Show More
@@ -1,425 +1,425 b'' | |||||
1 | """ |
|
1 | """ | |
2 | Module provides a class allowing to wrap communication over subprocess.Popen |
|
2 | Module provides a class allowing to wrap communication over subprocess.Popen | |
3 | input, output, error streams into a meaningfull, non-blocking, concurrent |
|
3 | input, output, error streams into a meaningfull, non-blocking, concurrent | |
4 | stream processor exposing the output data as an iterator fitting to be a |
|
4 | stream processor exposing the output data as an iterator fitting to be a | |
5 | return value passed by a WSGI applicaiton to a WSGI server per PEP 3333. |
|
5 | return value passed by a WSGI applicaiton to a WSGI server per PEP 3333. | |
6 |
|
6 | |||
7 | Copyright (c) 2011 Daniel Dotsenko <dotsa[at]hotmail.com> |
|
7 | Copyright (c) 2011 Daniel Dotsenko <dotsa[at]hotmail.com> | |
8 |
|
8 | |||
9 | This file is part of git_http_backend.py Project. |
|
9 | This file is part of git_http_backend.py Project. | |
10 |
|
10 | |||
11 | git_http_backend.py Project is free software: you can redistribute it and/or |
|
11 | git_http_backend.py Project is free software: you can redistribute it and/or | |
12 | modify it under the terms of the GNU Lesser General Public License as |
|
12 | modify it under the terms of the GNU Lesser General Public License as | |
13 | published by the Free Software Foundation, either version 2.1 of the License, |
|
13 | published by the Free Software Foundation, either version 2.1 of the License, | |
14 | or (at your option) any later version. |
|
14 | or (at your option) any later version. | |
15 |
|
15 | |||
16 | git_http_backend.py Project is distributed in the hope that it will be useful, |
|
16 | git_http_backend.py Project is distributed in the hope that it will be useful, | |
17 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
17 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
18 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
|
18 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
19 | GNU Lesser General Public License for more details. |
|
19 | GNU Lesser General Public License for more details. | |
20 |
|
20 | |||
21 | You should have received a copy of the GNU Lesser General Public License |
|
21 | You should have received a copy of the GNU Lesser General Public License | |
22 | along with git_http_backend.py Project. |
|
22 | along with git_http_backend.py Project. | |
23 | If not, see <http://www.gnu.org/licenses/>. |
|
23 | If not, see <http://www.gnu.org/licenses/>. | |
24 | """ |
|
24 | """ | |
25 | import os |
|
25 | import os | |
26 | import subprocess |
|
26 | import subprocess | |
27 | from kallithea.lib.vcs.utils.compat import deque, Event, Thread, _bytes, _bytearray |
|
27 | from kallithea.lib.vcs.utils.compat import deque, Event, Thread, _bytes, _bytearray | |
28 |
|
28 | |||
29 |
|
29 | |||
30 | class StreamFeeder(Thread): |
|
30 | class StreamFeeder(Thread): | |
31 | """ |
|
31 | """ | |
32 | Normal writing into pipe-like is blocking once the buffer is filled. |
|
32 | Normal writing into pipe-like is blocking once the buffer is filled. | |
33 | This thread allows a thread to seep data from a file-like into a pipe |
|
33 | This thread allows a thread to seep data from a file-like into a pipe | |
34 | without blocking the main thread. |
|
34 | without blocking the main thread. | |
35 | We close inpipe once the end of the source stream is reached. |
|
35 | We close inpipe once the end of the source stream is reached. | |
36 | """ |
|
36 | """ | |
37 |
|
37 | |||
38 | def __init__(self, source): |
|
38 | def __init__(self, source): | |
39 | super(StreamFeeder, self).__init__() |
|
39 | super(StreamFeeder, self).__init__() | |
40 | self.daemon = True |
|
40 | self.daemon = True | |
41 | filelike = False |
|
41 | filelike = False | |
42 | self.bytes = _bytes() |
|
42 | self.bytes = _bytes() | |
43 | if type(source) in (type(''), _bytes, _bytearray): # string-like |
|
43 | if type(source) in (type(''), _bytes, _bytearray): # string-like | |
44 | self.bytes = _bytes(source) |
|
44 | self.bytes = _bytes(source) | |
45 | else: # can be either file pointer or file-like |
|
45 | else: # can be either file pointer or file-like | |
46 | if type(source) in (int, long): # file pointer it is |
|
46 | if type(source) in (int, long): # file pointer it is | |
47 | ## converting file descriptor (int) stdin into file-like |
|
47 | ## converting file descriptor (int) stdin into file-like | |
48 | source = os.fdopen(source, 'rb', 16384) |
|
48 | source = os.fdopen(source, 'rb', 16384) | |
49 | # let's see if source is file-like by now |
|
49 | # let's see if source is file-like by now | |
50 | filelike = hasattr(source, 'read') |
|
50 | filelike = hasattr(source, 'read') | |
51 | if not filelike and not self.bytes: |
|
51 | if not filelike and not self.bytes: | |
52 | raise TypeError("StreamFeeder's source object must be a readable " |
|
52 | raise TypeError("StreamFeeder's source object must be a readable " | |
53 | "file-like, a file descriptor, or a string-like.") |
|
53 | "file-like, a file descriptor, or a string-like.") | |
54 | self.source = source |
|
54 | self.source = source | |
55 | self.readiface, self.writeiface = os.pipe() |
|
55 | self.readiface, self.writeiface = os.pipe() | |
56 |
|
56 | |||
57 | def run(self): |
|
57 | def run(self): | |
58 | t = self.writeiface |
|
58 | t = self.writeiface | |
59 | if self.bytes: |
|
59 | if self.bytes: | |
60 | os.write(t, self.bytes) |
|
60 | os.write(t, self.bytes) | |
61 | else: |
|
61 | else: | |
62 | s = self.source |
|
62 | s = self.source | |
63 | b = s.read(4096) |
|
63 | b = s.read(4096) | |
64 | while b: |
|
64 | while b: | |
65 | os.write(t, b) |
|
65 | os.write(t, b) | |
66 | b = s.read(4096) |
|
66 | b = s.read(4096) | |
67 | os.close(t) |
|
67 | os.close(t) | |
68 |
|
68 | |||
69 | @property |
|
69 | @property | |
70 | def output(self): |
|
70 | def output(self): | |
71 | return self.readiface |
|
71 | return self.readiface | |
72 |
|
72 | |||
73 |
|
73 | |||
74 | class InputStreamChunker(Thread): |
|
74 | class InputStreamChunker(Thread): | |
75 | def __init__(self, source, target, buffer_size, chunk_size): |
|
75 | def __init__(self, source, target, buffer_size, chunk_size): | |
76 |
|
76 | |||
77 | super(InputStreamChunker, self).__init__() |
|
77 | super(InputStreamChunker, self).__init__() | |
78 |
|
78 | |||
79 | self.daemon = True # die die die. |
|
79 | self.daemon = True # die die die. | |
80 |
|
80 | |||
81 | self.source = source |
|
81 | self.source = source | |
82 | self.target = target |
|
82 | self.target = target | |
83 | self.chunk_count_max = int(buffer_size / chunk_size) + 1 |
|
83 | self.chunk_count_max = int(buffer_size / chunk_size) + 1 | |
84 | self.chunk_size = chunk_size |
|
84 | self.chunk_size = chunk_size | |
85 |
|
85 | |||
86 | self.data_added = Event() |
|
86 | self.data_added = Event() | |
87 | self.data_added.clear() |
|
87 | self.data_added.clear() | |
88 |
|
88 | |||
89 | self.keep_reading = Event() |
|
89 | self.keep_reading = Event() | |
90 | self.keep_reading.set() |
|
90 | self.keep_reading.set() | |
91 |
|
91 | |||
92 | self.EOF = Event() |
|
92 | self.EOF = Event() | |
93 | self.EOF.clear() |
|
93 | self.EOF.clear() | |
94 |
|
94 | |||
95 | self.go = Event() |
|
95 | self.go = Event() | |
96 | self.go.set() |
|
96 | self.go.set() | |
97 |
|
97 | |||
98 | def stop(self): |
|
98 | def stop(self): | |
99 | self.go.clear() |
|
99 | self.go.clear() | |
100 | self.EOF.set() |
|
100 | self.EOF.set() | |
101 | try: |
|
101 | try: | |
102 | # this is not proper, but is done to force the reader thread let |
|
102 | # this is not proper, but is done to force the reader thread let | |
