ident

5.0 KB

1. #!/usr/bin/env python3
2. # tests/realdata-compare.py (ident mission 009, from gitoria's mission 002): compare two tests/realdata-baseline.mjs
3. # outputs. Masks only what is not the app's output:
4. # every file "at":<ms> (the seed's request time), "since":<ms> (a session's sign-in time), "who":<hex> (the connection), "site":"<file:l:c>" and Child@<line>:<col> (source positions of
5. # View nodes / mount keys: they move when lines move), ?v=<hash> and hl-shell-<hash> (hl:web's content hashes of the modules, the worker's cache name)
6. # 'w-' files (the writes) what is random or a clock per run: new mpackdb ids, rids / codes / pick ids / hashes (hex),
7. # API keys and secrets, invite tokens, new short ids, millisecond times, "YYYY-MM-DD HH:MM" stamps, six-digit codes;
8. # the stored state (w-store) is compared line by line, sorted after masking
9. # '-module-' files (the browser half) count as the same when their own code is the same with every source position
10. # masked and every imported name it uses is the same kind with the same body (moving a static to another
11. # file changes the module's tables, not the behaviour); `var x=` ↔ `const x=` is masked (the `let` step).
12. # python3 tests/realdata-compare.py <dirA> <dirB> [-v] → counts + the differing files; exit 1 on a real difference
13. import os, re, sys, difflib
14. a, b = sys.argv[1], sys.argv[2]; verbose = '-v' in sys.argv
15. masks = [(re.compile(r'\\"at\\":\d+'), r'\"at\":T'), (re.compile(r'"at":\d+'), '"at":T'),
16. (re.compile(r'\\"site\\":\\"[^"\\]*\\"'), r'\"site\":S'), (re.compile(r'"site":"[^"]*"'), '"site":S'),
17. (re.compile(r'\?v=[0-9a-f]+'), '?v=H'), (re.compile(r'hl-shell-[0-9a-f]+'), 'hl-shell-H'), (re.compile(r'([A-Z]\w*)@\d+:\d+'), r'\1@L:C'),
18. (re.compile(r'(\\?"since\\?":)\d+'), r'\1T'), (re.compile(r'(\\?"who\\?":\\?")[0-9a-f]+'), r'\1W')]
19. wmasks = [(re.compile(r'pk_[0-9a-f]{32}'), 'pk_KEY'), (re.compile(r'sk_[0-9a-f]{48}'), 'sk_SECRET'), (re.compile(r'\b0m[0-9a-z]{10}\b'), 'ID'), (re.compile(r'\b[0-9a-f]{64}\b'), 'H64'), (re.compile(r'\b[0-9a-f]{48}\b'), 'H48'),
20. (re.compile(r'\b[0-9a-f]{32}\b'), 'H32'),
21. (re.compile(r'\b1[789]\d{11}\b'), 'MS'), (re.compile(r'\d{4}-\d{2}-\d{2} \d{2}:\d{2}'), 'TIME'),
22. (re.compile(r'(?<![\d.])\d{6}(?![\d.])'), 'CODE'), (re.compile(r'(shortId\\?":\\?"|class="shortid">)[0-9a-z]{5}'), r'\1SHORT')]
23. POS = re.compile(r'[\w-]+\.hl:\d+:\d+|[A-Z]\w*@\d+:\d+|\b(?:var|const) (?=\w+=)')
24. def moduleParts(t):
25. t = re.sub(r'\?v=[0-9a-f]+', '?v=H', t)
26. kinds, own = {}, []
27. lines = t.split('\n'); i = 0
28. while i < len(lines):
29. l = lines[i]
30. m = re.match(r'\$_\d+\.(\w+)=(.*)$', l)
31. if m:
32. if re.match(r'\(\)=>_\d+\(', m.group(2)): kinds.setdefault(m.group(1), ('stub', ''))
33. else:
34. body = [m.group(2)]
35. while not lines[i].endswith('}));') and not (lines[i].endswith(';') and lines[i] is l and '=>{' not in l):
36. i += 1; body.append(lines[i])
37. kinds[m.group(1)] = ('fn', POS.sub('P', re.sub(r'\$_\d+', '$_N', '\n'.join(body))))
38. i += 1; continue
39. if l.startswith('import{') and '/__hl/hl-runtime.js' in l: i += 1; continue
40. if re.match(r'const \$_\d+=\{\};$', l) or re.match(r'const \w+=\$_\d+\.\w+;$', l) or re.match(r'const \w+=\(\)=>_\d+\(2,', l):
41. m2 = re.match(r'const (\w+)=\(\)=>_\d+\(2,', l)
42. if m2: kinds[m2.group(1)] = ('plugin', '')
43. i += 1; continue
44. own.append(POS.sub('P', re.sub(r'\$_\d+', '$_N', l))); i += 1
45. return kinds, '\n'.join(own)
47. def sameModule(x, y):
48. kx, ox = moduleParts(x); ky, oy = moduleParts(y)
49. if ox != oy: return False
50. words = set(re.findall(r'\b\w+\b', ox))
51. return all(kx.get(n) == ky.get(n) for n in words & (set(kx) | set(ky)))
53. def norm(t, w, store=False):
54. for r, s in masks: t = r.sub(s, t)
55. if w:
56. for r, s in wmasks: t = r.sub(s, t)
57. if store: t = '\n'.join(sorted(t.split('\n')))
58. return t
59. skip = {'0000-server-log'}
60. fa, fb = sorted(f for f in os.listdir(a) if f not in skip), sorted(f for f in os.listdir(b) if f not in skip)
61. if fa != fb: print('FILE LISTS DIFFER', sorted(set(fa) ^ set(fb))[:20]); sys.exit(1)
62. same = masked = modules = 0; bad = []
63. for f in fa:
64. x, y = open(os.path.join(a, f), errors='replace').read(), open(os.path.join(b, f), errors='replace').read()
65. if x == y: same += 1; continue
66. w = '-w-' in f; st = f.endswith('-w-store')
67. if norm(x, w, st) == norm(y, w, st): masked += 1; continue
68. if '-module-' in f and sameModule(x, y): modules += 1; continue
69. bad.append(f)
70. if verbose:
71. print('==== DIFF', f)
72. for l in list(difflib.unified_diff(norm(x, w, st).split('\n'), norm(y, w, st).split('\n'), lineterm='', n=0))[:12]: print(' ', l[:300])
73. reads = sum(1 for f in fa if '-w-' not in f)
74. print(f'{len(fa)} files ({reads} reads, {len(fa) - reads} writes): {same} byte-identical, {masked} identical after masking, {modules} browser modules the same (code, imported functions), {len(bad)} DIFFERENT')
75. for f in bad[:40]: print(' DIFF', f)
76. sys.exit(1 if bad else 0)