Merge pull request #32 from cortega26/codex/fix-publish-sub-routine-skipping-issue

Fix publish workflow skip reason visibility

Author: cortega26

.github/workflows/scrape.yml

@@ -117,11 +117,40 @@ jobs:
         id: decision
         run: |
           python - <<'PY'
-          import json, pathlib, os
-          payload = json.loads(pathlib.Path('artifacts/run_summary.json').read_text())
-          value = 'true' if payload.get('publish') else 'false'
-          with open(os.environ['GITHUB_OUTPUT'], 'a', encoding='utf-8') as handle:
-              handle.write(f"publish={value}\n")
+          """Expose publish decision metadata for downstream jobs."""
+
+          from __future__ import annotations
+
+          import json
+          import os
+          from pathlib import Path
+
+          summary_path = Path("artifacts/run_summary.json")
+          if not summary_path.exists():
+              raise SystemExit("Run summary not found; cannot determine publish decision")
+
+          payload = json.loads(summary_path.read_text(encoding="utf-8"))
+          publish_flag = bool(payload.get("publish"))
+          decision = payload.get("decision", {}) or {}
+          status = str(decision.get("status", "unknown"))
+          reason = str(payload.get("publish_reason") or decision.get("reason") or "unknown")
+
+          outputs_path = Path(os.environ["GITHUB_OUTPUT"])
+          with outputs_path.open("a", encoding="utf-8") as handle:
+              handle.write(f"publish={'true' if publish_flag else 'false'}\n")
+              handle.write(f"publish_reason={reason}\n")
+              handle.write(f"decision_status={status}\n")
+
+          summary_output = os.environ.get("GITHUB_STEP_SUMMARY")
+          if summary_output:
+              with Path(summary_output).open("a", encoding="utf-8") as handle:
+                  handle.write("### Publish decision\n")
+                  handle.write(f"- Publish: {'yes' if publish_flag else 'no'}\n")
+                  handle.write(f"- Status: {status}\n")
+                  handle.write(f"- Reason: {reason}\n")
+
+          if not publish_flag:
+              print(f"::notice::Publish job will be skipped because {reason} (status={status}).")
           PY
 
       - name: Save pipeline state
@@ -167,29 +196,40 @@ jobs:
 
   publish:
     needs: ingest
-    if: needs.ingest.outputs.publish == 'true'
+    if: needs.ingest.result == 'success'
     runs-on: ubuntu-latest
     steps:
+      - name: Publish decision summary
+        if: ${{ needs.ingest.outputs.publish != 'true' }}
+        run: |
+          echo "Publish step skipped: ${{ needs.ingest.outputs.publish_reason || 'no reason provided' }} (status=${{ needs.ingest.outputs.decision_status || 'unknown' }})"
+          exit 0
+
       - uses: actions/checkout@v4
+        if: ${{ needs.ingest.outputs.publish == 'true' }}
 
       - name: Set up Python
         uses: actions/setup-python@v5
+        if: ${{ needs.ingest.outputs.publish == 'true' }}
         with:
           python-version: '3.11'
           cache: 'pip'
           cache-dependency-path: 'requirements.txt'
 
       - name: Install dependencies
+        if: ${{ needs.ingest.outputs.publish == 'true' }}
         run: |
           python -m pip install --upgrade pip
           pip install -r requirements.txt
 
       - name: Download artifacts
+        if: ${{ needs.ingest.outputs.publish == 'true' }}
         uses: actions/download-artifact@v4
         with:
           name: alt-sources-artifacts
 
       - name: Publish to Google Sheets
+        if: ${{ needs.ingest.outputs.publish == 'true' }}
         env:
           GOOGLE_SERVICE_ACCOUNT_JSON: ${{ secrets.GOOGLE_SHEETS_CREDENTIALS }}
           GOOGLE_CREDENTIALS: ${{ secrets.GOOGLE_CREDENTIALS }}

docs/API.md

@@ -14,7 +14,7 @@ True
 
 - Sources: `"pozos"` (primary + fallback) or `"openloto"` (fallback only).
 - `source_overrides`: case‑insensitive mapping of `{ "openloto": url, "resultadoslotochile": url }`.
-- Returns a run summary with `publish` boolean and artifact paths.
+- Returns a run summary with `publish` boolean, `publish_reason` string, and artifact paths.
 
 Example:
 
