utils.py

#!/usr/bin/env python
"""
General Utilities
(part of web.py)
"""

__all__ = [
  "Storage", "storage", "storify", 
  "Counter", "counter",
  "iters", 
  "rstrips", "lstrips", "strips", 
  "safeunicode", "safestr", "utf8",
  "TimeoutError", "timelimit",
  "Memoize", "memoize",
  "re_compile", "re_subm",
  "group", "uniq", "iterview",
  "IterBetter", "iterbetter",
  "safeiter", "safewrite",
  "dictreverse", "dictfind", "dictfindall", "dictincr", "dictadd",
  "requeue", "restack",
  "listget", "intget", "datestr",
  "numify", "denumify", "commify", "dateify",
  "nthstr", "cond",
  "CaptureStdout", "capturestdout", "Profile", "profile",
  "tryall",
  "ThreadedDict", "threadeddict",
  "autoassign",
  "to36",
  "safemarkdown",
  "sendmail"
]

import re, sys, time, threading, itertools, traceback, os

try:
    import subprocess
except ImportError: 
    subprocess = None

try: import datetime
except ImportError: pass

try: set
except NameError:
    from sets import Set as set
    
try:
    from threading import local as threadlocal
except ImportError:
    from python23 import threadlocal

class Storage(dict):
    """
    A Storage object is like a dictionary except `obj.foo` can be used
    in addition to `obj['foo']`.
    
        >>> o = storage(a=1)
        >>> o.a
        1
        >>> o['a']
        1
        >>> o.a = 2
        >>> o['a']
        2
        >>> del o.a
        >>> o.a
        Traceback (most recent call last):
            ...
        AttributeError: 'a'
    
    """
    def __getattr__(self, key): 
        try:
            return self[key]
        except KeyError, k:
            raise AttributeError, k
    
    def __setattr__(self, key, value): 
        self[key] = value
    
    def __delattr__(self, key):
        try:
            del self[key]
        except KeyError, k:
            raise AttributeError, k
    
    def __repr__(self):     
        return '<Storage ' + dict.__repr__(self) + '>'

storage = Storage

def storify(mapping, *requireds, **defaults):
    """
    Creates a `storage` object from dictionary `mapping`, raising `KeyError` if
    d doesn't have all of the keys in `requireds` and using the default 
    values for keys found in `defaults`.

    For example, `storify({'a':1, 'c':3}, b=2, c=0)` will return the equivalent of
    `storage({'a':1, 'b':2, 'c':3})`.
    
    If a `storify` value is a list (e.g. multiple values in a form submission), 
    `storify` returns the last element of the list, unless the key appears in 
    `defaults` as a list. Thus:
    
        >>> storify({'a':[1, 2]}).a
        2
        >>> storify({'a':[1, 2]}, a=[]).a
        [1, 2]
        >>> storify({'a':1}, a=[]).a
        [1]
        >>> storify({}, a=[]).a
        []
    
    Similarly, if the value has a `value` attribute, `storify will return _its_
    value, unless the key appears in `defaults` as a dictionary.
    
        >>> storify({'a':storage(value=1)}).a
        1
        >>> storify({'a':storage(value=1)}, a={}).a
        <Storage {'value': 1}>
        >>> storify({}, a={}).a
        {}
        
    Optionally, keyword parameter `_unicode` can be passed to convert all values to unicode.
    
        >>> storify({'x': 'a'}, _unicode=True)
        <Storage {'x': u'a'}>
        >>> storify({'x': storage(value='a')}, x={}, _unicode=True)
        <Storage {'x': <Storage {'value': 'a'}>}>
        >>> storify({'x': storage(value='a')}, _unicode=True)
        <Storage {'x': u'a'}>
    """
    _unicode = defaults.pop('_unicode', False)

    # if _unicode is callable object, use it convert a string to unicode.
    to_unicode = safeunicode
    if _unicode is not False and hasattr(_unicode, "__call__"):
        to_unicode = _unicode
    
    def unicodify(s):
        if _unicode and isinstance(s, str): return to_unicode(s)
        else: return s
        
    def getvalue(x):
        if hasattr(x, 'file') and hasattr(x, 'value'):
            return x.value
        elif hasattr(x, 'value'):
            return unicodify(x.value)
        else:
            return unicodify(x)
    
    stor = Storage()
    for key in requireds + tuple(mapping.keys()):
        value = mapping[key]
        if isinstance(value, list):
            if isinstance(defaults.get(key), list):
                value = [getvalue(x) for x in value]
            else:
                value = value[-1]
        if not isinstance(defaults.get(key), dict):
            value = getvalue(value)
        if isinstance(defaults.get(key), list) and not isinstance(value, list):
            value = [value]
        setattr(stor, key, value)

    for (key, value) in defaults.iteritems():
        result = value
        if hasattr(stor, key): 
            result = stor[key]
        if value == () and not isinstance(result, tuple): 
            result = (result,)
        setattr(stor, key, result)
    
    return stor

class Counter(storage):
    """Keeps count of how many times something is added.
        
