[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: apt archive verification tool

On 13/04/17 15:36, peter green wrote:
The census code might be useful as an illustration but I think it
would be better to write a new tool for the purpose.

If the apt repository is managed by reprepro, you can use the checkpool command.
The main reason I want this is checking our historical snapshots are intact.

Unfortunately while these were created by reprepro the reprepro databases were not saved as part of the snapshot. So using checkpool is not an option.

I'm currently hacking together a quick and dirty python script.

Well here it is.


1. The script only looks at sha1 hashes. If there is no sha1 hash for a file it won't be checked.
2. The script relies on having uncompressed Packages and Sources files available and listed in the Release files.
3. It only checks files referenced from the release files. There is no checking of the release files themselves.

import os
import sys
import hashlib
import gzip
from sortedcontainers import SortedDict

def addfiletoverify(filestoverify,filename,sha1,size):
	size = int(size)
	sha1andsize = (sha1,size)
	if filename in filestoverify:
		if (sha1andsize != filestoverify[filename]):
			print('error: same file with different hash/size old:'+repr(filestoverify[filename])+' new:'+repr(sha1andsize))
		filestoverify[filename] = sha1andsize

dists = os.listdir('dists/')
filestoverify = SortedDict() #sorted to hopefully get better locality on file accesses.
for dist in dists:
	f = open('dists/'+dist+'/Release','rb')
	insha1 = False;
	for line in f:
		if (line == b'SHA1:\n'):
			insha1 = True
		elif ((line[0] == 32) and insha1):
			linesplit = line.split()
			filename = b'dists/'+dist.encode('ascii')+b'/'+linesplit[2]
			#if filename in filestoverify:
			#	if files
			if filename.endswith(b'Packages'):
				print('found packages file: '+filename.decode('ascii'))
				pf = open(filename,'rb')
				filename = None
				size = None
				sha1 = None
				for line in pf:
					linesplit = line.split()
					if (len(linesplit) == 0):
						if (filename != None):
						filename = None
						size = None
						sha1 = None
					elif (linesplit[0] == b'Filename:'):
						filename = linesplit[1]
					elif (linesplit[0] == b'Size:'):
						size = linesplit[1]
					elif (linesplit[0] == b'SHA1:'):
						sha1 = linesplit[1]
			elif filename.endswith(b'Sources'):
				print('found sources file: '+filename.decode('ascii'))
				pf = open(filename,'rb')
				filesfound = [];
				directory = None
				insha1p = False;
				for line in pf:
					linesplit = line.split()
					if (len(linesplit) == 0):
						for ls in filesfound:
						filesfound = [];
						directory = None
						insha1p = False
					elif ((line[0] == 32) and insha1p):
					elif (linesplit[0] == b'Directory:'):
						insha1p = False
						directory = linesplit[1]
					elif (linesplit[0] == b'Checksums-Sha1:'):
						insha1p = True
						insha1p = False
			insha1 = False


#incomplete code to descend into dscs, seems this is
#not actually needed as files depended on by dscs
#are listed in the Sources file directly.
#filessofar = filestoverify.copy();
#for filename, sha1andsize in filessofar.items():
#	if filename.endswith(b'dsc'):
#	f = open(filename,'rb')
#	insha1 = False
#	for line in f:
#		if (line == b'Checksums-Sha1::\n'):
#			insha1 = True
#		elif ((line[0] == 32) and insha1):
#		else:
#			insha1 = False
#	f.close()

for filename, sha1andsize in filestoverify.items():
	sha1,size = sha1andsize;
	print('verifying '+filename.decode('ascii'))
	if b'../' in filename:
		print('fucked up filename')
	if not os.path.isfile(filename):
		if not os.path.isfile(filename+b'.gz'):
			print('missing file '+ filename.decode('ascii'))
			#sometimes reprepro seems to create only a .gz file but includes the non-gzipped file in the index
			f = gzip.open(filename+b'.gz','rb')
		f = open(filename,'rb')
	data = f.read();
	sha1hash = hashlib.sha1(data)
	sha1hashed = sha1hash.hexdigest().encode('ascii')
	if (sha1 != sha1hashed):
		print('hash mismatch');
	filesize = len(data)
	if (size != filesize):
		print('size mismatch');


Reply to: