Repository navigation
Expand file tree
/
Copy pathcompare_bytecode.py
More file actions
673 lines (599 loc) · 21.7 KB
/
Copy pathcompare_bytecode.py
File metadata and controls
673 lines (599 loc) · 21.7 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
#!/usr/bin/env python3
"""Compare bytecode between CPython and RustPython.
Compiles all Python files under Lib/ with both interpreters and reports
differences in the generated bytecode instructions.
Usage:
python scripts/compare_bytecode.py
python scripts/compare_bytecode.py --detail
python scripts/compare_bytecode.py --filter "asyncio/*.py"
python scripts/compare_bytecode.py --summary-json report.json
"""
import argparse
import fnmatch
import json
import os
import random
import subprocess
import sys
import tempfile
from collections import defaultdict
SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
PROJECT_ROOT = os.path.dirname(SCRIPT_DIR)
DIS_DUMP = os.path.join(SCRIPT_DIR, "dis_dump.py")
DEFAULT_REPORT = os.path.join(PROJECT_ROOT, "compare_bytecode.report")
DUMP_TIMEOUT = 600
def find_rustpython():
"""Locate the RustPython binary, allowing release builds only."""
if "RUSTPYTHON" in os.environ:
path = os.environ["RUSTPYTHON"]
normalized = os.path.normpath(path)
debug_fragment = os.path.join("target", "debug", "rustpython")
if normalized.endswith(debug_fragment):
raise ValueError(
"RUSTPYTHON must point to a release binary, not target/debug/rustpython"
)
return path
path = os.path.join(PROJECT_ROOT, "target", "release", "rustpython")
if os.path.isfile(path) and os.access(path, os.X_OK):
return path
return None
def collect_targets(lib_dir, pattern=None):
"""Collect Python files to compare, relative to lib_dir."""
targets = []
for root, dirs, files in os.walk(lib_dir):
dirs[:] = sorted(
d for d in dirs if d != "__pycache__" and not d.startswith(".")
)
for fname in sorted(files):
if not fname.endswith(".py"):
continue
fpath = os.path.join(root, fname)
relpath = os.path.relpath(fpath, lib_dir)
if pattern and not fnmatch.fnmatch(relpath, pattern):
continue
targets.append((relpath, fpath))
return targets
def _start_one(interpreter, targets, base_dir):
"""Start a single dis_dump.py subprocess."""
env = os.environ.copy()
if interpreter != sys.executable:
env["RUSTPYTHONPATH"] = base_dir
files_file = None
output_file = None
try:
files_file = tempfile.NamedTemporaryFile(
mode="w",
encoding="utf-8",
delete=False,
prefix="compare-bytecode-files-",
)
output_file = tempfile.NamedTemporaryFile(
mode="w",
encoding="utf-8",
delete=False,
prefix="compare-bytecode-output-",
)
for _, path in targets:
files_file.write(path)
files_file.write("\n")
files_file.close()
output_file.close()
cmd = [
interpreter,
DIS_DUMP,
"--base-dir",
base_dir,
"--files-from",
files_file.name,
"--output",
output_file.name,
"--progress",
"10",
]
proc = subprocess.Popen(
cmd,
stdout=subprocess.PIPE,
stderr=None, # inherit stderr so progress dots appear on terminal
env=env,
cwd=PROJECT_ROOT,
)
return {
"proc": proc,
"files_file": files_file.name,
"output_file": output_file.name,
"targets": targets,
"interpreter": interpreter,
"base_dir": base_dir,
}
except Exception:
for handle in (files_file, output_file):
if handle is None:
continue
try:
handle.close()
finally:
if os.path.exists(handle.name):
os.unlink(handle.name)
raise
def _load_dump_output(output_file):
try:
with open(output_file, encoding="utf-8") as f:
content = f.read().strip()
except OSError as e:
print(" Failed to read dump output: %s" % e, file=sys.stderr)
return None
if not content:
return {}
try:
return json.loads(content)
