workflow.py

# encoding: utf-8
#
# Copyright (c) 2014 Dean Jackson <deanishe@deanishe.net>
#
# MIT Licence. See http://opensource.org/licenses/MIT
#
# Created on 2014-02-15
#

"""The :class:`Workflow` object is the main interface to this library.

:class:`Workflow` is targeted at Alfred 2. Use
:class:`~workflow.workflow3.Workflow3` if you want to use Alfred 3's new
features, such as :ref:`workflow variables <workflow-variables>` or
more powerful modifiers.

See :ref:`setup` in the :ref:`user-manual` for an example of how to set
up your Python script to best utilise the :class:`Workflow` object.

"""

from __future__ import print_function, unicode_literals

import binascii
from contextlib import contextmanager
import cPickle
from copy import deepcopy
import errno
import json
import logging
import logging.handlers
import os
import pickle
import plistlib
import re
import shutil
import signal
import string
import subprocess
import sys
import time
import unicodedata

try:
    import xml.etree.cElementTree as ET
except ImportError:  # pragma: no cover
    import xml.etree.ElementTree as ET


#: Sentinel for properties that haven't been set yet (that might
#: correctly have the value ``None``)
UNSET = object()

####################################################################
# Standard system icons
####################################################################

# These icons are default OS X icons. They are super-high quality, and
# will be familiar to users.
# This library uses `ICON_ERROR` when a workflow dies in flames, so
# in my own workflows, I use `ICON_WARNING` for less fatal errors
# (e.g. bad user input, no results etc.)

# The system icons are all in this directory. There are many more than
# are listed here

ICON_ROOT = '/System/Library/CoreServices/CoreTypes.bundle/Contents/Resources'

ICON_ACCOUNT = os.path.join(ICON_ROOT, 'Accounts.icns')
ICON_BURN = os.path.join(ICON_ROOT, 'BurningIcon.icns')
ICON_CLOCK = os.path.join(ICON_ROOT, 'Clock.icns')
ICON_COLOR = os.path.join(ICON_ROOT, 'ProfileBackgroundColor.icns')
ICON_COLOUR = ICON_COLOR  # Queen's English, if you please
ICON_EJECT = os.path.join(ICON_ROOT, 'EjectMediaIcon.icns')
# Shown when a workflow throws an error
ICON_ERROR = os.path.join(ICON_ROOT, 'AlertStopIcon.icns')
ICON_FAVORITE = os.path.join(ICON_ROOT, 'ToolbarFavoritesIcon.icns')
ICON_FAVOURITE = ICON_FAVORITE
ICON_GROUP = os.path.join(ICON_ROOT, 'GroupIcon.icns')
ICON_HELP = os.path.join(ICON_ROOT, 'HelpIcon.icns')
ICON_HOME = os.path.join(ICON_ROOT, 'HomeFolderIcon.icns')
ICON_INFO = os.path.join(ICON_ROOT, 'ToolbarInfo.icns')
ICON_NETWORK = os.path.join(ICON_ROOT, 'GenericNetworkIcon.icns')
ICON_NOTE = os.path.join(ICON_ROOT, 'AlertNoteIcon.icns')
ICON_SETTINGS = os.path.join(ICON_ROOT, 'ToolbarAdvanced.icns')
ICON_SWIRL = os.path.join(ICON_ROOT, 'ErasingIcon.icns')
ICON_SWITCH = os.path.join(ICON_ROOT, 'General.icns')
ICON_SYNC = os.path.join(ICON_ROOT, 'Sync.icns')
ICON_TRASH = os.path.join(ICON_ROOT, 'TrashIcon.icns')
ICON_USER = os.path.join(ICON_ROOT, 'UserIcon.icns')
ICON_WARNING = os.path.join(ICON_ROOT, 'AlertCautionIcon.icns')
ICON_WEB = os.path.join(ICON_ROOT, 'BookmarkIcon.icns')

####################################################################
# non-ASCII to ASCII diacritic folding.
# Used by `fold_to_ascii` method
####################################################################

ASCII_REPLACEMENTS = {
    'À': 'A',
    'Á': 'A',
    'Â': 'A',
    'Ã': 'A',
    'Ä': 'A',
    'Å': 'A',
    'Æ': 'AE',
    'Ç': 'C',
    'È': 'E',
    'É': 'E',
    'Ê': 'E',
    'Ë': 'E',
    'Ì': 'I',
    'Í': 'I',
    'Î': 'I',
    'Ï': 'I',
    'Ð': 'D',
    'Ñ': 'N',
    'Ò': 'O',
    'Ó': 'O',
    'Ô': 'O',
    'Õ': 'O',
    'Ö': 'O',
    'Ø': 'O',
    'Ù': 'U',
    'Ú': 'U',
    'Û': 'U',
    'Ü': 'U',
    'Ý': 'Y',
    'Þ': 'Th',
    'ß': 'ss',
    'à': 'a',
    'á': 'a',
    'â': 'a',
    'ã': 'a',
    'ä': 'a',
    'å': 'a',
    'æ': 'ae',
    'ç': 'c',
    'è': 'e',
    'é': 'e',
    'ê': 'e',
    'ë': 'e',
    'ì': 'i',
    'í': 'i',
    'î': 'i',
    'ï': 'i',
    'ð': 'd',
    'ñ': 'n',
    'ò': 'o',
    'ó': 'o',
    'ô': 'o',
    'õ': 'o',
    'ö': 'o',
    'ø': 'o',
    'ù': 'u',
    'ú': 'u',
    'û': 'u',
    'ü': 'u',
    'ý': 'y',
    'þ': 'th',
    'ÿ': 'y',
    'Ł': 'L',
    'ł': 'l',
    'Ń': 'N',
    'ń': 'n',
    'Ņ': 'N',
    'ņ': 'n',
    'Ň': 'N',
    'ň': 'n',
    'Ŋ': 'ng',
    'ŋ': 'NG',
    'Ō': 'O',
    'ō': 'o',
    'Ŏ': 'O',
    'ŏ': 'o',
    'Ő': 'O',
    'ő': 'o',
    'Œ': 'OE',
    'œ': 'oe',
    'Ŕ': 'R',
    'ŕ': 'r',
    'Ŗ': 'R',
    'ŗ': 'r',
    'Ř': 'R',
    'ř': 'r',
    'Ś': 'S',
    'ś': 's',
    'Ŝ': 'S',
    'ŝ': 's',
    'Ş': 'S',
    'ş': 's',
    'Š': 'S',
    'š': 's',
    'Ţ': 'T',
    'ţ': 't',
    'Ť': 'T',
    'ť': 't',
    'Ŧ': 'T',
    'ŧ': 't',
    'Ũ': 'U',
    'ũ': 'u',
    'Ū': 'U',
    'ū': 'u',
    'Ŭ': 'U',
    'ŭ': 'u',
    'Ů': 'U',
    'ů': 'u',
    'Ű': 'U',
    'ű': 'u',
    'Ŵ': 'W',
    'ŵ': 'w',
    'Ŷ': 'Y',
    'ŷ': 'y',
    'Ÿ': 'Y',
    'Ź': 'Z',
    'ź': 'z',
    'Ż': 'Z',
    'ż': 'z',
    'Ž': 'Z',
    'ž': 'z',
    'ſ': 's',
    'Α': 'A',
    'Β': 'B',
    'Γ': 'G',
    'Δ': 'D',
    'Ε': 'E',
    'Ζ': 'Z',
    'Η': 'E',
    'Θ': 'Th',
    'Ι': 'I',
    'Κ': 'K',
    'Λ': 'L',
    'Μ': 'M',
    'Ν': 'N',
    'Ξ': 'Ks',
    'Ο': 'O',
    'Π': 'P',
    'Ρ': 'R',
    'Σ': 'S',
    'Τ': 'T',
    'Υ': 'U',
    'Φ': 'Ph',
    'Χ': 'Kh',
    'Ψ': 'Ps',
    'Ω': 'O',
    'α': 'a',
    'β': 'b',
    'γ': 'g',
    'δ': 'd',
    'ε': 'e',
    'ζ': 'z',
    'η': 'e',
    'θ': 'th',
    'ι': 'i',
    'κ': 'k',
    'λ': 'l',
    'μ': 'm',
    'ν': 'n',
    'ξ': 'x',
    'ο': 'o',
    'π': 'p',
    'ρ': 'r',
    'ς': 's',
    'σ': 's',
    'τ': 't',
    'υ': 'u',
    'φ': 'ph',
    'χ': 'kh',
    'ψ': 'ps',
    'ω': 'o',
    'А': 'A',
    'Б': 'B',
    'В': 'V',
    'Г': 'G',
    'Д': 'D',
    'Е': 'E',
    'Ж': 'Zh',
    'З': 'Z',
    'И': 'I',
    'Й': 'I',
    'К': 'K',
    'Л': 'L',
    'М': 'M',
    'Н': 'N',
    'О': 'O',
    'П': 'P',
    'Р': 'R',
    'С': 'S',
    'Т': 'T',
    'У': 'U',
    'Ф': 'F',
    'Х': 'Kh',
    'Ц': 'Ts',
    'Ч': 'Ch',
    'Ш': 'Sh',
    'Щ': 'Shch',
    'Ъ': "'",
    'Ы': 'Y',
    'Ь': "'",
    'Э': 'E',
    'Ю': 'Iu',
    'Я': 'Ia',
    'а': 'a',
    'б': 'b',
    'в': 'v',
    'г': 'g',
    'д': 'd',
    'е': 'e',
    'ж': 'zh',
    'з': 'z',
    'и': 'i',
    'й': 'i',
    'к': 'k',
    'л': 'l',
    'м': 'm',
    'н': 'n',
    'о': 'o',
    'п': 'p',
    'р': 'r',
    'с': 's',
    'т': 't',
    'у': 'u',
    'ф': 'f',
    'х': 'kh',
    'ц': 'ts',
    'ч': 'ch',
    'ш': 'sh',
    'щ': 'shch',
    'ъ': "'",
    'ы': 'y',
    'ь': "'",
    'э': 'e',
    'ю': 'iu',
    'я': 'ia',
    # 'ᴀ': '',
    # 'ᴁ': '',
    # 'ᴂ': '',
    # 'ᴃ': '',
    # 'ᴄ': '',
    # 'ᴅ': '',
    # 'ᴆ': '',
    # 'ᴇ': '',
    # 'ᴈ': '',
    # 'ᴉ': '',
    # 'ᴊ': '',
    # 'ᴋ': '',
    # 'ᴌ': '',
    # 'ᴍ': '',
    # 'ᴎ': '',
    # 'ᴏ': '',
    # 'ᴐ': '',
    # 'ᴑ': '',
    # 'ᴒ': '',
    # 'ᴓ': '',
    # 'ᴔ': '',
    # 'ᴕ': '',
    # 'ᴖ': '',
    # 'ᴗ': '',
    # 'ᴘ': '',
    # 'ᴙ': '',
    # 'ᴚ': '',
    # 'ᴛ': '',
    # 'ᴜ': '',
    # 'ᴝ': '',
    # 'ᴞ': '',
    # 'ᴟ': '',
    # 'ᴠ': '',
    # 'ᴡ': '',
    # 'ᴢ': '',
    # 'ᴣ': '',
    # 'ᴤ': '',
    # 'ᴥ': '',
    'ᴦ': 'G',
    'ᴧ': 'L',
    'ᴨ': 'P',
    'ᴩ': 'R',
    'ᴪ': 'PS',
    'ẞ': 'Ss',
    'Ỳ': 'Y',
    'ỳ': 'y',
    'Ỵ': 'Y',
    'ỵ': 'y',
    'Ỹ': 'Y',
    'ỹ': 'y',
}

