Add ability for audit to queue ingest for missing

AlgoliaAuditService can now optionally trigger the ingest of missing
documents via the ClassificationService. From the CLI this is triggered
with the --reingest flag.

src/ook/cli.py

@@ -91,8 +91,9 @@ async def upload_doc_stub(dataset: Path) -> None:
 
 
 @main.command()
+@click.option("--reingest", is_flag=True, help="Reingest missing documents.")
 @run_with_asyncio
-async def audit() -> None:
+async def audit(*, reingest: bool = False) -> None:
     """Audit the Algolia document index and check if any documents are missing
     based on the listing of projects registered in the LTD Keeper service.
     """
@@ -109,4 +110,6 @@ async def audit() -> None:
         raise click.UsageError("Algolia credentials not set in environment.")
     async with Factory.create_standalone(logger=logger) as factory:
         algolia_audit_service = factory.create_algolia_audit_service()
-        await algolia_audit_service.audit_missing_documents()
+        await algolia_audit_service.audit_missing_documents(
+            ingest_missing=reingest
+        )

src/ook/factory.py

@@ -235,4 +235,5 @@ def create_algolia_audit_service(self) -> AlgoliaAuditService:
             http_client=self.http_client,
             algolia_search_client=self._process_context.algolia_client,
             logger=self._logger,
+            classification_service=self.create_classification_service(),
         )

src/ook/services/algoliaaudit.py

@@ -10,6 +10,7 @@
 from structlog.stdlib import BoundLogger
 
 from ..config import config
+from .classification import ClassificationService
 
 # Python regular expression pattern that matches an LTD document slug such as
 # "sqr-000".
@@ -57,13 +58,17 @@ def __init__(
         http_client: AsyncClient,
         logger: BoundLogger,
         algolia_search_client: SearchClient,
+        classification_service: ClassificationService,
     ) -> None:
         """Initialize the service."""
         self._http_client = http_client
         self._search_client = algolia_search_client
+        self._classifier = classification_service
         self._logger = logger
 
-    async def audit_missing_documents(self) -> list[LtdDocument]:
+    async def audit_missing_documents(
+        self, *, ingest_missing: bool = False
+    ) -> list[LtdDocument]:
         """Audit the Algolia indices for completeness of missing documents.
 
         A document is considered "missing" if it is registered in the LTD API,
@@ -97,14 +102,20 @@ async def audit_missing_documents(self) -> list[LtdDocument]:
                     published_url=expected_ltd_doc.published_url,
                 )
                 missing_docs.append(expected_ltd_doc)
+        missing_docs.sort()
 
         self._logger.info(
             "Audit complete.",
             found=len(expected_ltd_docs) - len(missing_docs),
             missing=len(missing_docs),
         )
 
-        missing_docs.sort()
+        if ingest_missing and len(missing_docs) > 0:
+            for doc in missing_docs:
+                await self._classifier.queue_ingest_for_ltd_product_slug(
+                    product_slug=doc.slug, edition_slug="main"
+                )
+
         return missing_docs
 
     async def _get_ltd_documents(self) -> list[LtdDocument]: