python (3.11.7)
1 # SPDX-FileCopyrightText: 2015 Eric Larson
2 #
3 # SPDX-License-Identifier: Apache-2.0
4
5 from tempfile import NamedTemporaryFile
6 import mmap
7
8
9 class ESC[4;38;5;81mCallbackFileWrapper(ESC[4;38;5;149mobject):
10 """
11 Small wrapper around a fp object which will tee everything read into a
12 buffer, and when that file is closed it will execute a callback with the
13 contents of that buffer.
14
15 All attributes are proxied to the underlying file object.
16
17 This class uses members with a double underscore (__) leading prefix so as
18 not to accidentally shadow an attribute.
19
20 The data is stored in a temporary file until it is all available. As long
21 as the temporary files directory is disk-based (sometimes it's a
22 memory-backed-``tmpfs`` on Linux), data will be unloaded to disk if memory
23 pressure is high. For small files the disk usually won't be used at all,
24 it'll all be in the filesystem memory cache, so there should be no
25 performance impact.
26 """
27
28 def __init__(self, fp, callback):
29 self.__buf = NamedTemporaryFile("rb+", delete=True)
30 self.__fp = fp
31 self.__callback = callback
32
33 def __getattr__(self, name):
34 # The vaguaries of garbage collection means that self.__fp is
35 # not always set. By using __getattribute__ and the private
36 # name[0] allows looking up the attribute value and raising an
37 # AttributeError when it doesn't exist. This stop thigns from
38 # infinitely recursing calls to getattr in the case where
39 # self.__fp hasn't been set.
40 #
41 # [0] https://docs.python.org/2/reference/expressions.html#atom-identifiers
42 fp = self.__getattribute__("_CallbackFileWrapper__fp")
43 return getattr(fp, name)
44
45 def __is_fp_closed(self):
46 try:
47 return self.__fp.fp is None
48
49 except AttributeError:
50 pass
51
52 try:
53 return self.__fp.closed
54
55 except AttributeError:
56 pass
57
58 # We just don't cache it then.
59 # TODO: Add some logging here...
60 return False
61
62 def _close(self):
63 if self.__callback:
64 if self.__buf.tell() == 0:
65 # Empty file:
66 result = b""
67 else:
68 # Return the data without actually loading it into memory,
69 # relying on Python's buffer API and mmap(). mmap() just gives
70 # a view directly into the filesystem's memory cache, so it
71 # doesn't result in duplicate memory use.
72 self.__buf.seek(0, 0)
73 result = memoryview(
74 mmap.mmap(self.__buf.fileno(), 0, access=mmap.ACCESS_READ)
75 )
76 self.__callback(result)
77
78 # We assign this to None here, because otherwise we can get into
79 # really tricky problems where the CPython interpreter dead locks
80 # because the callback is holding a reference to something which
81 # has a __del__ method. Setting this to None breaks the cycle
82 # and allows the garbage collector to do it's thing normally.
83 self.__callback = None
84
85 # Closing the temporary file releases memory and frees disk space.
86 # Important when caching big files.
87 self.__buf.close()
88
89 def read(self, amt=None):
90 data = self.__fp.read(amt)
91 if data:
92 # We may be dealing with b'', a sign that things are over:
93 # it's passed e.g. after we've already closed self.__buf.
94 self.__buf.write(data)
95 if self.__is_fp_closed():
96 self._close()
97
98 return data
99
100 def _safe_read(self, amt):
101 data = self.__fp._safe_read(amt)
102 if amt == 2 and data == b"\r\n":
103 # urllib executes this read to toss the CRLF at the end
104 # of the chunk.
105 return data
106
107 self.__buf.write(data)
108 if self.__is_fp_closed():
109 self._close()
110
111 return data
Browse
Usage
Versions