####################################################################
# Smart-to-dumb punctuation mapping
####################################################################

DUMB_PUNCTUATION = {
    '‘': "'",
    '’': "'",
    '‚': "'",
    '“': '"',
    '”': '"',
    '„': '"',
    '–': '-',
    '—': '-'
}


####################################################################
# Used by `Workflow.filter`
####################################################################

# Anchor characters in a name
#: Characters that indicate the beginning of a "word" in CamelCase
INITIALS = string.ascii_uppercase + string.digits

#: Split on non-letters, numbers
split_on_delimiters = re.compile('[^a-zA-Z0-9]').split

# Match filter flags
#: Match items that start with ``query``
MATCH_STARTSWITH = 1
#: Match items whose capital letters start with ``query``
MATCH_CAPITALS = 2
#: Match items with a component "word" that matches ``query``
MATCH_ATOM = 4
#: Match items whose initials (based on atoms) start with ``query``
MATCH_INITIALS_STARTSWITH = 8
#: Match items whose initials (based on atoms) contain ``query``
MATCH_INITIALS_CONTAIN = 16
#: Combination of :const:`MATCH_INITIALS_STARTSWITH` and
#: :const:`MATCH_INITIALS_CONTAIN`
MATCH_INITIALS = 24
#: Match items if ``query`` is a substring
MATCH_SUBSTRING = 32
#: Match items if all characters in ``query`` appear in the item in order
MATCH_ALLCHARS = 64
#: Combination of all other ``MATCH_*`` constants
MATCH_ALL = 127


####################################################################
# Used by `Workflow.check_update`
####################################################################

# Number of days to wait between checking for updates to the workflow
DEFAULT_UPDATE_FREQUENCY = 1


####################################################################
# Lockfile and Keychain access errors
####################################################################

class AcquisitionError(Exception):
    """Raised if a lock cannot be acquired."""


class KeychainError(Exception):
    """Raised for unknown Keychain errors.

    Raised by methods :meth:`Workflow.save_password`,
    :meth:`Workflow.get_password` and :meth:`Workflow.delete_password`
    when ``security`` CLI app returns an unknown error code.
    """


class PasswordNotFound(KeychainError):
    """Password not in Keychain.

    Raised by method :meth:`Workflow.get_password` when ``account``
    is unknown to the Keychain.
    """


class PasswordExists(KeychainError):
    """Raised when trying to overwrite an existing account password.

    You should never receive this error: it is used internally
    by the :meth:`Workflow.save_password` method to know if it needs
    to delete the old password first (a Keychain implementation detail).
    """


####################################################################
# Helper functions
####################################################################

def isascii(text):
    """Test if ``text`` contains only ASCII characters.

    :param text: text to test for ASCII-ness
    :type text: ``unicode``
    :returns: ``True`` if ``text`` contains only ASCII characters
    :rtype: ``Boolean``

    """
    try:
        text.encode('ascii')
    except UnicodeEncodeError:
        return False
    return True


####################################################################
# Implementation classes
####################################################################

class SerializerManager(object):
    """Contains registered serializers.

    .. versionadded:: 1.8

    A configured instance of this class is available at
    ``workflow.manager``.

    Use :meth:`register()` to register new (or replace
    existing) serializers, which you can specify by name when calling
    :class:`Workflow` data storage methods.

    See :ref:`manual-serialization` and :ref:`manual-persistent-data`
    for further information.

    """

    def __init__(self):
        """Create new SerializerManager object."""
        self._serializers = {}

    def register(self, name, serializer):
        """Register ``serializer`` object under ``name``.

        Raises :class:`AttributeError` if ``serializer`` in invalid.

        .. note::

            ``name`` will be used as the file extension of the saved files.

        :param name: Name to register ``serializer`` under
        :type name: ``unicode`` or ``str``
        :param serializer: object with ``load()`` and ``dump()``
            methods

        """
        # Basic validation
        getattr(serializer, 'load')
        getattr(serializer, 'dump')

        self._serializers[name] = serializer

    def serializer(self, name):
        """Return serializer object for ``name``.

        :param name: Name of serializer to return
        :type name: ``unicode`` or ``str``
        :returns: serializer object or ``None`` if no such serializer
            is registered.

        """
        return self._serializers.get(name)

    def unregister(self, name):
        """Remove registered serializer with ``name``.

        Raises a :class:`ValueError` if there is no such registered
        serializer.

        :param name: Name of serializer to remove
        :type name: ``unicode`` or ``str``
        :returns: serializer object

        """
        if name not in self._serializers:
            raise ValueError('No such serializer registered : {0}'.format(
                             name))

        serializer = self._serializers[name]
        del self._serializers[name]

        return serializer

    @property
    def serializers(self):
        """Return names of registered serializers."""
        return sorted(self._serializers.keys())


class JSONSerializer(object):
    """Wrapper around :mod:`json`. Sets ``indent`` and ``encoding``.

    .. versionadded:: 1.8

    Use this serializer if you need readable data files. JSON doesn't
    support Python objects as well as ``cPickle``/``pickle``, so be
    careful which data you try to serialize as JSON.

    """

    @classmethod
    def load(cls, file_obj):
        """Load serialized object from open JSON file.

        .. versionadded:: 1.8

        :param file_obj: file handle
        :type file_obj: ``file`` object
        :returns: object loaded from JSON file
        :rtype: object

        """
        return json.load(file_obj)

    @classmethod
    def dump(cls, obj, file_obj):
        """Serialize object ``obj`` to open JSON file.

        .. versionadded:: 1.8

        :param obj: Python object to serialize
        :type obj: JSON-serializable data structure
        :param file_obj: file handle
        :type file_obj: ``file`` object

        """
        return json.dump(obj, file_obj, indent=2, encoding='utf-8')


class CPickleSerializer(object):
    """Wrapper around :mod:`cPickle`. Sets ``protocol``.

    .. versionadded:: 1.8

    This is the default serializer and the best combination of speed and
    flexibility.

    """

    @classmethod
    def load(cls, file_obj):
        """Load serialized object from open pickle file.

        .. versionadded:: 1.8

        :param file_obj: file handle
        :type file_obj: ``file`` object
        :returns: object loaded from pickle file
        :rtype: object

        """
        return cPickle.load(file_obj)

    @classmethod
    def dump(cls, obj, file_obj):
        """Serialize object ``obj`` to open pickle file.

        .. versionadded:: 1.8

        :param obj: Python object to serialize
        :type obj: Python object
        :param file_obj: file handle
        :type file_obj: ``file`` object

        """
        return cPickle.dump(obj, file_obj, protocol=-1)


class PickleSerializer(object):
    """Wrapper around :mod:`pickle`. Sets ``protocol``.

    .. versionadded:: 1.8

    Use this serializer if you need to add custom pickling.

    """

    @classmethod
    def load(cls, file_obj):
        """Load serialized object from open pickle file.

        .. versionadded:: 1.8

        :param file_obj: file handle
        :type file_obj: ``file`` object
        :returns: object loaded from pickle file
        :rtype: object

        """
        return pickle.load(file_obj)

    @classmethod
    def dump(cls, obj, file_obj):
        """Serialize object ``obj`` to open pickle file.

        .. versionadded:: 1.8

        :param obj: Python object to serialize
        :type obj: Python object
        :param file_obj: file handle
        :type file_obj: ``file`` object

        """
        return pickle.dump(obj, file_obj, protocol=-1)


# Set up default manager and register built-in serializers
manager = SerializerManager()
manager.register('cpickle', CPickleSerializer)
manager.register('pickle', PickleSerializer)
manager.register('json', JSONSerializer)


class Item(object):
    """Represents a feedback item for Alfred.

    Generates Alfred-compliant XML for a single item.

    You probably shouldn't use this class directly, but via
    :meth:`Workflow.add_item`. See :meth:`~Workflow.add_item`
    for details of arguments.

    """

    def __init__(self, title, subtitle='', modifier_subtitles=None,
                 arg=None, autocomplete=None, valid=False, uid=None,
                 icon=None, icontype=None, type=None, largetext=None,
                 copytext=None, quicklookurl=None):
        """Same arguments as :meth:`Workflow.add_item`."""
        self.title = title
        self.subtitle = subtitle
        self.modifier_subtitles = modifier_subtitles or {}
        self.arg = arg
        self.autocomplete = autocomplete
        self.valid = valid
        self.uid = uid
        self.icon = icon
        self.icontype = icontype
        self.type = type
        self.largetext = largetext
        self.copytext = copytext
        self.quicklookurl = quicklookurl

    @property
    def elem(self):
        """Create and return feedback item for Alfred.

        :returns: :class:`ElementTree.Element <xml.etree.ElementTree.Element>`
            instance for this :class:`Item` instance.

