diff --git a/arkindex/documents/admin.py b/arkindex/documents/admin.py
index 2a00a591715a6d7eee001b4d9afac889d73afbb7..7a42a6b9f39ad17c4c081f95a2276f2049144295 100644
--- a/arkindex/documents/admin.py
+++ b/arkindex/documents/admin.py
@@ -74,9 +74,9 @@ class ElementAdmin(admin.ModelAdmin):
class TranscriptionAdmin(admin.ModelAdmin):
list_display = ('id', 'text', 'score', 'element', )
list_filter = [('type', EnumFieldListFilter), 'source']
- fields = ('id', 'text', 'score', 'element', 'zone', 'source', )
+ fields = ('id', 'text', 'score', 'element', 'source', )
readonly_fields = ('id', )
- raw_id_fields = ('element', 'zone', )
+ raw_id_fields = ('element', )
class MLClassAdmin(admin.ModelAdmin):
diff --git a/arkindex/documents/api/elements.py b/arkindex/documents/api/elements.py
index 547b88756653ee5f2ab4988de5bf2bb6dcdca45f..f1eb993b517acb4c5e8195e2ed5a30abe5f65f9f 100644
--- a/arkindex/documents/api/elements.py
+++ b/arkindex/documents/api/elements.py
@@ -3,7 +3,8 @@ from datetime import datetime, timezone
from psycopg2.extras import execute_values
from django.conf import settings
from django.db import transaction, connection
-from django.db.models import Q, Prefetch, Max, QuerySet
+from django.db.models import Q, Prefetch, Max, QuerySet, CharField
+from django.db.models.functions import Cast
from django.shortcuts import get_object_or_404
from django.utils.functional import cached_property
from rest_framework.exceptions import ValidationError, NotFound
@@ -790,13 +791,11 @@ class ElementTranscriptions(ListAPIView):
))
self.check_object_permissions(self.request, element)
+ # ORDER BY casting IDs as char to avoid the PostgreSQL optimizer's inefficient scan
queryset = Transcription.objects \
- .prefetch_related('zone__image__server', 'source') \
- .extra(
- # ORDER BY casting IDs as char to avoid PostgreSQL optimizer inefficient scan
- select={'char_id': 'CAST(id AS CHAR(36))'},
- order_by=['char_id']
- )
+ .prefetch_related('element__zone__image__server', 'source') \
+ .annotate(char_id=Cast('id', output_field=CharField())) \
+ .order_by('char_id')
if self.is_recursive:
queryset = queryset.filter(
diff --git a/arkindex/documents/api/entities.py b/arkindex/documents/api/entities.py
index cc69925745f4afda656150789830e793b38a9acb..8a5f54a13563fd210fe352683508988ccb2a7b67 100644
--- a/arkindex/documents/api/entities.py
+++ b/arkindex/documents/api/entities.py
@@ -109,14 +109,14 @@ class EntityElements(ListAPIView):
corpus__in=Corpus.objects.readable(self.request.user),
metadatas__entity_id=pk
) \
- .select_related('type') \
- .prefetch_related('metadatas__entity', 'metadatas__revision', 'corpus')
+ .select_related('type', 'corpus') \
+ .prefetch_related('metadatas__entity', 'metadatas__revision', 'zone__image__server')
transcription_elements = Element.objects \
.filter(
corpus__in=Corpus.objects.readable(self.request.user),
transcriptions__transcription_entities__entity_id=pk
- ).select_related('type') \
- .prefetch_related('metadatas__entity', 'metadatas__revision', 'corpus')
+ ).select_related('type', 'corpus') \
+ .prefetch_related('metadatas__entity', 'metadatas__revision', 'zone__image__server')
return metadata_elements.union(transcription_elements) \
.order_by('name', 'type')
diff --git a/arkindex/documents/api/ml.py b/arkindex/documents/api/ml.py
index 885372f661949bacee4b5235f7c858acd7c3f7c8..544ec623a5dd1adf101280b95734e3e06032f180 100644
--- a/arkindex/documents/api/ml.py
+++ b/arkindex/documents/api/ml.py
@@ -278,7 +278,6 @@ class ElementTranscriptionsBulk(CreateAPIView):
transcriptions.append(Transcription(
element=annotation['element'],
type=tr_type,
- zone=None,
worker_version=worker_version,
text=annotation['text'],
score=annotation['score']
diff --git a/arkindex/documents/consumers.py b/arkindex/documents/consumers.py
index 5632e2411d11b3e59dc1a19fadda45782117662c..d41888420276a0d66417b25707f324f7992bc492 100644
--- a/arkindex/documents/consumers.py
+++ b/arkindex/documents/consumers.py
@@ -75,29 +75,28 @@ class ReindexConsumer(SyncConsumer):
elif element_id or corpus_id:
if element_id:
# Pick this element, and all its children
- elements_queryset = list(Element.objects.get_descending(element_id))
- elements_queryset.append(Element.objects.get(id=element_id))
+ elements_queryset = Element.objects.filter(Q(id=element_id) | Q(paths__path__contains=[element_id]))
else:
# Pick all elements in the corpus
elements_queryset = Element.objects.filter(corpus_id=corpus_id)
- transcriptions_queryset = Transcription.objects.filter(
- element__in=elements_queryset,
- zone__isnull=False
- )
+ transcriptions_queryset = Transcription.objects.filter(element__in=elements_queryset)
entities_queryset = Entity.objects.filter(
Q(metadatas__element__in=elements_queryset)
| Q(transcriptions__element__in=elements_queryset)
)
else:
- transcriptions_queryset = Transcription.objects.filter(zone__isnull=False)
+ transcriptions_queryset = Transcription.objects.all()
elements_queryset = Element.objects.all()
entities_queryset = Entity.objects.all()
if transcriptions:
- indexer.run_index(transcriptions_queryset, bulk_size=400)
+ indexer.run_index(transcriptions_queryset.select_related('element'), bulk_size=400)
if elements:
- indexer.run_index(elements_queryset, bulk_size=100)
+ indexer.run_index(
+ elements_queryset.select_related('type').prefetch_related('metadatas', 'transcriptions'),
+ bulk_size=100,
+ )
if entities:
indexer.run_index(entities_queryset, bulk_size=400)
diff --git a/arkindex/documents/fixtures/data.json b/arkindex/documents/fixtures/data.json
index 0cdc384ccf43c0737888e344c1ac841e55aba6ae..43e9debed534a5920a0c161adca0b0c8e538e6e2 100644
--- a/arkindex/documents/fixtures/data.json
+++ b/arkindex/documents/fixtures/data.json
@@ -1,148 +1,159 @@
[
{
"model": "dataimport.repository",
- "pk": "d0cd8a85-8f30-43e1-ba19-7bd0dacb2041",
+ "pk": "1e022c37-4a8e-4624-8fbe-6c8a4d5b22a2",
"fields": {
"url": "http://gitlab/repo",
"type": "iiif",
"hook_token": "hook-token",
- "credentials": "d21e40e2-32d2-4903-ac4a-6672dc6c14ba",
+ "credentials": "b07d0cb3-65b7-45c4-9bf2-d7153f4d27cd",
"provider_name": "GitLabProvider"
}
},
{
"model": "dataimport.repository",
- "pk": "f020dbad-f6ba-4d95-9bba-35dc80272943",
+ "pk": "b6e463c3-0631-425f-8702-da7f32166b21",
"fields": {
"url": "http://my_repo.fake/workers/worker",
"type": "worker",
"hook_token": "worker-hook-token",
- "credentials": "d21e40e2-32d2-4903-ac4a-6672dc6c14ba",
+ "credentials": "b07d0cb3-65b7-45c4-9bf2-d7153f4d27cd",
"provider_name": "GitLabProvider"
}
},
{
"model": "dataimport.revision",
- "pk": "3ddfb063-eff0-4a9a-8b0f-000c0eca0f7a",
+ "pk": "bc9f203f-e870-4f79-96ce-4c0f3211691e",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "repo": "d0cd8a85-8f30-43e1-ba19-7bd0dacb2041",
- "hash": "42",
- "message": "a",
- "author": "me"
+ "repo": "b6e463c3-0631-425f-8702-da7f32166b21",
+ "hash": "1337",
+ "message": "My w0rk3r",
+ "author": "Test user"
}
},
{
"model": "dataimport.revision",
- "pk": "796847cd-9463-4496-a102-c1a5f825e50d",
+ "pk": "f353f467-9fb4-4f78-b962-03a8cd10cd85",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "repo": "f020dbad-f6ba-4d95-9bba-35dc80272943",
- "hash": "1337",
- "message": "My w0rk3r",
- "author": "Test user"
+ "repo": "1e022c37-4a8e-4624-8fbe-6c8a4d5b22a2",
+ "hash": "42",
+ "message": "a",
+ "author": "me"
}
},
{
"model": "dataimport.worker",
- "pk": "29a773f3-d140-48e3-b1e4-29325f9dbc6f",
+ "pk": "72cd3c6b-be61-439e-ac00-da0440a6dfab",
"fields": {
"name": "Recognizer",
"slug": "reco",
"type": "recognizer",
- "repository": "f020dbad-f6ba-4d95-9bba-35dc80272943"
+ "repository": "b6e463c3-0631-425f-8702-da7f32166b21"
}
},
{
"model": "dataimport.worker",
- "pk": "cf4243ca-5f42-4dbd-a2fe-14a8e75a33d8",
+ "pk": "dd3e3126-f07d-47d5-80cc-f9439aca0ad0",
"fields": {
"name": "Document layout analyser",
"slug": "dla",
"type": "dla",
- "repository": "f020dbad-f6ba-4d95-9bba-35dc80272943"
+ "repository": "b6e463c3-0631-425f-8702-da7f32166b21"
}
},
{
"model": "dataimport.workerversion",
- "pk": "8d2ca1e7-f4f9-4403-8569-20e9712a897a",
+ "pk": "060456c0-c109-4f05-b78d-4dbe13995d05",
"fields": {
- "worker": "29a773f3-d140-48e3-b1e4-29325f9dbc6f",
- "revision": "796847cd-9463-4496-a102-c1a5f825e50d",
+ "worker": "72cd3c6b-be61-439e-ac00-da0440a6dfab",
+ "revision": "bc9f203f-e870-4f79-96ce-4c0f3211691e",
"configuration": {
"test": 42
},
"state": "available",
- "docker_image": "71a4b60b-a2f2-4503-8f5d-9d3c38f66626",
+ "docker_image": "2f4c9981-6ee0-43a1-80f0-54cca23be3ff",
"docker_image_iid": null
}
},
{
"model": "dataimport.workerversion",
- "pk": "afa4a177-b8d2-4bcf-aaa7-b00196807e1d",
+ "pk": "67be8557-e683-43c5-b9d7-1b57b4516e09",
"fields": {
- "worker": "cf4243ca-5f42-4dbd-a2fe-14a8e75a33d8",
- "revision": "796847cd-9463-4496-a102-c1a5f825e50d",
+ "worker": "dd3e3126-f07d-47d5-80cc-f9439aca0ad0",
+ "revision": "bc9f203f-e870-4f79-96ce-4c0f3211691e",
"configuration": {
"test": 42
},
"state": "available",
- "docker_image": "71a4b60b-a2f2-4503-8f5d-9d3c38f66626",
+ "docker_image": "2f4c9981-6ee0-43a1-80f0-54cca23be3ff",
"docker_image_iid": null
}
},
{
"model": "documents.corpus",
- "pk": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
+ "pk": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
"name": "Unit Tests",
"description": "",
- "repository": "d0cd8a85-8f30-43e1-ba19-7bd0dacb2041",
+ "repository": "1e022c37-4a8e-4624-8fbe-6c8a4d5b22a2",
"public": true
}
},
{
"model": "documents.elementtype",
- "pk": "33d93a96-58a9-4a7c-be54-12d4691f4e81",
+ "pk": "1e98cd23-9b0b-44ed-84b5-3a9e4dec68b3",
"fields": {
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "slug": "text_line",
- "display_name": "Line",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "slug": "volume",
+ "display_name": "Volume",
+ "folder": true,
+ "allowed_transcription": null
+ }
+},
+{
+ "model": "documents.elementtype",
+ "pk": "6acab4ca-a475-485e-b9cf-fc8a8404bcd1",
+ "fields": {
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "slug": "word",
+ "display_name": "Word",
"folder": false,
- "allowed_transcription": "line"
+ "allowed_transcription": "word"
}
},
{
"model": "documents.elementtype",
- "pk": "3bd6b808-4ced-41f5-8ae0-bf00804fcbaa",
+ "pk": "6bc2cbbd-b98e-4689-bf9d-8a9095710632",
"fields": {
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "slug": "volume",
- "display_name": "Volume",
- "folder": true,
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "slug": "act",
+ "display_name": "Act",
+ "folder": false,
"allowed_transcription": null
}
},
{
"model": "documents.elementtype",
- "pk": "a8c815f0-258f-4025-8f9b-f3bd6a87904c",
+ "pk": "77739a6b-f044-4a5e-9c19-bc881caafbbf",
"fields": {
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "slug": "surface",
- "display_name": "Surface",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "slug": "text_line",
+ "display_name": "Line",
"folder": false,
- "allowed_transcription": null
+ "allowed_transcription": "line"
}
},
{
"model": "documents.elementtype",
- "pk": "aad1941d-d7dd-485f-8cf3-90487ac522c8",
+ "pk": "8ccc9061-0bc7-4d74-8a83-d427273a8c56",
"fields": {
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
"slug": "page",
"display_name": "Page",
"folder": false,
@@ -151,284 +162,435 @@
},
{
"model": "documents.elementtype",
- "pk": "bff45072-bc1b-408d-bd1f-b3ffd22d8796",
+ "pk": "f56fb6dd-afe4-49b3-a870-963467404935",
"fields": {
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "slug": "act",
- "display_name": "Act",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "slug": "surface",
+ "display_name": "Surface",
"folder": false,
"allowed_transcription": null
}
},
{
"model": "documents.elementpath",
- "pk": "0355c4b6-bb16-474e-9b7b-bb86077339f3",
+ "pk": "06d9848e-c2cd-4f1b-9d07-00ed04f69b9e",
"fields": {
- "element": "f3981e96-ed95-43a0-b1ea-fd288c55f5b3",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\", \"ccc2cfdb-2de5-4e42-8a1a-fc5e1041b21d\"]",
- "ordering": 0
+ "element": "bbd63fcc-23af-4593-bcb5-341c3008b7ab",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"35fb048c-f441-4e9d-bb69-78a8dfdb4598\"]",
+ "ordering": 2
+ }
+},
+{
+ "model": "documents.elementpath",
+ "pk": "07d8a649-9b5c-4338-9e42-20a3fc9c884e",
+ "fields": {
+ "element": "35fb048c-f441-4e9d-bb69-78a8dfdb4598",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\"]",
+ "ordering": 1
}
},
{
"model": "documents.elementpath",
- "pk": "0406c678-c1b1-41b2-a4bd-d85e83ccd230",
+ "pk": "0abd85c5-7d97-49ca-9b60-b67e7e7d215e",
"fields": {
- "element": "ccc2cfdb-2de5-4e42-8a1a-fc5e1041b21d",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\"]",
+ "element": "7945d738-4a6e-4a76-ad57-21c03d63625d",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\"]",
"ordering": 2
}
},
{
"model": "documents.elementpath",
- "pk": "0f73c6dc-07b3-494a-b2fd-692c9a19ce5a",