103 | # go of the input because, if successful, .close() will send EOF |
|
103 | # go of the input because, if successful, .close() will send EOF | |
104 | # down the pipe. |
|
104 | # down the pipe. | |
105 | self.source.close() |
|
105 | self.source.close() | |
106 | except: |
|
106 | except: | |
107 | pass |
|
107 | pass | |
108 |
|
108 | |||
109 | def run(self): |
|
109 | def run(self): | |
110 | s = self.source |
|
110 | s = self.source | |
111 | t = self.target |
|
111 | t = self.target | |
112 | cs = self.chunk_size |
|
112 | cs = self.chunk_size | |
113 | ccm = self.chunk_count_max |
|
113 | ccm = self.chunk_count_max | |
114 | kr = self.keep_reading |
|
114 | kr = self.keep_reading | |
115 | da = self.data_added |
|
115 | da = self.data_added | |
116 | go = self.go |
|
116 | go = self.go | |
117 |
|
117 | |||
118 | try: |
|
118 | try: | |
119 | b = s.read(cs) |
|
119 | b = s.read(cs) | |
120 | except ValueError: |
|
120 | except ValueError: | |
121 | b = '' |
|
121 | b = '' | |
122 |
|
122 | |||
123 | while b and go.is_set(): |
|
123 | while b and go.is_set(): | |
124 | if len(t) > ccm: |
|
124 | if len(t) > ccm: | |
125 | kr.clear() |
|
125 | kr.clear() | |
126 | kr.wait(2) |
|
126 | kr.wait(2) | |
127 | # # this only works on 2.7.x and up |
|
127 | # # this only works on 2.7.x and up | |
128 | # if not kr.wait(10): |
|
128 | # if not kr.wait(10): | |
129 | # raise Exception("Timed out while waiting for input to be read.") |
|
129 | # raise Exception("Timed out while waiting for input to be read.") | |
130 | # instead we'll use this |
|
130 | # instead we'll use this | |
131 | if len(t) > ccm + 3: |
|
131 | if len(t) > ccm + 3: | |
132 | raise IOError( |
|
132 | raise IOError( | |
133 | "Timed out while waiting for input from subprocess.") |
|
133 | "Timed out while waiting for input from subprocess.") | |
134 | t.append(b) |
|
134 | t.append(b) | |
135 | da.set() |
|
135 | da.set() | |
136 | try: |
|
136 | try: | |
137 | b = s.read(cs) |
|
137 | b = s.read(cs) | |
138 | except ValueError: # probably "I/O operation on closed file" |
|
138 | except ValueError: # probably "I/O operation on closed file" | |
139 | b = '' |
|
139 | b = '' | |
140 |
|
140 | |||
141 | self.EOF.set() |
|
141 | self.EOF.set() | |
142 | da.set() # for cases when done but there was no input. |
|
142 | da.set() # for cases when done but there was no input. | |
143 |
|
143 | |||
144 |
|
144 | |||
145 | class BufferedGenerator(object): |
|
145 | class BufferedGenerator(object): | |
146 | """ |
|
146 | """ | |
147 | Class behaves as a non-blocking, buffered pipe reader. |
|
147 | Class behaves as a non-blocking, buffered pipe reader. | |
148 | Reads chunks of data (through a thread) |
|
148 | Reads chunks of data (through a thread) | |
149 | from a blocking pipe, and attaches these to an array (Deque) of chunks. |
|
149 | from a blocking pipe, and attaches these to an array (Deque) of chunks. | |
150 | Reading is halted in the thread when max chunks is internally buffered. |
|
150 | Reading is halted in the thread when max chunks is internally buffered. | |
151 | The .next() may operate in blocking or non-blocking fashion by yielding |
|
151 | The .next() may operate in blocking or non-blocking fashion by yielding | |
152 | '' if no data is ready |
|
152 | '' if no data is ready | |
153 | to be sent or by not returning until there is some data to send |
|
153 | to be sent or by not returning until there is some data to send | |
154 | When we get EOF from underlying source pipe we raise the marker to raise |
|
154 | When we get EOF from underlying source pipe we raise the marker to raise | |
155 | StopIteration after the last chunk of data is yielded. |
|
155 | StopIteration after the last chunk of data is yielded. | |
156 | """ |
|
156 | """ | |
157 |
|
157 | |||
158 | def __init__(self, source, buffer_size=65536, chunk_size=4096, |
|