        """
        # Attributes on <item> element
        attr = {}
        if self.valid:
            attr['valid'] = 'yes'
        else:
            attr['valid'] = 'no'
        # Allow empty string for autocomplete. This is a useful value,
        # as TABing the result will revert the query back to just the
        # keyword
        if self.autocomplete is not None:
            attr['autocomplete'] = self.autocomplete

        # Optional attributes
        for name in ('uid', 'type'):
            value = getattr(self, name, None)
            if value:
                attr[name] = value

        root = ET.Element('item', attr)
        ET.SubElement(root, 'title').text = self.title
        ET.SubElement(root, 'subtitle').text = self.subtitle

        # Add modifier subtitles
        for mod in ('cmd', 'ctrl', 'alt', 'shift', 'fn'):
            if mod in self.modifier_subtitles:
                ET.SubElement(root, 'subtitle',
                              {'mod': mod}).text = self.modifier_subtitles[mod]

        # Add arg as element instead of attribute on <item>, as it's more
        # flexible (newlines aren't allowed in attributes)
        if self.arg:
            ET.SubElement(root, 'arg').text = self.arg

        # Add icon if there is one
        if self.icon:
            if self.icontype:
                attr = dict(type=self.icontype)
            else:
                attr = {}
            ET.SubElement(root, 'icon', attr).text = self.icon

        if self.largetext:
            ET.SubElement(root, 'text',
                          {'type': 'largetype'}).text = self.largetext

        if self.copytext:
            ET.SubElement(root, 'text',
                          {'type': 'copy'}).text = self.copytext

        if self.quicklookurl:
            ET.SubElement(root, 'quicklookurl').text = self.quicklookurl

        return root


class LockFile(object):
    """Context manager to create lock files."""

    def __init__(self, protected_path, timeout=0, delay=0.05):
        """Create new :class:`LockFile` object."""
        self.lockfile = protected_path + '.lock'
        self.timeout = timeout
        self.delay = delay
        self._locked = False

    @property
    def locked(self):
        """`True` if file is locked by this instance."""
        return self._locked

    def acquire(self, blocking=True):
        """Acquire the lock if possible.

        If the lock is in use and ``blocking`` is ``False``, return
        ``False``.

        Otherwise, check every `self.delay` seconds until it acquires
        lock or exceeds `self.timeout` and raises an exception.

        """
        start = time.time()
        while True:
            try:
                fd = os.open(self.lockfile, os.O_CREAT | os.O_EXCL | os.O_RDWR)
                with os.fdopen(fd, 'w') as fd:
                    fd.write('{0}'.format(os.getpid()))
                break
            except OSError as err:
                if err.errno != errno.EEXIST:  # pragma: no cover
                    raise
                if self.timeout and (time.time() - start) >= self.timeout:
                    raise AcquisitionError('Lock acquisition timed out.')
                if not blocking:
                    return False
                time.sleep(self.delay)

        self._locked = True
        return True

    def release(self):
        """Release the lock by deleting `self.lockfile`."""
        self._locked = False
        os.unlink(self.lockfile)

    def __enter__(self):
        """Acquire lock."""
        self.acquire()
        return self

    def __exit__(self, typ, value, traceback):
        """Release lock."""
        self.release()

    def __del__(self):
        """Clear up `self.lockfile`."""
        if self._locked:  # pragma: no cover
            self.release()


@contextmanager
def atomic_writer(file_path, mode):
    """Atomic file writer.

    :param file_path: path of file to write to.
    :type file_path: ``unicode``
    :param mode: sames as for `func:open`
    :type mode: string

    .. versionadded:: 1.12

    Context manager that ensures the file is only written if the write
    succeeds. The data is first written to a temporary file.

    """
    temp_suffix = '.aw.temp'
    temp_file_path = file_path + temp_suffix
    with open(temp_file_path, mode) as file_obj:
        try:
            yield file_obj
            os.rename(temp_file_path, file_path)
        finally:
            try:
                os.remove(temp_file_path)
            except (OSError, IOError):
                pass


class uninterruptible(object):
    """Decorator that postpones SIGTERM until wrapped function is complete.

    .. versionadded:: 1.12

    Since version 2.7, Alfred allows Script Filters to be killed. If
    your workflow is killed in the middle of critical code (e.g.
    writing data to disk), this may corrupt your workflow's data.

    Use this decorator to wrap critical functions that *must* complete.
    If the script is killed while a wrapped function is executing,
    the SIGTERM will be caught and handled after your function has
    finished executing.

    Alfred-Workflow uses this internally to ensure its settings, data
    and cache writes complete.

    .. important::

        This decorator is NOT thread-safe.

    """

    def __init__(self, func, class_name=''):
        """Decorate `func`."""
        self.func = func
        self._caught_signal = None

    def signal_handler(self, signum, frame):
        """Called when process receives SIGTERM."""
        self._caught_signal = (signum, frame)

    def __call__(self, *args, **kwargs):
        """Trap ``SIGTERM`` and call wrapped function."""
        self._caught_signal = None
        # Register handler for SIGTERM, then call `self.func`
        self.old_signal_handler = signal.getsignal(signal.SIGTERM)
        signal.signal(signal.SIGTERM, self.signal_handler)

        self.func(*args, **kwargs)

        # Restore old signal handler
        signal.signal(signal.SIGTERM, self.old_signal_handler)

        # Handle any signal caught during execution
        if self._caught_signal is not None:
            signum, frame = self._caught_signal
            if callable(self.old_signal_handler):
                self.old_signal_handler(signum, frame)
            elif self.old_signal_handler == signal.SIG_DFL:
                sys.exit(0)

    def __get__(self, obj=None, klass=None):
        """Decorator API."""
        return self.__class__(self.func.__get__(obj, klass),
                              klass.__name__)


class Settings(dict):
    """A dictionary that saves itself when changed.

    Dictionary keys & values will be saved as a JSON file
    at ``filepath``. If the file does not exist, the dictionary
    (and settings file) will be initialised with ``defaults``.

    :param filepath: where to save the settings
    :type filepath: :class:`unicode`
    :param defaults: dict of default settings
    :type defaults: :class:`dict`


    An appropriate instance is provided by :class:`Workflow` instances at
    :attr:`Workflow.settings`.

    """

    def __init__(self, filepath, defaults=None):
        """Create new :class:`Settings` object."""
        super(Settings, self).__init__()
        self._filepath = filepath
        self._nosave = False
        self._original = {}
        if os.path.exists(self._filepath):
            self._load()
        elif defaults:
            for key, val in defaults.items():
                self[key] = val
            self.save()  # save default settings

    def _load(self):
        """Load cached settings from JSON file `self._filepath`."""
        self._nosave = True
        d = {}
        with open(self._filepath, 'rb') as file_obj:
            for key, value in json.load(file_obj, encoding='utf-8').items():
                d[key] = value
        self.update(d)
        self._original = deepcopy(d)
        self._nosave = False

    @uninterruptible
    def save(self):
        """Save settings to JSON file specified in ``self._filepath``.

        If you're using this class via :attr:`Workflow.settings`, which
        you probably are, ``self._filepath`` will be ``settings.json``
        in your workflow's data directory (see :attr:`~Workflow.datadir`).
        """
        if self._nosave:
            return
        data = {}
        data.update(self)
        # for key, value in self.items():
        #     data[key] = value
        with LockFile(self._filepath):
            with atomic_writer(self._filepath, 'wb') as file_obj:
                json.dump(data, file_obj, sort_keys=True, indent=2,
                          encoding='utf-8')

    # dict methods
    def __setitem__(self, key, value):
        """Implement :class:`dict` interface."""
        if self._original.get(key) != value:
            super(Settings, self).__setitem__(key, value)
            self.save()

    def __delitem__(self, key):
        """Implement :class:`dict` interface."""
        super(Settings, self).__delitem__(key)
        self.save()

    def update(self, *args, **kwargs):
        """Override :class:`dict` method to save on update."""
        super(Settings, self).update(*args, **kwargs)
        self.save()

    def setdefault(self, key, value=None):
        """Override :class:`dict` method to save on update."""
        ret = super(Settings, self).setdefault(key, value)
        self.save()
        return ret


class Workflow(object):
    """Create new :class:`Workflow` instance.

    :param default_settings: default workflow settings. If no settings file
        exists, :class:`Workflow.settings` will be pre-populated with
        ``default_settings``.
    :type default_settings: :class:`dict`
    :param update_settings: settings for updating your workflow from GitHub.
        This must be a :class:`dict` that contains ``github_slug`` and
        ``version`` keys. ``github_slug`` is of the form ``username/repo``
        and ``version`` **must** correspond to the tag of a release. The
        boolean ``prereleases`` key is optional and if ``True`` will
        override the :ref:`magic argument <magic-arguments>` preference.
        This is only recommended when the installed workflow is a pre-release.
        See :ref:`updates` for more information.
    :type update_settings: :class:`dict`
    :param input_encoding: encoding of command line arguments
    :type input_encoding: :class:`unicode`
    :param normalization: normalisation to apply to CLI args.
        See :meth:`Workflow.decode` for more details.
    :type normalization: :class:`unicode`
    :param capture_args: capture and act on ``workflow:*`` arguments. See
        :ref:`Magic arguments <magic-arguments>` for details.
    :type capture_args: :class:`Boolean`
    :param libraries: sequence of paths to directories containing
        libraries. These paths will be prepended to ``sys.path``.
    :type libraries: :class:`tuple` or :class:`list`
    :param help_url: URL to webpage where a user can ask for help with
        the workflow, report bugs, etc. This could be the GitHub repo
        or a page on AlfredForum.com. If your workflow throws an error,
        this URL will be displayed in the log and Alfred's debugger. It can
        also be opened directly in a web browser with the ``workflow:help``
        :ref:`magic argument <magic-arguments>`.
    :type help_url: :class:`unicode` or :class:`str`

    """

    # Which class to use to generate feedback items. You probably
    # won't want to change this
    item_class = Item

    def __init__(self, default_settings=None, update_settings=None,
                 input_encoding='utf-8', normalization='NFC',
                 capture_args=True, libraries=None,
                 help_url=None):
        """Create new :class:`Workflow` object."""
        self._default_settings = default_settings or {}
        self._update_settings = update_settings or {}
        self._input_encoding = input_encoding
        self._normalizsation = normalization
        self._capture_args = capture_args
        self.help_url = help_url
        self._workflowdir = None
        self._settings_path = None
        self._settings = None
        self._bundleid = None
        self._debugging = None
        self._name = None
        self._cache_serializer = 'cpickle'
        self._data_serializer = 'cpickle'
        self._info = None
        self._info_loaded = False
        self._logger = None
        self._items = []
        self._alfred_env = None
        # Version number of the workflow
        self._version = UNSET
        # Version from last workflow run
        self._last_version_run = UNSET
        # Cache for regex patterns created for filter keys
        self._search_pattern_cache = {}
        # Magic arguments
        #: The prefix for all magic arguments. Default is ``workflow:``
        self.magic_prefix = 'workflow:'
        #: Mapping of available magic arguments. The built-in magic
        #: arguments are registered by default. To add your own magic arguments
        #: (or override built-ins), add a key:value pair where the key is
        #: what the user should enter (prefixed with :attr:`magic_prefix`)
        #: and the value is a callable that will be called when the argument
        #: is entered. If you would like to display a message in Alfred, the
        #: function should return a ``unicode`` string.
        #:
        #: By default, the magic arguments documented
        #: :ref:`here <magic-arguments>` are registered.
        self.magic_arguments = {}

        self._register_default_magic()

        if libraries:
            sys.path = libraries + sys.path

    ####################################################################
    # API methods
    ####################################################################

    # info.plist contents and alfred_* environment variables  ----------

    @property
    def alfred_version(self):
        """Alfred version as :class:`~workflow.update.Version` object."""
        from update import Version
        return Version(self.alfred_env.get('version'))

    @property
    def alfred_env(self):
        """Dict of Alfred's environmental variables minus ``alfred_`` prefix.