+ "pk": "1aefc8c1-1dba-4f8e-b2eb-b7b7c5bc0b01",
"fields": {
- "element": "9621c876-e7da-4d43-a5a7-82470dead663",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\"]",
+ "element": "7e788454-afed-4bef-a370-bccf0dd3b2b4",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\"]",
"ordering": 0
}
},
{
"model": "documents.elementpath",
- "pk": "208dd6c4-5510-4ceb-b8b4-fef27680df72",
+ "pk": "1f1a82a0-7835-465b-ae6d-d6db65e92696",
"fields": {
- "element": "cb69d995-2ce2-4404-a567-5598cf9dd231",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\"]",
+ "element": "ebdb0c60-e1e4-4f8a-a8b7-4127896ac194",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\"]",
"ordering": 1
}
},
{
"model": "documents.elementpath",
- "pk": "269ebcc7-7b11-43c9-a6cd-94d1ee99b03a",
+ "pk": "2f8c2139-9b01-415c-8a80-b708a7fd841d",
"fields": {
- "element": "dbfa2c38-8664-447a-bb09-215cc86c6ab4",
- "path": "[\"209e1e3b-5808-43bf-99b7-801959d6c807\"]",
- "ordering": 2
+ "element": "326843be-0112-4954-a2ea-da78aded01a9",
+ "path": "[\"07422db5-10c5-404b-9325-db2689871761\"]",
+ "ordering": 0
}
},
{
"model": "documents.elementpath",
- "pk": "287575b8-9b19-49bd-adab-a241385191f4",
+ "pk": "4beb2166-9bc1-462f-bbbb-34cb43ceeb8a",
"fields": {
- "element": "f5535f84-2ca7-4853-9ab6-44606a8dc462",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\", \"9621c876-e7da-4d43-a5a7-82470dead663\"]",
+ "element": "b9f39256-6ff0-4daf-b56a-a5a316964133",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"35fb048c-f441-4e9d-bb69-78a8dfdb4598\"]",
"ordering": 0
}
},
{
"model": "documents.elementpath",
- "pk": "36cc34f3-8d0b-43cb-9040-7169471a03d4",
+ "pk": "4e98abb3-e74c-4532-a831-11b41c00f2d5",
+ "fields": {
+ "element": "96893993-5b8d-4ed0-b305-08f745d8df01",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"35fb048c-f441-4e9d-bb69-78a8dfdb4598\"]",
+ "ordering": 1
+ }
+},
+{
+ "model": "documents.elementpath",
+ "pk": "4ed309a6-3af4-4fe4-975d-0a25ecd48bcf",
+ "fields": {
+ "element": "c4ccff42-93d3-4f66-a073-397dd85e1806",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\"]",
+ "ordering": 4
+ }
+},
+{
+ "model": "documents.elementpath",
+ "pk": "5980b951-7bcc-4213-a73b-e71659e8a8e1",
"fields": {
- "element": "564aa56f-b257-445c-9930-4b6110628c09",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\"]",
+ "element": "5130da8e-f2aa-477d-bf58-a2357efc8b2b",
+ "path": "[\"07422db5-10c5-404b-9325-db2689871761\"]",
"ordering": 2
}
},
{
"model": "documents.elementpath",
- "pk": "38df3e8b-d824-4862-b7b5-580065b2011b",
+ "pk": "7680a018-4523-4abc-8242-132958dd7f5a",
+ "fields": {
+ "element": "fcf303df-36a5-40f2-997f-82eb47541bcf",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"99de88f1-9655-4bae-9203-01d56d097ef7\"]",
+ "ordering": 1
+ }
+},
+{
+ "model": "documents.elementpath",
+ "pk": "7de05621-65f2-4dcb-8f44-9a88fdfb3a22",
"fields": {
- "element": "c0ec4456-f6a2-4253-a876-6eba4f96f863",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\"]",
+ "element": "cbafeb68-d66a-4e1c-bcd4-47dd88b02eec",
+ "path": "[\"07422db5-10c5-404b-9325-db2689871761\"]",
"ordering": 1
}
},
{
"model": "documents.elementpath",
- "pk": "3a6d0696-1ca2-4f83-a6dc-25a859482ebc",
+ "pk": "a3e6f951-64da-474b-bd61-d323ad71d5c9",
"fields": {
- "element": "8f92f79c-471a-4280-9530-66ec406ab8b2",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\"]",
+ "element": "9f81a475-ed1f-470d-b11a-9740d6bb24d2",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"99de88f1-9655-4bae-9203-01d56d097ef7\"]",
"ordering": 0
}
},
{
"model": "documents.elementpath",
- "pk": "582355b4-4352-4277-b6dd-62c6a8dfbf88",
+ "pk": "a7ad8b75-e9fd-44f9-acec-a283b90d9585",
"fields": {
- "element": "d7b6b144-88cb-4c47-a380-a84339773dda",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\", \"cb69d995-2ce2-4404-a567-5598cf9dd231\"]",
+ "element": "73b56c40-4136-499b-9c10-82c2f569c78b",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"99de88f1-9655-4bae-9203-01d56d097ef7\"]",
"ordering": 0
}
},
{
"model": "documents.elementpath",
- "pk": "7abb9de4-a62f-49e4-979f-8975fe2e1111",
+ "pk": "a9aec95c-fa0f-4b0d-b3ab-7e061481b7bb",
"fields": {
- "element": "02d0fa30-2a2d-4972-a069-626c629357a4",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\", \"d19ded5f-c50c-4cc6-a0cf-4463e2f2df68\"]",
+ "element": "1fcb137d-12aa-47ea-8d94-fffd821cc3da",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"b675ba10-f6ed-45cb-8a8c-449940471db5\"]",
"ordering": 0
}
},
{
"model": "documents.elementpath",
- "pk": "a30e4bb0-b337-4718-a46b-53fc390393d0",
+ "pk": "aa5478d1-be7c-48c2-90cf-ae0294981ad9",
"fields": {
- "element": "7c022411-eb2c-4f06-9387-5d8418a0bf33",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\"]",
- "ordering": 3
+ "element": "4d5191a9-5161-4587-a6bf-6430786f869a",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"99de88f1-9655-4bae-9203-01d56d097ef7\"]",
+ "ordering": 2
}
},
{
"model": "documents.elementpath",
- "pk": "a47f8cec-6c57-4d9d-a7f8-0ab9ab56ab44",
+ "pk": "ae105ab9-b082-44e3-8dd9-eb14c65f83a1",
"fields": {
- "element": "b09ebead-3440-4040-8b82-5dc2bae4d2d9",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\", \"8f92f79c-471a-4280-9530-66ec406ab8b2\"]",
+ "element": "cb1b2af4-9cd8-4ecb-99a3-a7f309198a7b",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"7e788454-afed-4bef-a370-bccf0dd3b2b4\"]",
"ordering": 0
}
},
{
"model": "documents.elementpath",
- "pk": "b1b9f9f3-f543-4c57-aa49-a0aa716f1b71",
+ "pk": "b5b1d6ea-c2ee-4c91-8f90-be745e05559d",
+ "fields": {
+ "element": "9644857d-6c33-4241-ad43-38054fba11a0",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"ebdb0c60-e1e4-4f8a-a8b7-4127896ac194\"]",
+ "ordering": 1
+ }
+},
+{
+ "model": "documents.elementpath",
+ "pk": "b5c3893e-2127-4a09-9325-8545962f325c",
"fields": {
- "element": "7e4d2f71-f52a-44a6-8cb5-f209a0b82408",
- "path": "[\"209e1e3b-5808-43bf-99b7-801959d6c807\"]",
+ "element": "ce358ae1-5d50-476e-b3bb-2f5abb3a9ce9",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"ebdb0c60-e1e4-4f8a-a8b7-4127896ac194\"]",
"ordering": 0
}
},
{
"model": "documents.elementpath",
- "pk": "bc93dbca-6436-4601-9ecd-0ab08972b309",
+ "pk": "c048eef3-0e7b-4f6a-836c-bd0cae40b509",
"fields": {
- "element": "51585cf0-35b6-4b13-a445-37bed0aae13b",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\", \"7c022411-eb2c-4f06-9387-5d8418a0bf33\"]",
+ "element": "72f64b68-afc4-41e0-8020-e590bdaa9f7c",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"7945d738-4a6e-4a76-ad57-21c03d63625d\"]",
+ "ordering": 1
+ }
+},
+{
+ "model": "documents.elementpath",
+ "pk": "d203709c-b528-4aa6-be17-56795c1d03fb",
+ "fields": {
+ "element": "5f20bff4-aa8b-4afa-ad96-e791c2433dc1",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"7945d738-4a6e-4a76-ad57-21c03d63625d\"]",
+ "ordering": 2
+ }
+},
+{
+ "model": "documents.elementpath",
+ "pk": "dff4ed00-697a-4e42-aff3-5d09f8f7785e",
+ "fields": {
+ "element": "99de88f1-9655-4bae-9203-01d56d097ef7",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\"]",
"ordering": 0
}
},
{
"model": "documents.elementpath",
- "pk": "d47680ce-4ab8-4eb4-8cbe-cf06373ed654",
+ "pk": "e456aee3-8cf4-45e3-83f4-b64e61733b9c",
"fields": {
- "element": "d19ded5f-c50c-4cc6-a0cf-4463e2f2df68",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\"]",
- "ordering": 4
+ "element": "3442efb5-4960-432e-b6f9-795a415186d8",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"c4ccff42-93d3-4f66-a073-397dd85e1806\"]",
+ "ordering": 0
}
},
{
"model": "documents.elementpath",
- "pk": "f593ed74-c825-44fa-a7da-a0d9776f2e0b",
+ "pk": "f35b7715-2102-4827-9589-57c1b68405b9",
"fields": {
- "element": "56a2302d-e027-4724-85be-c4113ff2a243",
- "path": "[\"3c4bce04-3ae8-4f17-86b0-46bc25408b8f\", \"cb69d995-2ce2-4404-a567-5598cf9dd231\"]",
- "ordering": 1
+ "element": "a493ea73-d17d-4f63-b6c9-1ebe04f36641",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"c53fc91f-58c7-4fda-a92c-72a5f4051176\"]",
+ "ordering": 0
}
},
{
"model": "documents.elementpath",
- "pk": "fc3dfb9d-1763-4892-95a9-1c8adeb8f41c",
+ "pk": "f4affadd-0195-41da-80c5-d36e0a6b064c",
"fields": {
- "element": "4e9de67b-6e4f-46b7-936e-34ed9692900b",
- "path": "[\"209e1e3b-5808-43bf-99b7-801959d6c807\"]",
- "ordering": 1
+ "element": "775da282-9bef-4dcd-b1fa-0a7ff555cb9d",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\", \"7945d738-4a6e-4a76-ad57-21c03d63625d\"]",
+ "ordering": 0
+ }
+},
+{
+ "model": "documents.elementpath",
+ "pk": "f95a476b-a0b2-484d-ae3f-b720095c3f37",
+ "fields": {
+ "element": "b675ba10-f6ed-45cb-8a8c-449940471db5",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\"]",
+ "ordering": 3
+ }
+},
+{
+ "model": "documents.elementpath",
+ "pk": "fc030c8c-7dce-477f-913b-520a3ae6e545",
+ "fields": {
+ "element": "c53fc91f-58c7-4fda-a92c-72a5f4051176",
+ "path": "[\"e001d1ee-fa90-4f96-b722-e2ba20b84097\"]",
+ "ordering": 2
+ }
+},
+{
+ "model": "documents.element",
+ "pk": "07422db5-10c5-404b-9325-db2689871761",
+ "fields": {
+ "created": "2020-02-02T01:23:45.678Z",
+ "updated": "2020-02-02T01:23:45.678Z",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "1e98cd23-9b0b-44ed-84b5-3a9e4dec68b3",
+ "name": "Volume 2",
+ "zone": null,
+ "source": null,
+ "worker_version": null
+ }
+},
+{
+ "model": "documents.element",
+ "pk": "1fcb137d-12aa-47ea-8d94-fffd821cc3da",
+ "fields": {
+ "created": "2020-02-02T01:23:45.678Z",
+ "updated": "2020-02-02T01:23:45.678Z",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "f56fb6dd-afe4-49b3-a870-963467404935",
+ "name": "Surface E",
+ "zone": "440f85ff-25d9-4488-8654-98c5916df96f",
+ "source": null,
+ "worker_version": null
+ }
+},
+{
+ "model": "documents.element",
+ "pk": "326843be-0112-4954-a2ea-da78aded01a9",
+ "fields": {
+ "created": "2020-02-02T01:23:45.678Z",
+ "updated": "2020-02-02T01:23:45.678Z",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "8ccc9061-0bc7-4d74-8a83-d427273a8c56",
+ "name": "Volume 2, page 1r",
+ "zone": "aefbd01b-3421-458f-a8e6-2b5ffdaccc31",
+ "source": null,
+ "worker_version": null
}
},
{
"model": "documents.element",
- "pk": "02d0fa30-2a2d-4972-a069-626c629357a4",
+ "pk": "3442efb5-4960-432e-b6f9-795a415186d8",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "a8c815f0-258f-4025-8f9b-f3bd6a87904c",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "f56fb6dd-afe4-49b3-a870-963467404935",
"name": "Surface F",
- "zone": "d2e7ee12-14fa-46c8-a360-f5b31939052a",
+ "zone": "b70d7b27-c973-44ba-93c5-602957ae491e",
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "209e1e3b-5808-43bf-99b7-801959d6c807",
+ "pk": "35fb048c-f441-4e9d-bb69-78a8dfdb4598",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "3bd6b808-4ced-41f5-8ae0-bf00804fcbaa",
- "name": "Volume 2",
- "zone": null,
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "8ccc9061-0bc7-4d74-8a83-d427273a8c56",
+ "name": "Volume 1, page 1v",
+ "zone": "0fb66d83-3d13-4d4e-82d1-1fd7fc82036f",
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "3c4bce04-3ae8-4f17-86b0-46bc25408b8f",
+ "pk": "4d5191a9-5161-4587-a6bf-6430786f869a",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "3bd6b808-4ced-41f5-8ae0-bf00804fcbaa",
- "name": "Volume 1",
- "zone": null,
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "6acab4ca-a475-485e-b9cf-fc8a8404bcd1",
+ "name": "DATUM",
+ "zone": "40edd01e-02e6-4fbf-b2c6-b7452a72c79e",
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "4e9de67b-6e4f-46b7-936e-34ed9692900b",
+ "pk": "5130da8e-f2aa-477d-bf58-a2357efc8b2b",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "aad1941d-d7dd-485f-8cf3-90487ac522c8",
- "name": "Volume 2, page 1v",
- "zone": "e67eba1f-4993-492e-9ca0-67451ceb094c",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "8ccc9061-0bc7-4d74-8a83-d427273a8c56",
+ "name": "Volume 2, page 2r",
+ "zone": "11bb6bba-8a2e-4f9e-97b1-0f48bc34103f",
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "51585cf0-35b6-4b13-a445-37bed0aae13b",
+ "pk": "5f20bff4-aa8b-4afa-ad96-e791c2433dc1",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "a8c815f0-258f-4025-8f9b-f3bd6a87904c",
- "name": "Surface E",
- "zone": "f3b62073-8fe4-46c9-b7bf-012c3014359b",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "6acab4ca-a475-485e-b9cf-fc8a8404bcd1",
+ "name": "DATUM",
+ "zone": "7e767702-f644-451f-91e8-05423bb8e074",
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "564aa56f-b257-445c-9930-4b6110628c09",
+ "pk": "72f64b68-afc4-41e0-8020-e590bdaa9f7c",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "aad1941d-d7dd-485f-8cf3-90487ac522c8",
- "name": "Volume 1, page 2r",
- "zone": "20f0abb7-3aac-460b-b1af-b75ed7dab550",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "6acab4ca-a475-485e-b9cf-fc8a8404bcd1",
+ "name": "ROY",
+ "zone": "0039b5f5-078f-4667-825c-6e620b92b59a",
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "56a2302d-e027-4724-85be-c4113ff2a243",
