Support element paths in bulk_transcriptions

arkindex/images/importer.py

 from arkindex.images.models import Zone, Image
-from arkindex.documents.models import Transcription, ElementLink, ElementType, Element, Page
+from arkindex.documents.models import Transcription, ElementLink, ElementType, Element, Page, ElementPath
 from arkindex.project.tools import BoundingBox
 from collections import namedtuple
 from abc import ABC, abstractmethod
@@ -60,10 +60,12 @@ def import_indexes(image, page, index_path, extension='jpg'):
     logger.info('Created {} transcriptions '.format(len(new_transcriptions)))
 
 
-def bulk_transcriptions(image, page, items):
+def bulk_transcriptions(image, parent, items):
     """
     Create transcriptions and zones in bulk
     """
+    assert isinstance(parent, Element)
+
     # Link a transcription data with a bounding box
     # This is hashable (box is hashable)
     TrBox = namedtuple('TrBox', 'box, line, text, score')
@@ -109,7 +111,7 @@ def bulk_transcriptions(image, page, items):
         # Raw elements
         elements = Element.objects.bulk_create(
             Element(
-                corpus=page.corpus,
+                corpus=parent.corpus,
                 type=ElementType.Word,
                 name=n.text,
                 zone_id=uuid.uuid4()
@@ -151,13 +153,23 @@ def bulk_transcriptions(image, page, items):
         )
 
     # Create all links between transcription and page
-    max_order_dl = ElementLink.objects.filter(parent=page).order_by('-order').first()
+    max_order_dl = ElementLink.objects.filter(parent=parent).order_by('-order').first()
     max_order = 0 if max_order_dl is None else max_order_dl.order + 1
     ElementLink.objects.bulk_create(
-        ElementLink(parent=page, child=elt, order=i)
+        ElementLink(parent=parent, child=elt, order=i)
         for i, elt in enumerate(elements, max_order)
     )
 
+    # Support ElementPath
+    paths = ElementPath.objects.filter(element=parent).values_list('path', flat=True)
+    if not paths:
+        paths = [[]] # Wonderful hack to handle no parents case
+    ElementPath.objects.bulk_create(
+        ElementPath(element=elt, path=[parent.id, ] + path)
+        for elt in elements
+        for path in paths
+    )
+
     return transcriptions
 
 

arkindex/images/tests.py

 from django.test import TestCase
-from arkindex.documents.models import Corpus, Page, Transcription
+from arkindex.documents.models import Corpus, Page, Transcription, Element, ElementType
 from arkindex.images.models import ImageServer, Image, Zone
 from arkindex.images.importer import bulk_transcriptions
 
@@ -57,6 +57,13 @@ class TestBulkTranscriptions(TestCase):
         self.assertEqual(out[0].zone.polygon, [(0, 0), (0, 100), (100, 100), (100, 0)])
         self.assertEqual(out[1].zone.polygon, [(20, 20), (20, 120), (120, 120), (120, 20)])
 
+        # Check path
+        children = Element.objects.get_descending(self.page.id)
+        self.assertEqual(children.count(), 2)
+        ids = children.values_list('id', flat=True)
+        self.assertIn(out[0].id, ids)
+        self.assertIn(out[1].id, ids)
+
     def test_bulk_transcriptions_unique(self):
         """Check bulk_transcriptions does not import the same transcriptions twice"""
         items = [