1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
|
from __future__ import generators
execfile("static.py")
import os, stat, types
#######################################################################
#
# lazy - Define some lazy data structures and functions acting on them
#
class Iter:
"""Hold static methods for the manipulation of lazy iterators"""
def filter(predicate, iterator):
"""Like filter in a lazy functional programming language"""
for i in iterator:
if predicate(i): yield i
def map(function, iterator):
"""Like map in a lazy functional programming language"""
for i in iterator: yield function(i)
def foreach(function, iterator):
"""Run function on each element in iterator"""
for i in iterator: function(i)
def cat(*iters):
"""Lazily concatenate iterators"""
for iter in iters:
for i in iter: yield i
def cat2(iter_of_iters):
"""Lazily concatenate iterators, iterated by big iterator"""
for iter in iter_of_iters:
for i in iter: yield i
def empty(iter):
"""True if iterator has length 0"""
for i in iter: return None
return 1
def equal(iter1, iter2, verbose = None, operator = lambda x, y: x == y):
"""True if iterator 1 has same elements as iterator 2
Use equality operator, or == if it is unspecified.
"""
for i1 in iter1:
try: i2 = iter2.next()
except StopIteration:
if verbose: print "End when i1 = %s" % i1
return None
if not operator(i1, i2):
if verbose: print "%s not equal to %s" % (i1, i2)
return None
try: i2 = iter2.next()
except StopIteration: return 1
if verbose: print "End when i2 = %s" % i2
return None
def Or(iter):
"""True if any element in iterator is true. Short circuiting"""
i = None
for i in iter:
if i: return i
return i
def And(iter):
"""True if all elements in iterator are true. Short circuiting"""
i = 1
for i in iter:
if not i: return i
return i
def len(iter):
"""Return length of iterator"""
i = 0
while 1:
try: iter.next()
except StopIteration: return i
i = i+1
def foldr(f, default, iter):
"""foldr the "fundamental list recursion operator"?"""
try: next = iter.next()
except StopIteration: return default
return f(next, Iter.foldr(f, default, iter))
def foldl(f, default, iter):
"""the fundamental list iteration operator.."""
while 1:
try: next = iter.next()
except StopIteration: return default
default = f(default, next)
def multiplex(iter, num_of_forks, final_func = None, closing_func = None):
"""Split a single iterater into a number of streams
The return val will be a list with length num_of_forks, each
of which will be an iterator like iter. final_func is the
function that will be called on each element in iter just as
it is being removed from the buffer. closing_func is called
when all the streams are finished.
"""
if num_of_forks == 2 and not final_func and not closing_func:
im2 = IterMultiplex2(iter)
return (im2.yielda(), im2.yieldb())
if not final_func: final_func = lambda i: None
if not closing_func: closing_func = lambda: None
# buffer is a list of elements that some iterators need and others
# don't
buffer = []
# buffer[forkposition[i]] is the next element yieled by iterator
# i. If it is -1, yield from the original iter
starting_forkposition = [-1] * num_of_forks
forkposition = starting_forkposition[:]
called_closing_func = [None]
def get_next(fork_num):
"""Return the next element requested by fork_num"""
if forkposition[fork_num] == -1:
try: buffer.insert(0, iter.next())
except StopIteration:
# call closing_func if necessary
if (forkposition == starting_forkposition and
not called_closing_func[0]):
closing_func()
called_closing_func[0] = None
raise StopIteration
for i in range(num_of_forks): forkposition[i] += 1
return_val = buffer[forkposition[fork_num]]
forkposition[fork_num] -= 1
blen = len(buffer)
if not (blen-1) in forkposition:
# Last position in buffer no longer needed
assert forkposition[fork_num] == blen-2
final_func(buffer[blen-1])
del buffer[blen-1]
return return_val
def make_iterator(fork_num):
while(1): yield get_next(fork_num)
return tuple(map(make_iterator, range(num_of_forks)))
MakeStatic(Iter)
class IterMultiplex2:
"""Multiplex an iterator into 2 parts
This is a special optimized case of the Iter.multiplex function,
used when there is no closing_func or final_func, and we only want
to split it into 2. By profiling, this is a time sensitive class.