        .. versionadded:: 1.7

        The variables Alfred 2.4+ exports are:

        ============================  =========================================
        Variable                      Description
        ============================  =========================================
        alfred_debug                  Set to ``1`` if Alfred's debugger is
                                      open, otherwise unset.
        alfred_preferences            Path to Alfred.alfredpreferences
                                      (where your workflows and settings are
                                      stored).
        alfred_preferences_localhash  Machine-specific preferences are stored
                                      in ``Alfred.alfredpreferences/preferences/local/<hash>``
                                      (see ``alfred_preferences`` above for
                                      the path to ``Alfred.alfredpreferences``)
        alfred_theme                  ID of selected theme
        alfred_theme_background       Background colour of selected theme in
                                      format ``rgba(r,g,b,a)``
        alfred_theme_subtext          Show result subtext.
                                      ``0`` = Always,
                                      ``1`` = Alternative actions only,
                                      ``2`` = Selected result only,
                                      ``3`` = Never
        alfred_version                Alfred version number, e.g. ``'2.4'``
        alfred_version_build          Alfred build number, e.g. ``277``
        alfred_workflow_bundleid      Bundle ID, e.g.
                                      ``net.deanishe.alfred-mailto``
        alfred_workflow_cache         Path to workflow's cache directory
        alfred_workflow_data          Path to workflow's data directory
        alfred_workflow_name          Name of current workflow
        alfred_workflow_uid           UID of workflow
        alfred_workflow_version       The version number specified in the
                                      workflow configuration sheet/info.plist
        ============================  =========================================

        **Note:** all values are Unicode strings except ``version_build`` and
        ``theme_subtext``, which are integers.

        :returns: ``dict`` of Alfred's environmental variables without the
            ``alfred_`` prefix, e.g. ``preferences``, ``workflow_data``.

        """
        if self._alfred_env is not None:
            return self._alfred_env

        data = {}

        for key in (
                'alfred_debug',
                'alfred_preferences',
                'alfred_preferences_localhash',
                'alfred_theme',
                'alfred_theme_background',
                'alfred_theme_subtext',
                'alfred_version',
                'alfred_version_build',
                'alfred_workflow_bundleid',
                'alfred_workflow_cache',
                'alfred_workflow_data',
                'alfred_workflow_name',
                'alfred_workflow_uid',
                'alfred_workflow_version'):

            value = os.getenv(key)

            if isinstance(value, str):
                if key in ('alfred_debug', 'alfred_version_build',
                           'alfred_theme_subtext'):
                    value = int(value)
                else:
                    value = self.decode(value)

            data[key[7:]] = value

        self._alfred_env = data

        return self._alfred_env

    @property
    def info(self):
        """:class:`dict` of ``info.plist`` contents."""
        if not self._info_loaded:
            self._load_info_plist()
        return self._info

    @property
    def bundleid(self):
        """Workflow bundle ID from environmental vars or ``info.plist``.

        :returns: bundle ID
        :rtype: ``unicode``

        """
        if not self._bundleid:
            if self.alfred_env.get('workflow_bundleid'):
                self._bundleid = self.alfred_env.get('workflow_bundleid')
            else:
                self._bundleid = unicode(self.info['bundleid'], 'utf-8')

        return self._bundleid

    @property
    def debugging(self):
        """Whether Alfred's debugger is open.

        :returns: ``True`` if Alfred's debugger is open.
        :rtype: ``bool``

        """
        if self._debugging is None:
            if self.alfred_env.get('debug') == 1:
                self._debugging = True
            else:
                self._debugging = False
        return self._debugging

    @property
    def name(self):
        """Workflow name from Alfred's environmental vars or ``info.plist``.

        :returns: workflow name
        :rtype: ``unicode``

        """
        if not self._name:
            if self.alfred_env.get('workflow_name'):
                self._name = self.decode(self.alfred_env.get('workflow_name'))
            else:
                self._name = self.decode(self.info['name'])

        return self._name

    @property
    def version(self):
        """Return the version of the workflow.

        .. versionadded:: 1.9.10

        Get the workflow version from environment variable,
        the ``update_settings`` dict passed on
        instantiation, the ``version`` file located in the workflow's
        root directory or ``info.plist``. Return ``None`` if none
        exists or :class:`ValueError` if the version number is invalid
        (i.e. not semantic).

        :returns: Version of the workflow (not Alfred-Workflow)
        :rtype: :class:`~workflow.update.Version` object

        """
        if self._version is UNSET:

            version = None
            # environment variable has priority
            if self.alfred_env.get('workflow_version'):
                version = self.alfred_env['workflow_version']

            # Try `update_settings`
            elif self._update_settings:
                version = self._update_settings.get('version')

            # `version` file
            if not version:
                filepath = self.workflowfile('version')

                if os.path.exists(filepath):
                    with open(filepath, 'rb') as fileobj:
                        version = fileobj.read()

            # info.plist
            if not version:
                version = self.info.get('version')

            if version:
                from update import Version
                version = Version(version)

            self._version = version

        return self._version

    # Workflow utility methods -----------------------------------------

    @property
    def args(self):
        """Return command line args as normalised unicode.

        Args are decoded and normalised via :meth:`~Workflow.decode`.

        The encoding and normalisation are the ``input_encoding`` and
        ``normalization`` arguments passed to :class:`Workflow` (``UTF-8``
        and ``NFC`` are the defaults).

        If :class:`Workflow` is called with ``capture_args=True``
        (the default), :class:`Workflow` will look for certain
        ``workflow:*`` args and, if found, perform the corresponding
        actions and exit the workflow.

        See :ref:`Magic arguments <magic-arguments>` for details.

        """
        msg = None
        args = [self.decode(arg) for arg in sys.argv[1:]]

        # Handle magic args
        if len(args) and self._capture_args:
            for name in self.magic_arguments:
                key = '{0}{1}'.format(self.magic_prefix, name)
                if key in args:
                    msg = self.magic_arguments[name]()

            if msg:
                self.logger.debug(msg)
                if not sys.stdout.isatty():  # Show message in Alfred
                    self.add_item(msg, valid=False, icon=ICON_INFO)
                    self.send_feedback()
                sys.exit(0)
        return args

    @property
    def cachedir(self):
        """Path to workflow's cache directory.

        The cache directory is a subdirectory of Alfred's own cache directory
        in ``~/Library/Caches``. The full path is:

        ``~/Library/Caches/com.runningwithcrayons.Alfred-X/Workflow Data/<bundle id>``

        ``Alfred-X`` may be ``Alfred-2`` or ``Alfred-3``.

        :returns: full path to workflow's cache directory
        :rtype: ``unicode``

        """
        if self.alfred_env.get('workflow_cache'):
            dirpath = self.alfred_env.get('workflow_cache')

        else:
            dirpath = self._default_cachedir

        return self._create(dirpath)

    @property
    def _default_cachedir(self):
        """Alfred 2's default cache directory."""
        return os.path.join(
            os.path.expanduser(
                '~/Library/Caches/com.runningwithcrayons.Alfred-2/'
                'Workflow Data/'),
            self.bundleid)

    @property
    def datadir(self):
        """Path to workflow's data directory.

        The data directory is a subdirectory of Alfred's own data directory in
        ``~/Library/Application Support``. The full path is:

        ``~/Library/Application Support/Alfred 2/Workflow Data/<bundle id>``

        :returns: full path to workflow data directory
        :rtype: ``unicode``

        """
        if self.alfred_env.get('workflow_data'):
            dirpath = self.alfred_env.get('workflow_data')

        else:
            dirpath = self._default_datadir

        return self._create(dirpath)

    @property
    def _default_datadir(self):
        """Alfred 2's default data directory."""
        return os.path.join(os.path.expanduser(
            '~/Library/Application Support/Alfred 2/Workflow Data/'),
            self.bundleid)

    @property
    def workflowdir(self):
        """Path to workflow's root directory (where ``info.plist`` is).

        :returns: full path to workflow root directory
        :rtype: ``unicode``

        """
        if not self._workflowdir:
            # Try the working directory first, then the directory
            # the library is in. CWD will be the workflow root if
            # a workflow is being run in Alfred
            candidates = [
                os.path.abspath(os.getcwdu()),
                os.path.dirname(os.path.abspath(os.path.dirname(__file__)))]

            # climb the directory tree until we find `info.plist`
            for dirpath in candidates:

                # Ensure directory path is Unicode
                dirpath = self.decode(dirpath)

                while True:
                    if os.path.exists(os.path.join(dirpath, 'info.plist')):
                        self._workflowdir = dirpath
                        break

                    elif dirpath == '/':
                        # no `info.plist` found
                        break

                    # Check the parent directory
                    dirpath = os.path.dirname(dirpath)

                # No need to check other candidates
                if self._workflowdir:
                    break

            if not self._workflowdir:
                raise IOError("'info.plist' not found in directory tree")

        return self._workflowdir

    def cachefile(self, filename):
        """Path to ``filename`` in workflow's cache directory.

        Return absolute path to ``filename`` within your workflow's
        :attr:`cache directory <Workflow.cachedir>`.

        :param filename: basename of file
        :type filename: ``unicode``
        :returns: full path to file within cache directory
        :rtype: ``unicode``

        """
        return os.path.join(self.cachedir, filename)

    def datafile(self, filename):
        """Path to ``filename`` in workflow's data directory.

        Return absolute path to ``filename`` within your workflow's
        :attr:`data directory <Workflow.datadir>`.

        :param filename: basename of file
        :type filename: ``unicode``
        :returns: full path to file within data directory
        :rtype: ``unicode``

        """
        return os.path.join(self.datadir, filename)

    def workflowfile(self, filename):
        """Return full path to ``filename`` in workflow's root directory.

        :param filename: basename of file
        :type filename: ``unicode``
        :returns: full path to file within data directory
        :rtype: ``unicode``

        """
        return os.path.join(self.workflowdir, filename)

    @property
    def logfile(self):
        """Path to logfile.

        :returns: path to logfile within workflow's cache directory
        :rtype: ``unicode``

        """
        return self.cachefile('%s.log' % self.bundleid)

    @property
    def logger(self):
        """Logger that logs to both console and a log file.