        >>> c = counter()
        >>> c.add('x')
        >>> c.add('x')
        >>> c.add('x')
        >>> c.add('x')
        >>> c.add('x')
        >>> c.add('y')
        >>> c
        <Counter {'y': 1, 'x': 5}>
        >>> c.most()
        ['x']
    """
    def add(self, n):
        self.setdefault(n, 0)
        self[n] += 1
    
    def most(self):
        """Returns the keys with maximum count."""
        m = max(self.itervalues())
        return [k for k, v in self.iteritems() if v == m]
        
    def least(self):
        """Returns the keys with mininum count."""
        m = min(self.itervalues())
        return [k for k, v in self.iteritems() if v == m]

    def percent(self, key):
       """Returns what percentage a certain key is of all entries.

           >>> c = counter()
           >>> c.add('x')
           >>> c.add('x')
           >>> c.add('x')
           >>> c.add('y')
           >>> c.percent('x')
           0.75
           >>> c.percent('y')
           0.25
       """
       return float(self[key])/sum(self.values())
             
    def sorted_keys(self):
        """Returns keys sorted by value.
             
             >>> c = counter()
             >>> c.add('x')
             >>> c.add('x')
             >>> c.add('y')
             >>> c.sorted_keys()
             ['x', 'y']
        """
        return sorted(self.keys(), key=lambda k: self[k], reverse=True)
    
    def sorted_values(self):
        """Returns values sorted by value.
            
            >>> c = counter()
            >>> c.add('x')
            >>> c.add('x')
            >>> c.add('y')
            >>> c.sorted_values()
            [2, 1]
        """
        return [self[k] for k in self.sorted_keys()]
    
    def sorted_items(self):
        """Returns items sorted by value.
            
            >>> c = counter()
            >>> c.add('x')
            >>> c.add('x')
            >>> c.add('y')
            >>> c.sorted_items()
            [('x', 2), ('y', 1)]
        """
        return [(k, self[k]) for k in self.sorted_keys()]
    
    def __repr__(self):
        return '<Counter ' + dict.__repr__(self) + '>'
       
counter = Counter

iters = [list, tuple]
import __builtin__
if hasattr(__builtin__, 'set'):
    iters.append(set)
if hasattr(__builtin__, 'frozenset'):
    iters.append(set)
if sys.version_info < (2,6): # sets module deprecated in 2.6
    try:
        from sets import Set
        iters.append(Set)
    except ImportError: 
        pass
    
class _hack(tuple): pass
iters = _hack(iters)
iters.__doc__ = """
A list of iterable items (like lists, but not strings). Includes whichever
of lists, tuples, sets, and Sets are available in this version of Python.
"""

def _strips(direction, text, remove):
    if isinstance(remove, iters):
        for subr in remove:
            text = _strips(direction, text, subr)
        return text
    
    if direction == 'l': 
        if text.startswith(remove): 
            return text[len(remove):]
    elif direction == 'r':
        if text.endswith(remove):   
            return text[:-len(remove)]
    else: 
        raise ValueError, "Direction needs to be r or l."
    return text

def rstrips(text, remove):
    """
    removes the string `remove` from the right of `text`

        >>> rstrips("foobar", "bar")
        'foo'
    
    """
    return _strips('r', text, remove)

def lstrips(text, remove):
    """
    removes the string `remove` from the left of `text`
    
        >>> lstrips("foobar", "foo")
        'bar'
        >>> lstrips('http://foo.org/', ['http://', 'https://'])
        'foo.org/'
        >>> lstrips('FOOBARBAZ', ['FOO', 'BAR'])
        'BAZ'
        >>> lstrips('FOOBARBAZ', ['BAR', 'FOO'])
        'BARBAZ'
    
    """
    return _strips('l', text, remove)

def strips(text, remove):
    """
    removes the string `remove` from the both sides of `text`

        >>> strips("foobarfoo", "foo")
        'bar'
    
    """
    return rstrips(lstrips(text, remove), remove)

def safeunicode(obj, encoding='utf-8'):
    r"""
    Converts any given object to unicode string.
    
        >>> safeunicode('hello')
        u'hello'
        >>> safeunicode(2)
        u'2'
        >>> safeunicode('\xe1\x88\xb4')
        u'\u1234'
    """
    t = type(obj)
    if t is unicode:
        return obj
    elif t is str:
        return obj.decode(encoding)
    elif t in [int, float, bool]:
        return unicode(obj)
    elif hasattr(obj, '__unicode__') or isinstance(obj, unicode):
        return unicode(obj)
    else:
        return str(obj).decode(encoding)
    
def safestr(obj, encoding='utf-8'):
    r"""
    Converts any given object to utf-8 encoded string. 
    
        >>> safestr('hello')
        'hello'
        >>> safestr(u'\u1234')
        '\xe1\x88\xb4'
        >>> safestr(2)
        '2'
    """
    if isinstance(obj, unicode):
        return obj.encode(encoding)
    elif isinstance(obj, str):
        return obj
    elif hasattr(obj, 'next'): # iterator
        return itertools.imap(safestr, obj)
    else:
        return str(obj)

# for backward-compatibility
utf8 = safestr
    
class TimeoutError(Exception): pass
def timelimit(timeout):
    """
    A decorator to limit a function to `timeout` seconds, raising `TimeoutError`
    if it takes longer.
    