158 | def __init__(self, source, buffer_size=65536, chunk_size=4096, | |
159 | starting_values=None, bottomless=False): |
|
159 | starting_values=None, bottomless=False): | |
160 | starting_values = starting_values or [] |
|
160 | starting_values = starting_values or [] | |
161 | if bottomless: |
|
161 | if bottomless: | |
162 | maxlen = int(buffer_size / chunk_size) |
|
162 | maxlen = int(buffer_size / chunk_size) | |
163 | else: |
|
163 | else: | |
164 | maxlen = None |
|
164 | maxlen = None | |
165 |
|
165 | |||
166 | self.data = deque(starting_values, maxlen) |
|
166 | self.data = deque(starting_values, maxlen) | |
167 | self.worker = InputStreamChunker(source, self.data, buffer_size, |
|
167 | self.worker = InputStreamChunker(source, self.data, buffer_size, | |
168 | chunk_size) |
|
168 | chunk_size) | |
169 | if starting_values: |
|
169 | if starting_values: | |
170 | self.worker.data_added.set() |
|
170 | self.worker.data_added.set() | |
171 | self.worker.start() |
|
171 | self.worker.start() | |
172 |
|
172 | |||
173 | #################### |
|
173 | #################### | |
174 | # Generator's methods |
|
174 | # Generator's methods | |
175 | #################### |
|
175 | #################### | |
176 |
|
176 | |||
177 | def __iter__(self): |
|
177 | def __iter__(self): | |
178 | return self |
|
178 | return self | |
179 |
|
179 | |||
180 | def next(self): |
|
180 | def next(self): | |
181 | while not len(self.data) and not self.worker.EOF.is_set(): |
|
181 | while not len(self.data) and not self.worker.EOF.is_set(): | |
182 | self.worker.data_added.clear() |
|
182 | self.worker.data_added.clear() | |
183 | self.worker.data_added.wait(0.2) |
|
183 | self.worker.data_added.wait(0.2) | |
184 | if len(self.data): |
|
184 | if len(self.data): | |
185 | self.worker.keep_reading.set() |
|
185 | self.worker.keep_reading.set() | |
186 | return _bytes(self.data.popleft()) |
|
186 | return _bytes(self.data.popleft()) | |
187 | elif self.worker.EOF.is_set(): |
|
187 | elif self.worker.EOF.is_set(): | |
188 | raise StopIteration |
|
188 | raise StopIteration | |
189 |
|
189 | |||
190 | def throw(self, type, value=None, traceback=None): |
|
190 | def throw(self, type, value=None, traceback=None): | |
191 | if not self.worker.EOF.is_set(): |
|
191 | if not self.worker.EOF.is_set(): | |
192 | raise type(value) |
|
192 | raise type(value) | |
193 |
|
193 | |||
194 | def start(self): |
|
194 | def start(self): | |
195 | self.worker.start() |
|
195 | self.worker.start() | |
196 |
|
196 | |||
197 | def stop(self): |
|
197 | def stop(self): | |
198 | self.worker.stop() |
|
198 | self.worker.stop() | |
199 |
|
199 | |||
200 | def close(self): |
|
200 | def close(self): | |
201 | try: |
|
201 | try: | |
202 | self.worker.stop() |
|
202 | self.worker.stop() | |
203 | self.throw(GeneratorExit) |
|
203 | self.throw(GeneratorExit) | |
204 | except (GeneratorExit, StopIteration): |
|
204 | except (GeneratorExit, StopIteration): | |
205 | pass |
|
205 | pass | |
206 |
|
206 | |||
207 | def __del__(self): |
|
207 | def __del__(self): | |
208 | self.close() |
|
208 | self.close() | |
209 |
|
209 | |||
210 | #################### |
|
210 | #################### | |
211 | # Threaded reader's infrastructure. |
|
211 | # Threaded reader's infrastructure. | |
212 | #################### |
|
212 | #################### | |
213 | @property |
|
213 | @property | |
214 | def input(self): |
|
214 | def input(self): | |
215 | return self.worker.w |
|
215 | return self.worker.w | |
216 |
|
216 | |||
217 | @property |
|
217 | @property | |
218 | def data_added_event(self): |
|
218 | def data_added_event(self): | |
219 | return self.worker.data_added |
|
219 | return self.worker.data_added | |
220 |
|
220 | |||
221 | @property |
|
221 | @property | |
222 | def data_added(self): |
|
222 | def data_added(self): | |
223 | return self.worker.data_added.is_set() |
|