+ "pk": "73b56c40-4136-499b-9c10-82c2f569c78b",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "a8c815f0-258f-4025-8f9b-f3bd6a87904c",
- "name": "Surface C",
- "zone": "dda78215-f30b-497f-aedb-fac49055b667",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "6acab4ca-a475-485e-b9cf-fc8a8404bcd1",
+ "name": "PARIS",
+ "zone": "72e1be99-f517-48c4-b45a-808dea668ec0",
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "7c022411-eb2c-4f06-9387-5d8418a0bf33",
+ "pk": "775da282-9bef-4dcd-b1fa-0a7ff555cb9d",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "bff45072-bc1b-408d-bd1f-b3ffd22d8796",
- "name": "Act 4",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "6acab4ca-a475-485e-b9cf-fc8a8404bcd1",
+ "name": "PARIS",
+ "zone": "272f32c2-29eb-47e7-b244-b0a6c2a772c0",
+ "source": null,
+ "worker_version": null
+ }
+},
+{
+ "model": "documents.element",
+ "pk": "7945d738-4a6e-4a76-ad57-21c03d63625d",
+ "fields": {
+ "created": "2020-02-02T01:23:45.678Z",
+ "updated": "2020-02-02T01:23:45.678Z",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "8ccc9061-0bc7-4d74-8a83-d427273a8c56",
+ "name": "Volume 1, page 2r",
+ "zone": "6a85fb05-f7d1-498d-b93c-62739562c3e3",
+ "source": null,
+ "worker_version": null
+ }
+},
+{
+ "model": "documents.element",
+ "pk": "7e788454-afed-4bef-a370-bccf0dd3b2b4",
+ "fields": {
+ "created": "2020-02-02T01:23:45.678Z",
+ "updated": "2020-02-02T01:23:45.678Z",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "6bc2cbbd-b98e-4689-bf9d-8a9095710632",
+ "name": "Act 1",
"zone": null,
"source": null,
"worker_version": null
@@ -436,41 +598,83 @@
},
{
"model": "documents.element",
- "pk": "7e4d2f71-f52a-44a6-8cb5-f209a0b82408",
+ "pk": "9644857d-6c33-4241-ad43-38054fba11a0",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "aad1941d-d7dd-485f-8cf3-90487ac522c8",
- "name": "Volume 2, page 1r",
- "zone": "541c027b-3e0b-42f6-acf6-abcc49ff5f1e",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "f56fb6dd-afe4-49b3-a870-963467404935",
+ "name": "Surface C",
+ "zone": "0fb66d83-3d13-4d4e-82d1-1fd7fc82036f",
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "8f92f79c-471a-4280-9530-66ec406ab8b2",
+ "pk": "96893993-5b8d-4ed0-b305-08f745d8df01",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "aad1941d-d7dd-485f-8cf3-90487ac522c8",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "6acab4ca-a475-485e-b9cf-fc8a8404bcd1",
+ "name": "ROY",
+ "zone": "db138d8c-b5fc-46e6-8dc6-483a1ffc8bf6",
+ "source": null,
+ "worker_version": null
+ }
+},
+{
+ "model": "documents.element",
+ "pk": "99de88f1-9655-4bae-9203-01d56d097ef7",
+ "fields": {
+ "created": "2020-02-02T01:23:45.678Z",
+ "updated": "2020-02-02T01:23:45.678Z",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "8ccc9061-0bc7-4d74-8a83-d427273a8c56",
"name": "Volume 1, page 1r",
- "zone": "5bd58b5d-8ed2-4e56-a466-b863101dfdc6",
+ "zone": "fb6dff0f-e117-48f0-b27a-9fa65f42a055",
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "9621c876-e7da-4d43-a5a7-82470dead663",
+ "pk": "9f81a475-ed1f-470d-b11a-9740d6bb24d2",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "bff45072-bc1b-408d-bd1f-b3ffd22d8796",
- "name": "Act 1",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "77739a6b-f044-4a5e-9c19-bc881caafbbf",
+ "name": "Text line",
+ "zone": "f96e2027-4568-4527-a469-bcf27f4fcbb2",
+ "source": null,
+ "worker_version": null
+ }
+},
+{
+ "model": "documents.element",
+ "pk": "a493ea73-d17d-4f63-b6c9-1ebe04f36641",
+ "fields": {
+ "created": "2020-02-02T01:23:45.678Z",
+ "updated": "2020-02-02T01:23:45.678Z",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "f56fb6dd-afe4-49b3-a870-963467404935",
+ "name": "Surface D",
+ "zone": "a236763a-6d34-4c41-a1e5-a5e19d115d50",
+ "source": null,
+ "worker_version": null
+ }
+},
+{
+ "model": "documents.element",
+ "pk": "b675ba10-f6ed-45cb-8a8c-449940471db5",
+ "fields": {
+ "created": "2020-02-02T01:23:45.678Z",
+ "updated": "2020-02-02T01:23:45.678Z",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "6bc2cbbd-b98e-4689-bf9d-8a9095710632",
+ "name": "Act 4",
"zone": null,
"source": null,
"worker_version": null
@@ -478,41 +682,41 @@
},
{
"model": "documents.element",
- "pk": "b09ebead-3440-4040-8b82-5dc2bae4d2d9",
+ "pk": "b9f39256-6ff0-4daf-b56a-a5a316964133",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "33d93a96-58a9-4a7c-be54-12d4691f4e81",
- "name": "Text line",
- "zone": "59c89a4e-b961-4158-8107-f24463b40fca",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "6acab4ca-a475-485e-b9cf-fc8a8404bcd1",
+ "name": "PARIS",
+ "zone": "c12fb064-b0df-4d8a-af13-55b2e9862e8d",
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "c0ec4456-f6a2-4253-a876-6eba4f96f863",
+ "pk": "bbd63fcc-23af-4593-bcb5-341c3008b7ab",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "aad1941d-d7dd-485f-8cf3-90487ac522c8",
- "name": "Volume 1, page 1v",
- "zone": "dda78215-f30b-497f-aedb-fac49055b667",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "6acab4ca-a475-485e-b9cf-fc8a8404bcd1",
+ "name": "DATUM",
+ "zone": "bea07b51-a5bd-400a-b5d0-b0320dc77bd1",
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "cb69d995-2ce2-4404-a567-5598cf9dd231",
+ "pk": "c4ccff42-93d3-4f66-a073-397dd85e1806",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "bff45072-bc1b-408d-bd1f-b3ffd22d8796",
- "name": "Act 2",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "6bc2cbbd-b98e-4689-bf9d-8a9095710632",
+ "name": "Act 5",
"zone": null,
"source": null,
"worker_version": null
@@ -520,12 +724,12 @@
},
{
"model": "documents.element",
- "pk": "ccc2cfdb-2de5-4e42-8a1a-fc5e1041b21d",
+ "pk": "c53fc91f-58c7-4fda-a92c-72a5f4051176",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "bff45072-bc1b-408d-bd1f-b3ffd22d8796",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "6bc2cbbd-b98e-4689-bf9d-8a9095710632",
"name": "Act 3",
"zone": null,
"source": null,
@@ -534,77 +738,91 @@
},
{
"model": "documents.element",
- "pk": "d19ded5f-c50c-4cc6-a0cf-4463e2f2df68",
+ "pk": "cb1b2af4-9cd8-4ecb-99a3-a7f309198a7b",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "bff45072-bc1b-408d-bd1f-b3ffd22d8796",
- "name": "Act 5",
- "zone": null,
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "f56fb6dd-afe4-49b3-a870-963467404935",
+ "name": "Surface A",
+ "zone": "9bae7b8e-b03b-40cf-a0d8-80393d76a304",
+ "source": null,
+ "worker_version": null
+ }
+},
+{
+ "model": "documents.element",
+ "pk": "cbafeb68-d66a-4e1c-bcd4-47dd88b02eec",
+ "fields": {
+ "created": "2020-02-02T01:23:45.678Z",
+ "updated": "2020-02-02T01:23:45.678Z",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "8ccc9061-0bc7-4d74-8a83-d427273a8c56",
+ "name": "Volume 2, page 1v",
+ "zone": "34f21948-295f-48a6-9644-41af328fdce4",
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "d7b6b144-88cb-4c47-a380-a84339773dda",
+ "pk": "ce358ae1-5d50-476e-b3bb-2f5abb3a9ce9",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "a8c815f0-258f-4025-8f9b-f3bd6a87904c",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "f56fb6dd-afe4-49b3-a870-963467404935",
"name": "Surface B",
- "zone": "b9cb9d9d-8de0-4fcf-8063-a5eb30634b8e",
+ "zone": "28c84829-8312-4f41-81e5-fa00fad02718",
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "dbfa2c38-8664-447a-bb09-215cc86c6ab4",
+ "pk": "e001d1ee-fa90-4f96-b722-e2ba20b84097",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "aad1941d-d7dd-485f-8cf3-90487ac522c8",
- "name": "Volume 2, page 2r",
- "zone": "543950e5-031d-41d1-83fd-7365c193c916",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "1e98cd23-9b0b-44ed-84b5-3a9e4dec68b3",
+ "name": "Volume 1",
+ "zone": null,
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "f3981e96-ed95-43a0-b1ea-fd288c55f5b3",
+ "pk": "ebdb0c60-e1e4-4f8a-a8b7-4127896ac194",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "a8c815f0-258f-4025-8f9b-f3bd6a87904c",
- "name": "Surface D",
- "zone": "3e6c6344-4dec-4609-b12d-14fd65c32664",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "6bc2cbbd-b98e-4689-bf9d-8a9095710632",
+ "name": "Act 2",
+ "zone": null,
"source": null,
"worker_version": null
}
},
{
"model": "documents.element",
- "pk": "f5535f84-2ca7-4853-9ab6-44606a8dc462",
+ "pk": "fcf303df-36a5-40f2-997f-82eb47541bcf",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "a8c815f0-258f-4025-8f9b-f3bd6a87904c",
- "name": "Surface A",
- "zone": "53ad7c0c-efcd-4001-be5b-298d1e8498b6",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "6acab4ca-a475-485e-b9cf-fc8a8404bcd1",
+ "name": "ROY",
+ "zone": "f96e2027-4568-4527-a469-bcf27f4fcbb2",
"source": null,
"worker_version": null
}
},
{
"model": "documents.datasource",
- "pk": "9ad06f06-000c-4d01-8038-1afc2e73c8e1",
+ "pk": "1eaefd28-7475-47bb-aff5-b53a225e26ec",
"fields": {
"type": "recognizer",
"slug": "test",
@@ -615,7 +833,7 @@
},
{
"model": "documents.datasource",
- "pk": "c357f678-9e18-4179-a9ee-4f35fdf26cd8",
+ "pk": "def9eadf-7b67-4f82-be03-5175ae644424",
"fields": {
"type": "classifier",
"slug": "test",
@@ -626,25 +844,23 @@
},
{
"model": "documents.transcription",
- "pk": "2578a1cf-3e57-4a15-b4cb-a0008c13a563",
+ "pk": "05f120a9-1c8a-4344-bf09-ef258cc04eca",
"fields": {
- "element": "8f92f79c-471a-4280-9530-66ec406ab8b2",
+ "element": "775da282-9bef-4dcd-b1fa-0a7ff555cb9d",
"type": "word",
- "zone": "59c89a4e-b961-4158-8107-f24463b40fca",
- "source": "9ad06f06-000c-4d01-8038-1afc2e73c8e1",
+ "source": "1eaefd28-7475-47bb-aff5-b53a225e26ec",
"worker_version": null,
- "text": "ROY",
+ "text": "PARIS",
"score": 1.0
}
},
{
"model": "documents.transcription",
- "pk": "4fc2945e-e4be-44d6-a839-c3149cbc66e0",
+ "pk": "094d1e1b-d35d-4e86-a2fd-4fabd60ba5fa",
"fields": {
- "element": "564aa56f-b257-445c-9930-4b6110628c09",
+ "element": "fcf303df-36a5-40f2-997f-82eb47541bcf",
"type": "word",
- "zone": "a2723514-109f-4b13-8336-e0128c49aee4",
- "source": "9ad06f06-000c-4d01-8038-1afc2e73c8e1",
+ "source": "1eaefd28-7475-47bb-aff5-b53a225e26ec",
"worker_version": null,
"text": "ROY",
"score": 1.0
@@ -652,64 +868,59 @@
},
{
"model": "documents.transcription",
- "pk": "51785975-0bf3-4bd1-8002-ae8c21dacbd4",
+ "pk": "0c7b5b2d-d1f6-494c-8ba4-169cdf255f5e",
"fields": {
- "element": "c0ec4456-f6a2-4253-a876-6eba4f96f863",
+ "element": "4d5191a9-5161-4587-a6bf-6430786f869a",
"type": "word",
- "zone": "2e3f07cc-d8cb-4735-bed0-63d4d109d893",
- "source": "9ad06f06-000c-4d01-8038-1afc2e73c8e1",
+ "source": "1eaefd28-7475-47bb-aff5-b53a225e26ec",
"worker_version": null,
- "text": "PARIS",
+ "text": "DATUM",
"score": 1.0
}
},
{
"model": "documents.transcription",
- "pk": "72409ca8-98ff-4055-98e8-5b5e8620e1fe",
+ "pk": "6c517add-09a0-41e4-bb03-0aea9ce1e628",
"fields": {
- "element": "c0ec4456-f6a2-4253-a876-6eba4f96f863",
+ "element": "73b56c40-4136-499b-9c10-82c2f569c78b",
"type": "word",
- "zone": "de6f5a26-52fd-4361-b30e-08b98b9d480a",
- "source": "9ad06f06-000c-4d01-8038-1afc2e73c8e1",
+ "source": "1eaefd28-7475-47bb-aff5-b53a225e26ec",
"worker_version": null,
- "text": "DATUM",
+ "text": "PARIS",
"score": 1.0
}
},
{
"model": "documents.transcription",
- "pk": "81e2a90b-d518-4280-b71b-7cdc24587472",
+ "pk": "802856d1-13a8-4d63-9af6-b0c8db086a89",
"fields": {
- "element": "564aa56f-b257-445c-9930-4b6110628c09",
- "type": "word",
- "zone": "658a1c75-a4de-4842-8a09-2a51c22c8269",
- "source": "9ad06f06-000c-4d01-8038-1afc2e73c8e1",
+ "element": "99de88f1-9655-4bae-9203-01d56d097ef7",
+ "type": "page",
+ "source": "1eaefd28-7475-47bb-aff5-b53a225e26ec",
"worker_version": null,
- "text": "PARIS",
+ "text": "Lorem ipsum dolor sit amet",
"score": 1.0
}
},
{
"model": "documents.transcription",
- "pk": "d4e38d93-48dc-4c46-881f-fe85ddfd01da",
+ "pk": "ab32afab-5d69-47d1-80f4-3ab1151462ef",
"fields": {
- "element": "8f92f79c-471a-4280-9530-66ec406ab8b2",
+ "element": "5f20bff4-aa8b-4afa-ad96-e791c2433dc1",
"type": "word",
- "zone": "2d6cbfa0-917e-4f3e-91db-3e050f236922",
- "source": "9ad06f06-000c-4d01-8038-1afc2e73c8e1",
+ "source": "1eaefd28-7475-47bb-aff5-b53a225e26ec",
"worker_version": null,
- "text": "PARIS",
+ "text": "DATUM",
"score": 1.0
}
},
{
"model": "documents.transcription",
- "pk": "d5297c59-e7b8-4c03-8b53-d84347ec726c",
+ "pk": "d5cc8936-0a81-4b01-83d8-1dd75b0d60c0",
"fields": {
- "element": "8f92f79c-471a-4280-9530-66ec406ab8b2",
+ "element": "bbd63fcc-23af-4593-bcb5-341c3008b7ab",
"type": "word",
- "zone": "1323086e-7117-4454-b6d2-d2e9b9b30251",
- "source": "9ad06f06-000c-4d01-8038-1afc2e73c8e1",
+ "source": "1eaefd28-7475-47bb-aff5-b53a225e26ec",
"worker_version": null,
"text": "DATUM",
"score": 1.0
@@ -717,12 +928,11 @@
},
{
"model": "documents.transcription",
- "pk": "daef6593-28c3-4b81-a693-e1ec81f52718",
+ "pk": "d7789e57-9fae-4465-9ef3-2f552a3d1e86",
"fields": {
- "element": "c0ec4456-f6a2-4253-a876-6eba4f96f863",
