| 1 |
1858
|
aaronmk
|
# Parallel processing
|
| 2 |
|
|
|
| 3 |
1873
|
aaronmk
|
import cPickle
|
| 4 |
1877
|
aaronmk
|
import itertools
|
| 5 |
1862
|
aaronmk
|
import Queue
|
| 6 |
1877
|
aaronmk
|
import rand
|
| 7 |
1873
|
aaronmk
|
import types
|
| 8 |
1858
|
aaronmk
|
import warnings
|
| 9 |
|
|
|
| 10 |
1877
|
aaronmk
|
import collection
|
| 11 |
|
|
import dicts
|
| 12 |
1873
|
aaronmk
|
import exc
|
| 13 |
|
|
from Runnable import Runnable
|
| 14 |
|
|
|
| 15 |
|
|
def try_pickle(value):
|
| 16 |
|
|
try: cPickle.dumps(value)
|
| 17 |
|
|
except Exception, e:
|
| 18 |
|
|
exc.add_msg(e, 'Tried to pickle: '+repr(value))
|
| 19 |
|
|
raise
|
| 20 |
|
|
|
| 21 |
1877
|
aaronmk
|
def vars_id_dict(locals_, globals_, *misc):
|
| 22 |
|
|
'''Usage: vars_id_dict(locals(), globals(), misc...)'''
|
| 23 |
|
|
vars_ = map(lambda v: v.values(), [locals_, globals_]) + list(misc)
|
| 24 |
|
|
return dicts.id_dict(vars_)
|
| 25 |
1873
|
aaronmk
|
|
| 26 |
1877
|
aaronmk
|
def prepickle(value, vars_id_dict_):
|
| 27 |
|
|
def filter_(value):
|
| 28 |
|
|
# Try pickling the value. If it fails, we'll get a full traceback here,
|
| 29 |
|
|
# which is not provided with pickling errors in multiprocessing's Pool.
|
| 30 |
|
|
try: try_pickle(value)
|
| 31 |
|
|
except Exception, e:
|
| 32 |
|
|
id_ = id(value)
|
| 33 |
|
|
if id_ in vars_id_dict_: value = id_
|
| 34 |
|
|
else: raise e
|
| 35 |
|
|
return value
|
| 36 |
|
|
return collection.rmap(filter_, value)
|
| 37 |
|
|
|
| 38 |
|
|
def post_unpickle(value, vars_id_dict_):
|
| 39 |
|
|
def filter_(value):
|
| 40 |
|
|
try: return vars_id_dict_[value] # value is an id()
|
| 41 |
|
|
except KeyError: return value
|
| 42 |
|
|
return collection.rmap(filter_, value)
|
| 43 |
|
|
|
| 44 |
1862
|
aaronmk
|
class SyncPool:
|
| 45 |
1861
|
aaronmk
|
'''A dummy synchronous Pool to use if multiprocessing is not available'''
|
| 46 |
1862
|
aaronmk
|
def __init__(self, processes=None): pass
|
| 47 |
|
|
|
| 48 |
1860
|
aaronmk
|
class Result:
|
| 49 |
|
|
def __init__(self, value): self.value = value
|
| 50 |
|
|
|
| 51 |
|
|
def get(timeout=None): return self.value
|
| 52 |
|
|
|
| 53 |
|
|
def wait(timeout=None): pass
|
| 54 |
|
|
|
| 55 |
|
|
def ready(): return True
|
| 56 |
|
|
|
| 57 |
|
|
def successful(): return True # TODO: False if raised exception
|
| 58 |
|
|
|
| 59 |
1877
|
aaronmk
|
def apply_async(self, func, args=(), kw_args={}, callback=None):
|
| 60 |
1858
|
aaronmk
|
if callback == None: callback = lambda v: None
|
| 61 |
|
|
|
| 62 |
1860
|
aaronmk
|
value = func(*args, **kw_args)
|
| 63 |
|
|
callback(value)
|
| 64 |
1863
|
aaronmk
|
return self.Result(value)
|
| 65 |
1858
|
aaronmk
|
|
| 66 |
1862
|
aaronmk
|
class MultiProducerPool:
|
| 67 |
|
|
'''A multi-producer pool. You must call pool.main_loop() in the thread that
|
| 68 |
|
|
created this to process new tasks.'''