        >>> import time
        >>> def meaningoflife():
        ...     time.sleep(.2)
        ...     return 42
        >>> 
        >>> timelimit(.1)(meaningoflife)()
        Traceback (most recent call last):
            ...
        TimeoutError: took too long
        >>> timelimit(1)(meaningoflife)()
        42

    _Caveat:_ The function isn't stopped after `timeout` seconds but continues 
    executing in a separate thread. (There seems to be no way to kill a thread.)

    inspired by <http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/473878>
    """
    def _1(function):
        def _2(*args, **kw):
            class Dispatch(threading.Thread):
                def __init__(self):
                    threading.Thread.__init__(self)
                    self.result = None
                    self.error = None

                    self.setDaemon(True)
                    self.start()

                def run(self):
                    try:
                        self.result = function(*args, **kw)
                    except:
                        self.error = sys.exc_info()

            c = Dispatch()
            c.join(timeout)
            if c.isAlive():
                raise TimeoutError, 'took too long'
            if c.error:
                raise c.error[0], c.error[1]
            return c.result
        return _2
    return _1

class Memoize:
    """
    'Memoizes' a function, caching its return values for each input.
    If `expires` is specified, values are recalculated after `expires` seconds.
    If `background` is specified, values are recalculated in a separate thread.
    
        >>> calls = 0
        >>> def howmanytimeshaveibeencalled():
        ...     global calls
        ...     calls += 1
        ...     return calls
        >>> fastcalls = memoize(howmanytimeshaveibeencalled)
        >>> howmanytimeshaveibeencalled()
        1
        >>> howmanytimeshaveibeencalled()
        2
        >>> fastcalls()
        3
        >>> fastcalls()
        3
        >>> import time
        >>> fastcalls = memoize(howmanytimeshaveibeencalled, .1, background=False)
        >>> fastcalls()
        4
        >>> fastcalls()
        4
        >>> time.sleep(.2)
        >>> fastcalls()
        5
        >>> def slowfunc():
        ...     time.sleep(.1)
        ...     return howmanytimeshaveibeencalled()
        >>> fastcalls = memoize(slowfunc, .2, background=True)
        >>> fastcalls()
        6
        >>> timelimit(.05)(fastcalls)()
        6
        >>> time.sleep(.2)
        >>> timelimit(.05)(fastcalls)()
        6
        >>> timelimit(.05)(fastcalls)()
        6
        >>> time.sleep(.2)
        >>> timelimit(.05)(fastcalls)()
        7
        >>> fastcalls = memoize(slowfunc, None, background=True)
        >>> threading.Thread(target=fastcalls).start()
        >>> time.sleep(.01)
        >>> fastcalls()
        9
    """
    def __init__(self, func, expires=None, background=True): 
        self.func = func
        self.cache = {}
        self.expires = expires
        self.background = background
        self.running = {}
    
    def __call__(self, *args, **keywords):
        key = (args, tuple(keywords.items()))
        if not self.running.get(key):
            self.running[key] = threading.Lock()
        def update(block=False):
            if self.running[key].acquire(block):
                try:
                    self.cache[key] = (self.func(*args, **keywords), time.time())
                finally:
                    self.running[key].release()
        
        if key not in self.cache: 
            update(block=True)
        elif self.expires and (time.time() - self.cache[key][1]) > self.expires:
            if self.background:
                threading.Thread(target=update).start()
            else:
                update()
        return self.cache[key][0]

memoize = Memoize

re_compile = memoize(re.compile) #@@ threadsafe?
re_compile.__doc__ = """
A memoized version of re.compile.
"""

class _re_subm_proxy:
    def __init__(self): 
        self.match = None
    def __call__(self, match): 
        self.match = match
        return ''

def re_subm(pat, repl, string):
    """
    Like re.sub, but returns the replacement _and_ the match object.
    
        >>> t, m = re_subm('g(oo+)fball', r'f\\1lish', 'goooooofball')
        >>> t
        'foooooolish'
        >>> m.groups()
        ('oooooo',)
    """
    compiled_pat = re_compile(pat)
    proxy = _re_subm_proxy()
    compiled_pat.sub(proxy.__call__, string)
    return compiled_pat.sub(repl, string), proxy.match

def group(seq, size): 
    """
    Returns an iterator over a series of lists of length size from iterable.

        >>> list(group([1,2,3,4], 2))
        [[1, 2], [3, 4]]
        >>> list(group([1,2,3,4,5], 2))
        [[1, 2], [3, 4], [5]]
    """
    def take(seq, n):
        for i in xrange(n):
            yield seq.next()

    if not hasattr(seq, 'next'):  
        seq = iter(seq)
    while True: 
        x = list(take(seq, size))
        if x:
            yield x
        else:
            break

def uniq(seq, key=None):
    """
    Removes duplicate elements from a list while preserving the order of the rest.

        >>> uniq([9,0,2,1,0])
        [9, 0, 2, 1]

    The value of the optional `key` parameter should be a function that
    takes a single argument and returns a key to test the uniqueness.