223 | return self.worker.data_added.is_set() | |
224 |
|
224 | |||
225 | @property |
|
225 | @property | |
226 | def reading_paused(self): |
|
226 | def reading_paused(self): | |
227 | return not self.worker.keep_reading.is_set() |
|
227 | return not self.worker.keep_reading.is_set() | |
228 |
|
228 | |||
229 | @property |
|
229 | @property | |
230 | def done_reading_event(self): |
|
230 | def done_reading_event(self): | |
231 | """ |
|
231 | """ | |
232 | Done_reading does not mean that the iterator's buffer is empty. |
|
232 | Done_reading does not mean that the iterator's buffer is empty. | |
233 | Iterator might have done reading from underlying source, but the read |
|
233 | Iterator might have done reading from underlying source, but the read | |
234 | chunks might still be available for serving through .next() method. |
|
234 | chunks might still be available for serving through .next() method. | |
235 |
|
235 | |||
236 | :returns: An Event class instance. |
|
236 | :returns: An Event class instance. | |
237 | """ |
|
237 | """ | |
238 | return self.worker.EOF |
|
238 | return self.worker.EOF | |
239 |
|
239 | |||
240 | @property |
|
240 | @property | |
241 | def done_reading(self): |
|
241 | def done_reading(self): | |
242 | """ |
|
242 | """ | |
243 | Done_reading does not mean that the iterator's buffer is empty. |
|
243 | Done_reading does not mean that the iterator's buffer is empty. | |
244 | Iterator might have done reading from underlying source, but the read |
|
244 | Iterator might have done reading from underlying source, but the read | |
245 | chunks might still be available for serving through .next() method. |
|
245 | chunks might still be available for serving through .next() method. | |
246 |
|
246 | |||
247 | :returns: An Bool value. |
|
247 | :returns: An Bool value. | |
248 | """ |
|
248 | """ | |
249 | return self.worker.EOF.is_set() |
|
249 | return self.worker.EOF.is_set() | |
250 |
|
250 | |||
251 | @property |
|
251 | @property | |
252 | def length(self): |
|
252 | def length(self): | |
253 | """ |
|
253 | """ | |
254 | returns int. |
|
254 | returns int. | |
255 |
|
255 | |||
256 | This is the length of the queue of chunks, not the length of |
|
256 | This is the length of the queue of chunks, not the length of | |
257 | the combined contents in those chunks. |
|
257 | the combined contents in those chunks. | |
258 |
|
258 | |||
259 | __len__() cannot be meaningfully implemented because this |
|
259 | __len__() cannot be meaningfully implemented because this | |
260 | reader is just flying through a bottomless pit content and |
|
260 | reader is just flying through a bottomless pit content and | |
261 | can only know the length of what it already saw. |
|
261 | can only know the length of what it already saw. | |
262 |
|
262 | |||
263 | If __len__() on WSGI server per PEP 3333 returns a value, |
|
263 | If __len__() on WSGI server per PEP 3333 returns a value, | |
264 | the response's length will be set to that. In order not to |
|
264 | the response's length will be set to that. In order not to | |
265 | confuse WSGI PEP3333 servers, we will not implement __len__ |
|
265 | confuse WSGI PEP3333 servers, we will not implement __len__ | |
266 | at all. |
|
266 | at all. | |
267 | """ |
|
267 | """ | |
268 | return len(self.data) |
|
268 | return len(self.data) | |
269 |
|
269 | |||
270 | def prepend(self, x): |
|
270 | def prepend(self, x): | |
271 | self.data.appendleft(x) |
|
271 | self.data.appendleft(x) | |
272 |
|
272 | |||
273 | def append(self, x): |
|
273 | def append(self, x): | |
274 | self.data.append(x) |
|
274 | self.data.append(x) | |
275 |
|
275 | |||
276 | def extend(self, o): |
|
276 | def extend(self, o): | |
277 | self.data.extend(o) |
|
277 | self.data.extend(o) | |
278 |
|
278 | |||
279 | def __getitem__(self, i): |
|
279 | def __getitem__(self, i): | |
