⚡ Refactor run_live_suite to reduce unnecessary list traversals

studies/nz-reconciliation/runner/src/nz_reconciliation/run_live_suite.py

@@ -49,20 +49,28 @@ def compare_live(
         domain = rs.get("domain")
         if domain == "income_tax":
             rs_out = {
-                "primary": rs.get("outputs", {}).get("individual_income_tax_before_credits")
+                "primary": rs.get("outputs", {}).get(
+                    "individual_income_tax_before_credits"
+                )
                 or next(iter(rs.get("outputs", {}).values()), None)
             }
             of_out = {
-                "primary": of.get("outputs", {}).get("individual_income_tax_before_credits")
+                "primary": of.get("outputs", {}).get(
+                    "individual_income_tax_before_credits"
+                )
             }
             pair = compare_pair(
                 {**rs, "outputs": {k: v for k, v in rs_out.items() if v}},
                 {**of, "outputs": {k: v for k, v in of_out.items() if v}},
                 tolerance=tolerance,
             )
         elif domain == "acc_earners_levy":
-            rs_val = rs.get("outputs", {}).get("acc_standard_earners_levy_including_gst")
-            of_val = of.get("outputs", {}).get("acc_standard_earners_levy_including_gst")
+            rs_val = rs.get("outputs", {}).get(
+                "acc_standard_earners_levy_including_gst"
+            )
+            of_val = of.get("outputs", {}).get(
+                "acc_standard_earners_levy_including_gst"
+            )
             pair = compare_pair(
                 {**rs, "outputs": {"primary": rs_val} if rs_val else {}},
                 {**of, "outputs": {"primary": of_val} if of_val else {}},
@@ -84,18 +92,20 @@ def compare_live(
 
 
 def build_report(
-    rulespec_rows: list[dict],
-    openfisca_rows: list[dict],
-    comparison_rows: list[dict],
+    agreed: list[dict],
+    non_agreed: list[dict],
+    total_cases: int,
+    rulespec_ok: int,
+    openfisca_ok: int,
 ) -> str:
-    agreed = [r for r in comparison_rows if r.get("agreement")]
+
     lines = [
         "# NZ reconciliation — live dual-engine report",
         "",
-        f"Cases: **{len(comparison_rows)}**",
+        f"Cases: **{total_cases}**",
         f"Numeric agreements (≤$0.02): **{len(agreed)}**",
-        f"RuleSpec oracle ok: **{sum(1 for r in rulespec_rows if r.get('status')=='ok')}**",
-        f"OpenFisca live ok: **{sum(1 for r in openfisca_rows if r.get('status')=='ok')}**",
+        f"RuleSpec oracle ok: **{rulespec_ok}**",
+        f"OpenFisca live ok: **{openfisca_ok}**",
         "",
         "## Agreements",
         "",
@@ -106,9 +116,7 @@ def build_report(
     else:
         lines.append("_None_")
     lines.extend(["", "## Non-agreements / gaps", ""])
-    for row in comparison_rows:
-        if row.get("agreement"):
-            continue
+    for row in non_agreed:
         lines.append(
             f"- `{row.get('caseId')}`: `{row.get('classification')}` "
             f"(rs={row.get('rulespec', {}).get('status')}, "
@@ -138,7 +146,9 @@ def build_report(
 
 
 def main(argv: list[str] | None = None) -> int:
-    parser = argparse.ArgumentParser(description="Live dual-engine NZ reconciliation suite.")
+    parser = argparse.ArgumentParser(
+        description="Live dual-engine NZ reconciliation suite."
+    )
     parser.add_argument("--results-dir", type=Path, default=DEFAULT_RESULTS)
     args = parser.parse_args(argv)
     results_dir = args.results_dir
@@ -148,22 +158,41 @@ def main(argv: list[str] | None = None) -> int:
     try:
         openfisca_rows = run_openfisca_live_suite()
     except Exception as exc:  # noqa: BLE001
-        print(json.dumps({"ok": False, "error": f"{type(exc).__name__}: {exc}"}, indent=2))
+        print(
+            json.dumps({"ok": False, "error": f"{type(exc).__name__}: {exc}"}, indent=2)
+        )
         return 2
 
     comparison_rows = compare_live(rulespec_rows, openfisca_rows)
     write_jsonl(results_dir / "rulespec-candidate-results.jsonl", rulespec_rows)
     write_jsonl(results_dir / "openfisca-aotearoa-live-results.jsonl", openfisca_rows)
     write_jsonl(results_dir / "comparison-live-results.jsonl", comparison_rows)
-    report = build_report(rulespec_rows, openfisca_rows, comparison_rows)
+    agreed = []
+    non_agreed = []
+    for r in comparison_rows:
+        if r.get("agreement"):
+            agreed.append(r)
+        else:
+            non_agreed.append(r)
+
+    rulespec_ok = sum(1 for r in rulespec_rows if r.get("status") == "ok")
+    openfisca_live_ok = sum(1 for r in openfisca_rows if r.get("status") == "ok")
+
+    report = build_report(
+        agreed,
+        non_agreed,
+        len(comparison_rows),
+        rulespec_ok,
+        openfisca_live_ok,
+    )
     (results_dir / "LIVE_DUAL_ENGINE_REPORT.md").write_text(report, encoding="utf-8")
 
     summary = {
         "ok": True,
         "cases": len(comparison_rows),
-        "agreements": sum(1 for r in comparison_rows if r.get("agreement")),
-        "rulespec_ok": sum(1 for r in rulespec_rows if r.get("status") == "ok"),
-        "openfisca_live_ok": sum(1 for r in openfisca_rows if r.get("status") == "ok"),
+        "agreements": len(agreed),
+        "rulespec_ok": rulespec_ok,
+        "openfisca_live_ok": openfisca_live_ok,
         "report": str(results_dir / "LIVE_DUAL_ENGINE_REPORT.md"),
     }
     print(json.dumps(summary, indent=2))