+ "element": "72f64b68-afc4-41e0-8020-e590bdaa9f7c",
"type": "word",
- "zone": "4ae281d2-de97-42dc-ac66-d3edc93bf92e",
- "source": "9ad06f06-000c-4d01-8038-1afc2e73c8e1",
+ "source": "1eaefd28-7475-47bb-aff5-b53a225e26ec",
"worker_version": null,
"text": "ROY",
"score": 1.0
@@ -730,65 +940,63 @@
},
{
"model": "documents.transcription",
- "pk": "e275ebac-db1d-48df-90d5-0513464f4ab0",
+ "pk": "dcd5a2be-5f89-4565-8acc-521c08672c65",
"fields": {
- "element": "564aa56f-b257-445c-9930-4b6110628c09",
+ "element": "96893993-5b8d-4ed0-b305-08f745d8df01",
"type": "word",
- "zone": "021f8b5e-4963-4638-a24f-fa99114ffd17",
- "source": "9ad06f06-000c-4d01-8038-1afc2e73c8e1",
+ "source": "1eaefd28-7475-47bb-aff5-b53a225e26ec",
"worker_version": null,
- "text": "DATUM",
+ "text": "ROY",
"score": 1.0
}
},
{
"model": "documents.transcription",
- "pk": "ea21f8c6-1ecf-4be8-848b-af333c8b83a1",
+ "pk": "f97dc9b7-15d5-4b3c-ac3e-621604fd5522",
"fields": {
- "element": "8f92f79c-471a-4280-9530-66ec406ab8b2",
- "type": "page",
- "zone": null,
- "source": "9ad06f06-000c-4d01-8038-1afc2e73c8e1",
+ "element": "b9f39256-6ff0-4daf-b56a-a5a316964133",
+ "type": "word",
+ "source": "1eaefd28-7475-47bb-aff5-b53a225e26ec",
"worker_version": null,
- "text": "Lorem ipsum dolor sit amet",
+ "text": "PARIS",
"score": 1.0
}
},
{
"model": "documents.allowedmetadata",
- "pk": "0f90ca4d-c698-4a14-89fc-32c153e6a5fd",
+ "pk": "2ae06c55-5609-4a22-9ae3-d3ec5fe78582",
"fields": {
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
"type": "text",
"name": "folio"
}
},
{
"model": "documents.allowedmetadata",
- "pk": "1ab75f14-57ca-4c05-a9fb-f10c082f4206",
+ "pk": "85f058a1-a3d3-43bc-8681-c11c5491e656",
"fields": {
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "location",
- "name": "location"
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "date",
+ "name": "date"
}
},
{
"model": "documents.allowedmetadata",
- "pk": "90cf5544-2b0a-465a-aae4-c7de6ade0592",
+ "pk": "d25ded48-d068-4fcd-a947-9a689ca271a8",
"fields": {
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
- "type": "date",
- "name": "date"
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
+ "type": "location",
+ "name": "location"
}
},
{
"model": "documents.metadata",
- "pk": "0be7aa65-fa4d-4cea-bba2-86748478b595",
+ "pk": "0706361e-7c9a-4dc0-b058-ad365f015d01",
"fields": {
- "element": "564aa56f-b257-445c-9930-4b6110628c09",
+ "element": "99de88f1-9655-4bae-9203-01d56d097ef7",
"name": "folio",
"type": "text",
- "value": "2r",
+ "value": "1r",
"revision": null,
"index": 0,
"entity": null
@@ -796,12 +1004,12 @@
},
{
"model": "documents.metadata",
- "pk": "12069432-7670-4378-a918-23daf27ea172",
+ "pk": "1000532c-4eb2-4d2b-91ac-1a8e7b743284",
"fields": {
- "element": "7c022411-eb2c-4f06-9387-5d8418a0bf33",
- "name": "number",
+ "element": "35fb048c-f441-4e9d-bb69-78a8dfdb4598",
+ "name": "folio",
"type": "text",
- "value": "4",
+ "value": "1v",
"revision": null,
"index": 0,
"entity": null
@@ -809,12 +1017,12 @@
},
{
"model": "documents.metadata",
- "pk": "14e7fd76-4c9b-4e45-b01c-f6ac04b4d17c",
+ "pk": "31edcdb5-5b70-4848-b890-fb88afe1d3e0",
"fields": {
- "element": "cb69d995-2ce2-4404-a567-5598cf9dd231",
+ "element": "c4ccff42-93d3-4f66-a073-397dd85e1806",
"name": "number",
"type": "text",
- "value": "2",
+ "value": "5",
"revision": null,
"index": 0,
"entity": null
@@ -822,12 +1030,12 @@
},
{
"model": "documents.metadata",
- "pk": "2f82730b-fc55-485c-869d-7aa2214aafef",
+ "pk": "6a916e68-e8f1-455e-b0ab-f5947355187c",
"fields": {
- "element": "c0ec4456-f6a2-4253-a876-6eba4f96f863",
- "name": "folio",
+ "element": "b675ba10-f6ed-45cb-8a8c-449940471db5",
+ "name": "number",
"type": "text",
- "value": "1v",
+ "value": "4",
"revision": null,
"index": 0,
"entity": null
@@ -835,9 +1043,9 @@
},
{
"model": "documents.metadata",
- "pk": "63ff193e-fd67-4c2d-b52d-3c99a49f1131",
+ "pk": "74fd04ea-9352-41c9-bdbf-24e36f49a305",
"fields": {
- "element": "dbfa2c38-8664-447a-bb09-215cc86c6ab4",
+ "element": "5130da8e-f2aa-477d-bf58-a2357efc8b2b",
"name": "folio",
"type": "text",
"value": "2r",
@@ -848,12 +1056,12 @@
},
{
"model": "documents.metadata",
- "pk": "759c2edc-6e7b-4f96-9cd9-0daa38d9757a",
+ "pk": "99164dd7-bc9c-49c9-89f6-a8ffc0be7779",
"fields": {
- "element": "4e9de67b-6e4f-46b7-936e-34ed9692900b",
- "name": "folio",
+ "element": "c53fc91f-58c7-4fda-a92c-72a5f4051176",
+ "name": "number",
"type": "text",
- "value": "1v",
+ "value": "3",
"revision": null,
"index": 0,
"entity": null
@@ -861,12 +1069,12 @@
},
{
"model": "documents.metadata",
- "pk": "79c508f3-4288-4a1b-8f1f-19af1a057e3e",
+ "pk": "9c5602e4-1fb8-444f-b23d-6f9175eab2c9",
"fields": {
- "element": "d19ded5f-c50c-4cc6-a0cf-4463e2f2df68",
- "name": "number",
+ "element": "cbafeb68-d66a-4e1c-bcd4-47dd88b02eec",
+ "name": "folio",
"type": "text",
- "value": "5",
+ "value": "1v",
"revision": null,
"index": 0,
"entity": null
@@ -874,12 +1082,12 @@
},
{
"model": "documents.metadata",
- "pk": "7f3688fb-3a16-4cba-b218-47a372eb5480",
+ "pk": "afce1c14-2ee8-46ed-9c78-4360562e91c2",
"fields": {
- "element": "9621c876-e7da-4d43-a5a7-82470dead663",
+ "element": "ebdb0c60-e1e4-4f8a-a8b7-4127896ac194",
"name": "number",
"type": "text",
- "value": "1",
+ "value": "2",
"revision": null,
"index": 0,
"entity": null
@@ -887,12 +1095,12 @@
},
{
"model": "documents.metadata",
- "pk": "d86e7d8d-2dc8-4ac7-ac5e-d8aff8df4548",
+ "pk": "cef63c02-d1b9-459b-88b2-771f3f9b7c87",
"fields": {
- "element": "7e4d2f71-f52a-44a6-8cb5-f209a0b82408",
- "name": "folio",
+ "element": "7e788454-afed-4bef-a370-bccf0dd3b2b4",
+ "name": "number",
"type": "text",
- "value": "1r",
+ "value": "1",
"revision": null,
"index": 0,
"entity": null
@@ -900,12 +1108,12 @@
},
{
"model": "documents.metadata",
- "pk": "d9c74f49-670d-4b37-a21e-60ca50059cda",
+ "pk": "e1850c59-5a05-438a-a49d-e8a99710671e",
"fields": {
- "element": "ccc2cfdb-2de5-4e42-8a1a-fc5e1041b21d",
- "name": "number",
+ "element": "326843be-0112-4954-a2ea-da78aded01a9",
+ "name": "folio",
"type": "text",
- "value": "3",
+ "value": "1r",
"revision": null,
"index": 0,
"entity": null
@@ -913,12 +1121,12 @@
},
{
"model": "documents.metadata",
- "pk": "f4103a83-94eb-4634-8cd3-32db7e94ac1e",
+ "pk": "edeaf010-9326-4a0f-9e31-f403b8363be6",
"fields": {
- "element": "8f92f79c-471a-4280-9530-66ec406ab8b2",
+ "element": "7945d738-4a6e-4a76-ad57-21c03d63625d",
"name": "folio",
"type": "text",
- "value": "1r",
+ "value": "2r",
"revision": null,
"index": 0,
"entity": null
@@ -942,12 +1150,12 @@
},
{
"model": "images.image",
- "pk": "0a18ca58-69a7-44f8-b2ac-739a7a978cf2",
+ "pk": "5f4f5d85-42d7-48d0-bae3-6be0dfa1762f",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
"server": 1,
- "path": "img3",
+ "path": "img6",
"width": 1000,
"height": 1000,
"hash": null,
@@ -956,7 +1164,7 @@
},
{
"model": "images.image",
- "pk": "7ce02a7c-877a-46bc-8557-cd87b6945d81",
+ "pk": "5f870efa-b012-4344-92ac-e5226c4cd26b",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
@@ -970,7 +1178,7 @@
},
{
"model": "images.image",
- "pk": "a415642f-af03-422d-acf2-a9fce1cf6ef1",
+ "pk": "68474756-161a-4199-96e1-e79dc216a8a0",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
@@ -984,12 +1192,12 @@
},
{
"model": "images.image",
- "pk": "c1f75796-3a1c-4e85-b8fa-43472a719312",
+ "pk": "75371cf1-e321-47ec-bbc3-16c7c3d8f89a",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
"server": 1,
- "path": "img5",
+ "path": "img3",
"width": 1000,
"height": 1000,
"hash": null,
@@ -998,7 +1206,7 @@
},
{
"model": "images.image",
- "pk": "ddbb369c-a6ed-488c-9108-e996ade67b29",
+ "pk": "8f7f17ff-9555-488f-acac-9cb6814e46a8",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
@@ -1012,12 +1220,12 @@
},
{
"model": "images.image",
- "pk": "e6df804d-82f7-420e-a562-fb3d265a2bec",
+ "pk": "996b53e8-4b88-473f-ad2e-11ea4848c14e",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
"server": 1,
- "path": "img6",
+ "path": "img5",
"width": 1000,
"height": 1000,
"hash": null,
@@ -1026,209 +1234,209 @@
},
{
"model": "images.zone",
- "pk": "021f8b5e-4963-4638-a24f-fa99114ffd17",
+ "pk": "0039b5f5-078f-4667-825c-6e620b92b59a",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "0a18ca58-69a7-44f8-b2ac-739a7a978cf2",
- "polygon": "LINEARRING (700 700, 700 800, 800 800, 800 700, 700 700)"
+ "image": "75371cf1-e321-47ec-bbc3-16c7c3d8f89a",
+ "polygon": "LINEARRING (400 400, 400 500, 500 500, 500 400, 400 400)"
}
},
{
"model": "images.zone",
- "pk": "1323086e-7117-4454-b6d2-d2e9b9b30251",
+ "pk": "0fb66d83-3d13-4d4e-82d1-1fd7fc82036f",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "ddbb369c-a6ed-488c-9108-e996ade67b29",
- "polygon": "LINEARRING (700 700, 700 800, 800 800, 800 700, 700 700)"
+ "image": "68474756-161a-4199-96e1-e79dc216a8a0",
+ "polygon": "LINEARRING (0 0, 0 1000, 1000 1000, 1000 0, 0 0)"
}
},
{
"model": "images.zone",
- "pk": "20f0abb7-3aac-460b-b1af-b75ed7dab550",
+ "pk": "11bb6bba-8a2e-4f9e-97b1-0f48bc34103f",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "0a18ca58-69a7-44f8-b2ac-739a7a978cf2",
+ "image": "5f4f5d85-42d7-48d0-bae3-6be0dfa1762f",
"polygon": "LINEARRING (0 0, 0 1000, 1000 1000, 1000 0, 0 0)"
}
},
{
"model": "images.zone",
- "pk": "2d6cbfa0-917e-4f3e-91db-3e050f236922",
+ "pk": "272f32c2-29eb-47e7-b244-b0a6c2a772c0",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "ddbb369c-a6ed-488c-9108-e996ade67b29",
+ "image": "75371cf1-e321-47ec-bbc3-16c7c3d8f89a",
"polygon": "LINEARRING (100 100, 100 200, 200 200, 200 100, 100 100)"
}
},
{
"model": "images.zone",
- "pk": "2e3f07cc-d8cb-4735-bed0-63d4d109d893",
+ "pk": "28c84829-8312-4f41-81e5-fa00fad02718",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "a415642f-af03-422d-acf2-a9fce1cf6ef1",
- "polygon": "LINEARRING (100 100, 100 200, 200 200, 200 100, 100 100)"
+ "image": "8f7f17ff-9555-488f-acac-9cb6814e46a8",
+ "polygon": "LINEARRING (600 600, 600 1000, 1000 1000, 1000 600, 600 600)"
}
},
{
"model": "images.zone",
- "pk": "3e6c6344-4dec-4609-b12d-14fd65c32664",
+ "pk": "34f21948-295f-48a6-9644-41af328fdce4",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "0a18ca58-69a7-44f8-b2ac-739a7a978cf2",
- "polygon": "LINEARRING (0 0, 0 300, 300 300, 300 0, 0 0)"
+ "image": "996b53e8-4b88-473f-ad2e-11ea4848c14e",
+ "polygon": "LINEARRING (0 0, 0 1000, 1000 1000, 1000 0, 0 0)"
}
},
{
"model": "images.zone",
- "pk": "4ae281d2-de97-42dc-ac66-d3edc93bf92e",
+ "pk": "40edd01e-02e6-4fbf-b2c6-b7452a72c79e",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "a415642f-af03-422d-acf2-a9fce1cf6ef1",
- "polygon": "LINEARRING (400 400, 400 500, 500 500, 500 400, 400 400)"
+ "image": "8f7f17ff-9555-488f-acac-9cb6814e46a8",
+ "polygon": "LINEARRING (700 700, 700 800, 800 800, 800 700, 700 700)"
}
},
{
"model": "images.zone",
- "pk": "53ad7c0c-efcd-4001-be5b-298d1e8498b6",
+ "pk": "440f85ff-25d9-4488-8654-98c5916df96f",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "ddbb369c-a6ed-488c-9108-e996ade67b29",
- "polygon": "LINEARRING (0 0, 0 600, 600 600, 600 0, 0 0)"
+ "image": "75371cf1-e321-47ec-bbc3-16c7c3d8f89a",
+ "polygon": "LINEARRING (300 300, 300 600, 600 600, 600 300, 300 300)"
}
},
{
"model": "images.zone",
- "pk": "541c027b-3e0b-42f6-acf6-abcc49ff5f1e",
+ "pk": "6a85fb05-f7d1-498d-b93c-62739562c3e3",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "7ce02a7c-877a-46bc-8557-cd87b6945d81",
+ "image": "75371cf1-e321-47ec-bbc3-16c7c3d8f89a",
"polygon": "LINEARRING (0 0, 0 1000, 1000 1000, 1000 0, 0 0)"
}
},
{
"model": "images.zone",
- "pk": "543950e5-031d-41d1-83fd-7365c193c916",
+ "pk": "72e1be99-f517-48c4-b45a-808dea668ec0",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "e6df804d-82f7-420e-a562-fb3d265a2bec",
- "polygon": "LINEARRING (0 0, 0 1000, 1000 1000, 1000 0, 0 0)"
+ "image": "8f7f17ff-9555-488f-acac-9cb6814e46a8",
+ "polygon": "LINEARRING (100 100, 100 200, 200 200, 200 100, 100 100)"
}
},
{
"model": "images.zone",
- "pk": "59c89a4e-b961-4158-8107-f24463b40fca",
+ "pk": "7e767702-f644-451f-91e8-05423bb8e074",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "ddbb369c-a6ed-488c-9108-e996ade67b29",
- "polygon": "LINEARRING (400 400, 400 500, 500 500, 500 400, 400 400)"
+ "image": "75371cf1-e321-47ec-bbc3-16c7c3d8f89a",
+ "polygon": "LINEARRING (700 700, 700 800, 800 800, 800 700, 700 700)"
}
},
{
"model": "images.zone",
- "pk": "5bd58b5d-8ed2-4e56-a466-b863101dfdc6",
+ "pk": "9bae7b8e-b03b-40cf-a0d8-80393d76a304",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "ddbb369c-a6ed-488c-9108-e996ade67b29",
- "polygon": "LINEARRING (0 0, 0 1000, 1000 1000, 1000 0, 0 0)"
+ "image": "8f7f17ff-9555-488f-acac-9cb6814e46a8",
+ "polygon": "LINEARRING (0 0, 0 600, 600 600, 600 0, 0 0)"
}
},
{
"model": "images.zone",
- "pk": "658a1c75-a4de-4842-8a09-2a51c22c8269",
+ "pk": "a236763a-6d34-4c41-a1e5-a5e19d115d50",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "0a18ca58-69a7-44f8-b2ac-739a7a978cf2",
