#!/usr/bin/env python3 import http.client import json import string import sys import xml.etree.ElementTree if len(sys.argv[1:]) != 3: print('Usage: ia-verify-file ITEM FILENAME SHA1', file = sys.stderr) print('Verifies that FILENAME is present in ITEM (on all servers) and has hash SHA1', file = sys.stderr) sys.exit(1) item, filename, sha1 = sys.argv[1:] if len(item) > 100 or item[0] not in (string.ascii_letters + string.digits) or item[1:].strip(string.ascii_letters + string.digits + '_-.') != '': print('Invalid identifier', file = sys.stderr) sys.exit(1) # Fetch https://archive.org/metadata/ITEM to get the two servers that store the item conn = http.client.HTTPSConnection('archive.org', timeout = 30) conn.request('GET', f'/metadata/{item}') r = conn.getresponse() if r.status != 200: print(f'Error fetching item metadata: {r.status} {r.reason}', file = sys.stderr) sys.exit(1) o = json.loads(r.read()) if not o: print(f'Empty item metadata returned (item might not exist)', file = sys.stderr) sys.exit(1) if 'workable_servers' not in o or 'dir' not in o: print(f'Malformed metadata returned: {o!r}', file = sys.stderr) sys.exit(1) if len(o['workable_servers']) < 2: print(f'Not enough servers: {o["workable_servers"]!r}', file = sys.stderr) sys.exit(1) # Fetch _files.xml from each server and compare hash for server in o['workable_servers']: conn = http.client.HTTPSConnection(server, timeout = 30) conn.request('GET', f'{o["dir"]}/{item}_files.xml') r = conn.getresponse() if r.status != 200: print(f'Error fetching item files from {server}: {r.status} {r.reason}', file = sys.stderr) sys.exit(1) root = xml.etree.ElementTree.fromstring(r.read().decode('utf-8')) if root.tag != 'files' or not all(c.tag == 'file' for c in root): print(f'Invalid XML received from {server}', file = sys.stderr) sys.exit(1) for file in root: if file.attrib['name'] == filename: for c in file: if c.tag == 'sha1': if c.text != sha1: print(f'SHA-1 mismatch on {server}: expected {sha1}, found {c.text}', file = sys.stderr) sys.exit(1) break else: print(f'No SHA-1 found on {server}', file = sys.stderr) sys.exit(1) break else: print(f'File not found on {server}', file = sys.stderr) sys.exit(1)