        If Alfred's debugger is open, log level will be ``DEBUG``,
        else it will be ``INFO``.

        Use :meth:`open_log` to open the log file in Console.

        :returns: an initialised :class:`~logging.Logger`

        """
        if self._logger:
            return self._logger

        # Initialise new logger and optionally handlers
        logger = logging.getLogger('workflow')

        if not len(logger.handlers):  # Only add one set of handlers

            fmt = logging.Formatter(
                '%(asctime)s %(filename)s:%(lineno)s'
                ' %(levelname)-8s %(message)s',
                datefmt='%H:%M:%S')

            logfile = logging.handlers.RotatingFileHandler(
                self.logfile,
                maxBytes=1024 * 1024,
                backupCount=1)
            logfile.setFormatter(fmt)
            logger.addHandler(logfile)

            console = logging.StreamHandler()
            console.setFormatter(fmt)
            logger.addHandler(console)

        if self.debugging:
            logger.setLevel(logging.DEBUG)
        else:
            logger.setLevel(logging.INFO)

        self._logger = logger

        return self._logger

    @logger.setter
    def logger(self, logger):
        """Set a custom logger.

        :param logger: The logger to use
        :type logger: `~logging.Logger` instance

        """
        self._logger = logger

    @property
    def settings_path(self):
        """Path to settings file within workflow's data directory.

        :returns: path to ``settings.json`` file
        :rtype: ``unicode``

        """
        if not self._settings_path:
            self._settings_path = self.datafile('settings.json')
        return self._settings_path

    @property
    def settings(self):
        """Return a dictionary subclass that saves itself when changed.

        See :ref:`manual-settings` in the :ref:`user-manual` for more
        information on how to use :attr:`settings` and **important
        limitations** on what it can do.

        :returns: :class:`~workflow.workflow.Settings` instance
            initialised from the data in JSON file at
            :attr:`settings_path` or if that doesn't exist, with the
            ``default_settings`` :class:`dict` passed to
            :class:`Workflow` on instantiation.
        :rtype: :class:`~workflow.workflow.Settings` instance

        """
        if not self._settings:
            self.logger.debug('Reading settings from `{0}` ...'.format(
                              self.settings_path))
            self._settings = Settings(self.settings_path,
                                      self._default_settings)
        return self._settings

    @property
    def cache_serializer(self):
        """Name of default cache serializer.

        .. versionadded:: 1.8

        This serializer is used by :meth:`cache_data()` and
        :meth:`cached_data()`

        See :class:`SerializerManager` for details.

        :returns: serializer name
        :rtype: ``unicode``

        """
        return self._cache_serializer

    @cache_serializer.setter
    def cache_serializer(self, serializer_name):
        """Set the default cache serialization format.

        .. versionadded:: 1.8

        This serializer is used by :meth:`cache_data()` and
        :meth:`cached_data()`

        The specified serializer must already by registered with the
        :class:`SerializerManager` at `~workflow.workflow.manager`,
        otherwise a :class:`ValueError` will be raised.

        :param serializer_name: Name of default serializer to use.
        :type serializer_name:

        """
        if manager.serializer(serializer_name) is None:
            raise ValueError(
                'Unknown serializer : `{0}`. Register your serializer '
                'with `manager` first.'.format(serializer_name))

        self.logger.debug(
            'default cache serializer set to `{0}`'.format(serializer_name))

        self._cache_serializer = serializer_name

    @property
    def data_serializer(self):
        """Name of default data serializer.

        .. versionadded:: 1.8

        This serializer is used by :meth:`store_data()` and
        :meth:`stored_data()`

        See :class:`SerializerManager` for details.

        :returns: serializer name
        :rtype: ``unicode``

        """
        return self._data_serializer

    @data_serializer.setter
    def data_serializer(self, serializer_name):
        """Set the default cache serialization format.

        .. versionadded:: 1.8

        This serializer is used by :meth:`store_data()` and
        :meth:`stored_data()`

        The specified serializer must already by registered with the
        :class:`SerializerManager` at `~workflow.workflow.manager`,
        otherwise a :class:`ValueError` will be raised.

        :param serializer_name: Name of serializer to use by default.

        """
        if manager.serializer(serializer_name) is None:
            raise ValueError(
                'Unknown serializer : `{0}`. Register your serializer '
                'with `manager` first.'.format(serializer_name))

        self.logger.debug(
            'default data serializer set to `{0}`'.format(serializer_name))

        self._data_serializer = serializer_name

    def stored_data(self, name):
        """Retrieve data from data directory.

        Returns ``None`` if there are no data stored under ``name``.

        .. versionadded:: 1.8

        :param name: name of datastore

        """
        metadata_path = self.datafile('.{0}.alfred-workflow'.format(name))

        if not os.path.exists(metadata_path):
            self.logger.debug('No data stored for `{0}`'.format(name))
            return None

        with open(metadata_path, 'rb') as file_obj:
            serializer_name = file_obj.read().strip()

        serializer = manager.serializer(serializer_name)

        if serializer is None:
            raise ValueError(
                'Unknown serializer `{0}`. Register a corresponding '
                'serializer with `manager.register()` '
                'to load this data.'.format(serializer_name))

        self.logger.debug('Data `{0}` stored in `{1}` format'.format(
            name, serializer_name))

        filename = '{0}.{1}'.format(name, serializer_name)
        data_path = self.datafile(filename)

        if not os.path.exists(data_path):
            self.logger.debug('No data stored for `{0}`'.format(name))
            if os.path.exists(metadata_path):
                os.unlink(metadata_path)

            return None

        with open(data_path, 'rb') as file_obj:
            data = serializer.load(file_obj)

        self.logger.debug('Stored data loaded from : {0}'.format(data_path))

        return data

    def store_data(self, name, data, serializer=None):
        """Save data to data directory.

        .. versionadded:: 1.8

        If ``data`` is ``None``, the datastore will be deleted.

        Note that the datastore does NOT support mutliple threads.

        :param name: name of datastore
        :param data: object(s) to store. **Note:** some serializers
            can only handled certain types of data.
        :param serializer: name of serializer to use. If no serializer
            is specified, the default will be used. See
            :class:`SerializerManager` for more information.
        :returns: data in datastore or ``None``

        """
        # Ensure deletion is not interrupted by SIGTERM
        @uninterruptible
        def delete_paths(paths):
            """Clear one or more data stores"""
            for path in paths:
                if os.path.exists(path):
                    os.unlink(path)
                    self.logger.debug('Deleted data file : {0}'.format(path))

        serializer_name = serializer or self.data_serializer

        # In order for `stored_data()` to be able to load data stored with
        # an arbitrary serializer, yet still have meaningful file extensions,
        # the format (i.e. extension) is saved to an accompanying file
        metadata_path = self.datafile('.{0}.alfred-workflow'.format(name))
        filename = '{0}.{1}'.format(name, serializer_name)
        data_path = self.datafile(filename)

        if data_path == self.settings_path:
            raise ValueError(
                'Cannot save data to' +
                '`{0}` with format `{1}`. '.format(name, serializer_name) +
                "This would overwrite Alfred-Workflow's settings file.")

        serializer = manager.serializer(serializer_name)

        if serializer is None:
            raise ValueError(
                'Invalid serializer `{0}`. Register your serializer with '
                '`manager.register()` first.'.format(serializer_name))

        if data is None:  # Delete cached data
            delete_paths((metadata_path, data_path))
            return

        # Ensure write is not interrupted by SIGTERM
        @uninterruptible
        def _store():
            # Save file extension
            with atomic_writer(metadata_path, 'wb') as file_obj:
                file_obj.write(serializer_name)

            with atomic_writer(data_path, 'wb') as file_obj:
                serializer.dump(data, file_obj)

        _store()

        self.logger.debug('Stored data saved at : {0}'.format(data_path))

    def cached_data(self, name, data_func=None, max_age=60):
        """Return cached data if younger than ``max_age`` seconds.

        Retrieve data from cache or re-generate and re-cache data if
        stale/non-existant. If ``max_age`` is 0, return cached data no
        matter how old.

        :param name: name of datastore
        :param data_func: function to (re-)generate data.
        :type data_func: ``callable``
        :param max_age: maximum age of cached data in seconds
        :type max_age: ``int``
        :returns: cached data, return value of ``data_func`` or ``None``
            if ``data_func`` is not set

        """
        serializer = manager.serializer(self.cache_serializer)

        cache_path = self.cachefile('%s.%s' % (name, self.cache_serializer))
        age = self.cached_data_age(name)

        if (age < max_age or max_age == 0) and os.path.exists(cache_path):

            with open(cache_path, 'rb') as file_obj:
                self.logger.debug('Loading cached data from : %s',
                                  cache_path)
                return serializer.load(file_obj)

        if not data_func:
            return None

        data = data_func()
        self.cache_data(name, data)

        return data

    def cache_data(self, name, data):
        """Save ``data`` to cache under ``name``.

        If ``data`` is ``None``, the corresponding cache file will be
        deleted.

        :param name: name of datastore
        :param data: data to store. This may be any object supported by
                the cache serializer

        """
        serializer = manager.serializer(self.cache_serializer)

        cache_path = self.cachefile('%s.%s' % (name, self.cache_serializer))

        if data is None:
            if os.path.exists(cache_path):
                os.unlink(cache_path)
                self.logger.debug('Deleted cache file : %s', cache_path)
            return

        with atomic_writer(cache_path, 'wb') as file_obj:
            serializer.dump(data, file_obj)

        self.logger.debug('Cached data saved at : %s', cache_path)

    def cached_data_fresh(self, name, max_age):
        """Whether cache `name` is less than `max_age` seconds old.

        :param name: name of datastore
        :param max_age: maximum age of data in seconds
        :type max_age: ``int``
        :returns: ``True`` if data is less than ``max_age`` old, else
            ``False``

        """
        age = self.cached_data_age(name)

        if not age:
            return False

        return age < max_age

    def cached_data_age(self, name):
        """Return age in seconds of cache `name` or 0 if cache doesn't exist.

        :param name: name of datastore
        :type name: ``unicode``
        :returns: age of datastore in seconds
        :rtype: ``int``

        """
        cache_path = self.cachefile('%s.%s' % (name, self.cache_serializer))

        if not os.path.exists(cache_path):
            return 0

        return time.time() - os.stat(cache_path).st_mtime

    def filter(self, query, items, key=lambda x: x, ascending=False,
               include_score=False, min_score=0, max_results=0,
               match_on=MATCH_ALL, fold_diacritics=True):
        """Fuzzy search filter. Returns list of ``items`` that match ``query``.

        ``query`` is case-insensitive. Any item that does not contain the
        entirety of ``query`` is rejected.

