From 875bfd06e1c650b94f01036258da8b0c907bd326 Mon Sep 17 00:00:00 2001
From: Andy Kohler <akohler@library.ucla.edu>
Date: Wed, 12 Jun 2024 13:56:58 -0800
Subject: [PATCH] SYS-1637: Remove master file URLs from OAI feed (#134)

---
 charts/prod-ohstaff-values.yaml                      | 2 +-
 oh_staff_ui/classes/OralHistoryMods.py               | 9 ++++++---
 oh_staff_ui/templates/oh_staff_ui/release_notes.html | 8 +++++++-
 oh_staff_ui/tests.py                                 | 6 ++++++
 4 files changed, 20 insertions(+), 5 deletions(-)

diff --git a/charts/prod-ohstaff-values.yaml b/charts/prod-ohstaff-values.yaml
index ff50880..75c43f3 100644
--- a/charts/prod-ohstaff-values.yaml
+++ b/charts/prod-ohstaff-values.yaml
@@ -6,7 +6,7 @@ replicaCount: 1
 
 image:
   repository: uclalibrary/oral-history-staff-ui
-  tag: v1.1.10
+  tag: v1.1.11
   pullPolicy: Always
 
 nameOverride: ""
diff --git a/oh_staff_ui/classes/OralHistoryMods.py b/oh_staff_ui/classes/OralHistoryMods.py
index 2ffd5a8..6b7e54b 100644
--- a/oh_staff_ui/classes/OralHistoryMods.py
+++ b/oh_staff_ui/classes/OralHistoryMods.py
@@ -179,7 +179,8 @@ def _create_relateditem_audio(self, mi: MediaFile) -> MODSv34:
         for ts in MediaFile.objects.filter(
             item=pi, file_type__file_code="text_master_transcript"
         ):
-            if ts.file_url != "" and ts.file_url.endswith(".xml"):
+            # Add only for submasters, the public-access copy
+            if ts.file_url.endswith("submaster.xml"):
                 # Due to legacy design the text_master_transcript can have 2 file types
                 # associated with it, we only want to show xml (TEI), and ignore html files
                 ri.locations.append(LocationOH(url=ts.file_url, usage="timed log"))
@@ -208,7 +209,8 @@ def _populate_interview_content(self):
         for f in MediaFile.objects.filter(
             item=self._item, file_type__file_code__in=fc_to_label.keys()
         ).order_by("sequence"):
-            if f.file_url != "":
+            # Add only for submasters, the public-access copy
+            if "submaster" in f.file_url:
                 label = fc_to_label[f.file_type.file_code]
                 usage = ""
 
@@ -219,7 +221,8 @@ def _populate_interview_content(self):
                         label = f"{label} (TEI/P5 XML)"
                         usage = "timed log"
 
-                # If our MediaFile is TEI/P5 XML, a usage attribute is populated and should be included
+                # If our MediaFile is TEI/P5 XML, a usage attribute is populated
+                # and should be included
                 if usage:
                     self.locations.append(
                         LocationOH(url=f.file_url, label=label, usage=usage)
diff --git a/oh_staff_ui/templates/oh_staff_ui/release_notes.html b/oh_staff_ui/templates/oh_staff_ui/release_notes.html
index e1f9b04..a65ecf1 100644
--- a/oh_staff_ui/templates/oh_staff_ui/release_notes.html
+++ b/oh_staff_ui/templates/oh_staff_ui/release_notes.html
@@ -3,10 +3,16 @@
 {% block content %}
 <h3>Release Notes</h3>
 <hr/>
+<h4>1.1.11</h4>
+<p><i>June 12, 2024</i></p>
+<ul>
+   <li>Fixed bug which included master file URLs in the OAI feed.</li>
+</ul>
+
 <h4>1.1.10</h4>
 <p><i>June 6, 2024</i></p>
 <ul>
-   <li>Fix alignment of item metadata form fields.</li>
+   <li>Fixed alignment of item metadata form fields.</li>
 </ul>
 
 <h4>1.1.9</h4>
diff --git a/oh_staff_ui/tests.py b/oh_staff_ui/tests.py
index b9f8e1c..616f812 100644
--- a/oh_staff_ui/tests.py
+++ b/oh_staff_ui/tests.py
@@ -1543,6 +1543,12 @@ def test_sealed_audio_metadata_is_excluded(self):
         # That title should not be present, since that item has been Sealed.
         self.assertNotIn("Fake audio", related_titles)
 
+    def test_master_file_urls_are_excluded(self):
+        # Confirm that master file URLs, which are only used in the staff UI,
+        # are not included in the OAI feed.
+        ohmods = self.get_mods_from_interview_item()
+        self.assertFalse(b"/oh_masters/" in ohmods.serializeDocument())
+
 
 class FileMetadataMigrationTestCase(SimpleTestCase):
     # Test logic not already covered by OralHistoryFile tests.