except json.JSONDecodeError as e:
print(" JSON parse error: %s" % e, file=sys.stderr)
return None
def _run_sync_dump(interpreter, targets, base_dir, timeout=DUMP_TIMEOUT):
job = _start_one(interpreter, targets, base_dir)
proc = job["proc"]
stdout = b""
timed_out = False
try:
stdout = proc.communicate(timeout=timeout)[0]
except subprocess.TimeoutExpired:
proc.kill()
proc.communicate()
print(f" Timeout ({timeout}s)", file=sys.stderr)
timed_out = True
try:
data = _load_dump_output(job["output_file"])
finally:
for path in (job["files_file"], job["output_file"]):
if os.path.exists(path):
os.unlink(path)
if timed_out:
return {}
if proc.returncode != 0:
print(" Warning: exited with code %d" % proc.returncode, file=sys.stderr)
stray = stdout.decode(errors="replace").strip()
if stray:
print(" Warning: unexpected stdout from dump helper", file=sys.stderr)
return data
def _rerun_missing_targets(interpreter, targets, base_dir):
recovered = {}
failed = []
empty = []
for target in targets:
relpath = target[0]
data = _run_sync_dump(interpreter, [target], base_dir)
if data is None:
failed.append(relpath)
recovered[relpath] = {
"status": "error",
"error": "dump helper failed while rerunning target",
}
elif data:
recovered.update(data)
else:
empty.append(relpath)
recovered[relpath] = {
"status": "error",
"error": "dump helper produced no data while rerunning target",
}
if failed:
print(
" Warning: rerun failed for %d file(s): %s"
% (len(failed), ", ".join(failed[:5])),
file=sys.stderr,
)
if empty:
print(
" Warning: rerun produced no data for %d file(s): %s"
% (len(empty), ", ".join(empty[:5])),
file=sys.stderr,
)
return recovered
def _finish_one(job, timeout=DUMP_TIMEOUT):
"""Wait for a single dis_dump.py process and return parsed JSON."""
proc = job["proc"]
expected = {relpath for relpath, _ in job["targets"]}
stdout = b""
try:
stdout = proc.communicate(timeout=timeout)[0]
except subprocess.TimeoutExpired:
proc.kill()
proc.communicate()
print(
f" Timeout ({timeout}s), retrying {len(job['targets'])} file(s) serially",
file=sys.stderr,
)
data = None
else:
data = _load_dump_output(job["output_file"])
finally:
for path in (job["files_file"], job["output_file"]):
if os.path.exists(path):
os.unlink(path)
if proc.returncode != 0:
print(" Warning: exited with code %d" % proc.returncode, file=sys.stderr)
stray = stdout.decode(errors="replace").strip()
if stray:
print(" Warning: unexpected stdout from dump helper", file=sys.stderr)
if data is None:
return _rerun_missing_targets(
job["interpreter"], job["targets"], job["base_dir"]
)
missing = [
target
for target in job["targets"]
if target[0] not in data and target[0] in expected
]
if missing:
print(
" Re-running %d missing file(s) serially" % len(missing),
file=sys.stderr,
)
data.update(
_rerun_missing_targets(job["interpreter"], missing, job["base_dir"])
)
return data
def start_dump(interpreter, targets, base_dir, num_workers=1):
"""Start dis_dump.py under the given interpreter, split across workers."""
if num_workers <= 1 or len(targets) <= num_workers:
return [_start_one(interpreter, targets, base_dir)]
chunks = [[] for _ in range(num_workers)]
for i, t in enumerate(targets):
chunks[i % num_workers].append(t)
return [_start_one(interpreter, chunk, base_dir) for chunk in chunks if chunk]
def finish_dump(procs):
"""Wait for all dis_dump.py processes and merge results."""
merged = {}
for job in procs:
merged.update(_finish_one(job))
return merged
def compare_insts(cp_insts, rp_insts):
"""Compare two instruction lists. Returns list of (index, cp, rp) diffs."""
