Automatically resume empty saved source reviews

arr_web/arr_data_executor.py

@@ -3,7 +3,7 @@ from datetime import date
 from pathlib import Path
 
 from arr_web.arr_download_handoff import outcome_from_handoff
-from arr_web.arr_downloads import DownloadOutcome, validate_request_id
+from arr_web.arr_downloads import DownloadOutcome, validate_report_date, validate_request_id
 from integrations.ohip.arr_data import ARRDataSource, VERSION as SOURCE_VERSION
 from integrations.ohip import processing_handoff as handoff
 from integrations.ohip.audit_arr_capture import protected_read
@@ -34,6 +34,23 @@ class DirectARRExecutor:
     def finalize_data_review(self, request_id, revision, actor):
         return self.data_reviews.finalize(request_id, revision, actor)
 
+    def recover_data_review(self, request_id):
+        # Recovery may resume only an already collected, exactly bound local
+        # request. A missing/conflicting checkpoint must not start another fetch.
+        validate_request_id(request_id)
+        directory = self.root / "requests" / request_id
+        stored = strict_json(protected_read(directory / "request.json", 65536))
+        report_date = validate_report_date(stored.get("report_date"))
+        identity = {**self.source_identity(), "request_id": request_id, "report_date": report_date}
+        require(fingerprint(stored) == fingerprint(identity), "data_executor_request_conflict")
+        checkpoint = strict_json(protected_read(directory / "collected.json", 65536))
+        require(fingerprint(checkpoint["identity"]) == fingerprint(identity), "data_collection_checkpoint_conflict")
+        payload, manifest = self.data_reviews.original(request_id)
+        require(handoff._hash(payload) == checkpoint["data_sha256"]
+                and manifest == checkpoint["manifest_sha256"]
+                and strict_json(payload).get("report_date") == report_date, "data_collection_checkpoint_changed")
+        return self.data_reviews.recover_ready(request_id)
+
     def source_identity(self):
         return {"version": "arr-direct-data-executor/v1", "source_version": SOURCE_VERSION,
                 "hotel_id": self.source.hotel_id, "service_url": SERVICE, "application_id": APPLICATION,
@@ -90,6 +107,7 @@ class DirectARRExecutor:
                                                 "manifest_sha256": original_manifest}, replace=False)
                 manifest_sha256 = original_manifest
                 if review:
+                    self.recover_data_review(request_id)
                     completed = self.data_reviews.payload(request_id)
                     if completed is None:
                         return DownloadOutcome("needs_data_review")