From fcfafb7301c6499c3c34793b846ecca3affb1510 Mon Sep 17 00:00:00 2001
From: Jay Varner <jayvarner@gmail.com>
Date: Mon, 12 Aug 2024 15:42:55 -0400
Subject: [PATCH] Add extra metadata and call task to add OCR

---
 readux_ingest_ecds/models.py | 16 ++++++++++++++--
 test_app/tests/test_s3.py    |  5 +++--
 2 files changed, 17 insertions(+), 4 deletions(-)

diff --git a/readux_ingest_ecds/models.py b/readux_ingest_ecds/models.py
index 0cf8298..9367ec1 100644
--- a/readux_ingest_ecds/models.py
+++ b/readux_ingest_ecds/models.py
@@ -419,10 +419,15 @@ class Meta:
         verbose_name_plural = "Amazon S3 Ingests"
 
     def ingest(self):
-        metadata = metadata_from_file(self.metadata_spreadsheet.path)
+        rows = metadata_from_file(self.metadata_spreadsheet.path)
 
-        for pid in [row["pid"] for row in metadata]:
+        for row in rows:
+            pid = row["pid"]
             manifest = create_manifest_from_pid(pid, self.image_server)
+            metadata = dict(row)
+            for key, value in metadata.items():
+                setattr(manifest, key, value)
+
             manifest.collections.set(self.collections.all())
             manifest.save()
             local_ingest = Local.objects.create(
@@ -452,5 +457,12 @@ def ingest(self):
                     t_file.write(f"{image_file}\n")
 
             local_ingest.create_canvases()
+            manifest.save()
+            from .tasks import add_ocr_task_local
+
+            if os.environ["DJANGO_ENV"] == "test":
+                add_ocr_task_local(str(local_ingest.id))
+            else:
+                add_ocr_task_local.delay(str(local_ingest.id))
 
         self.delete()
diff --git a/test_app/tests/test_s3.py b/test_app/tests/test_s3.py
index cee0b36..f0f5932 100644
--- a/test_app/tests/test_s3.py
+++ b/test_app/tests/test_s3.py
@@ -82,12 +82,12 @@ def create_pids(self, pid_count=1, image_count=1, include_pid_in_file=True):
         pids = []
         pid_file = os.path.join(self.fixture_path, self.fake.file_name(extension="csv"))
         with open(pid_file, "w", encoding="utf-8") as t_file:
-            t_file.write("PID\n")
+            t_file.write("PID,Label\n")
 
         for _ in range(pid_count):
             pid = self.fake.isbn10()
             with open(pid_file, "a", encoding="utf-8") as t_file:
-                t_file.write(f"{pid}\n")
+                t_file.write(f"{pid},{self.fake.name()}\n")
             pids.append(pid)
             self.create_source_images(
                 pid=pid, count=image_count, include_pid_in_file=include_pid_in_file
@@ -123,6 +123,7 @@ def test_s3_ingest_pid_not_in_filename(self):
             ]
             assert Manifest.objects.filter(pid=pid).exists()
             assert Manifest.objects.get(pid=pid).canvas_set.count() == 4
+            assert Manifest.objects.get(pid=pid).label is not None
             assert len(ingested_images) == 4
             assert len(ingested_ocr) == 4