From f4c5b737bf87fb852d422013b0657b34891984a2 Mon Sep 17 00:00:00 2001
From: joshuastegmaier <jstegmaier@loc.gov>
Date: Tue, 24 Sep 2024 09:10:50 -0400
Subject: [PATCH] Added ability to undo an OCR transcription even if it's the
 first transcription on an asset

---
 concordia/models.py |  7 +++++--
 concordia/views.py  | 17 +++++++++++++++--
 2 files changed, 20 insertions(+), 4 deletions(-)

diff --git a/concordia/models.py b/concordia/models.py
index 17de74b26..3821a79d0 100644
--- a/concordia/models.py
+++ b/concordia/models.py
@@ -662,7 +662,6 @@ def can_rollback(self):
         transcriptions = (
             self.transcription_set.exclude(rolled_forward=True)
             .exclude(source_of__rolled_forward=True)
-            .exclude(rolled_back=True)
             .exclude(pk__gte=latest_transcription.pk)
             .order_by("-pk")
         )
@@ -677,9 +676,13 @@ def can_rollback(self):
 
         transcription_to_rollback_to = None
         for transcription in transcriptions:
+            if transcription.source:
+                transcription_to_check = transcription.source
+            else:
+                transcription_to_check = transcription
             if (
                 latest_transcription.rolled_back is False
-                or latest_transcription.supersedes != transcription
+                or latest_transcription.supersedes != transcription_to_check
             ):
                 transcription_to_rollback_to = transcription
                 break
diff --git a/concordia/views.py b/concordia/views.py
index 0f1965de5..2c4621bf5 100644
--- a/concordia/views.py
+++ b/concordia/views.py
@@ -1568,8 +1568,21 @@ def generate_ocr_transcription(request, *, asset_pk):
     supersedes_pk = request.POST.get("supersedes")
     language = request.POST.get("language", None)
     superseded = get_transcription_superseded(asset, supersedes_pk)
-    if superseded and isinstance(superseded, HttpResponse):
-        return superseded
+    if superseded:
+        if isinstance(superseded, HttpResponse):
+            return superseded
+    else:
+        # This means this is the first transcription on this asset
+        # to enable undoing of the OCR transcription, we create
+        # an empty transcription for the OCR transcription to supersede
+        superseded = Transcription(
+            asset=asset,
+            user=get_anonymous_user(),
+            text="",
+        )
+        superseded.full_clean()
+        superseded.save()
+
     transcription_text = asset.get_ocr_transcript(language)
     transcription = Transcription(
         asset=asset,