280 | return self.data[i] |
|
280 | return self.data[i] | |
281 |
|
281 | |||
282 |
|
282 | |||
283 | class SubprocessIOChunker(object): |
|
283 | class SubprocessIOChunker(object): | |
284 | """ |
|
284 | """ | |
285 | Processor class wrapping handling of subprocess IO. |
|
285 | Processor class wrapping handling of subprocess IO. | |
286 |
|
286 | |||
287 | In a way, this is a "communicate()" replacement with a twist. |
|
287 | In a way, this is a "communicate()" replacement with a twist. | |
288 |
|
288 | |||
289 | - We are multithreaded. Writing in and reading out, err are all sep threads. |
|
289 | - We are multithreaded. Writing in and reading out, err are all sep threads. | |
290 | - We support concurrent (in and out) stream processing. |
|
290 | - We support concurrent (in and out) stream processing. | |
291 | - The output is not a stream. It's a queue of read string (bytes, not unicode) |
|
291 | - The output is not a stream. It's a queue of read string (bytes, not unicode) | |
292 | chunks. The object behaves as an iterable. You can "for chunk in obj:" us. |
|
292 | chunks. The object behaves as an iterable. You can "for chunk in obj:" us. | |
293 | - We are non-blocking in more respects than communicate() |
|
293 | - We are non-blocking in more respects than communicate() | |
294 | (reading from subprocess out pauses when internal buffer is full, but |
|
294 | (reading from subprocess out pauses when internal buffer is full, but | |
295 | does not block the parent calling code. On the flip side, reading from |
|
295 | does not block the parent calling code. On the flip side, reading from | |
296 | slow-yielding subprocess may block the iteration until data shows up. This |
|
296 | slow-yielding subprocess may block the iteration until data shows up. This | |
297 | does not block the parallel inpipe reading occurring parallel thread.) |
|
297 | does not block the parallel inpipe reading occurring parallel thread.) | |
298 |
|
298 | |||
299 | The purpose of the object is to allow us to wrap subprocess interactions into |
|
299 | The purpose of the object is to allow us to wrap subprocess interactions into | |
300 | an iterable that can be passed to a WSGI server as the application's return |
|
300 | an iterable that can be passed to a WSGI server as the application's return | |
301 | value. Because of stream-processing-ability, WSGI does not have to read ALL |
|
301 | value. Because of stream-processing-ability, WSGI does not have to read ALL | |
302 | of the subprocess's output and buffer it, before handing it to WSGI server for |
|
302 | of the subprocess's output and buffer it, before handing it to WSGI server for | |
303 | HTTP response. Instead, the class initializer reads just a bit of the stream |
|
303 | HTTP response. Instead, the class initializer reads just a bit of the stream | |
304 | to figure out if error occurred or likely to occur and if not, just hands the |
|
304 | to figure out if error occurred or likely to occur and if not, just hands the | |
305 | further iteration over subprocess output to the server for completion of HTTP |
|
305 | further iteration over subprocess output to the server for completion of HTTP | |
306 | response. |
|
306 | response. | |
307 |
|
307 | |||
308 | The real or perceived subprocess error is trapped and raised as one of |
|
308 | The real or perceived subprocess error is trapped and raised as one of | |
309 | EnvironmentError family of exceptions |
|
309 | EnvironmentError family of exceptions | |
310 |
|
310 | |||
311 | Example usage: |
|
311 | Example usage: | |
312 | # try: |
|
312 | # try: | |
313 | # answer = SubprocessIOChunker( |
|
313 | # answer = SubprocessIOChunker( | |
314 | # cmd, |
|
314 | # cmd, | |
315 | # input, |
|
315 | # input, | |
316 | # buffer_size = 65536, |
|
316 | # buffer_size = 65536, | |
317 | # chunk_size = 4096 |
|
317 | # chunk_size = 4096 | |
