Fix: Recode text from mailcap command from default encoding to charset

[mimedecode.git] / mimedecode.py

diff --git a/mimedecode.py b/mimedecode.py
index 797017a77bfda74b2a807ab5994c7166e24ea417..5459cbff73ab05dcc32ba5f7645ae5951a525cb8 100755 (executable)
--- a/mimedecode.py
+++ b/mimedecode.py
@@ -2,8 +2,13 @@
 """Decode MIME message"""
 
 import sys, os
+import subprocess
 from mimedecode_version import __version__, \
     __author__, __copyright__, __license__
+if sys.version_info[0] >= 3:
+    # Replace email.message._formatparam with _formatparam from Python 2.7
+    # to avoid re-encoding non-ascii params.
+    import formatparam_27
 
 me = os.path.basename(sys.argv[0])
 
@@ -32,7 +37,11 @@ def output_headers(msg):
     for key, value in msg.items():
         output(key)
         output(": ")
-        output(value)
+        value = value.split(';', 1)
+        output(value[0])
+        if len(value) == 2:
+            output(";")
+            output(_decode_header(value[1], strip=False))
         output(os.linesep)
     output(os.linesep) # End of headers
 
@@ -49,7 +58,7 @@ def recode_if_needed(s, charset):
     return s
 
 
-def _decode_header(s):
+def _decode_header(s, strip=True):
     """Return a decoded string according to RFC 2047.
     NOTE: This is almost the same as email.Utils.decode.
     """
@@ -62,15 +71,14 @@ def _decode_header(s):
 
     rtn = []
     for atom, charset in L:
-        if charset is None:
-            charset = g.default_encoding
-        rtn.append(recode_if_needed(atom, charset))
-        rtn.append(' ')
-    del rtn[-1] # remove the last space
+        atom = recode_if_needed(atom, charset or g.default_encoding)
+        if strip:
+            atom = atom.strip()
+        rtn.append(atom)
 
     # Now that we've decoded everything, we just need to join all the parts
     # together into the final string.
-    return ''.join(rtn)
+    return ' '.join(rtn)
 
 def decode_header(msg, header):
     "Decode mail header (if exists) and put it back, if it was encoded"
@@ -215,6 +223,10 @@ def decode_body(msg, s):
         caps = mailcap.getcaps()
 
     content_type = msg.get_content_type()
+    if content_type.startswith('text/'):
+        charset = msg.get_content_charset()
+    else:
+        charset = None
     filename = tempfile.mktemp()
     command = None
 
@@ -232,19 +244,28 @@ def decode_body(msg, s):
         return s
 
     outfile = open(filename, 'wb')
+    if charset and bytes is not str and isinstance(s, bytes):  # Python3
+        s = s.decode(charset, "replace")
     if not isinstance(s, bytes):
         s = s.encode(g.default_encoding, "replace")
     outfile.write(s)
     outfile.close()
 
-    pipe = os.popen(command, 'r')
-    s = pipe.read()
-    pipe.close()
+    pipe = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE)
+    new_s = pipe.stdout.read()
+    pipe.stdout.close()
+    if pipe.wait() == 0: # result=0, Ok
+        s = new_s
+        if bytes is not str and isinstance(s, bytes):  # Python3
+            s = s.decode(g.default_encoding, "replace")
+        if charset and not isinstance(s, bytes):
+            s = s.encode(charset, "replace")
+        set_content_type(msg, "text/plain")
+        msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, g.host_name, command.split()[0])
+    else:
+        msg["X-MIME-Autoconverted"] = "failed conversion from %s to text/plain by %s id %s" % (content_type, g.host_name, command.split()[0])
     os.remove(filename)
 
-    set_content_type(msg, "text/plain")
-    msg["X-MIME-Autoconverted"] = "from %s to text/plain by %s id %s" % (content_type, g.host_name, command.split()[0])
-
     return s