diffs = []
for i in range(max(len(cp_insts), len(rp_insts))):
cp = cp_insts[i] if i < len(cp_insts) else None
rp = rp_insts[i] if i < len(rp_insts) else None
if cp != rp:
diffs.append((i, cp, rp))
return diffs
def compare_code(cp_code, rp_code, path=""):
"""Recursively compare code objects. Returns [(code_path, diffs)]."""
results = []
name = cp_code["name"]
full = (path + "/" + name) if path else name
diffs = compare_insts(cp_code.get("insts", []), rp_code.get("insts", []))
if diffs:
results.append((full, diffs))
cp_ch = cp_code.get("children", [])
rp_ch = rp_code.get("children", [])
cp_by_name = defaultdict(list)
rp_by_name = defaultdict(list)
for c in cp_ch:
cp_by_name[c["name"]].append(c)
for c in rp_ch:
rp_by_name[c["name"]].append(c)
all_names = list(dict.fromkeys(c["name"] for c in cp_ch))
for c in rp_ch:
if c["name"] not in cp_by_name:
all_names.append(c["name"])
for name in all_names:
cp_list = cp_by_name.get(name, [])
rp_list = rp_by_name.get(name, [])
for i in range(max(len(cp_list), len(rp_list))):
if i < len(cp_list) and i < len(rp_list):
results.extend(compare_code(cp_list[i], rp_list[i], full))
elif i < len(cp_list):
results.append((full + "/" + name, [(-1, "extra in CPython", None)]))
else:
results.append((full + "/" + name, [(-1, None, "extra in RustPython")]))
return results
def compare_code_summary(cp_code, rp_code):
"""Recursively compare code objects and return summary counts."""
diff_code_objects = 0
diff_instructions = compare_insts_count(
cp_code.get("insts", []), rp_code.get("insts", [])
)
if diff_instructions:
diff_code_objects += 1
cp_ch = cp_code.get("children", [])
rp_ch = rp_code.get("children", [])
cp_by_name = defaultdict(list)
rp_by_name = defaultdict(list)
for child in cp_ch:
cp_by_name[child["name"]].append(child)
for child in rp_ch:
rp_by_name[child["name"]].append(child)
all_names = list(dict.fromkeys(child["name"] for child in cp_ch))
for child in rp_ch:
if child["name"] not in cp_by_name:
all_names.append(child["name"])
for name in all_names:
cp_list = cp_by_name.get(name, [])
rp_list = rp_by_name.get(name, [])
for i in range(max(len(cp_list), len(rp_list))):
if i < len(cp_list) and i < len(rp_list):
child_objects, child_insts = compare_code_summary(
cp_list[i], rp_list[i]
)
diff_code_objects += child_objects
diff_instructions += child_insts
else:
diff_code_objects += 1
diff_instructions += 1
return diff_code_objects, diff_instructions
def compare_insts_count(cp_insts, rp_insts):
"""Count mismatched instruction slots without storing the full diff."""
