notes

4.7 KB

1. #!/usr/bin/env python3
2. # tests/realdata-compare.py (notes mission 002, copied from gitoria mission 002): compare two tests/realdata-baseline.mjs outputs. Masks only what is
3. # not the app's output:
4. # every file "at":<ms> (the seed's request time), "site":"<file:l:c>" and Child@<line>:<col> (source positions of
5. # View nodes / mount keys: they move when lines move), ?v=<hash> (hl:web's content hashes of the modules),
6. # state=<nonce> in the tickets connect redirect (random per click)
7. # 'w-' files (the writes) the new mpackdb ids, the version stamps (`edited`, ms), the random login tags
8. # '-module-' files (the browser half: component modules, runtime, client) — a module is the component's own code + tables
9. # of the statics it imports (a real client-callable function, or a refusing server-only stub naming its
10. # file). Moving a static to another file changes those tables and the source positions, not the behaviour, so
11. # a module counts as the same when (1) its own code is the same with every source position masked and (2) every
12. # imported name its own code uses is the same kind (function / stub), a function with the same body. A
13. # declaration's keyword is masked too (`var x=` ↔ `const x=`): Hybriel compiles `let x` to `var` and a plain
14. # declaration to `const` — exactly what the `let` cleanup changes.
15. # python3 tests/realdata-compare.py <dirA> <dirB> [-v] → counts + the differing files; exit 1 on a real difference
16. import os, re, sys, difflib
17. a, b = sys.argv[1], sys.argv[2]; verbose = '-v' in sys.argv
18. masks = [(re.compile(r'\\"at\\":\d+'), r'\"at\":T'), (re.compile(r'"at":\d+'), '"at":T'),
19. (re.compile(r'\\"site\\":\\"[^"\\]*\\"'), r'\"site\":S'), (re.compile(r'"site":"[^"]*"'), '"site":S'),
20. (re.compile(r'\?v=[0-9a-f]+'), '?v=H'), (re.compile(r'([A-Z]\w*)@\d+:\d+'), r'\1@L:C'),
21. (re.compile(r'state%3D[0-9a-f]+|state=[0-9a-f]+'), 'state=NONCE')]
22. wmasks = [(re.compile(r'\b0m[0-9a-z]{10}\b'), 'ID'), (re.compile(r'\b1[6-9]\d{11}\b'), 'MS'), (re.compile(r'\b[0-9a-f]{32}\b'), 'TAG')]
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+)=(.*)$', l)
31. if m:
32. if m.group(2).startswith('()=>_55('): kinds.setdefault(m.group(1), ('stub', ''))
33. else:
34. body = [m.group(2)]
35. while not lines[i].endswith('}));') and not (lines[i].endswith(';') and lines[i] is l and '=>{' not in l):
36. i += 1; body.append(lines[i])
37. kinds[m.group(1)] = ('fn', POS.sub('P', re.sub(r'\$_\d+', '$_N', '\n'.join(body))))
38. i += 1; continue
39. if l.startswith('import{') and '/__hl/hl-runtime.js' in l: i += 1; continue # the runtime helpers the tables use too
40. if re.match(r'const \$_\d+=\{\};$', l) or re.match(r'const \w+=\$_\d+\.\w+;$', l) or re.match(r'const \w+=\(\)=>_55\(2,', l):
41. m2 = re.match(r'const (\w+)=\(\)=>_55\(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):
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. return t
58. fa, fb = sorted(f for f in os.listdir(a) if f != '0000-server-log'), sorted(f for f in os.listdir(b) if f != '0000-server-log')
59. if fa != fb: print('FILE LISTS DIFFER', sorted(set(fa) ^ set(fb))[:20]); sys.exit(1)
60. same = masked = modules = 0; bad = []
61. for f in fa:
62. x, y = open(os.path.join(a, f), errors='replace').read(), open(os.path.join(b, f), errors='replace').read()
63. if x == y: same += 1; continue
64. w = '-w-' in f
65. if norm(x, w) == norm(y, w): masked += 1; continue
66. if '-module-' in f and sameModule(x, y): modules += 1; continue
67. bad.append(f)
68. if verbose:
69. print('==== DIFF', f)
70. for l in list(difflib.unified_diff(norm(x, w).split('\n'), norm(y, w).split('\n'), lineterm='', n=0))[:12]: print(' ', l[:300])
71. reads = sum(1 for f in fa if '-w-' not in f)
72. 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')
73. for f in bad[:40]: print(' DIFF', f)
74. sys.exit(1 if bad else 0)