test: check pinned source manifests against their own commits

Historical-evidence pins (dfg_re100 / jet_circle_cv / refined_dfg manifests)
now compare each recorded digest with the file at the recorded commit, read
with git, instead of with today's working tree. Missing history (no git,
shallow clone, installed copy) raises HistoryUnavailable and the test is
reported as skipped with that reason, never passed. Receipt/plan logic tests
take the selected manifest directly, so their assertions run again; the
pinned modes' mismatch rejections stay as negative controls.

V1 (run_jet_circle_cv) keeps its pinned mode and gains --current-checkout:
it records HEAD, the src tree and per-file hashes labelled as current
checkout, never as the frozen qualification, and requires a clean tree.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Claude
2026-10-20 18:15:50 +00:00
co-authored by Claude Opus 5.5
parent 750e8a483e
commit 941eb28b79
6 changed files with 161 additions and 26 deletions
+41 -7
View File
@@ -3,6 +3,7 @@
import argparse
import hashlib
import subprocess
import json
import os
from pathlib import Path
@@ -220,6 +221,30 @@ def verify(repo):
return manifest
def verify_historical(repo):
"""The pinned manifest describes the PIN tree (read with git, not the working tree)."""
from source_provenance import historical_sha256
manifest = json.loads((HERE / "data/jet_circle_cv/source_manifest.json").read_text())
if manifest["revision"] != PIN:
raise RuntimeError("Wrong pinned revision")
if historical_sha256(repo, PIN, manifest["files"]) != manifest["files"]:
raise RuntimeError("Historical manifest does not match its pinned revision")
return manifest
def current_checkout(repo):
"""Provenance for a run on the current checkout; never labelled as the pinned source."""
files = subprocess.check_output(["git", "-C", str(repo), "ls-files", "src"],
text=True).splitlines()
return dict(source_mode="current_checkout",
source_capture="current checkout, not the frozen pinned qualification",
source_revision=checkout_provenance(repo)["checkout_revision"],
src_tree=subprocess.check_output(["git", "-C", str(repo), "rev-parse",
"HEAD:src"], text=True).strip(),
src_sha256={f: hashlib.sha256((repo / f).read_bytes()).hexdigest()
for f in files})
def run_case(c, ctx, output, provenance, continuation=False):
from CelerisLab.body.geometry.circle import CircleGeometry
from CelerisLab.body.objects import SimObject
@@ -530,8 +555,7 @@ def run_case(c, ctx, output, provenance, continuation=False):
recorder.flush()
payload = dict(
case=c,
source_revision=PIN,
**provenance,
**{"source_revision": PIN, **provenance},
status="PILOT_DIAGNOSTIC_ONLY",
windows=rows,
physics_validation_pass=False,
@@ -571,6 +595,9 @@ def main():
help="Fixed measurement campaign: Re100, scale1/2, 3200 reference steps",
)
p.add_argument("--plan", action="store_true")
p.add_argument("--current-checkout", action="store_true",
help="Run on the current clean checkout and record its provenance "
"instead of requiring the frozen pinned source")
a = p.parse_args()
if a.steps <= 0 or a.steps > 3200:
p.error("steps must be 1..3200")
@@ -595,7 +622,8 @@ def main():
)
plan = dict(
case=c,
required_revision=PIN,
required_revision=None if a.current_checkout else PIN,
source_mode="current_checkout" if a.current_checkout else "pinned",
required_free_device_mib=1024,
estimated_peak_device_mib=128 * a.scale**2,
estimated_output_mib=64 if a.continuation else 2,
@@ -611,10 +639,16 @@ def main():
):
p.error("Root must select one verified idle GPU via CUDA_VISIBLE_DEVICES")
repo = a.repo.resolve()
source = verify(repo)
provenance = dict(
source_experiment=source["experiment"], **checkout_provenance(repo)
)
if a.current_checkout:
provenance = dict(**checkout_provenance(repo), **current_checkout(repo))
if provenance["checkout_dirty"]:
raise RuntimeError("Current-checkout runs require a clean checkout")
else:
source = verify(repo)
provenance = dict(
source_experiment=source["experiment"], source_mode="pinned",
**checkout_provenance(repo)
)
sys.path.insert(0, str(repo / "src"))
import CelerisLab
+35
View File
@@ -13,3 +13,38 @@ def checkout_provenance(repo):
text=True,
)
return dict(checkout_revision=revision, checkout_dirty=bool(dirty))
class HistoryUnavailable(RuntimeError):
"""The clone cannot show the recorded commit (no git, shallow clone, installed copy)."""
def historical_sha256(repo, commit, paths):
"""SHA-256 of each path as stored in `commit`, read with git, never from the working tree.
Raises HistoryUnavailable when git or the commit object is missing, so callers can
report an unchecked integrity record explicitly instead of passing or failing silently.
"""
import hashlib
git = ["git", "-C", str(repo)]
try:
subprocess.run([*git, "cat-file", "-e", f"{commit}^{{commit}}"], check=True,
capture_output=True)
except (OSError, subprocess.CalledProcessError) as error:
raise HistoryUnavailable(f"commit {commit} is not available in {repo}") from error
digests = {}
for path in paths:
try:
blob = subprocess.run([*git, "show", f"{commit}:{path}"], check=True,
capture_output=True).stdout
except subprocess.CalledProcessError as error:
raise ValueError(f"{path} is not part of {commit}") from error
digests[path] = hashlib.sha256(blob).hexdigest()
return digests
def historical_tree(repo, commit, prefix="src"):
"""Tracked file names under `prefix` at `commit` (HistoryUnavailable if missing)."""
historical_sha256(repo, commit, [])
return subprocess.check_output(["git", "-C", str(repo), "ls-tree", "-r", "--name-only",
commit, "--", prefix], text=True).splitlines()