        >>> uniq(["Foo", "foo", "bar"], key=lambda s: s.lower())
        ['Foo', 'bar']
    """
    key = key or (lambda x: x)
    seen = set()
    result = []
    for v in seq:
        k = key(v)
        if k in seen:
            continue
        seen.add(k)
        result.append(v)
    return result

def iterview(x):
   """
   Takes an iterable `x` and returns an iterator over it
   which prints its progress to stderr as it iterates through.
   """
   WIDTH = 70

   def plainformat(n, lenx):
       return '%5.1f%% (%*d/%d)' % ((float(n)/lenx)*100, len(str(lenx)), n, lenx)

   def bars(size, n, lenx):
       val = int((float(n)*size)/lenx + 0.5)
       if size - val:
           spacing = ">" + (" "*(size-val))[1:]
       else:
           spacing = ""
       return "[%s%s]" % ("="*val, spacing)

   def eta(elapsed, n, lenx):
       if n == 0:
           return '--:--:--'
       if n == lenx:
           secs = int(elapsed)
       else:
           secs = int((elapsed/n) * (lenx-n))
       mins, secs = divmod(secs, 60)
       hrs, mins = divmod(mins, 60)

       return '%02d:%02d:%02d' % (hrs, mins, secs)

   def format(starttime, n, lenx):
       out = plainformat(n, lenx) + ' '
       if n == lenx:
           end = '     '
       else:
           end = ' ETA '
       end += eta(time.time() - starttime, n, lenx)
       out += bars(WIDTH - len(out) - len(end), n, lenx)
       out += end
       return out

   starttime = time.time()
   lenx = len(x)
   for n, y in enumerate(x):
       sys.stderr.write('\r' + format(starttime, n, lenx))
       yield y
   sys.stderr.write('\r' + format(starttime, n+1, lenx) + '\n')

class IterBetter:
    """
    Returns an object that can be used as an iterator 
    but can also be used via __getitem__ (although it 
    cannot go backwards -- that is, you cannot request 
    `iterbetter[0]` after requesting `iterbetter[1]`).
    
        >>> import itertools
        >>> c = iterbetter(itertools.count())
        >>> c[1]
        1
        >>> c[5]
        5
        >>> c[3]
        Traceback (most recent call last):
            ...
        IndexError: already passed 3

    For boolean test, IterBetter peeps at first value in the itertor without effecting the iteration.

        >>> c = iterbetter(iter(range(5)))
        >>> bool(c)
        True
        >>> list(c)
        [0, 1, 2, 3, 4]
        >>> c = iterbetter(iter([]))
        >>> bool(c)
        False
        >>> list(c)
        []
    """
    def __init__(self, iterator): 
        self.i, self.c = iterator, 0

    def __iter__(self): 
        if hasattr(self, "_head"):
            yield self._head

        while 1:    
            yield self.i.next()
            self.c += 1

    def __getitem__(self, i):
        #todo: slices
        if i < self.c: 
            raise IndexError, "already passed "+str(i)
        try:
            while i > self.c: 
                self.i.next()
                self.c += 1
            # now self.c == i
            self.c += 1
            return self.i.next()
        except StopIteration: 
            raise IndexError, str(i)
            
    def __nonzero__(self):
        if hasattr(self, "__len__"):
            return len(self) != 0
        elif hasattr(self, "_head"):
            return True
        else:
            try:
                self._head = self.i.next()
            except StopIteration:
                return False
            else:
                return True

iterbetter = IterBetter

def safeiter(it, cleanup=None, ignore_errors=True):
    """Makes an iterator safe by ignoring the exceptions occured during the iteration.
    """
    def next():
        while True:
            try:
                return it.next()
            except StopIteration:
                raise
            except:
                traceback.print_exc()

    it = iter(it)
    while True:
        yield next()

def safewrite(filename, content):
    """Writes the content to a temp file and then moves the temp file to 
    given filename to avoid overwriting the existing file in case of errors.
    """
    f = file(filename + '.tmp', 'w')
    f.write(content)
    f.close()
    os.rename(f.name, filename)

def dictreverse(mapping):
    """
    Returns a new dictionary with keys and values swapped.
    
        >>> dictreverse({1: 2, 3: 4})
        {2: 1, 4: 3}
    """
    return dict([(value, key) for (key, value) in mapping.iteritems()])

def dictfind(dictionary, element):
    """
    Returns a key whose value in `dictionary` is `element` 
    or, if none exists, None.
    
        >>> d = {1:2, 3:4}
        >>> dictfind(d, 4)
        3
        >>> dictfind(d, 5)
    """
    for (key, value) in dictionary.iteritems():
        if element is value: 
            return key

def dictfindall(dictionary, element):
    """
    Returns the keys whose values in `dictionary` are `element`
    or, if none exists, [].
    
        >>> d = {1:4, 3:4}
        >>> dictfindall(d, 4)
        [1, 3]
        >>> dictfindall(d, 5)
        []
    """
    res = []
    for (key, value) in dictionary.iteritems():
        if element is value:
            res.append(key)
    return res

def dictincr(dictionary, element):
    """
    Increments `element` in `dictionary`, 
    setting it to one if it doesn't exist.
    
        >>> d = {1:2, 3:4}
        >>> dictincr(d, 1)
        3
        >>> d[1]
        3
        >>> dictincr(d, 5)
        1
        >>> d[5]
        1
    """
    dictionary.setdefault(element, 0)
    dictionary[element] += 1
    return dictionary[element]

def dictadd(*dicts):
    """
    Returns a dictionary consisting of the keys in the argument dictionaries.
    If they share a key, the value from the last argument is used.
    