diff_count = 0
for i in range(max(len(cp_insts), len(rp_insts))):
cp = cp_insts[i] if i < len(cp_insts) else None
rp = rp_insts[i] if i < len(rp_insts) else None
if cp != rp:
diff_count += 1
return diff_count
def main():
parser = argparse.ArgumentParser(description="Compare CPython/RustPython bytecode")
parser.add_argument(
"--detail", action="store_true", help="Show per-file instruction differences"
)
parser.add_argument("--filter", default=None, help="Glob pattern to filter files")
parser.add_argument(
"--max-diffs",
type=int,
default=5,
help="Max diffs shown per code object (default: 5)",
)
parser.add_argument(
"--summary-json", default=None, help="Write summary as JSON to file"
)
parser.add_argument(
"--sample",
type=int,
default=None,
help="Compare a random sample of N matching files",
)
parser.add_argument(
"--seed",
type=int,
default=None,
help="Random seed used with --sample",
)
parser.add_argument(
"--list-limit",
type=int,
default=10,
help="Max differing files to print in non-detail mode (default: 10)",
)
parser.add_argument(
"--lib-dir",
default=os.path.join(PROJECT_ROOT, "Lib"),
help="Library directory to compare",
)
parser.add_argument(
"-j",
"--jobs",
type=int,
default=None,
help="Number of parallel workers per interpreter (default: cpu_count)",
)
parser.add_argument(
"-o",
"--output",
default=DEFAULT_REPORT,
help="Report output file (default: compare_bytecode.report)",
)
args = parser.parse_args()
try:
rp_bin = find_rustpython()
except ValueError as exc:
print("Error: %s" % exc, file=sys.stderr)
sys.exit(1)
if not rp_bin:
print("Error: RustPython binary not found.", file=sys.stderr)
print(" Build with: cargo build --release", file=sys.stderr)
print(" Or set RUSTPYTHON=/path/to/binary", file=sys.stderr)
sys.exit(1)
if not os.path.isfile(DIS_DUMP):
print("Error: disassembler helper not found: %s" % DIS_DUMP, file=sys.stderr)
print(
" Expected scripts/dis_dump.py from origin/bytecode-parity",
file=sys.stderr,
)
sys.exit(1)
targets = collect_targets(args.lib_dir, args.filter)
sample_seed = None
if args.sample is not None:
if args.sample < 1:
print("Error: --sample must be >= 1", file=sys.stderr)
sys.exit(1)
sample_seed = (
args.seed
if args.seed is not None
else random.SystemRandom().randrange(2**32)
)
rng = random.Random(sample_seed)
sample_size = min(args.sample, len(targets))
targets = sorted(rng.sample(targets, sample_size), key=lambda item: item[0])
if not targets:
print("Error: no Python files matched", file=sys.stderr)
sys.exit(1)
report_path = args.output
log = lambda *a, **kw: print(*a, file=sys.stderr, **kw)
log("Report: %s" % os.path.relpath(report_path, PROJECT_ROOT))
log("Targets: %d file(s)" % len(targets))
num_workers = args.jobs if args.jobs else os.cpu_count() or 4
log("Workers: %d per interpreter" % num_workers)
sys.stderr.write("Dumping bytecode ")
sys.stderr.flush()
cp_procs = start_dump(sys.executable, targets, args.lib_dir, num_workers)
rp_procs = start_dump(rp_bin, targets, args.lib_dir, num_workers)
cp_data = finish_dump(cp_procs)
rp_data = finish_dump(rp_procs)
sys.stderr.write("\n")
if not cp_data:
log("Error: CPython dump produced no data")
sys.exit(1)
# Phase 2: Compare
all_files = sorted(set(cp_data) | set(rp_data))
match = 0
differ = 0
cp_err = 0
rp_err = 0
both_err = 0
rp_miss = 0
diff_files = []
rp_error_files = []
diff_summaries = []
need_detailed_diffs = args.detail
for fp in all_files:
cp = cp_data.get(fp)
rp = rp_data.get(fp)
if rp is None:
rp_miss += 1
continue
cp_ok = cp and cp.get("status") == "ok"
rp_ok = rp and rp.get("status") == "ok"
if not cp_ok and not rp_ok:
both_err += 1
elif not cp_ok:
cp_err += 1
elif not rp_ok:
rp_err += 1
rp_error_files.append((fp, rp.get("error", "?")))
