Fix bulk_transcriptions

arkindex/images/importer.py

@@ -8,6 +8,7 @@ import re
 import gzip
 import logging
 import fnmatch
+import uuid
 
 REGEX_INDEX = re.compile(
     b'^(?:line_(\d+) )?(.+) \d+ ([\de\-\.]+) (\d+) (\d+) (\d+) (\d+)')
@@ -93,13 +94,12 @@ def bulk_transcriptions(image, page, items):
     # Build all TrBox from existing
     existing = {
         TrBox(
-            BoundingBox(zone.polygon),
+            BoundingBox(tr.zone.polygon),
             tr.line,
             tr.text,
             tr.score,
         )
-        for tr in Transcription.objects.filter(zones__image=image).prefetch_related('zones')
-        for zone in tr.zones.all()
+        for tr in Transcription.objects.filter(zone__image=image).prefetch_related('zone')
     }
 
     # Calc needed TrBox to build
@@ -111,7 +111,7 @@ def bulk_transcriptions(image, page, items):
 
         # Raw elements
         elements = Element.objects.bulk_create(
-            Element(type=ElementType.Transcription)
+            Element(type=ElementType.Transcription, zone_id=uuid.uuid4())
             for _ in needed
         )
 
@@ -125,7 +125,7 @@ def bulk_transcriptions(image, page, items):
                     score=n.score,
                 ),
                 Zone(
-                    element_id=elt.id,
+                    id=elt.zone_id,
                     image=image,
                     polygon=n.box.to_polygon(),
                 )
@@ -133,6 +133,9 @@ def bulk_transcriptions(image, page, items):
             for elt, n in zip(elements, needed)
         ])
 
+        # Create zones in bulk
+        Zone.objects.bulk_create(zones)
+
         # Create transcriptions using a low-level bulk_create
         # as multi table is not supported yet by Django
         Transcription.objects.none()._batched_insert(
@@ -145,9 +148,6 @@ def bulk_transcriptions(image, page, items):
             batch_size=None,
         )
 
-        # Create zones in bulk
-        Zone.objects.bulk_create(zones)
-
     # Create all links between transcription and page
     max_order_dl = ElementLink.objects.filter(parent=page).order_by('-order').first()
     max_order = 0 if max_order_dl is None else max_order_dl.order + 1