        >>> dictadd({1: 0, 2: 0}, {2: 1, 3: 1})
        {1: 0, 2: 1, 3: 1}
    """
    result = {}
    for dct in dicts:
        result.update(dct)
    return result

def requeue(queue, index=-1):
    """Returns the element at index after moving it to the beginning of the queue.

        >>> x = [1, 2, 3, 4]
        >>> requeue(x)
        4
        >>> x
        [4, 1, 2, 3]
    """
    x = queue.pop(index)
    queue.insert(0, x)
    return x

def restack(stack, index=0):
    """Returns the element at index after moving it to the top of stack.

           >>> x = [1, 2, 3, 4]
           >>> restack(x)
           1
           >>> x
           [2, 3, 4, 1]
    """
    x = stack.pop(index)
    stack.append(x)
    return x

def listget(lst, ind, default=None):
    """
    Returns `lst[ind]` if it exists, `default` otherwise.
    
        >>> listget(['a'], 0)
        'a'
        >>> listget(['a'], 1)
        >>> listget(['a'], 1, 'b')
        'b'
    """
    if len(lst)-1 < ind: 
        return default
    return lst[ind]

def intget(integer, default=None):
    """
    Returns `integer` as an int or `default` if it can't.
    
        >>> intget('3')
        3
        >>> intget('3a')
        >>> intget('3a', 0)
        0
    """
    try:
        return int(integer)
    except (TypeError, ValueError):
        return default

def datestr(then, now=None):
    """
    Converts a (UTC) datetime object to a nice string representation.
    
        >>> from datetime import datetime, timedelta
        >>> d = datetime(1970, 5, 1)
        >>> datestr(d, now=d)
        '0 microseconds ago'
        >>> for t, v in {
        ...   timedelta(microseconds=1): '1 microsecond ago',
        ...   timedelta(microseconds=2): '2 microseconds ago',
        ...   -timedelta(microseconds=1): '1 microsecond from now',
        ...   -timedelta(microseconds=2): '2 microseconds from now',
        ...   timedelta(microseconds=2000): '2 milliseconds ago',
        ...   timedelta(seconds=2): '2 seconds ago',
        ...   timedelta(seconds=2*60): '2 minutes ago',
        ...   timedelta(seconds=2*60*60): '2 hours ago',
        ...   timedelta(days=2): '2 days ago',
        ... }.iteritems():
        ...     assert datestr(d, now=d+t) == v
        >>> datestr(datetime(1970, 1, 1), now=d)
        'January  1'
        >>> datestr(datetime(1969, 1, 1), now=d)
        'January  1, 1969'
        >>> datestr(datetime(1970, 6, 1), now=d)
        'June  1, 1970'
        >>> datestr(None)
        ''
    """
    def agohence(n, what, divisor=None):
        if divisor: n = n // divisor

        out = str(abs(n)) + ' ' + what       # '2 day'
        if abs(n) != 1: out += 's'           # '2 days'
        out += ' '                           # '2 days '
        if n < 0:
            out += 'from now'
        else:
            out += 'ago'
        return out                           # '2 days ago'

    oneday = 24 * 60 * 60

    if not then: return ""
    if not now: now = datetime.datetime.utcnow()
    if type(now).__name__ == "DateTime":
        now = datetime.datetime.fromtimestamp(now)
    if type(then).__name__ == "DateTime":
        then = datetime.datetime.fromtimestamp(then)
    elif type(then).__name__ == "date":
        then = datetime.datetime(then.year, then.month, then.day)

    delta = now - then
    deltaseconds = int(delta.days * oneday + delta.seconds + delta.microseconds * 1e-06)
    deltadays = abs(deltaseconds) // oneday
    if deltaseconds < 0: deltadays *= -1 # fix for oddity of floor

    if deltadays:
        if abs(deltadays) < 4:
            return agohence(deltadays, 'day')

        try:
            out = then.strftime('%B %e') # e.g. 'June  3'
        except ValueError:
            # %e doesn't work on Windows.
            out = then.strftime('%B %d') # e.g. 'June 03'

        if then.year != now.year or deltadays < 0:
            out += ', %s' % then.year
        return out

    if int(deltaseconds):
        if abs(deltaseconds) > (60 * 60):
            return agohence(deltaseconds, 'hour', 60 * 60)
        elif abs(deltaseconds) > 60:
            return agohence(deltaseconds, 'minute', 60)
        else:
            return agohence(deltaseconds, 'second')

    deltamicroseconds = delta.microseconds
    if delta.days: deltamicroseconds = int(delta.microseconds - 1e6) # datetime oddity
    if abs(deltamicroseconds) > 1000:
        return agohence(deltamicroseconds, 'millisecond', 1000)

    return agohence(deltamicroseconds, 'microsecond')

def numify(string):
    """
    Removes all non-digit characters from `string`.
    
        >>> numify('800-555-1212')
        '8005551212'
        >>> numify('800.555.1212')
        '8005551212'
    
    """
    return ''.join([c for c in str(string) if c.isdigit()])

def denumify(string, pattern):
    """
    Formats `string` according to `pattern`, where the letter X gets replaced
    by characters from `string`.
    