|
| 69 |
1873
|
aaronmk
|
|
| 70 |
1877
|
aaronmk
|
def __init__(self, processes=None, locals_=None, globals_=None, *shared):
|
| 71 |
1862
|
aaronmk
|
'''
|
| 72 |
|
|
@param processes If 0, uses SyncPool
|
| 73 |
|
|
@post The # processes actually used is made available in self.process_ct
|
| 74 |
|
|
'''
|
| 75 |
1877
|
aaronmk
|
if locals_ == None: locals_ = locals()
|
| 76 |
|
|
if globals_ == None: globals_ = globals()
|
| 77 |
|
|
|
| 78 |
1862
|
aaronmk
|
try:
|
| 79 |
|
|
if processes == 0: raise ImportError('turned off')
|
| 80 |
|
|
import multiprocessing
|
| 81 |
|
|
import multiprocessing.pool
|
| 82 |
|
|
except ImportError, e:
|
| 83 |
|
|
warnings.warn(UserWarning('Not using parallel processing: '+str(e)))
|
| 84 |
|
|
processes = 1
|
| 85 |
|
|
Pool_ = SyncPool
|
| 86 |
|
|
Queue_ = Queue.Queue
|
| 87 |
|
|
else:
|
| 88 |
|
|
if processes == None: processes = multiprocessing.cpu_count()
|
| 89 |
|
|
Pool_ = multiprocessing.pool.Pool
|
| 90 |
|
|
Queue_ = multiprocessing.Queue
|
| 91 |
|
|
|
| 92 |
|
|
self.process_ct = processes
|
| 93 |
|
|
self.pool = Pool_(processes)
|
| 94 |
|
|
self.queue = Queue_()
|
| 95 |
1877
|
aaronmk
|
# Values that may be pickled by id()
|
| 96 |
|
|
self.vars_id_dict = vars_id_dict(locals_, globals_, *shared)
|
| 97 |
1862
|
aaronmk
|
|
| 98 |
1877
|
aaronmk
|
def share(self, value):
|
| 99 |
|
|
'''Call this on every value that that may be pickled by id()'''
|
| 100 |
|
|
self.vars_id_dict[id(value)] = value
|
| 101 |
|
|
|
| 102 |
1862
|
aaronmk
|
def main_loop(self):
|
| 103 |
|
|
'''@param pool Must be a pool returned by mk_pool()'''
|
| 104 |
|
|
try:
|
| 105 |
1873
|
aaronmk
|
while True:
|
| 106 |
|
|
# block=False raises Empty immediately if the queue is empty,
|
| 107 |
|
|
# which indicates that the program is done
|
| 108 |
|
|
call = self.queue.get(block=False)
|
| 109 |
1877
|
aaronmk
|
self.pool.apply_async(call.func, self.post_unpickle(call.args),
|
| 110 |
|
|
self.post_unpickle(call.kw_args), call.callback)
|
| 111 |
1862
|
aaronmk
|
except Queue.Empty: pass
|
| 112 |
|
|
|
| 113 |
|
|
class Result:
|
| 114 |
|
|
def get(timeout=None): raise NotImplementedError()
|
| 115 |
|
|
|
| 116 |
|
|
def wait(timeout=None): raise NotImplementedError()
|
| 117 |
|
|
|
| 118 |
|
|
def ready(): raise NotImplementedError()
|
| 119 |
|
|
|
| 120 |
|
|
def successful(): raise NotImplementedError()
|
| 121 |
|
|
|
| 122 |
1877
|
aaronmk
|
def apply_async(self, func, args=(), kw_args={}, callback=None):
|
| 123 |
|
|
assert callback == None, 'Callbacks not supported'
|
| 124 |
1873
|
aaronmk
|
|
| 125 |
1877
|
aaronmk
|
call = Runnable(func, *self.prepickle(args), **self.prepickle(kw_args))
|
| 126 |
1873
|
aaronmk
|
call.callback = callback # store this inside the Runnable
|
| 127 |
|
|
|
| 128 |
1862
|
aaronmk
|
self.queue.put_nowait(call)
|
| 129 |
1863
|
aaronmk
|
return self.Result()
|
| 130 |
1877
|
aaronmk
|
|
| 131 |
|
|
def prepickle(self, value): return prepickle(value, self.vars_id_dict)
|
| 132 |
|
|
|
| 133 |
|
|
def post_unpickle(self, value):
|
| 134 |
|
|
return post_unpickle(value, self.vars_id_dict)
|