else:
if need_detailed_diffs:
code_diffs = compare_code(cp["code"], rp["code"])
if code_diffs:
differ += 1
diff_files.append((fp, code_diffs))
else:
match += 1
else:
diff_code_objects, diff_instructions = compare_code_summary(
cp["code"], rp["code"]
)
if diff_code_objects:
differ += 1
diff_summaries.append(
{
"path": fp,
"diff_code_objects": diff_code_objects,
"diff_instructions": diff_instructions,
}
)
else:
match += 1
total = match + differ + cp_err + rp_err + both_err + rp_miss
def pct(n):
return "%.1f%%" % (100.0 * n / total) if total else "0%"
# Phase 3: Write report to file
with open(report_path, "w") as out:
p = lambda *a: print(*a, file=out)
p("CPython: %s (%s)" % (sys.executable, sys.version.split()[0]))
p("RustPython: %s" % rp_bin)
p("Lib: %s" % args.lib_dir)
if sample_seed is not None:
p("Sample: %s file(s), seed=%s" % (len(targets), sample_seed))
p()
p("=" * 60)
p(" Bytecode Comparison Report")
p("=" * 60)
p()
p(" Total files: %6d" % total)
p(" Match: %6d (%s)" % (match, pct(match)))
p(" Differ: %6d (%s)" % (differ, pct(differ)))
p(" RustPython error: %6d (%s)" % (rp_err, pct(rp_err)))
p(" CPython error: %6d (%s)" % (cp_err, pct(cp_err)))
p(" Both error: %6d (%s)" % (both_err, pct(both_err)))
if rp_miss:
p(" RustPython missing: %6d (%s)" % (rp_miss, pct(rp_miss)))
p()
if args.detail:
if rp_error_files:
p("-" * 60)
p(" RustPython Compile Errors")
p("-" * 60)
for fp, err in rp_error_files[:50]:
p(" %s: %s" % (fp, err))
if len(rp_error_files) > 50:
p(" ... and %d more" % (len(rp_error_files) - 50))
p()
if diff_files:
p("-" * 60)
p(" Bytecode Differences")
p("-" * 60)
for fp, code_diffs in diff_files:
p()
p(" %s:" % fp)
for code_path, diffs in code_diffs:
shown = min(len(diffs), args.max_diffs)
p(" %s: %d difference(s)" % (code_path, len(diffs)))
for idx, cp_inst, rp_inst in diffs[:shown]:
if idx == -1:
p(" %s" % (cp_inst or rp_inst))
else:
p(" [%3d] CPython: %s" % (idx, cp_inst))
p(" RustPython: %s" % rp_inst)
if len(diffs) > shown:
p(" ... and %d more" % (len(diffs) - shown))
p()
else:
list_limit = 0 if args.summary_json else max(args.list_limit, 0)
if diff_summaries and list_limit:
shown = min(list_limit, len(diff_summaries))
total = len(diff_summaries)
p(f"Top differing files ({shown} shown of {total}):")
top = sorted(
diff_summaries,
key=lambda item: (
item["diff_instructions"],
item["diff_code_objects"],
item["path"],
),
reverse=True,
)[:list_limit]
for item in top:
p(
" %s (%d code objects, %d instruction diffs)"
% (
item["path"],
item["diff_code_objects"],
item["diff_instructions"],
)
)
p()
p("Use --detail to see specific instruction differences.")
p()
# Summary JSON output
if args.summary_json:
summary = {
"total": total,
"sample": args.sample,
"sample_seed": sample_seed,
"match": match,
"differ": differ,
"rp_error": rp_err,
"cp_error": cp_err,
"both_error": both_err,
"rp_missing": rp_miss,
"match_pct": round(100.0 * match / total, 2) if total else 0,
"diff_files": [fp for fp, _ in diff_files]
if need_detailed_diffs
else [item["path"] for item in diff_summaries],
"top_diff_files": sorted(
diff_summaries,
key=lambda item: (
item["diff_instructions"],
item["diff_code_objects"],
item["path"],
),
reverse=True,
)[: min(20, len(diff_summaries))],
"rp_error_files": [fp for fp, _ in rp_error_files],
}
with open(args.summary_json, "w") as f:
json.dump(summary, f, indent=2)
log("Summary JSON: %s" % args.summary_json)
log("Done: %d match, %d differ, %d errors" % (match, differ, rp_err))
sys.exit(0 if differ == 0 and rp_err == 0 else 1)
if __name__ == "__main__":
main()