Mercurial > repos > guerler > springsuite
annotate planemo/lib/python3.7/site-packages/boto/glacier/utils.py @ 0:d30785e31577 draft
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
| author | guerler | 
|---|---|
| date | Fri, 31 Jul 2020 00:18:57 -0400 | 
| parents | |
| children | 
| rev | line source | 
|---|---|
| 0 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 1 # Copyright (c) 2012 Amazon.com, Inc. or its affiliates. All Rights Reserved | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 2 # | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 3 # Permission is hereby granted, free of charge, to any person obtaining a | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 4 # copy of this software and associated documentation files (the | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 5 # "Software"), to deal in the Software without restriction, including | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 6 # without limitation the rights to use, copy, modify, merge, publish, dis- | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 7 # tribute, sublicense, and/or sell copies of the Software, and to permit | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 8 # persons to whom the Software is furnished to do so, subject to the fol- | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 9 # lowing conditions: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 10 # | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 11 # The above copyright notice and this permission notice shall be included | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 12 # in all copies or substantial portions of the Software. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 13 # | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 14 # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 15 # OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABIL- | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 16 # ITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 17 # SHALL THE AUTHOR BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 18 # WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 19 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 20 # IN THE SOFTWARE. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 21 # | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 22 import hashlib | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 23 import math | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 24 import binascii | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 25 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 26 from boto.compat import six | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 27 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 28 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 29 _MEGABYTE = 1024 * 1024 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 30 DEFAULT_PART_SIZE = 4 * _MEGABYTE | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 31 MAXIMUM_NUMBER_OF_PARTS = 10000 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 32 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 33 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 34 def minimum_part_size(size_in_bytes, default_part_size=DEFAULT_PART_SIZE): | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 35 """Calculate the minimum part size needed for a multipart upload. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 36 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 37 Glacier allows a maximum of 10,000 parts per upload. It also | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 38 states that the maximum archive size is 10,000 * 4 GB, which means | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 39 the part size can range from 1MB to 4GB (provided it is one 1MB | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 40 multiplied by a power of 2). | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 41 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 42 This function will compute what the minimum part size must be in | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 43 order to upload a file of size ``size_in_bytes``. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 44 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 45 It will first check if ``default_part_size`` is sufficient for | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 46 a part size given the ``size_in_bytes``. If this is not the case, | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 47 then the smallest part size than can accomodate a file of size | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 48 ``size_in_bytes`` will be returned. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 49 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 50 If the file size is greater than the maximum allowed archive | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 51 size of 10,000 * 4GB, a ``ValueError`` will be raised. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 52 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 53 """ | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 54 # The default part size (4 MB) will be too small for a very large | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 55 # archive, as there is a limit of 10,000 parts in a multipart upload. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 56 # This puts the maximum allowed archive size with the default part size | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 57 # at 40,000 MB. We need to do a sanity check on the part size, and find | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 58 # one that works if the default is too small. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 59 part_size = _MEGABYTE | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 60 if (default_part_size * MAXIMUM_NUMBER_OF_PARTS) < size_in_bytes: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 61 if size_in_bytes > (4096 * _MEGABYTE * 10000): | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 62 raise ValueError("File size too large: %s" % size_in_bytes) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 63 min_part_size = size_in_bytes / 10000 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 64 power = 3 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 65 while part_size < min_part_size: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 66 part_size = math.ldexp(_MEGABYTE, power) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 67 power += 1 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 68 part_size = int(part_size) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 69 else: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 70 part_size = default_part_size | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 71 return part_size | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 72 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 73 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 74 def chunk_hashes(bytestring, chunk_size=_MEGABYTE): | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 75 chunk_count = int(math.ceil(len(bytestring) / float(chunk_size))) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 76 hashes = [] | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 77 for i in range(chunk_count): | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 78 start = i * chunk_size | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 79 end = (i + 1) * chunk_size | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 80 hashes.append(hashlib.sha256(bytestring[start:end]).digest()) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 81 if not hashes: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 82 return [hashlib.sha256(b'').digest()] | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 83 return hashes | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 84 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 85 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 86 def tree_hash(fo): | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 87 """ | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 88 Given a hash of each 1MB chunk (from chunk_hashes) this will hash | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 89 together adjacent hashes until it ends up with one big one. So a | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 90 tree of hashes. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 91 """ | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 92 hashes = [] | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 93 hashes.extend(fo) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 94 while len(hashes) > 1: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 95 new_hashes = [] | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 96 while True: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 97 if len(hashes) > 1: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 98 first = hashes.pop(0) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 99 second = hashes.pop(0) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 100 new_hashes.append(hashlib.sha256(first + second).digest()) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 101 elif len(hashes) == 1: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 102 only = hashes.pop(0) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 103 new_hashes.append(only) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 104 else: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 105 break | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 106 hashes.extend(new_hashes) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 107 return hashes[0] | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 108 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 109 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 110 def compute_hashes_from_fileobj(fileobj, chunk_size=1024 * 1024): | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 111 """Compute the linear and tree hash from a fileobj. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 112 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 113 This function will compute the linear/tree hash of a fileobj | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 114 in a single pass through the fileobj. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 115 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 116 :param fileobj: A file like object. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 117 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 118 :param chunk_size: The size of the chunks to use for the tree | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 119 hash. This is also the buffer size used to read from | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 120 `fileobj`. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 121 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 122 :rtype: tuple | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 123 :return: A tuple of (linear_hash, tree_hash). Both hashes | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 124 are returned in hex. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 125 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 126 """ | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 127 # Python 3+, not binary | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 128 if six.PY3 and hasattr(fileobj, 'mode') and 'b' not in fileobj.mode: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 129 raise ValueError('File-like object must be opened in binary mode!') | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 130 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 131 linear_hash = hashlib.sha256() | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 132 chunks = [] | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 133 chunk = fileobj.read(chunk_size) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 134 while chunk: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 135 # It's possible to get a file-like object that has no mode (checked | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 136 # above) and returns something other than bytes (e.g. str). So here | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 137 # we try to catch that and encode to bytes. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 138 if not isinstance(chunk, bytes): | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 139 chunk = chunk.encode(getattr(fileobj, 'encoding', '') or 'utf-8') | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 140 linear_hash.update(chunk) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 141 chunks.append(hashlib.sha256(chunk).digest()) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 142 chunk = fileobj.read(chunk_size) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 143 if not chunks: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 144 chunks = [hashlib.sha256(b'').digest()] | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 145 return linear_hash.hexdigest(), bytes_to_hex(tree_hash(chunks)) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 146 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 147 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 148 def bytes_to_hex(str_as_bytes): | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 149 return binascii.hexlify(str_as_bytes) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 150 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 151 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 152 def tree_hash_from_str(str_as_bytes): | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 153 """ | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 154 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 155 :type str_as_bytes: str | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 156 :param str_as_bytes: The string for which to compute the tree hash. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 157 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 158 :rtype: str | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 159 :return: The computed tree hash, returned as hex. | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 160 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 161 """ | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 162 return bytes_to_hex(tree_hash(chunk_hashes(str_as_bytes))) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 163 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 164 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 165 class ResettingFileSender(object): | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 166 def __init__(self, archive): | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 167 self._archive = archive | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 168 self._starting_offset = archive.tell() | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 169 | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 170 def __call__(self, connection, method, path, body, headers): | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 171 try: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 172 connection.request(method, path, self._archive, headers) | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 173 return connection.getresponse() | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 174 finally: | 
| 
d30785e31577
"planemo upload commit 6eee67778febed82ddd413c3ca40b3183a3898f1"
 guerler parents: diff
changeset | 175 self._archive.seek(self._starting_offset) | 
