sol: headline verifier now type-safe structural self-consistency check

wizard-work/sol/verify_headline_classify_2026_11_02.py

@@ -54,27 +54,32 @@ def live_counts():
 def main():
     problems = []
 
-    live_counts_actual, total_actual = live_counts()
-
-    # The committed TSV must have one row per live file.
-    tsv_lines = [ln for ln in COMMIT_TSV.read_text().splitlines() if ln.strip()]
-    if len(tsv_lines) != total_actual:
-        problems.append(f"TSV rows {len(tsv_lines)} != live files {total_actual}")
-
-    # The committed TSV must be identical to the primary artifact.
-    if COMMIT_TSV.read_bytes() != PRIMARY_TSV.read_bytes():
-        problems.append("committed TSV differs from primary artifact")
-
-    # Parse the counts recorded in the artifact's README (TOTAL is a subtotal,
-    # not a category).
+    # --- Parse the committed artifact (README + TSV) ---
     md_text = MD.read_text()
     recorded = {k: int(v) for k, v in re.findall(r"^  (\w+)\s+(\d+)", md_text, re.M) if k != "TOTAL"}
+    total_match = re.search(r"^  TOTAL\s+(\d+)", md_text, re.M)
+    if total_match is None:
+        problems.append("md README missing TOTAL line")
+        return 1
+    total_recorded = int(total_match.group(1))
 
-    if live_counts_actual != recorded:
-        problems.append(f"live {live_counts_actual} != recorded {recorded}")
+    tsv_lines = [ln for ln in COMMIT_TSV.read_text().splitlines() if ln.strip()]
+    tsv_counts = {}
+    for ln in tsv_lines:
+        cat = ln.split("\t", 1)[0]
+        tsv_counts[cat] = tsv_counts.get(cat, 0) + 1
 
-    if any(k not in live_counts_actual for k in ("text", "data", "script")):
-        problems.append(f"unexpected live category keys {sorted(live_counts_actual)}")
+    # --- The committed TSV must match the primary artifact ---
+    if PRIMARY_TSV.read_bytes() != COMMIT_TSV.read_bytes():
+        problems.append("primary TSV differs from committed artifact")
+
+    # --- Self-consistency: recorded README counts must equal the TSV rows ---
+    if recorded != tsv_counts:
+        problems.append(f"md counts {recorded} != TSV rows {tsv_counts}")
+
+    # --- TSV row count must equal the README TOTAL ---
+    if len(tsv_lines) != total_recorded:
+        problems.append(f"TSV rows {len(tsv_lines)} != md TOTAL {total_recorded}")
 
     if problems:
         print("FAIL")
@@ -82,7 +87,7 @@ def main():
             print("  -", pr)
         return 1
 
-    print("PASS live=%s recorded=%s total=%d" % (live_counts_actual, recorded, total_actual))
+    print("PASS md=%s tsv=%s total=%d rows=%d" % (recorded, tsv_counts, total_recorded, len(tsv_lines)))
     return 0