From 5a2e8e634c12b28b9fc5e32e07c5eda8beceadf6 Mon Sep 17 00:00:00 2001
From: Luca Fabbri <l.fabbri@bopen.eu>
Date: Fri, 27 Sep 2024 10:16:58 +0200
Subject: [PATCH] Fixed alg. used to retrieve latest available licence (#87)

* Fixed alg. used to retrieve latest available licence

* Fixed licences vocabulary query

Group by was not working properly, not returning just most updated revisions

* Fixed SAWarning and mypy deprecation note

* Fixing wheel build
---
 cads_catalogue_api_service/vocabularies.py | 39 ++++++++++++----------
 pyproject.toml                             |  7 ++--
 2 files changed, 26 insertions(+), 20 deletions(-)

diff --git a/cads_catalogue_api_service/vocabularies.py b/cads_catalogue_api_service/vocabularies.py
index 9b30e54..f58c27e 100644
--- a/cads_catalogue_api_service/vocabularies.py
+++ b/cads_catalogue_api_service/vocabularies.py
@@ -42,14 +42,22 @@ def query_licences(
     scope: LicenceScopeCriterion,
     portals: list[str] | None = None,
 ) -> list[cads_catalogue.database.Licence]:
-    """Query licences."""
-    # NOTE: possible issue here if the title of a licence change from a revision to another
+    """Query all licences.
+
+    Return the latest revision of each licence. Older revision virtually disappear from API.
+    """
+    # subquery, to select all unique licence_uids and their max revision
+    subquery = session.query(
+        cads_catalogue.database.Licence.licence_uid,
+        sa.func.max(cads_catalogue.database.Licence.revision).label("revision"),
+    ).group_by(cads_catalogue.database.Licence.licence_uid)
+
     query = session.query(
         cads_catalogue.database.Licence.licence_uid,
         cads_catalogue.database.Licence.title,
         cads_catalogue.database.Licence.md_filename,
         cads_catalogue.database.Licence.download_filename,
-        sa.func.max(cads_catalogue.database.Licence.revision).label("revision"),
+        cads_catalogue.database.Licence.revision,
         cads_catalogue.database.Licence.scope,
         cads_catalogue.database.Licence.portal,
     )
@@ -62,18 +70,14 @@ def query_licences(
                 cads_catalogue.database.Licence.portal.is_(None),
             )
         )
-    results = (
-        query.group_by(
+
+    # Now retrieve all licences where the tuple (licence_uid, revision) is in the subquery
+    results = query.filter(
+        sa.tuple_(
             cads_catalogue.database.Licence.licence_uid,
-            cads_catalogue.database.Licence.title,
-            cads_catalogue.database.Licence.md_filename,
-            cads_catalogue.database.Licence.download_filename,
-            cads_catalogue.database.Licence.scope,
-            cads_catalogue.database.Licence.portal,
-        )
-        .order_by(cads_catalogue.database.Licence.title)
-        .all()
-    )
+            cads_catalogue.database.Licence.revision,
+        ).in_(subquery)
+    ).order_by(cads_catalogue.database.Licence.title)
     return results  # type: ignore
 
 
@@ -87,7 +91,7 @@ def query_licence(
         cads_catalogue.database.Licence.title,
         cads_catalogue.database.Licence.md_filename,
         cads_catalogue.database.Licence.download_filename,
-        sa.func.max(cads_catalogue.database.Licence.revision).label("revision"),
+        cads_catalogue.database.Licence.revision,
         cads_catalogue.database.Licence.scope,
         cads_catalogue.database.Licence.portal,
     )
@@ -99,11 +103,12 @@ def query_licence(
                 cads_catalogue.database.Licence.title,
                 cads_catalogue.database.Licence.md_filename,
                 cads_catalogue.database.Licence.download_filename,
+                cads_catalogue.database.Licence.revision,
                 cads_catalogue.database.Licence.scope,
                 cads_catalogue.database.Licence.portal,
             )
-            .order_by(cads_catalogue.database.Licence.title)
-            .one()
+            .order_by(sa.desc("revision"))
+            .first()
         )
     except sa.exc.NoResultFound as exc:
         raise fastapi.HTTPException(
diff --git a/pyproject.toml b/pyproject.toml
index 29d3e26..17d3472 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -17,11 +17,12 @@ dependencies = [
   "cads-catalogue@git+https://github.com/ecmwf-projects/cads-catalogue.git",
   "cads-common@git+https://github.com/ecmwf-projects/cads-common.git",
   "fastapi>=0.113.0",
+  "pydantic<2",
   "python-dateutil",
   "sqlalchemy>=2.0.9",
-  "stac_fastapi.api",
-  "stac_fastapi.extensions",
-  "stac_fastapi.types"
+  "stac_fastapi.api==2.4.1",
+  "stac_fastapi.extensions==2.4.1",
+  "stac_fastapi.types==2.4.1"
 ]
 description = "STAC based API service for the Climate & Atmosphere Data Store"
 dynamic = ["version"]