From 0285d2507455546be0beb440f996bd69806634ec Mon Sep 17 00:00:00 2001
From: Jay Varner <jayvarner@gmail.com>
Date: Fri, 13 Dec 2024 09:13:10 -0500
Subject: [PATCH] Actually call the canvas check. Add thumbnails to warnings
 email

---
 readux_ingest_ecds/admin.py                    |  2 ++
 readux_ingest_ecds/mail.py                     | 18 +++++++++++++++---
 readux_ingest_ecds/models.py                   |  2 +-
 readux_ingest_ecds/services/ocr_services.py    |  2 +-
 readux_ingest_ecds/tasks.py                    | 17 ++---------------
 .../templates/ingest_ecds_success_email.html   | 14 +++++++++-----
 test_app/tests/test_ocr.py                     |  3 ++-
 7 files changed, 32 insertions(+), 26 deletions(-)

diff --git a/readux_ingest_ecds/admin.py b/readux_ingest_ecds/admin.py
index 83f81eb..29e825f 100644
--- a/readux_ingest_ecds/admin.py
+++ b/readux_ingest_ecds/admin.py
@@ -14,7 +14,9 @@
 class LocalAdmin(admin.ModelAdmin):
     """Django admin ingest.models.local resource."""
 
+    list_display = ["manifest"]
     fields = ("bundle", "image_server", "collections")
+    readonly_fields = ["manifest", "warnings"]
     show_save_and_add_another = False
 
     def save_model(self, request, obj, form, change):
diff --git a/readux_ingest_ecds/mail.py b/readux_ingest_ecds/mail.py
index b939c35..ad2efd7 100644
--- a/readux_ingest_ecds/mail.py
+++ b/readux_ingest_ecds/mail.py
@@ -47,9 +47,21 @@ def send_email_on_success(creator=None, manifest=None, warnings=None):
         ingest_warnings = (
             warnings if warnings is not None and len(warnings) > 10 else None
         )
-        context["warnings"] = (
-            ingest_warnings.split("$$$$") if ingest_warnings is not None else None
-        )
+        context["warnings"] = ""
+        context["html_warnings"] = []
+        if ingest_warnings is not None:
+            warnings = ingest_warnings.split(" | ")
+            context["warnings"] = [
+                f'{w} -- https://iip.readux.io/iiif/3/{w.split(" ")[1]}/full/250,/0/default.jpg'
+                for w in warnings
+            ]
+            context["html_warnings"] = [
+                [
+                    w,
+                    f'<img src="https://iip.readux.io/iiif/3/{w.split(" ")[1]}/full/250,/0/default.jpg" />',
+                ]
+                for w in warnings
+            ]
 
         html_email = get_template("ingest_ecds_success_email.html").render(context)
         text_email = get_template("ingest_ecds_success_email.txt").render(context)
diff --git a/readux_ingest_ecds/models.py b/readux_ingest_ecds/models.py
index c565778..5d13c79 100644
--- a/readux_ingest_ecds/models.py
+++ b/readux_ingest_ecds/models.py
@@ -242,6 +242,7 @@ def create_canvases(self):
                 new_canvases.append(new_canvas)
 
         Canvas.objects.bulk_create(new_canvases)
+        self.check_canvases()
 
         upload_trigger_file(self.trigger_file)
 
@@ -272,7 +273,6 @@ def success(self):
 
             index = ManifestDocument()
             index.update(self.manifest, True, "index")
-        self.delete()
 
     def failure(self, exc):
         LOGGER.info(f"FAIL!!! {self.manifest.pid}")
diff --git a/readux_ingest_ecds/services/ocr_services.py b/readux_ingest_ecds/services/ocr_services.py
index 0d9d908..edbe474 100644
--- a/readux_ingest_ecds/services/ocr_services.py
+++ b/readux_ingest_ecds/services/ocr_services.py
@@ -546,7 +546,7 @@ def add_ocr_to_canvases(manifest):
         elif ocr is not None:
             new_ocr_annotations += add_ocr_annotations(canvas, ocr)
         else:
-            warnings.append(f"No OCR for {canvas.pid}")
+            warnings.append(f"Canvas {canvas.pid} - No OCR")
 