        >>> denumify("8005551212", "(XXX) XXX-XXXX")
        '(800) 555-1212'
    
    """
    out = []
    for c in pattern:
        if c == "X":
            out.append(string[0])
            string = string[1:]
        else:
            out.append(c)
    return ''.join(out)

def commify(n):
    """
    Add commas to an integer `n`.

        >>> commify(1)
        '1'
        >>> commify(123)
        '123'
        >>> commify(1234)
        '1,234'
        >>> commify(1234567890)
        '1,234,567,890'
        >>> commify(123.0)
        '123.0'
        >>> commify(1234.5)
        '1,234.5'
        >>> commify(1234.56789)
        '1,234.56789'
        >>> commify('%.2f' % 1234.5)
        '1,234.50'
        >>> commify(None)
        >>>

    """
    if n is None: return None
    n = str(n)
    if '.' in n:
        dollars, cents = n.split('.')
    else:
        dollars, cents = n, None

    r = []
    for i, c in enumerate(str(dollars)[::-1]):
        if i and (not (i % 3)):
            r.insert(0, ',')
        r.insert(0, c)
    out = ''.join(r)
    if cents:
        out += '.' + cents
    return out

def dateify(datestring):
    """
    Formats a numified `datestring` properly.
    """
    return denumify(datestring, "XXXX-XX-XX XX:XX:XX")


def nthstr(n):
    """
    Formats an ordinal.
    Doesn't handle negative numbers.

        >>> nthstr(1)
        '1st'
        >>> nthstr(0)
        '0th'
        >>> [nthstr(x) for x in [2, 3, 4, 5, 10, 11, 12, 13, 14, 15]]
        ['2nd', '3rd', '4th', '5th', '10th', '11th', '12th', '13th', '14th', '15th']
        >>> [nthstr(x) for x in [91, 92, 93, 94, 99, 100, 101, 102]]
        ['91st', '92nd', '93rd', '94th', '99th', '100th', '101st', '102nd']
        >>> [nthstr(x) for x in [111, 112, 113, 114, 115]]
        ['111th', '112th', '113th', '114th', '115th']

    """
    
    assert n >= 0
    if n % 100 in [11, 12, 13]: return '%sth' % n
    return {1: '%sst', 2: '%snd', 3: '%srd'}.get(n % 10, '%sth') % n

def cond(predicate, consequence, alternative=None):
    """
    Function replacement for if-else to use in expressions.
        
        >>> x = 2
        >>> cond(x % 2 == 0, "even", "odd")
        'even'
        >>> cond(x % 2 == 0, "even", "odd") + '_row'
        'even_row'
    """
    if predicate:
        return consequence
    else:
        return alternative

class CaptureStdout:
    """
    Captures everything `func` prints to stdout and returns it instead.
    
        >>> def idiot():
        ...     print "foo"
        >>> capturestdout(idiot)()
        'foo\\n'
    
    **WARNING:** Not threadsafe!
    """
    def __init__(self, func): 
        self.func = func
    def __call__(self, *args, **keywords):
        from cStringIO import StringIO
        # Not threadsafe!
        out = StringIO()
        oldstdout = sys.stdout
        sys.stdout = out
        try: 
            self.func(*args, **keywords)
        finally: 
            sys.stdout = oldstdout
        return out.getvalue()

capturestdout = CaptureStdout

class Profile:
    """
    Profiles `func` and returns a tuple containing its output
    and a string with human-readable profiling information.
        
        >>> import time
        >>> out, inf = profile(time.sleep)(.001)
        >>> out
        >>> inf[:10].strip()
        'took 0.0'
    """
    def __init__(self, func): 
        self.func = func
    def __call__(self, *args): ##, **kw):   kw unused
        import hotshot, hotshot.stats, os, tempfile ##, time already imported
        f, filename = tempfile.mkstemp()
        os.close(f)
        
        prof = hotshot.Profile(filename)

        stime = time.time()
        result = prof.runcall(self.func, *args)
        stime = time.time() - stime
        prof.close()

        import cStringIO
        out = cStringIO.StringIO()
        stats = hotshot.stats.load(filename)
        stats.stream = out
        stats.strip_dirs()
        stats.sort_stats('time', 'calls')
        stats.print_stats(40)
        stats.print_callers()

        x =  '\n\ntook '+ str(stime) + ' seconds\n'
        x += out.getvalue()

        # remove the tempfile
        try:
            os.remove(filename)
        except IOError:
            pass
            
        return result, x

profile = Profile


import traceback
# hack for compatibility with Python 2.3:
if not hasattr(traceback, 'format_exc'):
    from cStringIO import StringIO
    def format_exc(limit=None):
        strbuf = StringIO()
        traceback.print_exc(limit, strbuf)
        return strbuf.getvalue()
    traceback.format_exc = format_exc

def tryall(context, prefix=None):
    """
    Tries a series of functions and prints their results. 
    `context` is a dictionary mapping names to values; 
    the value will only be tried if it's callable.
    
