Merge pull request #222 from rice-crc/develop

Merge develop into main

api/document/management/commands/iiif_generate_manifests.py

@@ -64,6 +64,8 @@ def handle(self, *args, **options):
 
 		manifest_sources=Source.objects.all().filter(~Q(page_connections__page=None))
 
+		print(manifest_sources)
+
 		#screen out sources that lack either pages  
 		sources = Source.objects \
 			.prefetch_related('page_connections') \
@@ -73,7 +75,8 @@ def handle(self, *args, **options):
 			.prefetch_related('source_enslaver_connections__enslaver') \
 			.prefetch_related('source_enslaved_connections__enslaved') \
 			.filter(
-				~Q(page_connections__page=None) and ~Q(manifest_content=None)
+				~Q(page_connections__page=None)
+# 				and ~Q(manifest_content=None)
 			)
 		print(f"Found {sources.count()} sources with page images.")
 
@@ -90,7 +93,9 @@ def handle(self, *args, **options):
 		generated_count=0
 		for source in sources:
 			with transaction.atomic():
+				print("--->",source.title)
 				content = source.manifest_content
+# 				print("------->",content)
 
 				#then do a final pass to ensure that we don't have "pages" without images
 				#some of those did sneak in during the process of indexing transkribus against the library collections
@@ -220,7 +225,10 @@ def handle(self, *args, **options):
 				# Append entity connections to metadata.
 				doc_links = {}
 
-				metadata = list(content['metadata'])
+				if content is not None:
+					metadata = list(content['metadata'])
+				else:
+					metadata=[]
 
 				#voyage ids
 				source_voyages=source.source_voyage_connections.all()

api/document/migrations/0006_page_is_british_library_page_transkribus_pageid_and_more.py

@@ -1,6 +1,7 @@
 # Generated by Django 4.2.1 on 2023-11-20 17:10
 
 from django.db import migrations, models
+import django.db.models.deletion
 
 
 class Migration(migrations.Migration):
@@ -10,6 +11,13 @@ class Migration(migrations.Migration):
     ]
 
     operations = [
+        migrations.CreateModel(
+            name='SourceType',
+            fields=[
+                ('id', models.BigAutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('name', models.CharField(max_length=255, unique=True)),
+            ],
+        ),
         migrations.AddField(
             model_name='page',
             name='is_british_library',
@@ -40,4 +48,14 @@ class Migration(migrations.Migration):
             name='zotero_grouplibrary_name',
             field=models.CharField(default='sv-docs', max_length=255),
         ),
+        migrations.CreateModel(
+            name='Transcription',
+            fields=[
+                ('id', models.BigAutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('language_code', models.CharField(max_length=20)),
+                ('text', models.TextField()),
+                ('is_translation', models.BooleanField()),
+                ('page', models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, related_name='transcriptions', to='document.page')),
+            ],
+        )
     ]

api/document/migrations/0007_remove_page_iiif_manifest_url_and_more.py

@@ -0,0 +1,68 @@
+# Generated by Django 4.2.1 on 2024-02-15 17:35
+
+import django.core.validators
+from django.db import migrations, models
+import django.db.models.deletion
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('document', '0006_page_is_british_library_page_transkribus_pageid_and_more'),
+    ]
+
+    operations = [
+        migrations.RemoveField(
+            model_name='page',
+            name='iiif_manifest_url',
+        ),
+        migrations.RemoveField(
+            model_name='page',
+            name='transcription',
+        ),
+        migrations.AddField(
+            model_name='source',
+            name='bib',
+            field=models.TextField(blank=True, help_text='Formatted bibliography for the Document', null=True),
+        ),
+        migrations.AddField(
+            model_name='source',
+            name='manifest_content',
+            field=models.JSONField(blank=True, help_text='DCTerms imported from Zotero -- NOT the full manifest', null=True),
+        ),
+        migrations.AddField(
+            model_name='source',
+            name='source_type',
+            field=models.ForeignKey(blank=True, null=True, on_delete=django.db.models.deletion.CASCADE, related_name='+', to='document.sourcetype'),
+        ),
+        migrations.AddField(
+            model_name='source',
+            name='thumbnail',
+            field=models.TextField(blank=True, help_text='URL for a thumbnail of the Document', null=True),
+        ),
+        migrations.AddField(
+            model_name='source',
+            name='zotero_url',
+            field=models.URLField(blank=True, max_length=400, null=True),
+        ),
+        migrations.AlterField(
+            model_name='docsparsedate',
+            name='year',
+            field=models.IntegerField(blank=True, null=True, validators=[django.core.validators.MinValueValidator(0), django.core.validators.MaxValueValidator(2050)]),
+        ),
+        migrations.AlterField(
+            model_name='source',
+            name='has_published_manifest',
+            field=models.BooleanField(default=False, verbose_name='Is there a published manifest?'),
+        ),
+        migrations.AlterField(
+            model_name='source',
+            name='item_url',
+            field=models.URLField(blank=True, max_length=400, null=True),
+        ),
+        migrations.AlterField(
+            model_name='source',
+            name='title',
+            field=models.CharField(max_length=1000, verbose_name='Title'),
+        ),
+    ]