particle-os/debian-forge
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

tools/tree-diff: Use hash for content diffs

Browse source 
We need to know the exact difference of modified files in both trees.
Outputting the whole files into a diff might make a huge diff file,
therefore only their hashes are written.
This commit is contained in:
Ondřej Budai 2019-09-26 14:56:49 +02:00 • committed by Lars Karlitski
parent 9fd9270c53
commit fd2a20d247
1 changed files with 19 additions and 13 deletions
Download patch file Download diff file Expand all files Collapse all files

32
tree-diff
View file

@ -1,10 +1,22 @@
#!/usr/bin/env python3
import argparse
import hashlib
import json
import os
def hash_file(fd):
BLOCK_SIZE = 4096
hasher = hashlib.sha256()
buf = os.read(fd, BLOCK_SIZE)
while len(buf) > 0:
hasher.update(buf)
buf = os.read(fd, BLOCK_SIZE)
return f"sha256:{hasher.hexdigest()}"
def stat_diff(stat1, stat2, path, differences):
if stat1.st_mode != stat2.st_mode:
props = differences.setdefault(path, {})
@ -32,11 +44,7 @@ def selinux_diff(path1, path2, path, differences):
return True
def content_diff(name, dir_fd1, dir_fd2, size1, size2, path, differences):
if size1 != size2:
props = differences.setdefault(path, {})
props["content"] = "different"
return
def content_diff(name, dir_fd1, dir_fd2, path, differences):
try:
fd1 = os.open(name, flags=os.O_RDONLY, dir_fd=dir_fd1)
except OSError:
@ -47,12 +55,12 @@ def content_diff(name, dir_fd1, dir_fd2, size1, size2, path, differences):
os.close(fd1)
return
try:
for (byte_block1, byte_block2) in zip(iter(lambda f=fd1: os.read(f, 4096), b""),
iter(lambda f=fd2: os.read(f, 4096), b"")):
if byte_block1 != byte_block2:
props = differences.setdefault(path, {})
props["content"] = "different"
break
hash1 = hash_file(fd1)
hash2 = hash_file(fd2)
if hash1 != hash2:
props = differences.setdefault(path, {})
props["content"] = [hash1, hash2]
finally:
os.close(fd1)
os.close(fd2)
@ -106,8 +114,6 @@ def diff_aux(dir_fd1, dir_fd2, path, report):
content_diff(dirent.name,
dir_fd1,
dir_fd2,
stat1.st_size,
stat2.st_size,
os.path.join(path, dirent.name),
report["differences"])
elif dirent.is_dir(follow_symlinks=False):