        >>> tryall(dict(j=lambda: True))
        j: True
        ----------------------------------------
        results:
           True: 1

    For example, you might have a file `test/stuff.py` 
    with a series of functions testing various things in it. 
    At the bottom, have a line:

        if __name__ == "__main__": tryall(globals())

    Then you can run `python test/stuff.py` and get the results of 
    all the tests.
    """
    context = context.copy() # vars() would update
    results = {}
    for (key, value) in context.iteritems():
        if not hasattr(value, '__call__'): 
            continue
        if prefix and not key.startswith(prefix): 
            continue
        print key + ':',
        try:
            r = value()
            dictincr(results, r)
            print r
        except:
            print 'ERROR'
            dictincr(results, 'ERROR')
            print '   ' + '\n   '.join(traceback.format_exc().split('\n'))
        
    print '-'*40
    print 'results:'
    for (key, value) in results.iteritems():
        print ' '*2, str(key)+':', value
        
class ThreadedDict(threadlocal):
    """
    Thread local storage.
    
        >>> d = ThreadedDict()
        >>> d.x = 1
        >>> d.x
        1
        >>> import threading
        >>> def f(): d.x = 2
        ...
        >>> t = threading.Thread(target=f)
        >>> t.start()
        >>> t.join()
        >>> d.x
        1
    """
    _instances = set()
    
    def __init__(self):
        ThreadedDict._instances.add(self)
        
    def __del__(self):
        ThreadedDict._instances.remove(self)
        
    def __hash__(self):
        return id(self)
    
    def clear_all():
        """Clears all ThreadedDict instances.
        """
        for t in list(ThreadedDict._instances):
            t.clear()
    clear_all = staticmethod(clear_all)
    
    # Define all these methods to more or less fully emulate dict -- attribute access
    # is built into threading.local.

    def __getitem__(self, key):
        return self.__dict__[key]

    def __setitem__(self, key, value):
        self.__dict__[key] = value

    def __delitem__(self, key):
        del self.__dict__[key]

    def __contains__(self, key):
        return key in self.__dict__

    has_key = __contains__
        
    def clear(self):
        self.__dict__.clear()

    def copy(self):
        return self.__dict__.copy()

    def get(self, key, default=None):
        return self.__dict__.get(key, default)

    def items(self):
        return self.__dict__.items()

    def iteritems(self):
        return self.__dict__.iteritems()

    def keys(self):
        return self.__dict__.keys()

    def iterkeys(self):
        return self.__dict__.iterkeys()

    iter = iterkeys

    def values(self):
        return self.__dict__.values()

    def itervalues(self):
        return self.__dict__.itervalues()

    def pop(self, key, *args):
        return self.__dict__.pop(key, *args)

    def popitem(self):
        return self.__dict__.popitem()

    def setdefault(self, key, default=None):
        return self.__dict__.setdefault(key, default)

    def update(self, *args, **kwargs):
        self.__dict__.update(*args, **kwargs)

    def __repr__(self):
        return '<ThreadedDict %r>' % self.__dict__

    __str__ = __repr__
    
threadeddict = ThreadedDict

def autoassign(self, locals):
    """
    Automatically assigns local variables to `self`.
    
        >>> self = storage()
        >>> autoassign(self, dict(a=1, b=2))
        >>> self
        <Storage {'a': 1, 'b': 2}>
    
    Generally used in `__init__` methods, as in:

        def __init__(self, foo, bar, baz=1): autoassign(self, locals())
    """
    for (key, value) in locals.iteritems():
        if key == 'self': 
            continue
        setattr(self, key, value)

def to36(q):
    """
    Converts an integer to base 36 (a useful scheme for human-sayable IDs).
    
        >>> to36(35)
        'z'
        >>> to36(119292)
        '2k1o'
        >>> int(to36(939387374), 36)
        939387374
        >>> to36(0)
        '0'
        >>> to36(-393)
        Traceback (most recent call last):
            ... 
        ValueError: must supply a positive integer
    
    """
    if q < 0: raise ValueError, "must supply a positive integer"
    letters = "0123456789abcdefghijklmnopqrstuvwxyz"
    converted = []
    while q != 0:
        q, r = divmod(q, 36)
        converted.insert(0, letters[r])
    return "".join(converted) or '0'


r_url = re_compile('(?<!\()(http://(\S+))')
def safemarkdown(text):
    """
    Converts text to HTML following the rules of Markdown, but blocking any
    outside HTML input, so that only the things supported by Markdown
    can be used. Also converts raw URLs to links.

    (requires [markdown.py](http://webpy.org/markdown.py))
    """
    from markdown import markdown
    if text:
        text = text.replace('<', '&lt;')
        # TODO: automatically get page title?
        text = r_url.sub(r'<\1>', text)
        text = markdown(text)
        return text

def sendmail(from_address, to_address, subject, message, headers=None, **kw):
    """
    Sends the email message `message` with mail and envelope headers
    for from `from_address_` to `to_address` with `subject`. 
    Additional email headers can be specified with the dictionary 
    `headers.
    
    Optionally cc, bcc and attachments can be specified as keyword arguments.
    Attachments must be an iterable and each attachment can be either a 
    filename or a file object or a dictionary with filename, content and 
    optionally content_type keys.