- "polygon": "LINEARRING (100 100, 100 200, 200 200, 200 100, 100 100)"
+ "image": "75371cf1-e321-47ec-bbc3-16c7c3d8f89a",
+ "polygon": "LINEARRING (0 0, 0 300, 300 300, 300 0, 0 0)"
}
},
{
"model": "images.zone",
- "pk": "a2723514-109f-4b13-8336-e0128c49aee4",
+ "pk": "aefbd01b-3421-458f-a8e6-2b5ffdaccc31",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "0a18ca58-69a7-44f8-b2ac-739a7a978cf2",
- "polygon": "LINEARRING (400 400, 400 500, 500 500, 500 400, 400 400)"
+ "image": "5f870efa-b012-4344-92ac-e5226c4cd26b",
+ "polygon": "LINEARRING (0 0, 0 1000, 1000 1000, 1000 0, 0 0)"
}
},
{
"model": "images.zone",
- "pk": "b9cb9d9d-8de0-4fcf-8063-a5eb30634b8e",
+ "pk": "b70d7b27-c973-44ba-93c5-602957ae491e",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "ddbb369c-a6ed-488c-9108-e996ade67b29",
+ "image": "75371cf1-e321-47ec-bbc3-16c7c3d8f89a",
"polygon": "LINEARRING (600 600, 600 1000, 1000 1000, 1000 600, 600 600)"
}
},
{
"model": "images.zone",
- "pk": "d2e7ee12-14fa-46c8-a360-f5b31939052a",
+ "pk": "bea07b51-a5bd-400a-b5d0-b0320dc77bd1",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "0a18ca58-69a7-44f8-b2ac-739a7a978cf2",
- "polygon": "LINEARRING (600 600, 600 1000, 1000 1000, 1000 600, 600 600)"
+ "image": "68474756-161a-4199-96e1-e79dc216a8a0",
+ "polygon": "LINEARRING (700 700, 700 800, 800 800, 800 700, 700 700)"
}
},
{
"model": "images.zone",
- "pk": "dda78215-f30b-497f-aedb-fac49055b667",
+ "pk": "c12fb064-b0df-4d8a-af13-55b2e9862e8d",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "a415642f-af03-422d-acf2-a9fce1cf6ef1",
- "polygon": "LINEARRING (0 0, 0 1000, 1000 1000, 1000 0, 0 0)"
+ "image": "68474756-161a-4199-96e1-e79dc216a8a0",
+ "polygon": "LINEARRING (100 100, 100 200, 200 200, 200 100, 100 100)"
}
},
{
"model": "images.zone",
- "pk": "de6f5a26-52fd-4361-b30e-08b98b9d480a",
+ "pk": "db138d8c-b5fc-46e6-8dc6-483a1ffc8bf6",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "a415642f-af03-422d-acf2-a9fce1cf6ef1",
- "polygon": "LINEARRING (700 700, 700 800, 800 800, 800 700, 700 700)"
+ "image": "68474756-161a-4199-96e1-e79dc216a8a0",
+ "polygon": "LINEARRING (400 400, 400 500, 500 500, 500 400, 400 400)"
}
},
{
"model": "images.zone",
- "pk": "e67eba1f-4993-492e-9ca0-67451ceb094c",
+ "pk": "f96e2027-4568-4527-a469-bcf27f4fcbb2",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "c1f75796-3a1c-4e85-b8fa-43472a719312",
- "polygon": "LINEARRING (0 0, 0 1000, 1000 1000, 1000 0, 0 0)"
+ "image": "8f7f17ff-9555-488f-acac-9cb6814e46a8",
+ "polygon": "LINEARRING (400 400, 400 500, 500 500, 500 400, 400 400)"
}
},
{
"model": "images.zone",
- "pk": "f3b62073-8fe4-46c9-b7bf-012c3014359b",
+ "pk": "fb6dff0f-e117-48f0-b27a-9fa65f42a055",
"fields": {
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
- "image": "0a18ca58-69a7-44f8-b2ac-739a7a978cf2",
- "polygon": "LINEARRING (300 300, 300 600, 600 600, 600 300, 300 300)"
+ "image": "8f7f17ff-9555-488f-acac-9cb6814e46a8",
+ "polygon": "LINEARRING (0 0, 0 1000, 1000 1000, 1000 0, 0 0)"
}
},
{
"model": "users.user",
"pk": 1,
"fields": {
- "password": "pbkdf2_sha256$216000$l8Q874FbHSd8$nuAro6QYXwWYcYcFP1hVKgZUjTYnmbAKLqvFBRpLizU=",
+ "password": "pbkdf2_sha256$216000$nxRE2ciWpZ7g$92fj1x7D47uAXNxwfbkSX8ffAeO03T28riqjHM5yJVk=",
"last_login": null,
"email": "root@root.fr",
"transkribus_email": null,
@@ -1243,7 +1451,7 @@
"model": "users.user",
"pk": 2,
"fields": {
- "password": "pbkdf2_sha256$216000$JbMVfYKIdmbH$HacY6ieeQBFMnMT32valEOgJnifIPuxNWx5LsaZSEN4=",
+ "password": "pbkdf2_sha256$216000$QEkgVJ40KttJ$98o0rhKrYu1vFJZ/+rOKk/D+oi52z6XGYIayxGO6P4I=",
"last_login": null,
"email": "internal@internal.fr",
"transkribus_email": null,
@@ -1258,7 +1466,7 @@
"model": "users.user",
"pk": 3,
"fields": {
- "password": "pbkdf2_sha256$216000$RurQ9pYDub7S$bhJUTPQbo1jcqBoDS69RNbdzM6cr0ydKcE9Dk7mQyyU=",
+ "password": "pbkdf2_sha256$216000$mWYa0QHsUxjL$z8BQ0LObKxjCWFE29wIKXyRuL+rEtgWyGuboxc5DL1Q=",
"last_login": null,
"email": "user@user.fr",
"transkribus_email": null,
@@ -1271,7 +1479,7 @@
},
{
"model": "users.oauthcredentials",
- "pk": "d21e40e2-32d2-4903-ac4a-6672dc6c14ba",
+ "pk": "b07d0cb3-65b7-45c4-9bf2-d7153f4d27cd",
"fields": {
"user": 3,
"provider_name": "gitlab",
@@ -1288,7 +1496,7 @@
"pk": 1,
"fields": {
"user": 3,
- "corpus": "1f7e82ec-4094-4397-b729-ec424ea7a37f",
+ "corpus": "78d45b37-36d1-4576-8bb2-8e5844f31feb",
"can_write": true,
"can_admin": true
}
@@ -2879,7 +3087,7 @@
},
{
"model": "ponos.workflow",
- "pk": "b391055a-84f5-431e-84ff-fec1836c4a5e",
+ "pk": "1fb1e3f8-67e4-45dd-acc8-1bdedf4e6f98",
"fields": {
"recipe": "tasks:\n docker_build:\n image: reco",
"created": "2020-02-02T01:23:45.678Z",
@@ -2888,7 +3096,7 @@
},
{
"model": "ponos.task",
- "pk": "71dd7910-8bef-4940-ab6a-50df883a2cb3",
+ "pk": "a9d5d4ce-188e-4659-9f35-4de56a3e3e1d",
"fields": {
"run": 0,
"depth": 0,
@@ -2901,7 +3109,7 @@
"has_docker_socket": false,
"image_artifact": null,
"agent": null,
- "workflow": "b391055a-84f5-431e-84ff-fec1836c4a5e",
+ "workflow": "1fb1e3f8-67e4-45dd-acc8-1bdedf4e6f98",
"container": null,
"created": "2020-02-02T01:23:45.678Z",
"updated": "2020-02-02T01:23:45.678Z",
@@ -2910,9 +3118,9 @@
},
{
"model": "ponos.artifact",
- "pk": "71a4b60b-a2f2-4503-8f5d-9d3c38f66626",
+ "pk": "2f4c9981-6ee0-43a1-80f0-54cca23be3ff",
"fields": {
- "task": "71dd7910-8bef-4940-ab6a-50df883a2cb3",
+ "task": "a9d5d4ce-188e-4659-9f35-4de56a3e3e1d",
"path": "/path/to/docker_build",
"size": 42000,
"content_type": "application/octet-stream",
diff --git a/arkindex/documents/management/commands/build_fixtures.py b/arkindex/documents/management/commands/build_fixtures.py
index 55d5ea4161003e8bd70441c1510de1275b8c576b..24bd321d0c783f18c968b3643fd6f507547ae292 100644
--- a/arkindex/documents/management/commands/build_fixtures.py
+++ b/arkindex/documents/management/commands/build_fixtures.py
@@ -2,7 +2,7 @@
from django.core.management.base import BaseCommand
from arkindex_common.ml_tool import MLToolType
from arkindex_common.enums import TranscriptionType, MetaType
-from arkindex.documents.models import Corpus, Element, Transcription, DataSource, MetaData
+from arkindex.documents.models import Corpus, Element, DataSource, MetaData
from arkindex.dataimport.models import RepositoryType, WorkerVersion, WorkerVersionState, Workflow
from arkindex.images.models import ImageServer, Image, Zone
from arkindex.users.models import User, CorpusRight
@@ -150,6 +150,11 @@ class Command(BaseCommand):
# Allows manual transcriptions of type 'line' for text lines
allowed_transcription=TranscriptionType.Line
)
+ word_type = corpus.types.create(
+ slug='word',
+ display_name='Word',
+ allowed_transcription=TranscriptionType.Word
+ )
# Create 2 volumes
vol1 = Element.objects.create(
@@ -217,21 +222,24 @@ class Command(BaseCommand):
# Create transcriptions on images of volume 1
for page in (p1_1, p1_2, p1_3):
for word, pos in [("PARIS", 100), ("ROY", 400), ("DATUM", 700)]:
- Transcription.objects.create(
- element=page,
+ element = corpus.elements.create(
+ type=word_type,
+ name=word,
+ zone=makezone(page.zone.image, pos, pos + 100)
+ )
+ element.add_parent(page)
+ element.transcriptions.create(
source=recognizer_source,
text=word,
type=TranscriptionType.Word,
- zone=makezone(page.zone.image, pos, pos + 100),
score=1.0,
)
- # Create a page transcriptions on page 1 with no zone
+ # Create a page transcription on page 1
p1_1.transcriptions.create(
source=recognizer_source,
text='Lorem ipsum dolor sit amet',
type=TranscriptionType.Page,
- zone=None,
score=1.0,
)
diff --git a/arkindex/documents/management/commands/reindex.py b/arkindex/documents/management/commands/reindex.py
index 528ff0f2c791c1b952c9cd3a3079c91190aaf0e7..3632b44ac042bef105467e2efd6eba012e944d88 100644
--- a/arkindex/documents/management/commands/reindex.py
+++ b/arkindex/documents/management/commands/reindex.py
@@ -16,13 +16,15 @@ logger = logging.getLogger(__name__)
def get_transcriptions(corpus=None, folder=None):
if folder:
# Lookup all the transcriptions linked to a folder
- return Transcription.objects.filter(
+ queryset = Transcription.objects.filter(
element__in=Element.objects.get_descending(folder.id)
).distinct()
elif corpus:
- return Transcription.objects.filter(element__corpus=corpus)
+ queryset = Transcription.objects.filter(element__corpus=corpus)
+ else:
+ queryset = Transcription.objects.all()
- return Transcription.objects.all()
+ return queryset.select_related('element')
def get_elements(corpus=None, folder=None):
@@ -33,7 +35,7 @@ def get_elements(corpus=None, folder=None):
else:
queryset = Element.objects.all()
- return queryset.prefetch_related('metadatas', 'transcriptions')
+ return queryset.select_related('type').prefetch_related('metadatas', 'transcriptions')
def get_entities(corpus=None, folder=None):
diff --git a/arkindex/documents/migrations/0021_move_transcriptions.py b/arkindex/documents/migrations/0021_move_transcriptions.py
new file mode 100644
index 0000000000000000000000000000000000000000..728be9aa5f17ccaa1c2a1ec864d7b02660c3925e
--- /dev/null
+++ b/arkindex/documents/migrations/0021_move_transcriptions.py
@@ -0,0 +1,132 @@
+# Generated by Django 3.1 on 2020-09-01 07:48
+
+from django.db import migrations, models
+from arkindex_common.enums import TranscriptionType
+
+
+def preflight_checks(apps, schema_editor):
+ ElementType = apps.get_model('documents', 'ElementType')
+ existing_types = []
+
+ for ts_type in TranscriptionType:
+ if ElementType.objects.filter(slug=f'transcription_{ts_type.value}').exists():
+ existing_types.append(f'`transcription_{ts_type.value}`')
+
+ if existing_types:
+ raise AssertionError(
+ 'This migration could not be run because one or more element types use the reserved slug(s) '
+ + ', '.join(existing_types)
+ )
+
+
+FORWARD_SQL = [
+ 'CREATE EXTENSION IF NOT EXISTS "uuid-ossp";',
+ # Early handling for the edge case of transcriptions already on the correct element
+ """
+ UPDATE documents_transcription transcription
+ SET zone_id = NULL
+ FROM documents_element element
+ WHERE transcription.element_id = element.id
+ AND transcription.zone_id IS NOT NULL
+ AND transcription.zone_id = element.zone_id;
+ """,
+ # Create element types starting with transcription_* as needed
+ """
+ INSERT INTO documents_elementtype (id, corpus_id, slug, display_name, folder, allowed_transcription)
+ SELECT
+ uuid_generate_v4(),
+ element.corpus_id,
+ 'transcription_' || transcription.type,
+ initcap(transcription.type) || ' Transcription',
+ FALSE,
+ transcription.type
+ FROM documents_transcription transcription
+ INNER JOIN documents_element element ON (element.id = transcription.element_id)
+ WHERE transcription.zone_id IS NOT NULL
+ GROUP BY element.corpus_id, transcription.type;
+ """,
+ # Create new elements
+ """
+ INSERT INTO documents_element (id, corpus_id, type_id, name, zone_id, source_id, worker_version_id, created, updated)
+ SELECT
+ transcription.id,
+ element.corpus_id,
+ type.id,
+ (ROW_NUMBER() OVER (
+ PARTITION BY
+ transcription.element_id,
+ transcription.source_id,
+ transcription.worker_version_id,
+ transcription.type
+ ORDER BY
+ ST_Y(ST_StartPoint(polygon)),
+ ST_X(ST_StartPoint(polygon))
+ ))::varchar,
+ transcription.zone_id,
+ transcription.source_id,
+ transcription.worker_version_id,
+ NOW(),
+ NOW()
+ FROM
+ documents_transcription transcription
+ INNER JOIN documents_element element on (transcription.element_id = element.id)
+ INNER JOIN documents_elementtype type ON (type.corpus_id = element.corpus_id AND type.slug = 'transcription_' || transcription.type)
+ INNER JOIN images_zone zone ON (transcription.zone_id = zone.id);
+ """,
+ # Create element paths
+ # Append to existing parent paths of the parent element, or create one new element path with the parent element itself in it
+ """
+ INSERT INTO documents_elementpath (id, element_id, path, ordering)
+ SELECT
+ uuid_generate_v4(),
+ transcription.id,
+ COALESCE(path.path, ARRAY[]::uuid[]) || transcription.element_id,
+ ROW_NUMBER() OVER (
+ PARTITION BY transcription.element_id
+ ORDER BY
+ ST_Y(ST_StartPoint(polygon)),
+ ST_X(ST_StartPoint(polygon))
+ )
+ FROM
+ documents_transcription transcription
+ INNER JOIN images_zone zone ON (zone.id = transcription.zone_id)
+ LEFT JOIN documents_elementpath path ON (path.element_id = transcription.element_id);
+ """,
+ # Move transcriptions to their new elements
+ """
+ UPDATE documents_transcription
+ SET element_id = id
+ WHERE zone_id IS NOT NULL;
+ """,
+ # At this point, we can drop the zone column, but this would fail due to 'pending trigger events'
+ # Postgres does not allow editing the schema *after* editing the data in the same transcription;
+ # This migration is continued in documents.0021 to allow a new database transaction to happen.
