add helper split_commandstring

that splits a command string into a list of strings to pass on to subprocess.Popen and the like.

This helper is now used throughout the application instead of calling shlex.split directly as this
is a potential source for errors because shlex is not yet able to properly deal with unicode
bytestrings.

7 files changed, 26 insertions, 19 deletions

diff --git a/alot/account.py b/alot/account.py
index b65bf3cb..29ec768b 100644
--- a/alot/account.py
+++ b/alot/account.py
@@ -4,9 +4,9 @@ import time
 import email
 import os
 import glob
-import shlex
 
 from alot.helper import call_cmd_async
+from alot.helper import split_commandstring
 import alot.crypto as crypto
 
 
@@ -151,7 +151,7 @@ class SendmailAccount(Account):
 
     def send_mail(self, mail):
         mail['Date'] = email.utils.formatdate(time.time(), True)
-        cmdlist = shlex.split(self.cmd.encode('utf-8', errors='ignore'))
+        cmdlist = split_commandstring(self.cmd)
 
         def cb(out):
             logging.info('sent mail successfully')
diff --git a/alot/addressbooks.py b/alot/addressbooks.py
index f527492f..16818a39 100644
--- a/alot/addressbooks.py
+++ b/alot/addressbooks.py
@@ -1,9 +1,9 @@
 import re
 import os
-import shlex
 
 from alot.settings.utils import read_config
 from helper import call_cmd
+from alot.helper import split_commandstring
 
 
 class AddressBook(object):
@@ -73,7 +73,7 @@ class MatchSdtoutAddressbook(AddressBook):
         return self.lookup('\'\'')
 
     def lookup(self, prefix):
-        cmdlist = shlex.split(self.command.encode('utf-8', errors='ignore'))
+        cmdlist = split_commandstring(self.command)
         resultstring, errmsg, retval = call_cmd(cmdlist + [prefix])
         if not resultstring:
             return []
diff --git a/alot/commands/__init__.py b/alot/commands/__init__.py
index ff9b7951..b65af5e3 100644
--- a/alot/commands/__init__.py
+++ b/alot/commands/__init__.py
@@ -1,12 +1,12 @@
 import os
 import re
 import glob
-import shlex
 import logging
 import argparse
 
 from alot.settings import settings
 import alot.helper
+from alot.helper import split_commandstring
 
 
 class Command(object):
@@ -168,7 +168,7 @@ def commandfactory(cmdline, mode='global'):
         cmdline = 'shellescape \'%s\'' % cmdline[1:]
     cmdline = re.sub(r'"(.*)"', r'"\\"\1\\""', cmdline)
     try:
-        args = shlex.split(cmdline.encode('utf-8'))
+        args = split_commandstring(cmdline)
     except ValueError, e:
         raise CommandParseError(e.message)
     args = map(lambda x: alot.helper.string_decode(x, 'utf-8'), args)
diff --git a/alot/commands/globals.py b/alot/commands/globals.py
index bc1e779a..b8722185 100644
--- a/alot/commands/globals.py
+++ b/alot/commands/globals.py
@@ -2,7 +2,6 @@ import os
 import code
 from twisted.internet import threads
 import subprocess
-import shlex
 import email
 import urwid
 from twisted.internet.defer import inlineCallbacks
@@ -26,6 +25,7 @@ from alot.db.envelope import Envelope
 from alot import commands
 from alot.settings import settings
 from alot.errors import GPGProblem
+from alot.helper import split_commandstring
 
 MODE = 'global'
 
@@ -188,7 +188,7 @@ class ExternalCommand(Command):
             term_cmd = settings.get('terminal_cmd', '')
             term_cmd = term_cmd.encode('utf-8', errors='ignore')
             logging.info('spawn in terminal: %s' % term_cmd)
-            termcmdlist = shlex.split(term_cmd)
+            termcmdlist = split_commandstring(term_cmd)
             logging.info('term cmdlist: %s' % termcmdlist)
             self.cmdlist = termcmdlist + self.cmdlist
 