318 | # ) |
|
318 | # ) | |
319 | # except (EnvironmentError) as e: |
|
319 | # except (EnvironmentError) as e: | |
320 | # print str(e) |
|
320 | # print str(e) | |
321 | # raise e |
|
321 | # raise e | |
322 | # |
|
322 | # | |
323 | # return answer |
|
323 | # return answer | |
324 |
|
324 | |||
325 |
|
325 | |||
326 | """ |
|
326 | """ | |
327 |
|
327 | |||
328 | def __init__(self, cmd, inputstream=None, buffer_size=65536, |
|
328 | def __init__(self, cmd, inputstream=None, buffer_size=65536, | |
329 | chunk_size=4096, starting_values=None, **kwargs): |
|
329 | chunk_size=4096, starting_values=None, **kwargs): | |
330 | """ |
|
330 | """ | |
331 | Initializes SubprocessIOChunker |
|
331 | Initializes SubprocessIOChunker | |
332 |
|
332 | |||
333 | :param cmd: A Subprocess.Popen style "cmd". Can be string or array of strings |
|
333 | :param cmd: A Subprocess.Popen style "cmd". Can be string or array of strings | |
334 | :param inputstream: (Default: None) A file-like, string, or file pointer. |
|
334 | :param inputstream: (Default: None) A file-like, string, or file pointer. | |
335 | :param buffer_size: (Default: 65536) A size of total buffer per stream in bytes. |
|
335 | :param buffer_size: (Default: 65536) A size of total buffer per stream in bytes. | |
336 | :param chunk_size: (Default: 4096) A max size of a chunk. Actual chunk may be smaller. |
|
336 | :param chunk_size: (Default: 4096) A max size of a chunk. Actual chunk may be smaller. | |
337 | :param starting_values: (Default: []) An array of strings to put in front of output que. |
|
337 | :param starting_values: (Default: []) An array of strings to put in front of output que. | |
338 | """ |
|
338 | """ | |
339 | starting_values = starting_values or [] |
|
339 | starting_values = starting_values or [] | |
340 | if inputstream: |
|
340 | if inputstream: | |
341 | input_streamer = StreamFeeder(inputstream) |
|
341 | input_streamer = StreamFeeder(inputstream) | |
342 | input_streamer.start() |
|
342 | input_streamer.start() | |
343 | inputstream = input_streamer.output |
|
343 | inputstream = input_streamer.output | |
344 |
|
344 | |||
345 | # Note: fragile cmd mangling has been removed for use in Kallithea |
|
345 | # Note: fragile cmd mangling has been removed for use in Kallithea | |
346 | assert isinstance(cmd, list), cmd |
|
346 | assert isinstance(cmd, list), cmd | |
347 |
|
347 | |||
348 | _p = subprocess.Popen(cmd, bufsize=-1, |
|
348 | _p = subprocess.Popen(cmd, bufsize=-1, | |
349 | stdin=inputstream, |
|
349 | stdin=inputstream, | |
350 | stdout=subprocess.PIPE, |
|
350 | stdout=subprocess.PIPE, | |
351 | stderr=subprocess.PIPE, |
|
351 | stderr=subprocess.PIPE, | |
352 | **kwargs) |
|
352 | **kwargs) | |
353 |
|
353 | |||
354 | bg_out = BufferedGenerator(_p.stdout, buffer_size, chunk_size, |
|
354 | bg_out = BufferedGenerator(_p.stdout, buffer_size, chunk_size, | |
355 | starting_values) |
|
355 | starting_values) | |
356 | bg_err = BufferedGenerator(_p.stderr, 16000, 1, bottomless=True) |
|
356 | bg_err = BufferedGenerator(_p.stderr, 16000, 1, bottomless=True) | |
357 |
|
357 | |||
358 |
while not bg_out.done_reading and not bg_out.reading_paused |
|
358 | while not bg_out.done_reading and not bg_out.reading_paused: | |
359 | # doing this until we reach either end of file, or end of buffer. |
|
359 | # doing this until we reach either end of file, or end of buffer. | |
360 | bg_out.data_added_event.wait(1) |
|
360 | bg_out.data_added_event.wait(1) | |
361 | bg_out.data_added_event.clear() |
|
361 | bg_out.data_added_event.clear() | |
362 |
|
362 | |||
363 | # at this point it's still ambiguous if we are done reading or just full buffer. |
|
363 | # at this point it's still ambiguous if we are done reading or just full buffer. | |
364 | # Either way, if error (returned by ended process, or implied based on |
|