        .. warning::

            If ``query`` is an empty string or contains only whitespace,
            a :class:`ValueError` will be raised.

        :param query: query to test items against
        :type query: ``unicode``
        :param items: iterable of items to test
        :type items: ``list`` or ``tuple``
        :param key: function to get comparison key from ``items``.
            Must return a ``unicode`` string. The default simply returns
            the item.
        :type key: ``callable``
        :param ascending: set to ``True`` to get worst matches first
        :type ascending: ``Boolean``
        :param include_score: Useful for debugging the scoring algorithm.
            If ``True``, results will be a list of tuples
            ``(item, score, rule)``.
        :type include_score: ``Boolean``
        :param min_score: If non-zero, ignore results with a score lower
            than this.
        :type min_score: ``int``
        :param max_results: If non-zero, prune results list to this length.
        :type max_results: ``int``
        :param match_on: Filter option flags. Bitwise-combined list of
            ``MATCH_*`` constants (see below).
        :type match_on: ``int``
        :param fold_diacritics: Convert search keys to ASCII-only
            characters if ``query`` only contains ASCII characters.
        :type fold_diacritics: ``Boolean``
        :returns: list of ``items`` matching ``query`` or list of
            ``(item, score, rule)`` `tuples` if ``include_score`` is ``True``.
            ``rule`` is the ``MATCH_*`` rule that matched the item.
        :rtype: ``list``

        **Matching rules**

        By default, :meth:`filter` uses all of the following flags (i.e.
        :const:`MATCH_ALL`). The tests are always run in the given order:

        1. :const:`MATCH_STARTSWITH` : Item search key startswith
            ``query``(case-insensitive).
        2. :const:`MATCH_CAPITALS` : The list of capital letters in item
            search key starts with ``query`` (``query`` may be
            lower-case). E.g., ``of`` would match ``OmniFocus``,
            ``gc`` would match ``Google Chrome``.
        3. :const:`MATCH_ATOM` : Search key is split into "atoms" on
            non-word characters (.,-,' etc.). Matches if ``query`` is
            one of these atoms (case-insensitive).
        4. :const:`MATCH_INITIALS_STARTSWITH` : Initials are the first
            characters of the above-described "atoms" (case-insensitive).
        5. :const:`MATCH_INITIALS_CONTAIN` : ``query`` is a substring of
            the above-described initials.
        6. :const:`MATCH_INITIALS` : Combination of (4) and (5).
        7. :const:`MATCH_SUBSTRING` : Match if ``query`` is a substring
            of item search key (case-insensitive).
        8. :const:`MATCH_ALLCHARS` : Matches if all characters in
            ``query`` appear in item search key in the same order
            (case-insensitive).
        9. :const:`MATCH_ALL` : Combination of all the above.


        :const:`MATCH_ALLCHARS` is considerably slower than the other
        tests and provides much less accurate results.

        **Examples:**

        To ignore :const:`MATCH_ALLCHARS` (tends to provide the worst
        matches and is expensive to run), use
        ``match_on=MATCH_ALL ^ MATCH_ALLCHARS``.

        To match only on capitals, use ``match_on=MATCH_CAPITALS``.

        To match only on startswith and substring, use
        ``match_on=MATCH_STARTSWITH | MATCH_SUBSTRING``.

        **Diacritic folding**

        .. versionadded:: 1.3

        If ``fold_diacritics`` is ``True`` (the default), and ``query``
        contains only ASCII characters, non-ASCII characters in search keys
        will be converted to ASCII equivalents (e.g. **ü** -> **u**,
        **ß** -> **ss**, **é** -> **e**).

        See :const:`ASCII_REPLACEMENTS` for all replacements.

        If ``query`` contains non-ASCII characters, search keys will not be
        altered.

        """
        if not query:
            raise ValueError('Empty `query`')

        # Remove preceding/trailing spaces
        query = query.strip()

        if not query:
            raise ValueError('`query` contains only whitespace')

        # Use user override if there is one
        fold_diacritics = self.settings.get('__workflow_diacritic_folding',
                                            fold_diacritics)

        results = []

        for item in items:
            skip = False
            score = 0
            words = [s.strip() for s in query.split(' ')]
            value = key(item).strip()
            if value == '':
                continue
            for word in words:
                if word == '':
                    continue
                s, rule = self._filter_item(value, word, match_on,
                                            fold_diacritics)

                if not s:  # Skip items that don't match part of the query
                    skip = True
                score += s

            if skip:
                continue

            if score:
                # use "reversed" `score` (i.e. highest becomes lowest) and
                # `value` as sort key. This means items with the same score
                # will be sorted in alphabetical not reverse alphabetical order
                results.append(((100.0 / score, value.lower(), score),
                                (item, score, rule)))

        # sort on keys, then discard the keys
        results.sort(reverse=ascending)
        results = [t[1] for t in results]

        if min_score:
            results = [r for r in results if r[1] > min_score]

        if max_results and len(results) > max_results:
            results = results[:max_results]

        # return list of ``(item, score, rule)``
        if include_score:
            return results
        # just return list of items
        return [t[0] for t in results]

    def _filter_item(self, value, query, match_on, fold_diacritics):
        """Filter ``value`` against ``query`` using rules ``match_on``.

        :returns: ``(score, rule)``

        """
        query = query.lower()

        if not isascii(query):
            fold_diacritics = False

        if fold_diacritics:
            value = self.fold_to_ascii(value)

        # pre-filter any items that do not contain all characters
        # of ``query`` to save on running several more expensive tests
        if not set(query) <= set(value.lower()):

            return (0, None)

        # item starts with query
        if match_on & MATCH_STARTSWITH and value.lower().startswith(query):
            score = 100.0 - (len(value) / len(query))

            return (score, MATCH_STARTSWITH)

        # query matches capitalised letters in item,
        # e.g. of = OmniFocus
        if match_on & MATCH_CAPITALS:
            initials = ''.join([c for c in value if c in INITIALS])
            if initials.lower().startswith(query):
                score = 100.0 - (len(initials) / len(query))

                return (score, MATCH_CAPITALS)

        # split the item into "atoms", i.e. words separated by
        # spaces or other non-word characters
        if (match_on & MATCH_ATOM or
                match_on & MATCH_INITIALS_CONTAIN or
                match_on & MATCH_INITIALS_STARTSWITH):
            atoms = [s.lower() for s in split_on_delimiters(value)]
            # print('atoms : %s  -->  %s' % (value, atoms))
            # initials of the atoms
            initials = ''.join([s[0] for s in atoms if s])

        if match_on & MATCH_ATOM:
            # is `query` one of the atoms in item?
            # similar to substring, but scores more highly, as it's
            # a word within the item
            if query in atoms:
                score = 100.0 - (len(value) / len(query))

                return (score, MATCH_ATOM)

        # `query` matches start (or all) of the initials of the
        # atoms, e.g. ``himym`` matches "How I Met Your Mother"
        # *and* "how i met your mother" (the ``capitals`` rule only
        # matches the former)
        if (match_on & MATCH_INITIALS_STARTSWITH and
                initials.startswith(query)):
            score = 100.0 - (len(initials) / len(query))

            return (score, MATCH_INITIALS_STARTSWITH)

        # `query` is a substring of initials, e.g. ``doh`` matches
        # "The Dukes of Hazzard"
        elif (match_on & MATCH_INITIALS_CONTAIN and
                query in initials):
            score = 95.0 - (len(initials) / len(query))

            return (score, MATCH_INITIALS_CONTAIN)

        # `query` is a substring of item
        if match_on & MATCH_SUBSTRING and query in value.lower():
            score = 90.0 - (len(value) / len(query))

            return (score, MATCH_SUBSTRING)

        # finally, assign a score based on how close together the
        # characters in `query` are in item.
        if match_on & MATCH_ALLCHARS:
            search = self._search_for_query(query)
            match = search(value)
            if match:
                score = 100.0 / ((1 + match.start()) *
                                 (match.end() - match.start() + 1))

                return (score, MATCH_ALLCHARS)

        # Nothing matched
        return (0, None)

    def _search_for_query(self, query):
        if query in self._search_pattern_cache:
            return self._search_pattern_cache[query]

        # Build pattern: include all characters
        pattern = []
        for c in query:
            # pattern.append('[^{0}]*{0}'.format(re.escape(c)))
            pattern.append('.*?{0}'.format(re.escape(c)))
        pattern = ''.join(pattern)
        search = re.compile(pattern, re.IGNORECASE).search

        self._search_pattern_cache[query] = search
        return search

    def run(self, func, text_errors=False):
        """Call ``func`` to run your workflow.

        :param func: Callable to call with ``self`` (i.e. the :class:`Workflow`
            instance) as first argument.
        :param text_errors: Emit error messages in plain text, not in
            Alfred's XML/JSON feedback format. Use this when you're not
            running Alfred-Workflow in a Script Filter and would like
            to pass the error message to, say, a notification.
        :type text_errors: ``Boolean``

        ``func`` will be called with :class:`Workflow` instance as first
        argument.

        ``func`` should be the main entry point to your workflow.

        Any exceptions raised will be logged and an error message will be
        output to Alfred.

        """
        start = time.time()

        # Call workflow's entry function/method within a try-except block
        # to catch any errors and display an error message in Alfred
        try:

            if self.version:
                self.logger.debug(
                    'Workflow version : {0}'.format(self.version))

            # Run update check if configured for self-updates.
            # This call has to go in the `run` try-except block, as it will
            # initialise `self.settings`, which will raise an exception
            # if `settings.json` isn't valid.

            if self._update_settings:
                self.check_update()

            # Run workflow's entry function/method
            func(self)

            # Set last version run to current version after a successful
            # run
            self.set_last_version()

        except Exception as err:
            self.logger.exception(err)
            if self.help_url:
                self.logger.info(
                    'For assistance, see: {0}'.format(self.help_url))

            if not sys.stdout.isatty():  # Show error in Alfred
                if text_errors:
                    print(unicode(err).encode('utf-8'), end='')
                else:
                    self._items = []
                    if self._name:
                        name = self._name
                    elif self._bundleid:
                        name = self._bundleid
                    else:  # pragma: no cover
                        name = os.path.dirname(__file__)
                    self.add_item("Error in workflow '%s'" % name,
                                  unicode(err),
                                  icon=ICON_ERROR)
                    self.send_feedback()
            return 1

        finally:
            self.logger.debug('Workflow finished in {0:0.3f} seconds.'.format(
                time.time() - start))

        return 0

    # Alfred feedback methods ------------------------------------------

    def add_item(self, title, subtitle='', modifier_subtitles=None, arg=None,
                 autocomplete=None, valid=False, uid=None, icon=None,
                 icontype=None, type=None, largetext=None, copytext=None,
                 quicklookurl=None):
        """Add an item to be output to Alfred.

