Allow optional shlex arg parsing for command plugins

README.rst

@@ -84,6 +84,12 @@ basic helga settings, as outlined below:
   and 'PASSWORD' if the IRC server requires authentication.
 - ``LOG_LEVEL``: String for the default logging level (default: 'DEBUG')
 - ``LOG_FILE``: If set, a string indicating the log file for python logs
+- ``COMMAND_ARGS_SHLEX``: Control the behavior of argument parsing for command plugins
+  By default this is a naive str.split(' '), however a plugin may need this behavior to be a bit more
+  robust. By setting this value to True, shlex.split() will be used instead so that commands
+  like ``helga foo bar "baz qux"`` will yield an argument list like ['bar', 'baz qux'] instead
+  of ['bar', '"baz', 'qux"']. Shlex splits will be the default and only supported behavior
+  in a future version.
 - ``CHANNEL_LOGGING``: If True, enable conversation logging on all channels (default: False)
 - ``CHANNEL_LOGGING_DIR``: If using channel logs, the directory to which channel logs should
   logs should be written. A new directory will be created for each channel in which the

helga/plugins/core.py

@@ -2,15 +2,17 @@
 import pkg_resources
 import random
 import re
+import shlex
 import sys
+import warnings
 
 from collections import defaultdict
 from itertools import ifilter, imap
 
 import smokesignal
 
 from helga import log, settings
-from helga.util.encodings import to_unicode
+from helga.util.encodings import from_unicode, to_unicode
 
 
 logger = log.getLogger(__name__)
@@ -53,6 +55,10 @@
 PRIORITY_HIGH = 75
 
 
+if not settings.COMMAND_ARGS_SHLEX:
+    warnings.warn(u'Command arg parsing will default to shlex in a future version', FutureWarning)
+
+
 def random_ack():
     return random.choice(ACKS)
 
@@ -410,7 +416,22 @@ def parse(self, botnick, message):
             # FIXME: Log here?
             return u'', []
 
-        return cmd, filter(bool, argstr.strip().split(' '))
+        return cmd, filter(bool, self._parse_argstr(argstr))
+
+    def _parse_argstr(self, argstr):
+        """
+        Parse an argument string for this command. If COMMAND_ARGS_SHLEX
+        is set to False, then naive whitespace splitting is performed on the argument
+        string. If not, a more robust shlex.split() is performed. For example, given
+        the message 'helga foo bar "baz qux"', the former would produce arguments
+        ['bar', '"baz', 'qux"'] while the latter would produce ['bar', 'baz qux']
+        """
+        if settings.COMMAND_ARGS_SHLEX:
+            argv = shlex.split(from_unicode(argstr.strip()))
+        else:
+            argv = argstr.strip().split(' ')
+
+        return map(to_unicode, argv)
 
     def run(self, client, channel, nick, message, command, args):
         """

helga/settings.py

@@ -9,6 +9,15 @@
 
 LOG_LEVEL = 'DEBUG'
 
+# Control the behavior of argument parsing for command plugins
+# By default this is a naive str.split(' '), however a plugin
+# may need this behavior to be a bit more robust. By setting this value
+# to True, shlex.split() will be used instead so that commands
+# like `helga foo bar "baz qux"` will yield an argument list like
+# ['bar', 'baz qux'] instead of ['bar', '"baz', 'qux"']. Shlex splits
+# will be the default and only supported behavior in a future version.
+COMMAND_ARGS_SHLEX = False
+
 # Channel logging. Set CHANNEL_LOGGING_DIR
 CHANNEL_LOGGING = False
 CHANNEL_LOGGING_DIR = '.logs'

helga/tests/plugins/test_core.py

@@ -3,7 +3,8 @@
 Tests for helga.plugins
 """
 from collections import defaultdict
-from unittest import TestCase
+
+import pytest
 
 from mock import Mock, patch
 from pretend import stub
@@ -20,9 +21,9 @@
                                 registry)
 
 
-class RegistryTestCase(TestCase):
+class TestRegistry(object):
 
-    def setUp(self):
+    def setup(self):
         registry.plugins = {}
         registry.enabled_plugins = defaultdict(set)
 
@@ -158,7 +159,8 @@ def test_registry_is_singleton(self):
 
     def test_register_raises_typeerror(self):
         for plugin in self.invalid_plugins:
-            self.assertRaises(TypeError, registry.register, 'invalid', plugin)
+            with pytest.raises(TypeError):
+                registry.register('invalid', plugin)
 
     def test_register_valid_plugins(self):
         for plugin in self.valid_plugins:
@@ -167,7 +169,8 @@ def test_register_valid_plugins(self):
 
     def test_register_plugin_handles_unicode(self):
         for plugin in self.invalid_plugins:
-            self.assertRaises(TypeError, registry.register, 'invalid', plugin)
+            with pytest.raises(TypeError):
+                registry.register('invalid', plugin)
 
         for plugin in self.valid_plugins:
             name = u'{0}-{1}'.format(self.snowman, repr(plugin))
@@ -295,9 +298,9 @@ def test_preprocess(self):
             assert plugins[1].preprocess.called
 
 
-class PluginTestCase(TestCase):
+class TestPlugin(object):
 
-    def setUp(self):
+    def setup(self):
         self.plugin = Plugin()
         self.client = Mock(nickname='helga')
 
@@ -335,9 +338,9 @@ def test_base_plugin_process_calls_run(self):
             run.assert_called_with('foo', 'bar', 'baz', 'qux')
 
 
-class CommandTestCase(TestCase):
+class TestCommand(object):
 
-    def setUp(self):
+    def setup(self):
         self.cmd = Command('foo', aliases=('bar', 'baz'), help='foo cmd')
         self.client = Mock(nickname='helga')
 
@@ -392,6 +395,24 @@ def test_parse_handles_unicode(self):
         self.cmd.command = snowman
         assert (snowman, [disapproval]) == self.cmd.parse('helga', cmd)
 
+    @pytest.mark.parametrize('argstr,expected', [
+        ('foo bar "baz qux"', ['foo', 'bar', 'baz qux']),
+        (u'foo "bar ☃"', ['foo', u'bar ☃']),
+    ])
+    def test_parse_argstr_shlex(self, argstr, expected):
+        with patch('helga.plugins.core.settings') as settings:
+            settings.COMMAND_ARGS_SHLEX = True
+            assert self.cmd._parse_argstr(argstr) == expected
+
+    @pytest.mark.parametrize('argstr,expected', [
+        ('foo bar "baz qux"', ['foo', 'bar', '"baz', 'qux"']),
+        (u'foo "bar ☃"', ['foo', '"bar', u'☃"']),
+    ])
+    def test_parse_argstr_whitespace(self, argstr, expected):
+        with patch('helga.plugins.core.settings') as settings:
+            settings.COMMAND_ARGS_SHLEX = False
+            assert self.cmd._parse_argstr(argstr) == expected
+
     def test_process_for_different_command_returns_none(self):
         assert self.cmd.process(self.client, '#bots', 'me', 'helga qux') is None
 
@@ -429,9 +450,9 @@ def foo(client, chan, nick, msg, cmd, args):
         assert 'bar' == foo._plugins[0](self.client, '#bots', 'me', 'helga baz')
 
 
-class MatchTestCase(TestCase):
+class TestMatch(object):
 
-    def setUp(self):
+    def setup(self):
         self.match = Match('foo')
         self.client = Mock()