"""
def __init__(self, iter):
self.a_leading_by = 0 # How many places a is ahead of b
self.buffer = []
self.iter = iter
def yielda(self):
"""Return first iterator"""
buf, iter = self.buffer, self.iter
while(1):
if self.a_leading_by >= 0: # a is in front, add new element
elem = iter.next() # exception will be passed
buf.append(elem)
else: elem = buf.pop(0) # b is in front, subtract an element
self.a_leading_by += 1
yield elem
def yieldb(self):
"""Return second iterator"""
buf, iter = self.buffer, self.iter
while(1):
if self.a_leading_by <= 0: # b is in front, add new element
elem = iter.next() # exception will be passed
buf.append(elem)
else: elem = buf.pop(0) # a is in front, subtract an element
self.a_leading_by -= 1
yield elem
class IterTreeReducer:
"""Tree style reducer object for iterator
The indicies of a RORPIter form a tree type structure. This class
can be used on each element of an iter in sequence and the result
will be as if the corresponding tree was reduced. This tries to
bridge the gap between the tree nature of directories, and the
iterator nature of the connection between hosts and the temporal
order in which the files are processed.
The elements of the iterator are required to have a tuple-style
.index, called "indexed elem" below.
"""
def __init__(self, base_init, branch_reducer,
branch_base, base_final, initial_state = None):
"""ITR initializer
base_init is a function of one argument, an indexed elem. It
is called immediately on any elem in the iterator. It should
return some value type A.
branch_reducer and branch_base are used to form a value on a
bunch of reduced branches, in the way that a linked list of
type C can be folded to form a value type B.
base_final is called when leaving a tree. It takes three
arguments, the indexed elem, the output (type A) of base_init,
the output of branch_reducer on all the branches (type B) and
returns a value type C.
"""
self.base_init = base_init
self.branch_reducer = branch_reducer
self.base_final = base_final
self.branch_base = branch_base
if initial_state: self.setstate(initial_state)
else:
self.state = IterTreeReducerState(branch_base)
self.subreducer = None
def setstate(self, state):
"""Update with new state, recursive if necessary"""
self.state = state
if state.substate: self.subreducer = self.newinstance(state.substate)
else: self.subreducer = None
def getstate(self): return self.state
def getresult(self):
"""Return results of calculation"""
if not self.state.calculated: self.calculate_final_val()
return self.state.final_val
def intree(self, index):
"""Return true if index is still in current tree"""
return self.state.base_index == index[:len(self.state.base_index)]
def newinstance(self, state = None):
"""Return reducer of same type as self
If state is None, sets substate of self.state, otherwise
assume this is already set.
"""
new = self.__class__(self.base_init, self.branch_reducer,
self.branch_base, self.base_final, state)
if state is None: self.state.substate = new.state
return new
def process_w_subreducer(self, indexed_elem):
"""Give object to subreducer, if necessary update branch_val"""
if not self.subreducer:
self.subreducer = self.newinstance()
if not self.subreducer(indexed_elem):
self.state.branch_val = self.branch_reducer(self.state.branch_val,
self.subreducer.getresult())
self.subreducer = self.newinstance()
assert self.subreducer(indexed_elem)
def calculate_final_val(self):
"""Set final value"""
if self.subreducer:
self.state.branch_val = self.branch_reducer(self.state.branch_val,
self.subreducer.getresult())
if self.state.current_index is None:
# No input, set None as default value
self.state.final_val = None
else:
self.state.final_val = self.base_final(self.state.base_elem,
self.state.base_init_val,
self.state.branch_val)
self.state.calculated = 1
def __call__(self, indexed_elem):
"""Process elem, current position in iterator
Returns true if elem successfully processed, false if elem is
not in the current tree and thus the final result is
available.
"""
index = indexed_elem.index
assert type(index) is types.TupleType
if self.state.current_index is None: # must be at base
self.state.base_init_val = self.base_init(indexed_elem)
# Do most crash-prone op first, so we don't leave inconsistent
self.state.current_index = index
self.state.base_index = index
self.state.base_elem = indexed_elem
return 1
elif not index > self.state.current_index:
Log("Warning: oldindex %s >= newindex %s" %
(self.state.current_index, index), 2)
if not self.intree(index):
self.calculate_final_val()
return None
else:
self.process_w_subreducer(indexed_elem)
self.state.current_index = index
return 1
class IterTreeReducerState:
"""Holds the state for IterTreeReducers
An IterTreeReducer cannot be pickled directly because it holds
some anonymous functions. This class contains the relevant data
that is likely to be picklable, so the ITR can be saved and loaded
if the associated functions are known.
"""
def __init__(self, branch_base):
"""ITRS initializer
Class variables:
self.current_index - last index processing started on, or None
self.base_index - index of first element processed
self.base_elem - first element processed
self.branch_val - default branch reducing value
self.calculated - true iff the final value has been calculated
self.base_init_val - return value of base_init function
self.final_val - Final value once it's calculated
self.substate - IterTreeReducerState when subreducer active
"""
self.current_index = None
self.calculated = None
self.branch_val = branch_base
self.substate = None
|