From 7175b64e5ae8928ce7e03b60179b8a347b38c192 Mon Sep 17 00:00:00 2001
From: James B <james.baster@opendataservices.coop>
Date: Wed, 26 Mar 2025 10:40:00 +0000
Subject: [PATCH] Data Package should contain a publishers.jl file too

https://github.com/ThreeSixtyGiving/grantnav/issues/862
---
 .../commands/create_data_package.py           |  4 ++
 .../commands/manage_entities_data.py          | 40 ++++++++++++-------
 datastore/tests/test_commands.py              |  4 ++
 3 files changed, 34 insertions(+), 14 deletions(-)

diff --git a/datastore/db/management/commands/create_data_package.py b/datastore/db/management/commands/create_data_package.py
index 57833eae..f9e9ffbc 100644
--- a/datastore/db/management/commands/create_data_package.py
+++ b/datastore/db/management/commands/create_data_package.py
@@ -55,6 +55,7 @@ def handle(self, *args, **options):
         data_all_file = "%s/data_all.json" % options["dir"]
         recipients_file = "%s/recipients.jl" % options["dir"]
         funders_file = "%s/funders.jl" % options["dir"]
+        publishers_file = "%s/publishers.jl" % options["dir"]
 
         with open(funders_file, "w") as funders_fp:
             create_orgs_list("funder", funders_fp)
@@ -62,6 +63,9 @@ def handle(self, *args, **options):
         with open(recipients_file, "w") as recipients_fp:
             create_orgs_list("recipient", recipients_fp)
 
+        with open(publishers_file, "w") as recipients_fp:
+            create_orgs_list("publisher", recipients_fp)
+
         def flatten_grant(in_grant):
             """Add the additional_data inside grant object"""
             out_grant = {}
diff --git a/datastore/db/management/commands/manage_entities_data.py b/datastore/db/management/commands/manage_entities_data.py
index 3ab0c654..91d3050f 100644
--- a/datastore/db/management/commands/manage_entities_data.py
+++ b/datastore/db/management/commands/manage_entities_data.py
@@ -87,20 +87,32 @@ def create_orgs_list(entity_type, output=sys.stdout):
     entity_type: publisher, recipient, funder
     output: io
     """
-    query = f"""
-        SELECT DISTINCT
-        db_{entity_type}.org_id as "id",
-        db_{entity_type}.non_primary_org_ids as "non_primary_org_ids",
-        db_{entity_type}.name as name,
-        db_{entity_type}."aggregate" as "aggregate",
-        db_{entity_type}.additional_data as "additionalData",
-        additional_data_orginfocache.data as "ftcData",
-        db_publisher.name as "publisherName",
-        db_publisher.prefix as "publisherPrefix"
-        FROM db_{entity_type}
-        LEFT OUTER JOIN additional_data_orginfocache on db_{entity_type}.org_id = additional_data_orginfocache.org_id
-        LEFT OUTER JOIN db_publisher on db_{entity_type}.org_id = db_publisher.org_id OR db_publisher.org_id = ANY(db_{entity_type}.non_primary_org_ids)
-    """
+    if entity_type == "publisher":
+        query = f"""
+            SELECT DISTINCT
+            db_{entity_type}.org_id as "id",
+            db_{entity_type}.name as name,
+            db_{entity_type}."aggregate" as "aggregate",
+            db_{entity_type}.additional_data as "additionalData",
+            additional_data_orginfocache.data as "ftcData"
+            FROM db_{entity_type}
+            LEFT OUTER JOIN additional_data_orginfocache on db_{entity_type}.org_id = additional_data_orginfocache.org_id
+        """
+    else:
+        query = f"""
+            SELECT DISTINCT
+            db_{entity_type}.org_id as "id",
+            db_{entity_type}.non_primary_org_ids as "non_primary_org_ids",
+            db_{entity_type}.name as name,
+            db_{entity_type}."aggregate" as "aggregate",
+            db_{entity_type}.additional_data as "additionalData",
+            additional_data_orginfocache.data as "ftcData",
+            db_publisher.name as "publisherName",
+            db_publisher.prefix as "publisherPrefix"
+            FROM db_{entity_type}
+            LEFT OUTER JOIN additional_data_orginfocache on db_{entity_type}.org_id = additional_data_orginfocache.org_id
+            LEFT OUTER JOIN db_publisher on db_{entity_type}.org_id = db_publisher.org_id OR db_publisher.org_id = ANY(db_{entity_type}.non_primary_org_ids)
+        """
 
     def parse_data_in_result(result, col_types):
         # work around for https://github.com/ThreeSixtyGiving/datastore/issues/125
diff --git a/datastore/tests/test_commands.py b/datastore/tests/test_commands.py
index d42e63e6..e00e3550 100644
--- a/datastore/tests/test_commands.py
+++ b/datastore/tests/test_commands.py
@@ -34,6 +34,10 @@ def test_create_and_load_data_package(self):
             with open(os.path.join(tmpdir, "recipients.jl")) as recipients_fp:
                 json.loads(recipients_fp.readline().strip())
 
+            # Check the output json lines file by parsing the first line
+            with open(os.path.join(tmpdir, "publishers.jl")) as recipients_fp:
+                json.loads(recipients_fp.readline().strip())
+
             call_command("load_data_package", tmpdir, stderr=err_out)
             self.assertEqual(
                 len(err_out.getvalue()), 0, "Errors output by load command"