        :param title: Title shown in Alfred
        :type title: ``unicode``
        :param subtitle: Subtitle shown in Alfred
        :type subtitle: ``unicode``
        :param modifier_subtitles: Subtitles shown when modifier
            (CMD, OPT etc.) is pressed. Use a ``dict`` with the lowercase
            keys ``cmd``, ``ctrl``, ``shift``, ``alt`` and ``fn``
        :type modifier_subtitles: ``dict``
        :param arg: Argument passed by Alfred as ``{query}`` when item is
            actioned
        :type arg: ``unicode``
        :param autocomplete: Text expanded in Alfred when item is TABbed
        :type autocomplete: ``unicode``
        :param valid: Whether or not item can be actioned
        :type valid: ``Boolean``
        :param uid: Used by Alfred to remember/sort items
        :type uid: ``unicode``
        :param icon: Filename of icon to use
        :type icon: ``unicode``
        :param icontype: Type of icon. Must be one of ``None`` , ``'filetype'``
           or ``'fileicon'``. Use ``'filetype'`` when ``icon`` is a filetype
           such as ``'public.folder'``. Use ``'fileicon'`` when you wish to
           use the icon of the file specified as ``icon``, e.g.
           ``icon='/Applications/Safari.app', icontype='fileicon'``.
           Leave as `None` if ``icon`` points to an actual
           icon file.
        :type icontype: ``unicode``
        :param type: Result type. Currently only ``'file'`` is supported
            (by Alfred). This will tell Alfred to enable file actions for
            this item.
        :type type: ``unicode``
        :param largetext: Text to be displayed in Alfred's large text box
            if user presses CMD+L on item.
        :type largetext: ``unicode``
        :param copytext: Text to be copied to pasteboard if user presses
            CMD+C on item.
        :type copytext: ``unicode``
        :param quicklookurl: URL to be displayed using Alfred's Quick Look
            feature (tapping ``SHIFT`` or ``⌘+Y`` on a result).
        :type quicklookurl: ``unicode``
        :returns: :class:`Item` instance

        See the :ref:`script-filter-results` section of the documentation
        for a detailed description of what the various parameters do and how
        they interact with one another.

        See :ref:`icons` for a list of the supported system icons.

        .. note::

            Although this method returns an :class:`Item` instance, you don't
            need to hold onto it or worry about it. All generated :class:`Item`
            instances are also collected internally and sent to Alfred when
            :meth:`send_feedback` is called.

            The generated :class:`Item` is only returned in case you want to
            edit it or do something with it other than send it to Alfred.

        """
        item = self.item_class(title, subtitle, modifier_subtitles, arg,
                               autocomplete, valid, uid, icon, icontype, type,
                               largetext, copytext, quicklookurl)
        self._items.append(item)
        return item

    def send_feedback(self):
        """Print stored items to console/Alfred as XML."""
        root = ET.Element('items')
        for item in self._items:
            root.append(item.elem)
        sys.stdout.write('<?xml version="1.0" encoding="utf-8"?>\n')
        sys.stdout.write(ET.tostring(root).encode('utf-8'))
        sys.stdout.flush()

    ####################################################################
    # Updating methods
    ####################################################################

    @property
    def first_run(self):
        """Return ``True`` if it's the first time this version has run.

        .. versionadded:: 1.9.10

        Raises a :class:`ValueError` if :attr:`version` isn't set.

        """
        if not self.version:
            raise ValueError('No workflow version set')

        if not self.last_version_run:
            return True

        return self.version != self.last_version_run

    @property
    def last_version_run(self):
        """Return version of last version to run (or ``None``).

        .. versionadded:: 1.9.10

        :returns: :class:`~workflow.update.Version` instance
            or ``None``

        """
        if self._last_version_run is UNSET:

            version = self.settings.get('__workflow_last_version')
            if version:
                from update import Version
                version = Version(version)

            self._last_version_run = version

        self.logger.debug('Last run version : {0}'.format(
                          self._last_version_run))

        return self._last_version_run

    def set_last_version(self, version=None):
        """Set :attr:`last_version_run` to current version.

        .. versionadded:: 1.9.10

        :param version: version to store (default is current version)
        :type version: :class:`~workflow.update.Version` instance
            or ``unicode``
        :returns: ``True`` if version is saved, else ``False``

        """
        if not version:
            if not self.version:
                self.logger.warning(
                    "Can't save last version: workflow has no version")
                return False

            version = self.version

        if isinstance(version, basestring):
            from update import Version
            version = Version(version)

        self.settings['__workflow_last_version'] = str(version)

        self.logger.debug('Set last run version : {0}'.format(version))

        return True

    @property
    def update_available(self):
        """Whether an update is available.

        .. versionadded:: 1.9

        See :ref:`manual-updates` in the :ref:`user-manual` for detailed
        information on how to enable your workflow to update itself.

        :returns: ``True`` if an update is available, else ``False``

        """
        update_data = self.cached_data('__workflow_update_status', max_age=0)
        self.logger.debug('update_data : {0}'.format(update_data))

        if not update_data or not update_data.get('available'):
            return False

        return update_data['available']

    @property
    def prereleases(self):
        """Whether workflow should update to pre-release versions.

        .. versionadded:: 1.16

        :returns: ``True`` if pre-releases are enabled with the :ref:`magic
            argument <magic-arguments>` or the ``update_settings`` dict, else
            ``False``.

        """
        if self._update_settings.get('prereleases'):
            return True

        return self.settings.get('__workflow_prereleases') or False

    def check_update(self, force=False):
        """Call update script if it's time to check for a new release.

        .. versionadded:: 1.9

        The update script will be run in the background, so it won't
        interfere in the execution of your workflow.

        See :ref:`manual-updates` in the :ref:`user-manual` for detailed
        information on how to enable your workflow to update itself.

        :param force: Force update check
        :type force: ``Boolean``

        """
        frequency = self._update_settings.get('frequency',
                                              DEFAULT_UPDATE_FREQUENCY)

        if not force and not self.settings.get('__workflow_autoupdate', True):
            self.logger.debug('Auto update turned off by user')
            return

        # Check for new version if it's time
        if (force or not self.cached_data_fresh(
                '__workflow_update_status', frequency * 86400)):

            github_slug = self._update_settings['github_slug']
            # version = self._update_settings['version']
            version = str(self.version)

            from background import run_in_background

            # update.py is adjacent to this file
            update_script = os.path.join(os.path.dirname(__file__),
                                         b'update.py')

            cmd = ['/usr/bin/python', update_script, 'check', github_slug,
                   version]

            if self.prereleases:
                cmd.append('--prereleases')

            self.logger.info('Checking for update ...')

            run_in_background('__workflow_update_check', cmd)

        else:
            self.logger.debug('Update check not due')

    def start_update(self):
        """Check for update and download and install new workflow file.

        .. versionadded:: 1.9

        See :ref:`manual-updates` in the :ref:`user-manual` for detailed
        information on how to enable your workflow to update itself.

        :returns: ``True`` if an update is available and will be
            installed, else ``False``

        """
        import update

        github_slug = self._update_settings['github_slug']
        # version = self._update_settings['version']
        version = str(self.version)

        if not update.check_update(github_slug, version, self.prereleases):
            return False

        from background import run_in_background

        # update.py is adjacent to this file
        update_script = os.path.join(os.path.dirname(__file__),
                                     b'update.py')

        cmd = ['/usr/bin/python', update_script, 'install', github_slug,
               version]

        if self.prereleases:
            cmd.append('--prereleases')

        self.logger.debug('Downloading update ...')
        run_in_background('__workflow_update_install', cmd)

        return True

    ####################################################################
    # Keychain password storage methods
    ####################################################################

    def save_password(self, account, password, service=None):
        """Save account credentials.

        If the account exists, the old password will first be deleted
        (Keychain throws an error otherwise).

        If something goes wrong, a :class:`KeychainError` exception will
        be raised.

        :param account: name of the account the password is for, e.g.
            "Pinboard"
        :type account: ``unicode``
        :param password: the password to secure
        :type password: ``unicode``
        :param service: Name of the service. By default, this is the
            workflow's bundle ID
        :type service: ``unicode``

        """
        if not service:
            service = self.bundleid

        try:
            self._call_security('add-generic-password', service, account,
                                '-w', password)
            self.logger.debug('Saved password : %s:%s', service, account)

        except PasswordExists:
            self.logger.debug('Password exists : %s:%s', service, account)
            current_password = self.get_password(account, service)

            if current_password == password:
                self.logger.debug('Password unchanged')

            else:
                self.delete_password(account, service)
                self._call_security('add-generic-password', service,
                                    account, '-w', password)
                self.logger.debug('save_password : %s:%s', service, account)

    def get_password(self, account, service=None):
        """Retrieve the password saved at ``service/account``.

        Raise :class:`PasswordNotFound` exception if password doesn't exist.

        :param account: name of the account the password is for, e.g.
            "Pinboard"
        :type account: ``unicode``
        :param service: Name of the service. By default, this is the workflow's
                        bundle ID
        :type service: ``unicode``
        :returns: account password
        :rtype: ``unicode``

        """
        if not service:
            service = self.bundleid

        output = self._call_security('find-generic-password', service,
                                     account, '-g')

        # Parsing of `security` output is adapted from python-keyring
        # by Jason R. Coombs
        # https://pypi.python.org/pypi/keyring
        m = re.search(
            r'password:\s*(?:0x(?P<hex>[0-9A-F]+)\s*)?(?:"(?P<pw>.*)")?',
            output)

        if m:
            groups = m.groupdict()
            h = groups.get('hex')
            password = groups.get('pw')
            if h:
                password = unicode(binascii.unhexlify(h), 'utf-8')

        self.logger.debug('Got password : %s:%s', service, account)

        return password

    def delete_password(self, account, service=None):
        """Delete the password stored at ``service/account``.

        Raise :class:`PasswordNotFound` if account is unknown.