+]
+
+
+class Migration(migrations.Migration):
+
+ dependencies = [
+ ('documents', '0020_remove_source_xor_version_constraint'),
+ ('images', '0005_polygon_index')
+ ]
+
+ operations = [
+ migrations.AddConstraint(
+ model_name='transcription',
+ constraint=models.CheckConstraint(
+ check=~models.Q(source_id__isnull=False, worker_version_id__isnull=False),
+ name='transcription_source_not_worker_version',
+ )
+ ),
+ migrations.RunPython(
+ preflight_checks,
+ reverse_code=migrations.RunPython.noop,
+ elidable=True,
+ ),
+ migrations.RunSQL(
+ FORWARD_SQL,
+ reverse_sql=migrations.RunSQL.noop,
+ elidable=True,
+ ),
+ ]
diff --git a/arkindex/documents/migrations/0022_remove_transcription_zone.py b/arkindex/documents/migrations/0022_remove_transcription_zone.py
new file mode 100644
index 0000000000000000000000000000000000000000..edb333dab4f65cef1068ba654291e1468a32a201
--- /dev/null
+++ b/arkindex/documents/migrations/0022_remove_transcription_zone.py
@@ -0,0 +1,80 @@
+# Generated by Django 3.1 on 2020-09-01 07:48
+
+from django.db import migrations
+
+FORWARD_SQL = [
+ # Use a temporary table here to iterate over transcriptions just once before deleting,
+ # causing this migration to only take a few minutes
+ # Note the strange join conditions as either source_id or worker_version_id are NULL,
+ # which causes a NATURAL JOIN or a JOIN … USING to fail since comparing NULLs returns NULL.
+ """
+ CREATE TEMPORARY TABLE duplicate_ids AS
+ WITH filters AS (
+ SELECT
+ sub.*,
+ FIRST_VALUE(id) OVER (
+ PARTITION BY
+ transcription.element_id,
+ transcription.source_id,
+ transcription.worker_version_id
+ ) AS keep_id
+ FROM documents_transcription transcription
+ INNER JOIN (
+ SELECT element_id, source_id, worker_version_id
+ FROM documents_transcription
+ GROUP BY element_id, source_id, worker_version_id
+ HAVING COUNT(*) > 1
+ ) sub ON (
+ sub.element_id = transcription.element_id AND (
+ sub.source_id = transcription.source_id
+ OR sub.worker_version_id = transcription.worker_version_id
+ )
+ )
+ )
+ SELECT id
+ FROM documents_transcription transcription
+ INNER JOIN filters ON (
+ filters.element_id = transcription.element_id AND (
+ filters.source_id = transcription.source_id
+ OR filters.worker_version_id = transcription.worker_version_id
+ )
+ )
+ WHERE keep_id != id;
+ """,
+ # Remove any TranscriptionEntity that could be linked to the duplicate transcriptions
+ """
+ DELETE FROM documents_transcriptionentity transcriptionentity
+ USING duplicate_ids
+ WHERE transcriptionentity.transcription_id = duplicate_ids.id;
+ """,
+ # Remove duplicate transcriptions
+ """
+ DELETE FROM documents_transcription transcription
+ USING duplicate_ids
+ WHERE transcription.id = duplicate_ids.id;
+ """,
+ 'DROP TABLE duplicate_ids;',
+]
+
+
+class Migration(migrations.Migration):
+
+ dependencies = [
+ ('documents', '0021_move_transcriptions'),
+ ]
+
+ operations = [
+ migrations.RemoveField(
+ model_name='transcription',
+ name='zone',
+ ),
+ # Remove the few remaining transcriptions that would break the unique constraints we will add in documents.0023.
+ # Those are transcriptions from the same source, on the same element, with the exact same zones.
+ # This query is rather complex as we want to only remove duplicates, and window functions have their limits,
+ # but the GROUP BY…HAVING will quickly exclude most of the table so it isn't slow.
+ migrations.RunSQL(
+ FORWARD_SQL,
+ reverse_sql=migrations.RunSQL.noop,
+ elidable=True,
+ ),
+ ]
diff --git a/arkindex/documents/models.py b/arkindex/documents/models.py
index 7f77e4d49446607b318f429da5d881962d9fafb2..9f4386df43feb4289c556b49773abf7684f1c12a 100644
--- a/arkindex/documents/models.py
+++ b/arkindex/documents/models.py
@@ -440,12 +440,6 @@ class Transcription(models.Model):
max_length=50,
db_index=True,
)
- zone = models.ForeignKey(
- 'images.Zone',
- on_delete=models.PROTECT,
- related_name='transcriptions',
- null=True,
- )
source = models.ForeignKey(
DataSource,
on_delete=models.CASCADE,
@@ -468,17 +462,18 @@ class Transcription(models.Model):
related_name='transcriptions',
)
- class Meta:
- # The following index was attempted with md5(text) in a manual migration
- # but it causes too many performance issues.
- # unique_together = (
- # ('element', 'zone', 'text')
- # )
- pass
-
def __str__(self):
return 'Transcription: {}'.format(self.text[:20])
+ class Meta:
+ constraints = [
+ # Require either a source, a worker version, or none (manual), but not both at once
+ models.CheckConstraint(
+ check=~Q(source_id__isnull=False, worker_version_id__isnull=False),
+ name='transcription_source_not_worker_version',
+ )
+ ]
+
class TranscriptionEntity(models.Model):
"""
diff --git a/arkindex/documents/search.py b/arkindex/documents/search.py
index bfc09fcd2ae8236cfe353317f032089c3e7ce1d5..6c69a9713cd353ed392abc38e152b6ecbd0432d1 100644
--- a/arkindex/documents/search.py
+++ b/arkindex/documents/search.py
@@ -1,5 +1,6 @@
-from arkindex.documents.models import Transcription, Element, Entity
from itertools import chain
+from django.db.models import prefetch_related_objects
+from arkindex.documents.models import Transcription, Element, Entity
import uuid
@@ -12,7 +13,7 @@ def search_transcriptions_post(data):
ts = Transcription.objects \
.filter(id__in=transcription_ids) \
.order_by('-score') \
- .prefetch_related('zone__image__server', 'element', 'source')
+ .prefetch_related('element__zone__image__server', 'source')
element_ids = list(ts.values_list('element_id', flat=True))
all_parent_paths = Element.objects.get_ascendings_paths(*element_ids)
for trans in ts:
@@ -62,7 +63,7 @@ def search_elements_post(data):
transcriptions = {
t.id: t
- for t in Transcription.objects.filter(id__in=tr_ids).prefetch_related('zone__image__server', 'source')
+ for t in Transcription.objects.filter(id__in=tr_ids).prefetch_related('source')
}
elts_tr_ids = {
@@ -78,11 +79,15 @@ def search_elements_post(data):
for result in data
}
- elts = list(Element.objects.filter(id__in=elt_ids).prefetch_related('corpus', 'type'))
+ elts = list(Element.objects.filter(id__in=elt_ids).prefetch_related('corpus', 'type', 'zone__image__server'))
# Preserve the ordering given by ElasticSearch
ordered_elts = list(filter(None, map(lambda eid: next((e for e in elts if e.id == eid), None), elt_ids)))
all_paths = Element.objects.get_ascendings_paths(*(e.id for e in ordered_elts))
+ prefetch_related_objects(
+ [element for paths in all_paths.values() for path in paths for element in path],
+ 'type',
+ )
for elt in ordered_elts:
elt.transcriptions_results = list(filter(None, [transcriptions.get(tid) for tid in elts_tr_ids[elt.id]]))
diff --git a/arkindex/documents/serializers/iiif/annotations.py b/arkindex/documents/serializers/iiif/annotations.py
index 9b21d19f1c87d8532f8ae274691041aefc011d3b..0fdbb9fa99bf3dbfbc2efc970caddb1e421e2d51 100644
--- a/arkindex/documents/serializers/iiif/annotations.py
+++ b/arkindex/documents/serializers/iiif/annotations.py
@@ -1,5 +1,6 @@
from abc import ABC, abstractmethod
from django.conf import settings
+from django.db.models import Q
from rest_framework import serializers
from arkindex.documents.models import Element, Transcription
from arkindex.project.tools import build_absolute_url, bounding_box
@@ -77,7 +78,7 @@ class TranscriptionSearchAnnotationSerializer(TranscriptionAnnotationSerializer)
def get_target(self, ts):
assert isinstance(ts, Transcription)
url = build_absolute_url(ts.element, self.context['request'], 'api:iiif-canvas')
- x, y, w, h = bounding_box(ts.zone.polygon)
+ x, y, w, h = bounding_box(ts.element.zone.polygon)
return f'{url}#xywh={x},{y},{w},{h}'
@@ -103,8 +104,11 @@ class AnnotationListSerializer(serializers.BaseSerializer):
}
def get_elements(self, element):
- return element.transcriptions.all()
"Get a list of elements to serialize as annotations."
+ return Transcription.objects.filter(
+ Q(element=element)
+ | Q(element__in=Element.objects.get_descending(element.id))
+ )
class ElementAnnotationListSerializer(AnnotationListSerializer):
diff --git a/arkindex/documents/serializers/ml.py b/arkindex/documents/serializers/ml.py
index 65a34c6e7d267d75baddd25d5aa2ab8371ecc91d..a72388fb09beea768adbf1072fee2730a917cb38 100644
--- a/arkindex/documents/serializers/ml.py
+++ b/arkindex/documents/serializers/ml.py
@@ -10,7 +10,6 @@ from arkindex.documents.models import (
Corpus, Element, ElementType, Transcription, DataSource, MLClass, Classification, ClassificationState
)
from arkindex.project.serializer_fields import EnumField, LinearRingField
-from arkindex.images.serializers import ZoneSerializer
from arkindex.documents.serializers.light import ElementZoneSerializer
import uuid
@@ -232,18 +231,16 @@ class TranscriptionSerializer(serializers.ModelSerializer):
Serialises a Transcription
"""
type = EnumField(TranscriptionType, read_only=True)
- zone = ZoneSerializer(read_only=True)
source = DataSourceSerializer(read_only=True)
class Meta:
model = Transcription
- read_only_fields = ('id', 'type', 'score', 'zone', 'source')
+ read_only_fields = ('id', 'type', 'score', 'source')
fields = (
'id',
'type',
'text',
'score',
- 'zone',
'source',
'worker_version_id',
)
diff --git a/arkindex/documents/serializers/search.py b/arkindex/documents/serializers/search.py
index 9bd27fbbe31554d82febad4050b17102c551e920..3b6a6dfce020d6f07a9e5a34ab92a170b1a486a2 100644
--- a/arkindex/documents/serializers/search.py
+++ b/arkindex/documents/serializers/search.py
@@ -2,10 +2,10 @@ from django.conf import settings
from rest_framework import serializers
from arkindex_common.enums import EntityType, TranscriptionType
from arkindex.project.serializer_fields import EnumField
+from arkindex.images.serializers import ZoneSerializer
from arkindex.documents.date_parser import parse_date
from arkindex.documents.models import Element, ElementType, Entity
-from arkindex.documents.serializers.light import CorpusLightSerializer
-from arkindex.documents.serializers.elements import ElementSlimSerializer
+from arkindex.documents.serializers.light import CorpusLightSerializer, ElementLightSerializer
from arkindex.documents.serializers.ml import TranscriptionSerializer
import math
@@ -84,16 +84,16 @@ class IIIFSearchQuerySerializer(serializers.Serializer):
q = serializers.CharField(source='query', max_length=settings.ES_QUERY_STRING_MAX_LENGTH)
-class ElementSearchResultSerializer(serializers.ModelSerializer):
+class ElementSearchResultSerializer(ElementLightSerializer):
"""
A page search result with nested transcriptions
"""
- type = serializers.SlugRelatedField(slug_field='slug', read_only=True)
+ zone = ZoneSerializer()
transcriptions = TranscriptionSerializer(many=True, source='transcriptions_results')
total_transcriptions = serializers.IntegerField()
parent_paths = serializers.ListField(
child=serializers.ListField(
- child=ElementSlimSerializer()
+ child=ElementLightSerializer()
),
)
corpus = CorpusLightSerializer()
@@ -106,6 +106,7 @@ class ElementSearchResultSerializer(serializers.ModelSerializer):
'id',
'name',
'type',
+ 'zone',
'transcriptions',
'total_transcriptions',
'parent_paths',
diff --git a/arkindex/documents/tests/consumers/test_corpus_consumer.py b/arkindex/documents/tests/consumers/test_corpus_consumer.py
index f9252283e0cef1b5c2a86779f28f057a199f888d..e677e2c4f2e0159452c97502bf2adf9674592f7e 100644
--- a/arkindex/documents/tests/consumers/test_corpus_consumer.py
+++ b/arkindex/documents/tests/consumers/test_corpus_consumer.py
@@ -71,7 +71,6 @@ class TestDeleteCorpus(FixtureTestCase):
revision='-1',
internal=False,
),
- zone=cls.zone,
type=TranscriptionType.Word,
text='hi',
score=0.75,
@@ -122,7 +121,6 @@ class TestDeleteCorpus(FixtureTestCase):
ts = self.page.transcriptions.get()
self.assertEqual(ts.source.slug, 'reco')
- self.assertEqual(ts.zone, self.zone)
self.assertEqual(ts.type, TranscriptionType.Word)
self.assertEqual(ts.text, 'hi')
self.assertEqual(ts.score, 0.75)
diff --git a/arkindex/documents/tests/consumers/test_ml_results_consumer.py b/arkindex/documents/tests/consumers/test_ml_results_consumer.py
index c66f8524e031742f7d680b7fd26146cabf1842cc..1612370bffddb3ab3ec5ffcfe28271acef3ed7fe 100644
--- a/arkindex/documents/tests/consumers/test_ml_results_consumer.py
+++ b/arkindex/documents/tests/consumers/test_ml_results_consumer.py
@@ -64,7 +64,6 @@ class TestMLResultsConsumer(FixtureTestCase):
text='some text',
type=TranscriptionType.Line,
source=source,
- zone=cls.page2.zone,
).transcription_entities.create(
offset=0,
length=1,
diff --git a/arkindex/documents/tests/consumers/test_reindex_consumer.py b/arkindex/documents/tests/consumers/test_reindex_consumer.py
index 784908176fbed4ab38e4f3d91c804179bf11e288..bd5acb84965ae98a8863ccf9d870e9c0dcbab796 100644
--- a/arkindex/documents/tests/consumers/test_reindex_consumer.py
+++ b/arkindex/documents/tests/consumers/test_reindex_consumer.py
@@ -41,7 +41,6 @@ class TestReindexConsumer(FixtureTestCase):
text='something',
type=TranscriptionType.Word,
source=source,
- zone=zone,
)
ts.transcription_entities.create(
entity=corpus2.entities.create(
@@ -74,8 +73,7 @@ class TestReindexConsumer(FixtureTestCase):
(queryset, ), kwargs = call_args
self.assertQuerysetEqual(
queryset,
- # Only transcriptions with a zone may be added to transcriptions index
- Transcription.objects.filter(zone__isnull=False)
+ Transcription.objects.all()
)
self.assertDictEqual(kwargs, {'bulk_size': 400})
@@ -150,7 +148,7 @@ class TestReindexConsumer(FixtureTestCase):
(queryset, ), kwargs = ts_call
self.assertQuerysetEqual(
queryset,
- Transcription.objects.filter(element__corpus_id=self.corpus.id, zone__isnull=False)
+ Transcription.objects.filter(element__corpus_id=self.corpus.id)
)
self.assertDictEqual(kwargs, {'bulk_size': 400})
@@ -160,7 +158,7 @@ class TestReindexConsumer(FixtureTestCase):
})
self.assertEqual(mock().drop_index.call_count, 0)
self.assertEqual(mock().run_index.call_count, 3)
- entities_call, ts_call, elements_call = sorted(mock().run_index.call_args_list, key=repr)
+ elements_call, entities_call, ts_call = sorted(mock().run_index.call_args_list, key=repr)
elements_list = list(Element.objects.get_descending(self.folder.id))
elements_list.append(self.folder)
@@ -182,21 +180,6 @@ class TestReindexConsumer(FixtureTestCase):
(queryset, ), kwargs = ts_call
self.assertQuerysetEqual(
queryset,
- Transcription.objects.filter(element__in=elements_list, zone__isnull=False)