    If `web.config.smtp_server` is set, it will send the message
    to that SMTP server. Otherwise it will look for 
    `/usr/sbin/sendmail`, the typical location for the sendmail-style
    binary. To use sendmail from a different path, set `web.config.sendmail_path`.
    """
    attachments = kw.pop("attachments", [])
    mail = _EmailMessage(from_address, to_address, subject, message, headers, **kw)

    for a in attachments:
        if isinstance(a, dict):
            mail.attach(a['filename'], a['content'], a.get('content_type'))
        elif hasattr(a, 'read'): # file
            filename = os.path.basename(getattr(a, "name", ""))
            content_type = getattr(a, 'content_type', None)
            mail.attach(filename, a.read(), content_type)
        elif isinstance(a, basestring):
            f = open(a, 'rb')
            content = f.read()
            f.close()
            filename = os.path.basename(a)
            mail.attach(filename, content, None)
        else:
            raise ValueError, "Invalid attachment: %s" % repr(a)
            
    mail.send()

class _EmailMessage:
    def __init__(self, from_address, to_address, subject, message, headers=None, **kw):
        def listify(x):
            if not isinstance(x, list):
                return [safestr(x)]
            else:
                return [safestr(a) for a in x]
    
        subject = safestr(subject)
        message = safestr(message)

        from_address = safestr(from_address)
        to_address = listify(to_address)    
        cc = listify(kw.get('cc', []))
        bcc = listify(kw.get('bcc', []))
        recipients = to_address + cc + bcc

        import email.Utils
        self.from_address = email.Utils.parseaddr(from_address)[1]
        self.recipients = [email.Utils.parseaddr(r)[1] for r in recipients]        
    
        self.headers = dictadd({
          'From': from_address,
          'To': ", ".join(to_address),
          'Subject': subject
        }, headers or {})

        if cc:
            self.headers['Cc'] = ", ".join(cc)
    
        self.message = self.new_message()
        self.message.add_header("Content-Transfer-Encoding", "7bit")
        self.message.add_header("Content-Disposition", "inline")
        self.message.add_header("MIME-Version", "1.0")
        self.message.set_payload(message, 'utf-8')
        self.multipart = False
        
    def new_message(self):
        from email.Message import Message
        return Message()
        
    def attach(self, filename, content, content_type=None):
        if not self.multipart:
            msg = self.new_message()
            msg.add_header("Content-Type", "multipart/mixed")
            msg.attach(self.message)
            self.message = msg
            self.multipart = True
                        
        import mimetypes
        try:
            from email import encoders
        except:
            from email import Encoders as encoders
            
        content_type = content_type or mimetypes.guess_type(filename)[0] or "applcation/octet-stream"
        
        msg = self.new_message()
        msg.set_payload(content)
        msg.add_header('Content-Type', content_type)
        msg.add_header('Content-Disposition', 'attachment', filename=filename)
        
        if not content_type.startswith("text/"):
            encoders.encode_base64(msg)
            
        self.message.attach(msg)

    def prepare_message(self):
        for k, v in self.headers.iteritems():
            if k.lower() == "content-type":
                self.message.set_type(v)
            else:
                self.message.add_header(k, v)

        self.headers = {}

    def send(self):
        try:
            import webapi
        except ImportError:
            webapi = Storage(config=Storage())

        self.prepare_message()
        message_text = self.message.as_string()
    
        if webapi.config.get('smtp_server'):
            server = webapi.config.get('smtp_server')
            port = webapi.config.get('smtp_port', 0)
            username = webapi.config.get('smtp_username') 
            password = webapi.config.get('smtp_password')
            debug_level = webapi.config.get('smtp_debuglevel', None)
            starttls = webapi.config.get('smtp_starttls', False)

            import smtplib
            smtpserver = smtplib.SMTP(server, port)

            if debug_level:
                smtpserver.set_debuglevel(debug_level)

            if starttls:
                smtpserver.ehlo()
                smtpserver.starttls()
                smtpserver.ehlo()

            if username and password:
                smtpserver.login(username, password)

            smtpserver.sendmail(self.from_address, self.recipients, message_text)
            smtpserver.quit()
        elif webapi.config.get('email_engine') == 'aws':
            import boto.ses
            c = boto.ses.SESConnection(
              aws_access_key_id=webapi.config.get('aws_access_key_id'),
              aws_secret_access_key=web.api.config.get('aws_secret_access_key'))
            c.send_raw_email(self.from_address, message_text, self.recipients)
        else:
            sendmail = webapi.config.get('sendmail_path', '/usr/sbin/sendmail')
        
            assert not self.from_address.startswith('-'), 'security'
            for r in self.recipients:
                assert not r.startswith('-'), 'security'
                
            cmd = [sendmail, '-f', self.from_address] + self.recipients

            if subprocess:
                p = subprocess.Popen(cmd, stdin=subprocess.PIPE)
                p.stdin.write(message_text)
                p.stdin.close()
                p.wait()
            else:
                i, o = os.popen2(cmd)
                i.write(message)
                i.close()
                o.close()
                del i, o
                
    def __repr__(self):
        return "<EmailMessage>"
    
    def __str__(self):
        return self.message.as_string()

if __name__ == "__main__":
    import doctest
    doctest.testmod()