Applied minor fix

Author: dhondta

src/codext/__main__.py

@@ -3,6 +3,7 @@
 
 """
 import argparse
+import sys
 
 from .__common__ import *
 from .__info__ import __author__, __copyright__, __email__, __license__, __source__, __version__
@@ -27,6 +28,15 @@ def __format_list(items, include=True):
     return d
 
 
+def __is_binary(data):
+    """ Bytes are binary data if they are neither valid UTF-8 nor latin-1 text (as returned by text codecs). """
+    try:
+        data.decode("utf-8")
+        return False
+    except UnicodeDecodeError:
+        return any(b < 32 and b not in b"\t\n\r" or 127 <= b < 160 for b in data)
+
+
 def __print_tabular(lst, space=4):
     try:
         cols, _ = os.get_terminal_size()
@@ -200,6 +210,10 @@ def _format_action_invocation(self, action):
             if args.outfile:
                 with codecs.open(args.outfile, 'wb') as f:
                     f.write(c)
+            elif isinstance(c, bytes) and __is_binary(c):
+                # write binary data as is, as printing it would re-encode as UTF-8 the bytes that are not valid UTF-8
+                sys.stdout.buffer.write(c)
+                sys.stdout.flush()
             else:
                 print(ensure_str(c or "Could not %scode :-(" % ["en", "de"][args.command == "decode"]), end="")
         elif args.command == "guess":

tests/test_common.py

@@ -271,4 +271,21 @@ def test_handle_macros(self):
         self.assertIsNone(codext.add_macro("VALID-MACRO", "lzma", "base64"))
         self.assertIsNone(codext.remove("VALID-MACRO"))
         self.assertRaises(ValueError, codext.add_macro, "SHALL-FAIL", "base26", "sms", "letter-indices")
-
+    
+    def test_main_binary_output(self):
+        import os, subprocess
+        src = os.path.join(os.path.dirname(os.path.dirname(os.path.abspath(__file__))), "src")
+        env = {**os.environ, 'PYTHONPATH': src + os.pathsep + os.environ.get('PYTHONPATH', "")}
+        cli = lambda inp, *args: subprocess.run([sys.executable, "-m", "codext"] + list(args), input=inp, env=env,
+                                                capture_output=True, check=True).stdout
+        # binary output (not valid UTF-8) is written as is, not re-encoded as UTF-8 (the gzip header holding the time,
+        #  at bytes 4 to 7)
+        skip = lambda codec, data: data[:4] + data[8:] if codec == "gzip" else data
+        for codec in ["gzip", "piet"]:
+            out = cli(b"test", "encode", codec)
+            self.assertEqual(skip(codec, out), skip(codec, codext.encode(b"test", codec)))
+            self.assertEqual(cli(out, "decode", codec), b"test")
+        # text output is unchanged, including latin-1 text, printed as UTF-8
+        self.assertEqual(cli(b"test", "encode", "base64"), b"dGVzdA==")
+        self.assertEqual(cli("caf\u00e9".encode(), "encode", "reverse"), "\u00e9fac".encode())
+        self.assertEqual(cli(b"test", "encode", "base100"), codext.encode(b"test", "base100"))