tickets

1.6 KB

1. #!/usr/bin/env python3
2. # tests/realdata-compare-writes.py (mission 010): compare the WRITE part ('w-' files) of two realdata-baseline.mjs
3. # runs. Besides realdata-compare.py's masks it masks
4. # what a write makes new each run: mpackdb ids that the read part of the run never showed, epoch ms of today, Vienna
5. # stamps "YYYY-MM-DD HH:MM" of today, the API token. Each new id is numbered in order of appearance, so the SAME
6. # record must stand at the same places in both runs.
7. import os, re, sys, time
8. a, b = sys.argv[1], sys.argv[2]
9. src = open(os.path.join(os.path.dirname(os.path.abspath(__file__)), 'realdata-compare.py')).read()
10. exec(src[src.index('masks = '):src.index('fa, fb')])
11. known = set()
12. for f in os.listdir(a):
13. if '-w-' not in f: known |= set(re.findall(r'\b0m[0-9a-z]{10}\b', open(os.path.join(a, f)).read()))
14. today = time.strftime('%Y-%m-%d')
15. def wnorm(t, ids):
16. t = norm(t)
17. t = re.sub(r'tkt_[0-9a-f]{48}', 'TOKEN', t)
18. t = re.sub(r'\b17\d{11}\b', 'MS', t)
19. t = re.sub(today + r' \d\d:\d\d', 'STAMP', t)
20. t = re.sub(r'\b0m[0-9a-z]{10}\b', lambda m: m.group(0) if m.group(0) in known else ids.setdefault(m.group(0), 'NEW%d' % len(ids)), t)
21. return t
22. fs = sorted(f for f in os.listdir(a) if '-w-' in f)
23. assert fs == sorted(f for f in os.listdir(b) if '-w-' in f), 'file lists differ'
24. ia, ib = {}, {}; bad = []
25. for f in fs:
26. x, y = wnorm(open(os.path.join(a, f)).read(), ia), wnorm(open(os.path.join(b, f)).read(), ib)
27. if x != y: bad.append(f)
28. print(f'{len(fs)} write files: {len(fs) - len(bad)} identical after masking, {len(bad)} DIFFERENT; new ids {len(ia)} / {len(ib)}')
29. for f in bad[:20]: print(' DIFF', f)
30. sys.exit(1 if bad else 0)