-
Notifications
You must be signed in to change notification settings - Fork 1
/
chunksum.py
185 lines (140 loc) · 5.08 KB
/
chunksum.py
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
"""
Hash files by chunks parallelly
Aaron Tsang
MIT License
"""
from __future__ import annotations
import asyncio
import concurrent.futures
import dataclasses
import hashlib
import logging
import os.path
import re
from argparse import ArgumentParser
from dataclasses import dataclass
from typing import Awaitable, Dict, List, Tuple, Optional
_logger = logging.getLogger(__name__)
def parse_binary_number(value):
match = re.match(r'^(?P<size>[1-9][0-9]*)(?:(?P<unit>[kmgtpezy])b?)?$', value, re.IGNORECASE)
if not match:
raise ValueError(value)
size = int(match.group('size'))
multiplier = {
'k': 1024,
'm': 1024 ** 2,
'g': 1024 ** 3,
't': 1024 ** 4,
'p': 1024 ** 5,
'e': 1024 ** 6,
'z': 1024 ** 7,
'y': 1024 ** 8,
}.get(match.group('unit').lower(), 1)
assert isinstance(multiplier, int) and multiplier >= 1
return size * multiplier
def parse():
p = ArgumentParser()
p.add_argument('--verbose', '-v', action='count', default=0,
help='Increase the verbose level')
p.add_argument('--threads', type=int, default=4)
p.add_argument('--algorithm', type=str, default='sha1')
p.add_argument('--chunksize', type=parse_binary_number, default='4G')
p.add_argument('inputs', nargs='+')
ns = p.parse_args()
return ns
@dataclass
class FileEntry:
path: str
size: int
chunksize: int
nchunks: int = dataclasses.field(init=False)
def __post_init__(self):
q, r = divmod(self.size, self.chunksize)
self.nchunks = q + (1 if r else 0)
@dataclass
class FileHashes:
path: str
hashes: List[Optional[bytes]]
def compute_hash(algo, entry: FileEntry, chunk_id: int):
constructor = getattr(hashlib, algo, None)
if constructor is None or True:
md = hashlib.new(algo)
else:
md = constructor()
offset, length = chunk_range(entry.size, entry.chunksize, chunk_id)
_logger.debug('%s chunk_id=%d, offset=%d, length=%d', entry.path, chunk_id, offset, length)
with open(entry.path, 'rb', buffering=0) as handle:
buf = bytearray(1024 * 1024 * 32)
handle.seek(offset)
count = 0
while count < length:
nread = handle.readinto(buf)
if not nread:
raise ValueError(nread)
md.update(buf[:nread])
count += nread
# import mmap
# mm = mmap.mmap(handle.fileno(), length, prot=mmap.PROT_READ, offset=offset)
# md.update(mm[:])
# del mm
del handle
return entry.path, chunk_id, md.digest()
def chunk_range(size, chunksize, chunk_id):
assert size >= 0
assert chunksize >= 1
assert chunk_id >= 0
offset = chunk_id * chunksize
assert offset < size
length = chunksize
if offset + length > size:
length = size - offset
return offset, length
async def main():
ns = parse()
assert isinstance(ns.threads, int) and ns.threads >= 1
logging.root.handlers[:] = ()
if ns.verbose >= 2:
logging.basicConfig(level=logging.DEBUG)
elif ns.verbose >= 1:
logging.basicConfig(level=logging.INFO)
else:
logging.basicConfig()
_logger.debug('%r', ns)
chunksize = ns.chunksize
assert chunksize >= 1
assert chunksize % (64 * 1024) == 0
if ns.algorithm not in hashlib.algorithms_available:
raise ValueError(f'Algorithm {ns.algorithm} is not supported')
entries: Dict[str, FileEntry] = {s: FileEntry(s, os.path.getsize(s), ns.chunksize) for s in ns.inputs}
hashes: Dict[str, FileHashes] = {s: FileHashes(s, [None] * entries[s].nchunks) for s in ns.inputs}
loop = asyncio.get_running_loop()
tasks = []
# Seem that ThreadPoolExecutor is good enough
# ProcessPoolExecutor probably bounded by IO
with concurrent.futures.ThreadPoolExecutor(ns.threads) as pool:
def generator():
"""Schedule tasks by hashing different file concurrently"""
for path_, entry_ in entries.items():
for chunk_id_ in range(entry_.nchunks):
yield path_, chunk_id_
for path, chunk_id in sorted(generator(), key=lambda s: s[::-1]):
entry = entries.get(path)
tasks.append(loop.run_in_executor(
pool, compute_hash,
ns.algorithm, entry, chunk_id,
))
#
# for path, entry in entries.items():
# for chunk_id in range(entry.nchunks):
# tasks.append(loop.run_in_executor(
# pool, compute_hash,
# ns.algorithm, entry, chunk_id))
for future in asyncio.as_completed(tasks): # type: Awaitable[Tuple[str, int, bytes]]
path, chunk_id, h = await future
hashes[path].hashes[chunk_id] = h
entry = entries[path]
offset, length = chunk_range(entry.size, entry.chunksize, chunk_id)
_logger.info('hash(%s#%d) = %s', path, chunk_id, h.hex())
print(f'{h.hex()} {path}#{chunk_id:d} 0x{length:x}+{offset:x}', flush=True)
if __name__ == '__main__':
asyncio.run(main())