@@ -251,11 +251,9 @@ class EditCommand(ExternalCommand):
         if '%s' in editor_cmdstring:
             cmdstring = editor_cmdstring.replace('%s',
                                                  helper.shell_quote(path))
-            cmdstring = cmdstring.encode('utf-8', errors='ignore')
-            self.cmdlist = shlex.split(cmdstring)
+            self.cmdlist = split_commandstring(cmdstring)
         else:
-            cmdstring = editor_cmdstring.encode('utf-8', errors='ignore')
-            self.cmdlist = shlex.split(cmdstring) + [path]
+            self.cmdlist = split_commandstring(editor_cmdstring) + [path]
 
         logging.debug(self.cmdlist)
 
diff --git a/alot/commands/thread.py b/alot/commands/thread.py
index c304fa0c..be6c10d4 100644
--- a/alot/commands/thread.py
+++ b/alot/commands/thread.py
@@ -2,7 +2,6 @@ import os
 import logging
 import tempfile
 from twisted.internet.defer import inlineCallbacks
-import shlex
 import re
 import subprocess
 from email.Utils import parseaddr
@@ -22,10 +21,10 @@ from alot.db.utils import extract_headers
 from alot.db.utils import extract_body
 from alot.db.envelope import Envelope
 from alot.db.attachment import Attachment
-
 from alot.db.errors import DatabaseROError
 from alot.settings import settings
 from alot.helper import parse_mailcap_nametemplate
+from alot.helper import split_commandstring
 
 MODE = 'thread'
 
@@ -402,7 +401,7 @@ class PipeCommand(Command):
         """
         Command.__init__(self, **kwargs)
         if isinstance(cmd, unicode):
-            cmd = shlex.split(cmd.encode('UTF-8'))
+            cmd = split_commandstring(cmd)
         self.cmd = cmd
         self.whole_thread = all
         self.separately = separately
@@ -701,8 +700,7 @@ class OpenAttachmentCommand(Command):
             handler_cmd = mailcap.subst(handler_raw_commandstring, mimetype,
                                         filename=tempfile_name, plist=parms)
 
-            handler_cmd = handler_cmd.encode('utf-8', errors='ignore')
-            handler_cmdlist = shlex.split(handler_cmd)
+            handler_cmdlist = split_commandstring(handler_cmd)
 
             # 'needsterminal' makes handler overtake the terminal
             nt = entry.get('needsterminal', None)
diff --git a/alot/db/utils.py b/alot/db/utils.py
index 55073e45..e9d40c21 100644
--- a/alot/db/utils.py
+++ b/alot/db/utils.py
@@ -2,7 +2,6 @@ import os
 import email
 import tempfile
 import re
-import shlex
 from email.header import Header
 import email.charset as charset
 charset.add_charset('utf-8', charset.QP, charset.QP, 'utf-8')
@@ -15,6 +14,7 @@ from alot.settings import settings
 from alot.helper import string_sanitize
 from alot.helper import string_decode
 from alot.helper import parse_mailcap_nametemplate
+from alot.helper import split_commandstring
 
 
 def extract_headers(mail, headers=None):
@@ -99,7 +99,7 @@ def extract_body(mail, types=None):
                                     filename=tmpfile.name, plist=parms)
                 logging.debug('command: %s' % cmd)
                 logging.debug('parms: %s' % str(parms))
-                cmdlist = shlex.split(cmd.encode('utf-8', errors='ignore'))
+                cmdlist = split_commandstring(cmd)
                 # call handler
                 rendered_payload, errmsg, retval = helper.call_cmd(cmdlist)
                 # remove tempfile
diff --git a/alot/helper.py b/alot/helper.py
index 8d596994..f94e03f2 100644
--- a/alot/helper.py
+++ b/alot/helper.py
@@ -4,6 +4,7 @@ from datetime import datetime
 from collections import deque
 from string import strip
 import subprocess
+import shlex
 import email
 import os
 import re
@@ -21,6 +22,16 @@ import logging
 import tempfile
 
 
+def split_commandstring(cmdstring):
+    """
+    split command string into a list of strings to pass on to subprocess.Popen
+    and the like. This simply calls shlex.split but works also with unicode
+    bytestrings.
+    """
+    if isinstance(cmdstring, unicode):
+        cmdstring = cmdstring.encode('utf-8', errors='ignore')
+    return shlex.split(cmdstring)
+
 def safely_get(clb, E, on_error=''):
     """
     returns result of :func:`clb` and falls back to `on_error`