+ Transcription.objects.filter(element__in=elements_list)
)
self.assertDictEqual(kwargs, {'bulk_size': 400})
-
- def test_reindex_transcriptions_without_zone(self, mock):
- """
- Transcriptions with no zone may not be indexed in transcriptions index
- """
- ReindexConsumer({}).reindex_start({
- 'transcriptions': True,
- 'entities': False,
- 'elements': False,
- })
- transcription = Transcription.objects.filter(zone__isnull=True).first()
- self.assertNotEqual(transcription, None)
- self.assertEqual(mock().run_index.call_count, 1)
- (queryset, ), kwargs = mock().run_index.call_args
- self.assertFalse(queryset.filter(id=transcription.id).exists())
diff --git a/arkindex/documents/tests/test_bulk_element_transcriptions.py b/arkindex/documents/tests/test_bulk_element_transcriptions.py
index ee81ca3fc470a29667573ede1164a810e283c9e4..94f66aad33c783b9dc76f58881729142098f0df5 100644
--- a/arkindex/documents/tests/test_bulk_element_transcriptions.py
+++ b/arkindex/documents/tests/test_bulk_element_transcriptions.py
@@ -7,8 +7,7 @@ from rest_framework import status
from arkindex.project.tests import FixtureAPITestCase
from arkindex_common.enums import TranscriptionType
from arkindex.dataimport.models import WorkerVersion
-from arkindex.documents.models import Element, Corpus, DataSource
-from arkindex_common.ml_tool import MLToolType
+from arkindex.documents.models import Element, Corpus
import uuid
@@ -22,7 +21,6 @@ class TestBulkElementTranscriptions(FixtureAPITestCase):
super().setUpTestData()
cls.page = cls.corpus.elements.get(name='Volume 1, page 1r')
cls.vol = cls.corpus.elements.get(name='Volume 1')
- cls.src = DataSource.objects.get(slug='test', type=MLToolType.Recognizer)
cls.line = cls.corpus.elements.filter(type__slug='text_line').first()
cls.private_corpus = Corpus.objects.create(name='Private')
cls.private_page = cls.private_corpus.elements.create(type=cls.page.type)
@@ -34,8 +32,16 @@ class TestBulkElementTranscriptions(FixtureAPITestCase):
Bulk creates a list of element with an attached transcription generated by a worker_version
"""
get_layer_mock.return_value.send = AsyncMock()
- self.src.internal = True
- self.src.save()
+
+ # Create a manual transcription on the element
+ self.line.transcriptions.create(
+ text='A manual transcription',
+ worker_version=self.worker_version,
+ type=TranscriptionType.Line
+ )
+ self.assertEqual(self.line.transcriptions.count(), 1)
+
+ existing_element_ids = list(Element.objects.get_descending(self.page.id).values_list('id', flat=True))
transcriptions = [
([[13, 37], [133, 37], [133, 137], [13, 137], [13, 37]], 'Hello world !', 0.1337),
@@ -59,7 +65,7 @@ class TestBulkElementTranscriptions(FixtureAPITestCase):
)
self.assertEqual(response.status_code, status.HTTP_201_CREATED)
- created_elts = Element.objects.get_descending(self.page.id).exclude(id=self.line.id)
+ created_elts = Element.objects.get_descending(self.page.id).exclude(id__in=existing_element_ids)
self.assertEqual(created_elts.count(), 2)
self.assertTrue(all(map(lambda elt: elt.zone.image == self.page.zone.image, created_elts)))
self.assertListEqual(
@@ -73,10 +79,10 @@ class TestBulkElementTranscriptions(FixtureAPITestCase):
]
)
self.assertCountEqual(
- created_elts.values_list('transcriptions__type', 'transcriptions__text', 'transcriptions__zone', 'transcriptions__source', 'transcriptions__worker_version'),
+ created_elts.values_list('transcriptions__type', 'transcriptions__text', 'transcriptions__source', 'transcriptions__worker_version'),
[
- (TranscriptionType.Line, ('Hello world !'), None, None, self.worker_version.id),
- (TranscriptionType.Line, ('I <3 JavaScript'), None, None, self.worker_version.id)
+ (TranscriptionType.Line, ('Hello world !'), None, self.worker_version.id),
+ (TranscriptionType.Line, ('I <3 JavaScript'), None, self.worker_version.id)
]
)
get_layer_mock().send.assert_called_once_with('reindex', {
@@ -95,9 +101,6 @@ class TestBulkElementTranscriptions(FixtureAPITestCase):
"""
Sending 100 transcriptions should not increase the number of DB queries
"""
- self.src.internal = True
- self.src.save()
-
# Create 100 transcriptions
transcriptions = [(
[[i, i], [i, i + 20], [i + 20, i + 20], [i + 20, i], [i, i]],
@@ -139,8 +142,6 @@ class TestBulkElementTranscriptions(FixtureAPITestCase):
Does not erase present transcriptions, even from the same worker
"""
get_layer_mock.return_value.send = AsyncMock()
- self.src.internal = True
- self.src.save()
# Create a manual transcription on the element
self.line.transcriptions.create(
@@ -150,6 +151,7 @@ class TestBulkElementTranscriptions(FixtureAPITestCase):
)
self.assertEqual(self.line.transcriptions.count(), 1)
+ existing_element_ids = list(Element.objects.get_descending(self.page.id).values_list('id', flat=True))
transcriptions = [
([[13, 37], [133, 37], [133, 137], [13, 137], [13, 37]], 'Hello world !', 0.1337),
# Use line zone to create the second transcription
@@ -175,13 +177,12 @@ class TestBulkElementTranscriptions(FixtureAPITestCase):
)
self.assertEqual(response.status_code, status.HTTP_201_CREATED)
- page_elts = Element.objects.get_descending(self.page.id)
+ created_elts = Element.objects.get_descending(self.page.id).exclude(id__in=existing_element_ids)
# The existing text line has been reused
- self.assertEqual(page_elts.count(), 2)
- # There are now two transcriptions on the line
- self.assertCountEqual(self.line.transcriptions.values_list('type', 'text', 'zone'), [
- (TranscriptionType.Line, 'A manual transcription', None),
- (TranscriptionType.Line, 'I <3 JavaScript', None)
+ self.assertEqual(created_elts.count(), 1)
+ self.assertCountEqual(self.line.transcriptions.values_list('type', 'text'), [
+ (TranscriptionType.Line, 'A manual transcription'),
+ (TranscriptionType.Line, 'I <3 JavaScript')
])
def test_bulk_transcriptions_requires_verified(self):
@@ -268,14 +269,9 @@ class TestBulkElementTranscriptions(FixtureAPITestCase):
def test_bulk_transcriptions_return_ordered_elements(self):
"""
return_elements field allows to serialize an ordered list of sub-elements holding the transcriptions
- Returned elements may be duplicated in case the same zone is used for multiple transcriptions
"""
- self.src.internal = True
- self.src.save()
-
transcriptions = [
([[13, 37], [133, 37], [133, 137], [13, 137], [13, 37]], 'stuck', 0.13),
- ([[13, 37], [133, 37], [133, 137], [13, 137], [13, 37]], 'stick', 0.33),
([[24, 42], [64, 42], [64, 142], [24, 142], [24, 42]], 'stock', 0.37),
# The existing line has the smallest ordering
(self.line.zone.polygon.coords, 'stack', 0.42),
@@ -302,16 +298,13 @@ class TestBulkElementTranscriptions(FixtureAPITestCase):
self.assertEqual(response.status_code, status.HTTP_201_CREATED)
created_elts = Element.objects.get_descending(self.page.id).exclude(id=self.line.id)
- self.assertEqual(created_elts.count(), 2)
+ self.assertEqual(created_elts.count(), 5)
first_elt = created_elts.get(transcriptions__text='stuck')
second_elt = created_elts.get(transcriptions__text='stock')
self.assertListEqual(response.json(), [{
'id': str(first_elt.id),
'created': True
- }, {
- 'id': str(first_elt.id),
- 'created': False
}, {
'id': str(second_elt.id),
'created': True
diff --git a/arkindex/documents/tests/test_children_elements.py b/arkindex/documents/tests/test_children_elements.py
index 2423ae98229de620251581715943d1b0613ca58e..f46f6faeec64a92bb7c80c43317a931156937814 100644
--- a/arkindex/documents/tests/test_children_elements.py
+++ b/arkindex/documents/tests/test_children_elements.py
@@ -12,11 +12,11 @@ class TestChildrenElements(FixtureAPITestCase):
def setUpTestData(cls):
super().setUpTestData()
cls.vol = cls.corpus.elements.get(name='Volume 1')
- cls.element = Element.objects.get(name='Volume 1, page 2r')
cls.worker_version = WorkerVersion.objects.get(worker__slug='dla')
def test_element_children(self):
- response = self.client.get(reverse('api:elements-children', kwargs={'pk': str(self.element.id)}))
+ lonely_element = Element.objects.get(name='Volume 2, page 2r')
+ response = self.client.get(reverse('api:elements-children', kwargs={'pk': str(lonely_element.id)}))
self.assertEqual(response.status_code, status.HTTP_200_OK)
self.assertListEqual(response.json()['results'], [])
@@ -47,10 +47,10 @@ class TestChildrenElements(FixtureAPITestCase):
response = self.client.get(reverse('api:elements-children', kwargs={'pk': str(self.vol.id)}) + '?recursive')
self.assertEqual(response.status_code, status.HTTP_200_OK)
- self.assertEqual(len(response.json()['results']), 15)
+ self.assertEqual(len(response.json()['results']), 20)
self.assertSetEqual(
{r['type'] for r in response.json()['results']},
- {'act', 'page', 'surface', 'text_line'},
+ {'act', 'page', 'surface', 'text_line', 'word'},
)
response = self.client.get(reverse('api:elements-children', kwargs={'pk': str(self.vol.id)}) + '?type=page')
@@ -154,6 +154,13 @@ class TestChildrenElements(FixtureAPITestCase):
)
def test_children_with_has_children(self):
+ # Create a childless element
+ element = self.corpus.elements.create(
+ name='Spearow',
+ type=self.corpus.types.get(slug='surface')
+ )
+ element.add_parent(self.vol)
+
with self.assertNumQueries(9):
response = self.client.get(
reverse('api:elements-children', kwargs={'pk': str(self.vol.id)}),
@@ -169,8 +176,9 @@ class TestChildrenElements(FixtureAPITestCase):
'Act 4': True,
'Act 5': True,
'Volume 1, page 1r': True,
- 'Volume 1, page 1v': False,
- 'Volume 1, page 2r': False,
+ 'Volume 1, page 1v': True,
+ 'Volume 1, page 2r': True,
+ 'Spearow': False,
}
)
diff --git a/arkindex/documents/tests/test_corpus.py b/arkindex/documents/tests/test_corpus.py
index 8552c664339c8209c95af5c78f4279b2c36993a1..f46df8feb0d0725726e25ae427be00902775a4fb 100644
--- a/arkindex/documents/tests/test_corpus.py
+++ b/arkindex/documents/tests/test_corpus.py
@@ -15,6 +15,45 @@ FAKE_NOW = datetime.datetime.now()
# Fake DB fixtures creation date
DB_CREATED = '2020-02-02T01:23:45.678000Z'
+EXPECTED_CORPUS_TYPES = [
+ {
+ 'slug': 'volume',
+ 'display_name': 'Volume',
+ 'folder': True,
+ 'allowed_transcription': None
+ },
+ {
+ 'slug': 'page',
+ 'display_name': 'Page',
+ 'folder': False,
+ 'allowed_transcription': None
+ },
+ {
+ 'slug': 'act',
+ 'display_name': 'Act',
+ 'folder': False,
+ 'allowed_transcription': None
+ },
+ {
+ 'slug': 'surface',
+ 'display_name': 'Surface',
+ 'folder': False,
+ 'allowed_transcription': None
+ },
+ {
+ 'slug': 'text_line',
+ 'display_name': 'Line',
+ 'folder': False,
+ 'allowed_transcription': 'line'
+ },
+ {
+ 'slug': 'word',
+ 'display_name': 'Word',
+ 'folder': False,
+ 'allowed_transcription': 'word'
+ }
+]
+
class TestCorpus(FixtureAPITestCase):
@@ -43,38 +82,7 @@ class TestCorpus(FixtureAPITestCase):
types = data[0].pop('types')
for t in types:
del t['id']
- self.assertCountEqual(types, [
- {
- 'slug': 'volume',
- 'display_name': 'Volume',
- 'folder': True,
- 'allowed_transcription': None
- },
- {
- 'slug': 'page',
- 'display_name': 'Page',
- 'folder': False,
- 'allowed_transcription': None
- },
- {
- 'slug': 'act',
- 'display_name': 'Act',
- 'folder': False,
- 'allowed_transcription': None
- },
- {
- 'slug': 'surface',
- 'display_name': 'Surface',
- 'folder': False,
- 'allowed_transcription': None
- },
- {
- 'slug': 'text_line',
- 'display_name': 'Line',
- 'folder': False,
- 'allowed_transcription': 'line'
- }
- ])
+ self.assertCountEqual(types, EXPECTED_CORPUS_TYPES)
self.assertListEqual(
data,
@@ -103,38 +111,7 @@ class TestCorpus(FixtureAPITestCase):
types = data[1].pop('types')
for t in types:
del t['id']
- self.assertCountEqual(types, [
- {
- 'slug': 'volume',
- 'display_name': 'Volume',
- 'folder': True,
- 'allowed_transcription': None
- },
- {
- 'slug': 'page',
- 'display_name': 'Page',
- 'folder': False,
- 'allowed_transcription': None
- },
- {
- 'slug': 'act',
- 'display_name': 'Act',
- 'folder': False,
- 'allowed_transcription': None
- },
- {
- 'slug': 'surface',
- 'display_name': 'Surface',
- 'folder': False,
- 'allowed_transcription': None
- },
- {
- 'slug': 'text_line',
- 'display_name': 'Line',
- 'folder': False,
- 'allowed_transcription': 'line'
- }
- ])
+ self.assertCountEqual(types, EXPECTED_CORPUS_TYPES)
self.assertListEqual(
data,
@@ -173,38 +150,7 @@ class TestCorpus(FixtureAPITestCase):
types = data[2].pop('types')
for t in types:
del t['id']
- self.assertCountEqual(types, [
- {
- 'slug': 'volume',
- 'display_name': 'Volume',
- 'folder': True,
- 'allowed_transcription': None
- },
- {
- 'slug': 'page',
- 'display_name': 'Page',
- 'folder': False,
- 'allowed_transcription': None
- },
- {
- 'slug': 'act',
- 'display_name': 'Act',
- 'folder': False,
- 'allowed_transcription': None
- },
- {
- 'slug': 'surface',
- 'display_name': 'Surface',
- 'folder': False,
- 'allowed_transcription': None
- },
- {
- 'slug': 'text_line',
- 'display_name': 'Line',
- 'folder': False,
- 'allowed_transcription': 'line'
- }
- ])
+ self.assertCountEqual(types, EXPECTED_CORPUS_TYPES)
self.assertListEqual(
data,
@@ -336,38 +282,7 @@ class TestCorpus(FixtureAPITestCase):
types = data.pop('types')
for t in types:
del t['id']
- self.assertCountEqual(types, [
- {
- 'slug': 'volume',
- 'display_name': 'Volume',
- 'folder': True,
- 'allowed_transcription': None
- },
- {
- 'slug': 'page',
- 'display_name': 'Page',
- 'folder': False,
- 'allowed_transcription': None
- },
- {
- 'slug': 'act',
- 'display_name': 'Act',
- 'folder': False,
- 'allowed_transcription': None
- },
- {
- 'slug': 'surface',
- 'display_name': 'Surface',
- 'folder': False,
- 'allowed_transcription': None
- },
- {
- 'slug': 'text_line',
- 'display_name': 'Line',
- 'folder': False,
- 'allowed_transcription': 'line'
- }
- ])
+ self.assertCountEqual(types, EXPECTED_CORPUS_TYPES)
self.assertDictEqual(data, {
'id': str(self.corpus_public.id),
'name': self.corpus_public.name,
@@ -491,14 +406,14 @@ class TestCorpus(FixtureAPITestCase):
Ensure corpus deletion deletes elements first to prevent a ProtectedError
"""
self.assertEqual(Corpus.objects.count(), 3)
- self.assertEqual(Element.objects.count(), 20)
+ self.assertEqual(Element.objects.count(), 29)
# Use Django's default deletion method and expect a ProtectedError
with self.assertRaises(ProtectedError):
Model.delete(self.corpus)
self.assertEqual(Corpus.objects.count(), 3)
- self.assertEqual(Element.objects.count(), 20)
+ self.assertEqual(Element.objects.count(), 29)
