annotate contrib/packaging/hgpackaging/downloads.py @ 48848:17d5e25b8e78

packaging: remove py2exe / Python 2.7 support This commit started by deleting references to py2exe (which is only used on Python 2). After pulling the thread, quite a lot of code was orphaned and was deleted. Differential Revision: https://phab.mercurial-scm.org/D12265
author Gregory Szorc <gregory.szorc@gmail.com>
date Sat, 19 Feb 2022 22:13:11 -0700
parents 2372284d9457
children a7fe96abcf56
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
41952
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
1 # downloads.py - Code for downloading dependencies.
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
2 #
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
3 # Copyright 2019 Gregory Szorc <gregory.szorc@gmail.com>
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
4 #
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
5 # This software may be used and distributed according to the terms of the
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
6 # GNU General Public License version 2 or any later version.
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
7
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
8 # no-check-code because Python 3 native.
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
9
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
10 import gzip
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
11 import hashlib
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
12 import pathlib
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
13 import urllib.request
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
14
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
15
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
16 DOWNLOADS = {
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
17 'gettext': {
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
18 'url': 'https://versaweb.dl.sourceforge.net/project/gnuwin32/gettext/0.14.4/gettext-0.14.4-bin.zip',
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
19 'size': 1606131,
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
20 'sha256': '60b9ef26bc5cceef036f0424e542106cf158352b2677f43a01affd6d82a1d641',
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
21 'version': '0.14.4',
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
22 },
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
23 'gettext-dep': {
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
24 'url': 'https://versaweb.dl.sourceforge.net/project/gnuwin32/gettext/0.14.4/gettext-0.14.4-dep.zip',
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
25 'size': 715086,
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
26 'sha256': '411f94974492fd2ecf52590cb05b1023530aec67e64154a88b1e4ebcd9c28588',
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
27 },
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
28 }
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
29
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
30
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
31 def hash_path(p: pathlib.Path):
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
32 h = hashlib.sha256()
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
33
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
34 with p.open('rb') as fh:
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
35 while True:
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
36 chunk = fh.read(65536)
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
37 if not chunk:
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
38 break
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
39
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
40 h.update(chunk)
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
41
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
42 return h.hexdigest()
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
43
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
44
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
45 class IntegrityError(Exception):
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
46 """Represents an integrity error when downloading a URL."""
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
47
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
48
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
49 def secure_download_stream(url, size, sha256):
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
50 """Securely download a URL to a stream of chunks.
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
51
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
52 If the integrity of the download fails, an IntegrityError is
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
53 raised.
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
54 """
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
55 h = hashlib.sha256()
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
56 length = 0
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
57
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
58 with urllib.request.urlopen(url) as fh:
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 41952
diff changeset
59 if (
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 41952
diff changeset
60 not url.endswith('.gz')
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 41952
diff changeset
61 and fh.info().get('Content-Encoding') == 'gzip'
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 41952
diff changeset
62 ):
41952
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
63 fh = gzip.GzipFile(fileobj=fh)
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
64
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
65 while True:
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
66 chunk = fh.read(65536)
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
67 if not chunk:
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
68 break
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
69
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
70 h.update(chunk)
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
71 length += len(chunk)
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
72
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
73 yield chunk
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
74
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
75 digest = h.hexdigest()
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
76
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
77 if length != size:
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 41952
diff changeset
78 raise IntegrityError(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 41952
diff changeset
79 'size mismatch on %s: wanted %d; got %d' % (url, size, length)
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 41952
diff changeset
80 )
41952
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
81
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
82 if digest != sha256:
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 41952
diff changeset
83 raise IntegrityError(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 41952
diff changeset
84 'sha256 mismatch on %s: wanted %s; got %s' % (url, sha256, digest)
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 41952
diff changeset
85 )
41952
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
86
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
87
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
88 def download_to_path(url: str, path: pathlib.Path, size: int, sha256: str):
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
89 """Download a URL to a filesystem path, possibly with verification."""
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
90
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
91 # We download to a temporary file and rename at the end so there's
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
92 # no chance of the final file being partially written or containing
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
93 # bad data.
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
94 print('downloading %s to %s' % (url, path))
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
95
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
96 if path.exists():
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
97 good = True
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
98
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
99 if path.stat().st_size != size:
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
100 print('existing file size is wrong; removing')
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
101 good = False
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
102
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
103 if good:
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
104 if hash_path(path) != sha256:
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
105 print('existing file hash is wrong; removing')
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
106 good = False
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
107
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
108 if good:
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
109 print('%s exists and passes integrity checks' % path)
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
110 return
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
111
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
112 path.unlink()
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
113
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
114 tmp = path.with_name('%s.tmp' % path.name)
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
115
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
116 try:
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
117 with tmp.open('wb') as fh:
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
118 for chunk in secure_download_stream(url, size, sha256):
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
119 fh.write(chunk)
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
120 except IntegrityError:
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
121 tmp.unlink()
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
122 raise
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
123
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
124 tmp.rename(path)
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
125 print('successfully downloaded %s' % url)
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
126
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
127
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 41952
diff changeset
128 def download_entry(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 41952
diff changeset
129 name: dict, dest_path: pathlib.Path, local_name=None
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 41952
diff changeset
130 ) -> pathlib.Path:
41952
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
131 entry = DOWNLOADS[name]
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
132
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
133 url = entry['url']
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
134
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 41952
diff changeset
135 local_name = local_name or url[url.rindex('/') + 1 :]
41952
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
136
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
137 local_path = dest_path / local_name
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
138 download_to_path(url, local_path, entry['size'], entry['sha256'])
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
139
b83de9150c1c packaging: convert files to LF
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41921
diff changeset
140 return local_path, entry