ledger: fix census snapshot to a8f8c39 (2253 commits); pin check

wizard-work/ledger/2026-10-30-git-author-census.md

@@ -1,7 +1,7 @@
 # Git author census (2026-10-30)
 
-Source: `git log` of the troupe worktree (`d9e4a66`, 2253 commits
-reachable from HEAD). Extracted 2026-10-30.
+Snapshot: `git log` of the troupe worktree at `a8f8c39` (2253 commits
+reachable from HEAD before this census was committed). Extracted 2026-10-30.
 
 Identity key is the author name plus email in `Name <email>` form as
 recorded by git; no canonicalization (e.g. `Timmy` vs `timmy`, or a

wizard-work/ledger/check_git_author_census.py

@@ -1,9 +1,14 @@
 #!/usr/bin/env python3
-"""Git author census: extract author identities + commit counts from the worktree
-git history, write a table, and (as 'check') verify the artifact against a fresh
-recompile from `git log`. Pure extraction; no fabricated numbers.
+"""Git author census: extract author identities + commit counts from a snapshot of the
+worktree git history, write a table, and (as 'check') verify the artifact against a fresh
+recompile from `git log` of the pre-census snapshot (the commit the artifact represents).
+
+The census is a report ON the repo, committed after it. To keep the check honest (not a
+self-referential tautology), the artifact is verified against the git snapshot it describes
+(HEAD^), not against HEAD which includes the census commit itself.
 """
 import csv
+import os
 import subprocess
 import sys
 
@@ -12,19 +17,28 @@ LEDGER_DIR = REPO + "/wizard-work/ledger"
 CSV_NAME = "2026-10-30-git-author-census.csv"
 MD_NAME = "2026-10-30-git-author-census.md"
 ASN_DATE = "2026-10-30"
+# Snapshot the census artifact describes: the pre-census tip of the worktree, i.e. the
+# commit the artifact was generated from (so the census commit's own author is not
+# counted in itself). Pinned to a fixed hash so the check stays stable as the repo
+# keeps evolving.
+SNAPSHOT = "a8f8c39"
 
 
-def git_author_census():
-    """Recompute from git; returns {author_identity: count} with identity = 'Name <email>'."""
+def git_log(base):
     proc = subprocess.run(
-        ["git", "log", "--format=%an|%ae"],
+        ["git", "log", base, "--format=%an|%ae"],
         cwd=REPO,
         check=True,
         capture_output=True,
         text=True,
     )
+    return proc.stdout
+
+
+def git_author_census(base):
+    """Recompute from git; {author_identity: count}, identity = 'Name <email>'."""
     counts = {}
-    for raw in proc.stdout.splitlines():
+    for raw in git_log(base).splitlines():
         if not raw:
             continue
         name, _, email = raw.partition("|")
@@ -36,8 +50,7 @@ def git_author_census():
 def write_artifacts(counts):
     total = sum(counts.values())
     rows = sorted(counts.items(), key=lambda kv: (-kv[1], kv[0]))
-    os_dir = __import__("os")
-    os_dir.makedirs(LEDGER_DIR, exist_ok=True)
+    os.makedirs(LEDGER_DIR, exist_ok=True)
 
     csv_path = LEDGER_DIR + "/" + CSV_NAME
     with open(csv_path, "w", newline="") as f:
@@ -49,8 +62,10 @@ def write_artifacts(counts):
     md_path = LEDGER_DIR + "/" + MD_NAME
     with open(md_path, "w") as f:
         f.write("# Git author census ({date})\n\n".format(date=ASN_DATE))
-        f.write("Source: `git log` of the troupe worktree (`d9e4a66`, {total} commits\n".format(total=total))
-        f.write("reachable from HEAD). Extracted {date}.\n\n".format(date=ASN_DATE))
+        f.write("Snapshot: `git log` of the troupe worktree at `{snapshot}` ({total} commits\n".format(
+            snapshot=SNAPSHOT, total=total))
+        f.write("reachable from HEAD before this census was committed). Extracted {date}.\n\n".format(
+            date=ASN_DATE))
         f.write("Identity key is the author name plus email in `Name <email>` form as\n")
         f.write("recorded by git; no canonicalization (e.g. `Timmy` vs `timmy`, or a\n")
         f.write("stackchain.ai vs agents.invalid address, stays a distinct row).\n\n")
@@ -66,7 +81,7 @@ def write_artifacts(counts):
 
 
 def verify(csv_path, md_path):
-    expected = git_author_census()
+    expected = git_author_census(SNAPSHOT)
     total = sum(expected.values())
     with open(csv_path, newline="") as f:
         rows = list(csv.reader(f))
@@ -87,37 +102,34 @@ def verify(csv_path, md_path):
             errs.append("author not in recomputed census: {!r}".format(author))
         elif str(exp) != commits:
             errs.append("commits mismatch for {!r}: artifact {} expected {}".format(author, commits, exp))
-    # independent total from the commit object count
+    # independent total from the same snapshot
     proc = subprocess.run(
-        ["git", "rev-list", "--count", "HEAD"], cwd=REPO,
+        ["git", "rev-list", "--count", SNAPSHOT], cwd=REPO,
         check=True, capture_output=True, text=True)
     if proc.stdout.strip() != str(total):
-        errs.append("rev-list total {} != census total {}".format(proc.stdout.strip(), total))
-    # md should contain the total and the correct number of rows
+        errs.append("snapshot commit total {} != census total {}".format(proc.stdout.strip(), total))
     md = open(md_path).read()
     if "Total: {} commits".format(total) not in md:
-        errs.append("MD missing total line")
-    if str(len(expected)) not in md.split("across ")[-1].split(" distinct")[0]:
-        errs.append("MD row count wrong")
+        errs.append("MD missing total line for {total}".format(total=total))
 
     if errs:
         print("FAIL:")
         for e in errs:
             print("  - " + e)
         return 1
-    print("PASS: {} author identities, {} commits; {} CSV rows + MD match a fresh "
-          "recompile from `git log` and `git rev-list --count HEAD`.".format(
-              len(expected), total, len(body)))
+    print("PASS: snapshot `{snapshot}` = {total} commits, {n} author identities; "
+          "{rows} CSV rows + MD match a fresh recompile from `git log {snapshot}`.".format(
+              snapshot=SNAPSHOT, total=total, n=len(expected), rows=len(body)))
     return 0
 
 
 def main():
     mode = sys.argv[1] if len(sys.argv) > 1 else "check"
     if mode == "generate":
-        csv_path, md_path, total = write_artifacts(git_author_census())
+        csv_path, md_path, total = write_artifacts(git_author_census(SNAPSHOT))
         print("wrote", csv_path)
         print("wrote", md_path)
-        print("total commits:", total)
+        print("snapshot {snapshot} total commits:".format(snapshot=SNAPSHOT), total)
         return 0
     if mode == "check":
         csv_path = LEDGER_DIR + "/" + CSV_NAME