     chunks = divide_chunks(new_ocr_annotations, 100)
     for chunk in list(chunks):
diff --git a/readux_ingest_ecds/tasks.py b/readux_ingest_ecds/tasks.py
index f6668bc..9298d01 100644
--- a/readux_ingest_ecds/tasks.py
+++ b/readux_ingest_ecds/tasks.py
@@ -130,6 +130,7 @@ def add_canvases_task(ingest_id, manifest_pid, *args, **kwargs):
 
 @app.task(
     name="add_ocr_task_local_ecds",
+    base=FinalTask,
     autoretry_for=(Manifest.DoesNotExist,),
     retry_backoff=5,
 )
@@ -139,24 +140,10 @@ def add_ocr_task_local(ingest_id, manifest_pid, *args, **kwargs):
     local_ingest = Local.objects.get(pk=ingest_id)
     manifest = Manifest.objects.get(pk=local_ingest.manifest.pk)
     warnings = add_ocr_to_canvases(manifest)
-    local_ingest.warnings = "$$$$".join(warnings)
+    local_ingest.warnings = " | ".join(warnings)
     local_ingest.save()
 
 
-@app.task(
-    name="verify_canvases_task",
-    base=FinalTask,
-    autoretry_for=(Exception,),
-    retry_backoff=True,
-    max_retries=2,
-)
-def verify_canvases_task(ingest_id, *args, **kwargs):
-    """Task to call check for duplicate canvases"""
-    local_ingest = Local.objects.get(pk=ingest_id)
-    LOGGER.info(f"Checking canvases {local_ingest.manifest.pid}")
-    local_ingest.check_canvases()
-
-
 @app.task(
     name="s3_ingest_task_ecds",
     autoretry_for=(Exception,),
diff --git a/readux_ingest_ecds/templates/ingest_ecds_success_email.html b/readux_ingest_ecds/templates/ingest_ecds_success_email.html
index 4748d24..2c2cb01 100644
--- a/readux_ingest_ecds/templates/ingest_ecds_success_email.html
+++ b/readux_ingest_ecds/templates/ingest_ecds_success_email.html
@@ -12,11 +12,15 @@ <h3 style="color: green;">Ingest complete: {{ manifest_pid }}</h3>
     <li><a href="{{ volume_url }}">Link to volume</a></li>
   </ul>
 </p>
-{% if warnings is not None %}
+{% if html_warnings is not None %}
   <p>Warnings</p>
-  <ul>
-    {% for warning in warnings %}
-      <li>{{ warning }}</li>
+  <table>
+    {% for warning in html_warnings %}
+      <tr>
+        {% for part in warning %}
+          <td>{{ part }}</td>
+        {% endfor %}
+      </tr>
     {% endfor %}
-  </ul>
+  </table>
 {% endif %}
\ No newline at end of file
diff --git a/test_app/tests/test_ocr.py b/test_app/tests/test_ocr.py
index 8bb526a..0520f9e 100644
--- a/test_app/tests/test_ocr.py
+++ b/test_app/tests/test_ocr.py
@@ -38,6 +38,7 @@ def test_empty_xml(self):
         local.refresh_from_db()
         local.success()
         assert local.warnings.startswith(
-            f"Canvas {canvas.pid} - XMLSyntaxError: Document is empty, line 1, column 1 (<string>, line 1)$$$$"
+            f"Canvas {canvas.pid} - XMLSyntaxError: Document is empty, line 1, column 1 (<string>, line 1) | "
         )
         assert "XMLSyntaxError" in mail.outbox[0].body
+        assert "iip" in mail.outbox[0].body