.buffer.write(f'{ts} PART '.encode('ascii') + words[0] + b' leaves' + reason)
+			elif words[0] == b'Parts:': # PART
+				reason = (b' [' + b' '.join(words[3:])[1:-1] + b']') if len(words) > 4 or words[3] != b'()' else b''
+				sys.stdout.buffer.write(f'{ts} PART '.encode('ascii') + words[1] + b' leaves' + reason)
+			elif words[1:4] == [b'has', b'quit', b'IRC']: # QUIT
+				reason = (b' [' + b' '.join(words[4:])[1:-1] + b']') if len(words) > 4 or words[4] != b'()' else b''
+				sys.stdout.buffer.write(f'{ts} QUIT '.encode('ascii') + words[0] + b' quits' + reason)
+			elif words[0] == b'Quits:': # QUIT
+				reason = (b' [' + b' '.join(words[3:])[1:-1] + b']') if len(words) > 4 or words[3] != b'()' else b''
+				sys.stdout.buffer.write(f'{ts} QUIT '.encode('ascii') + words[0] + b' quits' + reason)
+			elif words[1:4] == [b'was', b'kicked', b'by']: # KICK
+				sys.stdout.buffer.write(f'{ts} KICK '.encode('ascii') + words[0] + b' is kicked by ' + words[4] + b' [' + b' '.join(words[5:])[1:-1] + b']')
+			elif words[1:3] == [b'sets', b'mode:']: # MODE
+				sys.stdout.buffer.write(f'{ts} MODE '.encode('ascii') + line[4:])
+			elif words[1:4] == [b'changes', b'topic', b'to:']: # TOPIC
+				sys.stdout.buffer.write(f'{ts} TOPIC '.encode('ascii') + words[0] + b' sets the topic to: ' + b' '.join(words[4:]))
+			elif words[1:4] == [b'changes', b'topic', b'to']: # TOPIC
+				sys.stdout.buffer.write(f'{ts} TOPIC '.encode('ascii') + words[0] + b' sets the topic to: ' + b' '.join(words[4:])[1:-1])
+			elif words[1:5] == [b'is', b'now', b'known', b'as']: # NICK
+				sys.stdout.buffer.write(f'{ts} NICK '.encode('ascii') + line[4:])
+			elif words[1:3] == [b'starts', b'logging']: # Silently ignore (there's already a JOIN)
 				pass
 			else:
 				print(f'MALFORMED LINE: {origLine!r}', file = sys.stderr)