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
|
"""Example showing how to access a PyTables file from multiple processes using
queues."""
import time
import queue
import random
import multiprocessing
from pathlib import Path
import numpy as np
import tables as tb
# this creates an HDF5 file with one array containing n rows
def make_file(file_path, n):
with tb.open_file(file_path, "w") as fobj:
array = fobj.create_carray("/", "array", tb.Int64Atom(), (n, n))
for i in range(n):
array[i, :] = i
# All access to the file goes through a single instance of this class.
# It contains several queues that are used to communicate with other
# processes.
# The read_queue is used for requests to read data from the HDF5 file.
# A list of result_queues is used to send data back to client processes.
# The write_queue is used for requests to modify the HDF5 file.
# One end of a pipe (shutdown) is used to signal the process to terminate.
class FileAccess(multiprocessing.Process):
def __init__(
self, h5_path, read_queue, result_queues, write_queue, shutdown
):
self.h5_path = h5_path
self.read_queue = read_queue
self.result_queues = result_queues
self.write_queue = write_queue
self.shutdown = shutdown
self.block_period = 0.01
super().__init__()
def run(self):
self.h5_file = tb.open_file(self.h5_path, "r+")
self.array = self.h5_file.get_node("/array")
another_loop = True
while another_loop:
# Check if the process has received the shutdown signal.
if self.shutdown.poll():
another_loop = False
# Check for any data requests in the read_queue.
try:
row_num, proc_num = self.read_queue.get(
True, self.block_period
)
# look up the appropriate result_queue for this data processor
# instance
result_queue = self.result_queues[proc_num]
print(
"processor {} reading from row {}".format(
proc_num, row_num
)
)
result_queue.put(self.read_data(row_num))
another_loop = True
except queue.Empty:
pass
# Check for any write requests in the write_queue.
try:
row_num, data = self.write_queue.get(True, self.block_period)
print("writing row", row_num)
self.write_data(row_num, data)
another_loop = True
except queue.Empty:
pass
# close the HDF5 file before shutting down
self.h5_file.close()
def read_data(self, row_num):
return self.array[row_num, :]
def write_data(self, row_num, data):
self.array[row_num, :] = data
# This class represents a process that does work by reading and writing to the
# HDF5 file. It does this by sending requests to the FileAccess class instance
# through its read and write queues. The data results are sent back through
# the result_queue.
# Its actions are logged to a text file.
class DataProcessor(multiprocessing.Process):
def __init__(
self,
read_queue,
result_queue,
write_queue,
proc_num,
array_size,
output_file,
):
self.read_queue = read_queue
self.result_queue = result_queue
self.write_queue = write_queue
self.proc_num = proc_num
self.array_size = array_size
self.output_file = output_file
super().__init__()
def run(self):
self.output_file = open(self.output_file, "w")
# read a random row from the file
row_num = random.randrange(self.array_size)
self.read_queue.put((row_num, self.proc_num))
self.output_file.write(str(row_num) + "\n")
self.output_file.write(str(self.result_queue.get()) + "\n")
# modify a random row to equal 11 * (self.proc_num + 1)
row_num = random.randrange(self.array_size)
new_data = np.zeros((1, self.array_size), "i8") + 11 * (
self.proc_num + 1
)
self.write_queue.put((row_num, new_data))
# pause, then read the modified row
time.sleep(0.015)
self.read_queue.put((row_num, self.proc_num))
self.output_file.write(str(row_num) + "\n")
self.output_file.write(str(self.result_queue.get()) + "\n")
self.output_file.close()
# this function starts the FileAccess class instance and
# sets up all the queues used to communicate with it
def make_queues(num_processors):
read_queue = multiprocessing.Queue()
write_queue = multiprocessing.Queue()
shutdown_recv, shutdown_send = multiprocessing.Pipe(False)
result_queues = [multiprocessing.Queue() for i in range(num_processors)]
file_access = FileAccess(
file_path, read_queue, result_queues, write_queue, shutdown_recv
)
file_access.start()
return read_queue, result_queues, write_queue, shutdown_send
if __name__ == "__main__":
# See the discussion in :issue:`790`.
multiprocessing.set_start_method("spawn")
file_path = "test.h5"
n = 10
make_file(file_path, n)
num_processors = 3
(read_queue, result_queues, write_queue, shutdown_send) = make_queues(
num_processors
)
processors = []
output_files = []
for i in range(num_processors):
result_queue = result_queues[i]
output_file = str(i)
processor = DataProcessor(
read_queue, result_queue, write_queue, i, n, output_file
)
processors.append(processor)
output_files.append(output_file)
# start all DataProcessor instances
for processor in processors:
processor.start()
# wait for all DataProcessor instances to finish
for processor in processors:
processor.join()
# shut down the FileAccess instance
shutdown_send.send(0)
# print out contents of log files and delete them
print()
for output_file in output_files:
print()
print(f"contents of log file {output_file}")
print(open(output_file).read())
Path(output_file).unlink()
Path("test.h5").unlink()
|