364 | # Either way, if error (returned by ended process, or implied based on | |
365 | # presence of stuff in stderr output) we error out. |
|
365 | # presence of stuff in stderr output) we error out. | |
366 | # Else, we are happy. |
|
366 | # Else, we are happy. | |
367 | _returncode = _p.poll() |
|
367 | _returncode = _p.poll() | |
368 | if _returncode or (_returncode is None and bg_err.length): |
|
368 | if _returncode or (_returncode is None and bg_err.length): | |
369 | try: |
|
369 | try: | |
370 | _p.terminate() |
|
370 | _p.terminate() | |
371 | except Exception: |
|
371 | except Exception: | |
372 | pass |
|
372 | pass | |
373 | bg_out.stop() |
|
373 | bg_out.stop() | |
374 | out = ''.join(bg_out) |
|
374 | out = ''.join(bg_out) | |
375 | bg_err.stop() |
|
375 | bg_err.stop() | |
376 | err = ''.join(bg_err) |
|
376 | err = ''.join(bg_err) | |
377 | if (err.strip() == 'fatal: The remote end hung up unexpectedly' and |
|
377 | if (err.strip() == 'fatal: The remote end hung up unexpectedly' and | |
378 | out.startswith('0034shallow ')): |
|
378 | out.startswith('0034shallow ')): | |
379 | # hack inspired by https://github.com/schacon/grack/pull/7 |
|
379 | # hack inspired by https://github.com/schacon/grack/pull/7 | |
380 | bg_out = iter([out]) |
|
380 | bg_out = iter([out]) | |
381 | _p = None |
|
381 | _p = None | |
382 | elif err: |
|
382 | elif err: | |
383 | raise EnvironmentError( |
|
383 | raise EnvironmentError( | |
384 | "Subprocess exited due to an error:\n" + err) |
|
384 | "Subprocess exited due to an error:\n" + err) | |
385 | else: |
|
385 | else: | |
386 | raise EnvironmentError( |
|
386 | raise EnvironmentError( | |
387 | "Subprocess exited with non 0 ret code:%s" % _returncode) |
|
387 | "Subprocess exited with non 0 ret code:%s" % _returncode) | |
388 | self.process = _p |
|
388 | self.process = _p | |
389 | self.output = bg_out |
|
389 | self.output = bg_out | |
390 | self.error = bg_err |
|
390 | self.error = bg_err | |
391 | self.inputstream = inputstream |
|
391 | self.inputstream = inputstream | |
392 |
|
392 | |||
393 | def __iter__(self): |
|
393 | def __iter__(self): | |
394 | return self |
|
394 | return self | |
395 |
|
395 | |||
396 | def next(self): |
|
396 | def next(self): | |
397 | if self.process and self.process.poll(): |
|
397 | if self.process and self.process.poll(): | |
398 | err = ''.join(self.error) |
|
398 | err = ''.join(self.error) | |
399 | raise EnvironmentError("Subprocess exited due to an error:\n" + err) |
|
399 | raise EnvironmentError("Subprocess exited due to an error:\n" + err) | |
400 | return self.output.next() |
|
400 | return self.output.next() | |
401 |
|
401 | |||
402 | def throw(self, type, value=None, traceback=None): |
|
402 | def throw(self, type, value=None, traceback=None): | |
403 | if self.output.length or not self.output.done_reading: |
|
403 | if self.output.length or not self.output.done_reading: | |
404 | raise type(value) |
|
404 | raise type(value) | |
405 |
|
405 | |||
406 | def close(self): |
|
406 | def close(self): | |
407 | try: |
|
407 | try: | |
408 | self.process.terminate() |
|
408 | self.process.terminate() | |
409 | except: |
|
409 | except: | |
410 | pass |
|
410 | pass | |
411 | try: |
|
411 | try: | |
412 | self.output.close() |
|
412 | self.output.close() | |
413 | except: |
|
413 | except: | |
414 | pass |
|
414 | pass | |
415 | try: |
|
415 | try: | |
416 | self.error.close() |
|
416 | self.error.close() | |
417 | except: |
|
417 | except: | |
418 | pass |
|
418 | pass | |
419 | try: |
|
419 | try: | |
420 | os.close(self.inputstream) |
|
420 | os.close(self.inputstream) | |
421 | except: |
|
421 | except: | |
422 | pass |
|
422 | pass | |
423 |
|
423 | |||
424 | def __del__(self): |
|
424 | def __del__(self): | |
425 | self.close() |
|
425 | self.close() |
General Comments 0
You need to be logged in to leave comments.
Login now