@@ -38,6 +38,7 @@ summary = run_pipeline(
     include_pozos=True,
 )
 print(summary["publish"])  # True/False
+print(summary["publish_reason"])  # e.g. "updated_or_new_amounts"
 ```
 
 ## Publishing

polla_app/pipeline.py

@@ -298,20 +298,21 @@ def _compute_unchanged(
     return False
 
 
-def _build_report_payload(
-    *,
-    run_id: str,
-    generated_at: str,
-    requested_sources: Sequence[str],
-    timeout: int,
-    retries: int,
-    fail_fast: bool,
-    sorteo: Any,
-    fecha: Any,
-    merged_pozos: Mapping[str, Any],
-    record_source: Any,
-    decision_status: str,
-) -> dict[str, Any]:
+def _build_report_payload(
+    *,
+    run_id: str,
+    generated_at: str,
+    requested_sources: Sequence[str],
+    timeout: int,
+    retries: int,
+    fail_fast: bool,
+    sorteo: Any,
+    fecha: Any,
+    merged_pozos: Mapping[str, Any],
+    record_source: Any,
+    decision_status: str,
+    decision_reason: str,
+) -> dict[str, Any]:
     return {
         "run": {
             "id": run_id,
@@ -321,41 +322,44 @@ def _build_report_payload(
             "retries": retries,
             "fail_fast": fail_fast,
         },
-        "last_draw": {"sorteo": sorteo, "fecha": fecha},
-        "decision": {
-            "status": decision_status,
-            "total_categories": len(merged_pozos),
-            "mismatched_categories": 0,
-        },
-        "prizes_changed": decision_status != "skip",
-        "mismatches": [],
-        "sources": {"pozos": {"url": record_source, "premios": 0}},
-        "failures": [],
+        "last_draw": {"sorteo": sorteo, "fecha": fecha},
+        "decision": {
+            "status": decision_status,
+            "total_categories": len(merged_pozos),
+            "mismatched_categories": 0,
+            "reason": decision_reason,
+        },
+        "prizes_changed": decision_status != "skip",
+        "mismatches": [],
+        "sources": {"pozos": {"url": record_source, "premios": 0}},
+        "failures": [],
     }
 
 
-def _build_summary_payload(
-    *,
-    run_id: str,
-    generated_at: str,
-    decision: Mapping[str, Any],
-    normalized_path: Path,
-    comparison_report_path: Path,
-    raw_dir: Path,
-    state_path: Path,
-    publish_flag: bool,
-) -> dict[str, Any]:
-    return {
-        "run_id": run_id,
-        "generated_at": generated_at,
-        "decision": dict(decision),
-        "prizes_changed": bool(decision.get("status") != "skip"),
-        "normalized_path": str(normalized_path),
-        "comparison_report": str(comparison_report_path),
-        "raw_dir": str(raw_dir),
-        "state_path": str(state_path),
-        "publish": publish_flag,
-    }
+def _build_summary_payload(
+    *,
+    run_id: str,
+    generated_at: str,
+    decision: Mapping[str, Any],
+    normalized_path: Path,
+    comparison_report_path: Path,
+    raw_dir: Path,
+    state_path: Path,
+    publish_flag: bool,
+    publish_reason: str,
+) -> dict[str, Any]:
+    return {
+        "run_id": run_id,
+        "generated_at": generated_at,
+        "decision": dict(decision),
+        "prizes_changed": bool(decision.get("status") != "skip"),
+        "normalized_path": str(normalized_path),
+        "comparison_report": str(comparison_report_path),
+        "raw_dir": str(raw_dir),
+        "state_path": str(state_path),
+        "publish": publish_flag,
+        "publish_reason": publish_reason,
+    }
 
 
 def _handle_pozos_only(
@@ -401,39 +405,48 @@ def _handle_pozos_only(
     _write_jsonl(normalized_path, [record])
     _write_jsonl(state_path, [record])
 
-    decision_status = "skip" if unchanged else "publish"
-    publish_flag = not unchanged
-    if force_publish and unchanged:
-        decision_status = "publish_forced"
-        publish_flag = True
+    if unchanged:
+        decision_status = "skip"
+        publish_flag = False
+        publish_reason = "sorteo_fecha_and_amounts_unchanged"
+    else:
+        decision_status = "publish"
+        publish_flag = True
+        publish_reason = "updated_or_new_amounts"
+    if force_publish and unchanged:
+        decision_status = "publish_forced"
+        publish_flag = True
+        publish_reason = "force_publish_requested"
 
     generated_at = datetime.now(timezone.utc).isoformat()
-    report_payload = _build_report_payload(
-        run_id=run_id,
-        generated_at=generated_at,
-        requested_sources=requested_sources,
-        timeout=timeout,
-        retries=retries,
-        fail_fast=fail_fast,
-        sorteo=sorteo,
-        fecha=fecha,
-        merged_pozos=merged_pozos,
-        record_source=record["fuente"],
-        decision_status=decision_status,
-    )
+    report_payload = _build_report_payload(
+        run_id=run_id,
+        generated_at=generated_at,
+        requested_sources=requested_sources,
+        timeout=timeout,
+        retries=retries,
+        fail_fast=fail_fast,
+        sorteo=sorteo,
+        fecha=fecha,
+        merged_pozos=merged_pozos,
+        record_source=record["fuente"],
+        decision_status=decision_status,
+        decision_reason=publish_reason,
+    )
     report_payload["api_version"] = API_VERSION
     _write_json(comparison_report_path, report_payload)
 
-    summary_payload = _build_summary_payload(
-        run_id=run_id,
-        generated_at=generated_at,
-        decision=report_payload["decision"],
-        normalized_path=normalized_path,
-        comparison_report_path=comparison_report_path,
-        raw_dir=raw_dir,
-        state_path=state_path,
-        publish_flag=publish_flag,
-    )
+    summary_payload = _build_summary_payload(
+        run_id=run_id,
+        generated_at=generated_at,
+        decision=report_payload["decision"],
+        normalized_path=normalized_path,
+        comparison_report_path=comparison_report_path,
+        raw_dir=raw_dir,
+        state_path=state_path,
+        publish_flag=publish_flag,
+        publish_reason=publish_reason,
+    )
     summary_payload["api_version"] = API_VERSION
 
     log_event(
@@ -449,14 +462,12 @@ def _handle_pozos_only(
         {
             "event": "pipeline_complete",
             "run_id": run_id,
-            "decision": decision_status,
-            "mismatch_ratio": 0.0,
-            "prizes_changed": not unchanged,
-            "reason": (
-                "sorteo_fecha_and_amounts_unchanged" if unchanged else "updated_or_new_amounts"
-            ),
-        }
-    )
+            "decision": decision_status,
+            "mismatch_ratio": 0.0,
+            "prizes_changed": not unchanged,
+            "reason": publish_reason,
+        }
+    )
     metric(
         "pipeline_run",
         log_event,