        :param account: name of the account the password is for, e.g.
            "Pinboard"
        :type account: ``unicode``
        :param service: Name of the service. By default, this is the workflow's
                        bundle ID
        :type service: ``unicode``

        """
        if not service:
            service = self.bundleid

        self._call_security('delete-generic-password', service, account)

        self.logger.debug('Deleted password : %s:%s', service, account)

    ####################################################################
    # Methods for workflow:* magic args
    ####################################################################

    def _register_default_magic(self):
        """Register the built-in magic arguments."""
        # TODO: refactor & simplify
        # Wrap callback and message with callable
        def callback(func, msg):
            def wrapper():
                func()
                return msg

            return wrapper

        self.magic_arguments['delcache'] = callback(self.clear_cache,
                                                    'Deleted workflow cache')
        self.magic_arguments['deldata'] = callback(self.clear_data,
                                                   'Deleted workflow data')
        self.magic_arguments['delsettings'] = callback(
            self.clear_settings, 'Deleted workflow settings')
        self.magic_arguments['reset'] = callback(self.reset,
                                                 'Reset workflow')
        self.magic_arguments['openlog'] = callback(self.open_log,
                                                   'Opening workflow log file')
        self.magic_arguments['opencache'] = callback(
            self.open_cachedir, 'Opening workflow cache directory')
        self.magic_arguments['opendata'] = callback(
            self.open_datadir, 'Opening workflow data directory')
        self.magic_arguments['openworkflow'] = callback(
            self.open_workflowdir, 'Opening workflow directory')
        self.magic_arguments['openterm'] = callback(
            self.open_terminal, 'Opening workflow root directory in Terminal')

        # Diacritic folding
        def fold_on():
            self.settings['__workflow_diacritic_folding'] = True
            return 'Diacritics will always be folded'

        def fold_off():
            self.settings['__workflow_diacritic_folding'] = False
            return 'Diacritics will never be folded'

        def fold_default():
            if '__workflow_diacritic_folding' in self.settings:
                del self.settings['__workflow_diacritic_folding']
            return 'Diacritics folding reset'

        self.magic_arguments['foldingon'] = fold_on
        self.magic_arguments['foldingoff'] = fold_off
        self.magic_arguments['foldingdefault'] = fold_default

        # Updates
        def update_on():
            self.settings['__workflow_autoupdate'] = True
            return 'Auto update turned on'

        def update_off():
            self.settings['__workflow_autoupdate'] = False
            return 'Auto update turned off'

        def prereleases_on():
            self.settings['__workflow_prereleases'] = True
            return 'Prerelease updates turned on'

        def prereleases_off():
            self.settings['__workflow_prereleases'] = False
            return 'Prerelease updates turned off'

        def do_update():
            if self.start_update():
                return 'Downloading and installing update ...'
            else:
                return 'No update available'

        self.magic_arguments['autoupdate'] = update_on
        self.magic_arguments['noautoupdate'] = update_off
        self.magic_arguments['prereleases'] = prereleases_on
        self.magic_arguments['noprereleases'] = prereleases_off
        self.magic_arguments['update'] = do_update

        # Help
        def do_help():
            if self.help_url:
                self.open_help()
                return 'Opening workflow help URL in browser'
            else:
                return 'Workflow has no help URL'

        def show_version():
            if self.version:
                return 'Version: {0}'.format(self.version)
            else:
                return 'This workflow has no version number'

        def list_magic():
            """Display all available magic args in Alfred."""
            isatty = sys.stderr.isatty()
            for name in sorted(self.magic_arguments.keys()):
                if name == 'magic':
                    continue
                arg = '{0}{1}'.format(self.magic_prefix, name)
                self.logger.debug(arg)

                if not isatty:
                    self.add_item(arg, icon=ICON_INFO)

            if not isatty:
                self.send_feedback()

        self.magic_arguments['help'] = do_help
        self.magic_arguments['magic'] = list_magic
        self.magic_arguments['version'] = show_version

    def clear_cache(self, filter_func=lambda f: True):
        """Delete all files in workflow's :attr:`cachedir`.

        :param filter_func: Callable to determine whether a file should be
            deleted or not. ``filter_func`` is called with the filename
            of each file in the data directory. If it returns ``True``,
            the file will be deleted.
            By default, *all* files will be deleted.
        :type filter_func: ``callable``
        """
        self._delete_directory_contents(self.cachedir, filter_func)

    def clear_data(self, filter_func=lambda f: True):
        """Delete all files in workflow's :attr:`datadir`.

        :param filter_func: Callable to determine whether a file should be
            deleted or not. ``filter_func`` is called with the filename
            of each file in the data directory. If it returns ``True``,
            the file will be deleted.
            By default, *all* files will be deleted.
        :type filter_func: ``callable``
        """
        self._delete_directory_contents(self.datadir, filter_func)

    def clear_settings(self):
        """Delete workflow's :attr:`settings_path`."""
        if os.path.exists(self.settings_path):
            os.unlink(self.settings_path)
            self.logger.debug('Deleted : %r', self.settings_path)

    def reset(self):
        """Delete workflow settings, cache and data.

        File :attr:`settings <settings_path>` and directories
        :attr:`cache <cachedir>` and :attr:`data <datadir>` are deleted.

        """
        self.clear_cache()
        self.clear_data()
        self.clear_settings()

    def open_log(self):
        """Open :attr:`logfile` in default app (usually Console.app)."""
        subprocess.call(['open', self.logfile])

    def open_cachedir(self):
        """Open the workflow's :attr:`cachedir` in Finder."""
        subprocess.call(['open', self.cachedir])

    def open_datadir(self):
        """Open the workflow's :attr:`datadir` in Finder."""
        subprocess.call(['open', self.datadir])

    def open_workflowdir(self):
        """Open the workflow's :attr:`workflowdir` in Finder."""
        subprocess.call(['open', self.workflowdir])

    def open_terminal(self):
        """Open a Terminal window at workflow's :attr:`workflowdir`."""
        subprocess.call(['open', '-a', 'Terminal',
                        self.workflowdir])

    def open_help(self):
        """Open :attr:`help_url` in default browser."""
        subprocess.call(['open', self.help_url])

        return 'Opening workflow help URL in browser'

    ####################################################################
    # Helper methods
    ####################################################################

    def decode(self, text, encoding=None, normalization=None):
        """Return ``text`` as normalised unicode.

        If ``encoding`` and/or ``normalization`` is ``None``, the
        ``input_encoding``and ``normalization`` parameters passed to
        :class:`Workflow` are used.

        :param text: string
        :type text: encoded or Unicode string. If ``text`` is already a
            Unicode string, it will only be normalised.
        :param encoding: The text encoding to use to decode ``text`` to
            Unicode.
        :type encoding: ``unicode`` or ``None``
        :param normalization: The nomalisation form to apply to ``text``.
        :type normalization: ``unicode`` or ``None``
        :returns: decoded and normalised ``unicode``

        :class:`Workflow` uses "NFC" normalisation by default. This is the
        standard for Python and will work well with data from the web (via
        :mod:`~workflow.web` or :mod:`json`).

        OS X, on the other hand, uses "NFD" normalisation (nearly), so data
        coming from the system (e.g. via :mod:`subprocess` or
        :func:`os.listdir`/:mod:`os.path`) may not match. You should either
        normalise this data, too, or change the default normalisation used by
        :class:`Workflow`.

        """
        encoding = encoding or self._input_encoding
        normalization = normalization or self._normalizsation
        if not isinstance(text, unicode):
            text = unicode(text, encoding)
        return unicodedata.normalize(normalization, text)

    def fold_to_ascii(self, text):
        """Convert non-ASCII characters to closest ASCII equivalent.

        .. versionadded:: 1.3

        .. note:: This only works for a subset of European languages.

        :param text: text to convert
        :type text: ``unicode``
        :returns: text containing only ASCII characters
        :rtype: ``unicode``

        """
        if isascii(text):
            return text
        text = ''.join([ASCII_REPLACEMENTS.get(c, c) for c in text])
        return unicode(unicodedata.normalize('NFKD',
                       text).encode('ascii', 'ignore'))

    def dumbify_punctuation(self, text):
        """Convert non-ASCII punctuation to closest ASCII equivalent.

        This method replaces "smart" quotes and n- or m-dashes with their
        workaday ASCII equivalents. This method is currently not used
        internally, but exists as a helper method for workflow authors.

        .. versionadded: 1.9.7

        :param text: text to convert
        :type text: ``unicode``
        :returns: text with only ASCII punctuation
        :rtype: ``unicode``

        """
        if isascii(text):
            return text

        text = ''.join([DUMB_PUNCTUATION.get(c, c) for c in text])
        return text

    def _delete_directory_contents(self, dirpath, filter_func):
        """Delete all files in a directory.

        :param dirpath: path to directory to clear
        :type dirpath: ``unicode`` or ``str``
        :param filter_func function to determine whether a file shall be
            deleted or not.
        :type filter_func ``callable``

        """
        if os.path.exists(dirpath):
            for filename in os.listdir(dirpath):
                if not filter_func(filename):
                    continue
                path = os.path.join(dirpath, filename)
                if os.path.isdir(path):
                    shutil.rmtree(path)
                else:
                    os.unlink(path)
                self.logger.debug('Deleted : %r', path)

    def _load_info_plist(self):
        """Load workflow info from ``info.plist``."""
        # info.plist should be in the directory above this one
        self._info = plistlib.readPlist(self.workflowfile('info.plist'))
        self._info_loaded = True

    def _create(self, dirpath):
        """Create directory `dirpath` if it doesn't exist.

        :param dirpath: path to directory
        :type dirpath: ``unicode``
        :returns: ``dirpath`` argument
        :rtype: ``unicode``

        """
        if not os.path.exists(dirpath):
            os.makedirs(dirpath)
        return dirpath

    def _call_security(self, action, service, account, *args):
        """Call ``security`` CLI program that provides access to keychains.

        May raise `PasswordNotFound`, `PasswordExists` or `KeychainError`
        exceptions (the first two are subclasses of `KeychainError`).

        :param action: The ``security`` action to call, e.g.
                           ``add-generic-password``
        :type action: ``unicode``
        :param service: Name of the service.
        :type service: ``unicode``
        :param account: name of the account the password is for, e.g.
            "Pinboard"
        :type account: ``unicode``
        :param password: the password to secure
        :type password: ``unicode``
        :param *args: list of command line arguments to be passed to
                      ``security``
        :type *args: `list` or `tuple`
        :returns: ``(retcode, output)``. ``retcode`` is an `int`, ``output`` a
                  ``unicode`` string.
        :rtype: `tuple` (`int`, ``unicode``)

        """
        cmd = ['security', action, '-s', service, '-a', account] + list(args)
        p = subprocess.Popen(cmd, stdout=subprocess.PIPE,
                             stderr=subprocess.STDOUT)
        stdout, _ = p.communicate()
        if p.returncode == 44:  # password does not exist
            raise PasswordNotFound()
        elif p.returncode == 45:  # password already exists
            raise PasswordExists()
        elif p.returncode > 0:
            err = KeychainError('Unknown Keychain error : %s' % stdout)
            err.retcode = p.returncode
            raise err
        return stdout.strip().decode('utf-8')