# Use the corpus deletion method and it works!
self.corpus.delete()
diff --git a/arkindex/documents/tests/test_corpus_elements.py b/arkindex/documents/tests/test_corpus_elements.py
index e8dedd55c88ac95ff418c34e1af6ca17d13a5d27..2a348376fff5294d6d1ad2e78919a92107b63a4a 100644
--- a/arkindex/documents/tests/test_corpus_elements.py
+++ b/arkindex/documents/tests/test_corpus_elements.py
@@ -68,7 +68,6 @@ class TestListElements(FixtureAPITestCase):
)
self.assertEqual(response.status_code, status.HTTP_200_OK)
data = response.json()
- self.assertIsNone(data['next'])
self.assertDictEqual(
{element['name']: element['has_children'] for element in data['results']},
@@ -87,8 +86,8 @@ class TestListElements(FixtureAPITestCase):
'Surface F': False,
'Volume 1': True,
'Volume 1, page 1r': True,
- 'Volume 1, page 1v': False,
- 'Volume 1, page 2r': False,
+ 'Volume 1, page 1v': True,
+ 'Volume 1, page 2r': True,
'Volume 2': True,
'Volume 2, page 1r': False,
'Volume 2, page 1v': False,
diff --git a/arkindex/documents/tests/test_create_transcriptions.py b/arkindex/documents/tests/test_create_transcriptions.py
index 999fc147aa70bd419a38c9fa09af249de36d07eb..f3415b7063d5b0b72646971a64def664ae180b57 100644
--- a/arkindex/documents/tests/test_create_transcriptions.py
+++ b/arkindex/documents/tests/test_create_transcriptions.py
@@ -81,11 +81,9 @@ class TestTranscriptionCreate(FixtureAPITestCase):
'type': 'line',
'source': None,
'worker_version_id': None,
- 'zone': None
})
- new_ts = Transcription.objects.get(text='A perfect day in a perfect place', type=TranscriptionType.Line.value)
- self.assertIsNone(new_ts.zone)
+ new_ts = Transcription.objects.get(text='A perfect day in a perfect place', type=TranscriptionType.Line)
self.assertIsNone(new_ts.score)
self.assertEqual(new_ts.worker_version, None)
self.assertTrue(self.line.transcriptions.filter(pk=new_ts.id).exists())
@@ -101,28 +99,6 @@ class TestTranscriptionCreate(FixtureAPITestCase):
'drop': False,
})
- @patch('arkindex.project.triggers.get_channel_layer')
- def test_create_transcription_takes_no_zone(self, get_layer_mock):
- """
- TranscriptionCreate endpoint do not create any zone for the transcription
- """
- get_layer_mock.return_value.send = AsyncMock()
-
- self.client.force_login(self.user)
- response = self.client.post(
- reverse('api:transcription-create', kwargs={'pk': self.line.id}),
- format='json',
- data={
- 'type': 'line',
- 'polygon': [(0, 0), (42, 0), (42, 42), (0, 42), (0, 0)],
- 'text': 'SQUARE',
- }
- )
- self.assertEqual(response.status_code, status.HTTP_201_CREATED)
-
- new_ts = Transcription.objects.get(text='SQUARE', type=TranscriptionType.Line.value)
- self.assertEqual(new_ts.zone, None)
-
@patch('arkindex.project.triggers.get_channel_layer')
def test_create_duplicated_transcription(self, get_layer_mock):
"""
@@ -222,7 +198,6 @@ class TestTranscriptionCreate(FixtureAPITestCase):
'text': 'NEKUDOTAYIM',
'type': 'word',
'worker_version_id': str(self.worker_version.id),
- 'zone': None
})
get_layer_mock().send.assert_called_once_with('reindex', {
diff --git a/arkindex/documents/tests/test_edit_transcriptions.py b/arkindex/documents/tests/test_edit_transcriptions.py
index 37b88f76f51cb51998f39b52dd3305620c029075..d2f0678a06712a10a25de1f7f20d3799d868fee2 100644
--- a/arkindex/documents/tests/test_edit_transcriptions.py
+++ b/arkindex/documents/tests/test_edit_transcriptions.py
@@ -2,7 +2,7 @@ from django.urls import reverse
from rest_framework import status
from arkindex.project.tests import FixtureAPITestCase
from arkindex_common.enums import TranscriptionType
-from arkindex.documents.models import Corpus, Transcription, DataSource
+from arkindex.documents.models import Corpus, Element, Transcription, DataSource
from arkindex.users.models import User
from arkindex_common.ml_tool import MLToolType
from uuid import uuid4
@@ -16,16 +16,17 @@ class TestEditTranscription(FixtureAPITestCase):
@classmethod
def setUpTestData(cls):
super().setUpTestData()
- cls.page = cls.corpus.elements.get(name='Volume 1, page 1r')
+ page = cls.corpus.elements.get(name='Volume 1, page 1r')
+ element = Element.objects.get_descending(page.id).filter(type__slug='word', name='PARIS').get()
cls.line = cls.corpus.elements.filter(type__slug='text_line').first()
- cls.ml_transcription = cls.page.transcriptions.get(text='PARIS')
- cls.private_corpus = Corpus.objects.create(name='Private')
- cls.private_page = cls.private_corpus.elements.create(type=cls.page.type)
+ cls.ml_transcription = element.transcriptions.get(text='PARIS')
+ private_corpus = Corpus.objects.create(name='Private')
+ cls.private_page = private_corpus.elements.create(type=page.type)
# Create an user with a read right only on the private corpus
cls.private_read_user = User.objects.create_user('a@bc.de', 'a')
cls.private_read_user.verified_email = True
cls.private_read_user.save()
- cls.private_corpus.corpus_right.create(user=cls.private_read_user)
+ private_corpus.corpus_right.create(user=cls.private_read_user)
def setUp(self):
self.manual_source = DataSource.objects.create(type=MLToolType.Recognizer, slug='manual', internal=False)
@@ -63,7 +64,6 @@ class TestEditTranscription(FixtureAPITestCase):
'text': 'A manual transcription',
'type': 'line',
'worker_version_id': None,
- 'zone': None
})
def test_transcription_retrieve_ml(self):
@@ -135,7 +135,6 @@ class TestEditTranscription(FixtureAPITestCase):
'text': 'a knight was living lonely',
'type': 'line',
'worker_version_id': None,
- 'zone': None
})
def test_transcription_patch_write_right(self):
diff --git a/arkindex/documents/tests/test_entities_api.py b/arkindex/documents/tests/test_entities_api.py
index b538a335a94eb484f4327d28300791827f10d2c1..b6e201e63a416ad39efeff54091ba853f4d29a8d 100644
--- a/arkindex/documents/tests/test_entities_api.py
+++ b/arkindex/documents/tests/test_entities_api.py
@@ -30,10 +30,11 @@ class TestEntitiesAPI(FixtureAPITestCase):
cls.source = DataSource.objects.get(slug='test', type=MLToolType.Recognizer)
cls.private_corpus = Corpus.objects.create(name='private')
cls.worker_version = WorkerVersion.objects.get(worker__slug='reco')
+ cls.page = cls.corpus.elements.get(name='Volume 1, page 1r')
+ cls.element_type = cls.corpus.types.get(slug='text_line')
def setUp(self):
super().setUp()
- self.page = self.corpus.elements.get(name='Volume 1, page 1r')
self.entity = Entity.objects.create(
type=EntityType.Person,
corpus=self.corpus,
@@ -56,17 +57,22 @@ class TestEntitiesAPI(FixtureAPITestCase):
zone, _ = self.page.zone.image.zones.get_or_create(
polygon=LinearRing((0, 0), (0, 42), (42, 42), (42, 0), (0, 0))
)
- self.transcription = self.page.transcriptions.create(
- text='Some transcribed text',
+ self.element = self.corpus.elements.create(
+ type=self.element_type,
+ name='Transcription',
zone=zone,
- source_id=self.source.id,
+ source=self.source,
+ )
+ self.transcription = self.element.transcriptions.create(
+ text='Some transcribed text',
+ source=self.source,
type=TranscriptionType.Line,
)
- self.metadata = self.page.metadatas.create(
+ self.metadata = self.element.metadatas.create(
name='test 1',
type=MetaType.Text,
value='Blah',
- element=self.page,
+ element=self.element,
entity=self.entity,
)
self.transcriptionentity = TranscriptionEntity.objects.create(
@@ -91,19 +97,19 @@ class TestEntitiesAPI(FixtureAPITestCase):
self.assertEqual(data['name'], self.entity.name)
def test_get_entity_elements(self):
- with self.assertNumQueries(9):
+ with self.assertNumQueries(8):
response = self.client.get(reverse('api:entity-elements', kwargs={'pk': str(self.entity.id)}))
self.assertEqual(response.status_code, status.HTTP_200_OK)
data = response.json()
results = data['results']
self.assertEqual(len(results), 1)
- self.assertEqual(results[0]['id'], str(self.page.id))
- self.assertEqual(results[0]['name'], self.page.name)
+ self.assertEqual(results[0]['id'], str(self.element.id))
+ self.assertEqual(results[0]['name'], self.element.name)
def test_get_entity_elements_corpus_acl(self):
self.client.force_login(self.user)
- self.page.corpus = self.private_corpus
- self.page.save()
+ self.element.corpus = self.private_corpus
+ self.element.save()
response = self.client.get(reverse('api:entity-elements', kwargs={'pk': str(self.entity.id)}))
self.assertEqual(response.status_code, status.HTTP_200_OK)
self.assertListEqual(response.json().get('results'), [])
@@ -115,10 +121,11 @@ class TestEntitiesAPI(FixtureAPITestCase):
elt = self.corpus.elements.create(
type=self.corpus.types.get(slug='act'),
name='001',
+ zone=zone,
)
- elt_tr = elt.transcriptions.create(source_id=self.source.id, text='goodbye', zone=zone)
+ elt_tr = elt.transcriptions.create(source_id=self.source.id, text='goodbye')
TranscriptionEntity.objects.create(transcription=elt_tr, entity=self.entity, offset=42, length=7)
- with self.assertNumQueries(9):
+ with self.assertNumQueries(8):
response = self.client.get(reverse('api:entity-elements', kwargs={'pk': str(self.entity.id)}))
self.assertEqual(response.status_code, status.HTTP_200_OK)
data = response.json()
@@ -160,8 +167,8 @@ class TestEntitiesAPI(FixtureAPITestCase):
'url': 'http://server',
}
}
- } if e.zone else None,
- } for e in (elt, self.page)]
+ },
+ } for e in (elt, self.element)]
)
def test_get_role_in_corpus(self):
@@ -405,8 +412,8 @@ class TestEntitiesAPI(FixtureAPITestCase):
def test_create_transcription_entity_wrong_acl(self):
self.client.force_login(self.user)
- self.page.corpus = self.private_corpus
- self.page.save()
+ self.element.corpus = self.private_corpus
+ self.element.save()
response = self.client.post(
reverse('api:transcription-entity-create', kwargs={'pk': str(self.transcription.id)}),
data=self.tr_entities_sample,
@@ -559,19 +566,19 @@ class TestEntitiesAPI(FixtureAPITestCase):
"""
List an element's entities sorted depending on their origin (metadata or transcriptions)
"""
- md = self.page.metadatas.create(name='some_metadata', type=MetaType.Location, value='something')
+ md = self.element.metadatas.create(name='some_metadata', type=MetaType.Location, value='something')
md.entity = self.entity_bis
md.save()
with self.assertNumQueries(9):
- response = self.client.get(reverse('api:element-entities', kwargs={'pk': str(self.page.id)}))
+ response = self.client.get(reverse('api:element-entities', kwargs={'pk': str(self.element.id)}))
self.assertEqual(response.status_code, status.HTTP_200_OK)
data = response.json()
self.assertDictEqual(
data,
{
- 'id': str(self.page.id),
- 'type': self.page.type.slug,
- 'name': self.page.name,
+ 'id': str(self.element.id),
+ 'type': self.element.type.slug,
+ 'name': self.element.name,
'transcriptions': [{
'entity': {
'id': str(t.entity.id),
@@ -617,14 +624,14 @@ class TestEntitiesAPI(FixtureAPITestCase):
'value': m.value,
'revision': m.revision,
'dates': []
- } for m in self.page.metadatas.exclude(entity=None)]
+ } for m in self.element.metadatas.exclude(entity=None)]
}
)
def test_list_element_entities_worker_version_validation(self):
with self.assertNumQueries(1):
response = self.client.get(
- reverse('api:element-entities', kwargs={'pk': str(self.page.id)}),
+ reverse('api:element-entities', kwargs={'pk': str(self.element.id)}),
data={'worker_version': 'blah'}
)
self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
@@ -632,7 +639,7 @@ class TestEntitiesAPI(FixtureAPITestCase):
with self.assertNumQueries(2):
response = self.client.get(
- reverse('api:element-entities', kwargs={'pk': str(self.page.id)}),
+ reverse('api:element-entities', kwargs={'pk': str(self.element.id)}),
data={'worker_version': str(uuid.uuid4())}
)
self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
@@ -645,7 +652,7 @@ class TestEntitiesAPI(FixtureAPITestCase):
with self.assertNumQueries(6):
response = self.client.get(
- reverse('api:element-entities', kwargs={'pk': str(self.page.id)}),
+ reverse('api:element-entities', kwargs={'pk': str(self.element.id)}),
data={'worker_version': str(self.worker_version.id)}
)
self.assertEqual(response.status_code, status.HTTP_200_OK)
@@ -653,9 +660,9 @@ class TestEntitiesAPI(FixtureAPITestCase):
self.assertDictEqual(
response.json(),
{
- 'id': str(self.page.id),
- 'name': 'Volume 1, page 1r',
- 'type': 'page',
+ 'id': str(self.element.id),
+ 'name': 'Transcription',
+ 'type': 'text_line',
'transcriptions': [],
'metadata': [
{
@@ -681,9 +688,9 @@ class TestEntitiesAPI(FixtureAPITestCase):
)
def test_list_element_entities_wrong_acl(self):
- self.page.corpus = self.private_corpus
- self.page.save()
- response = self.client.get(reverse('api:element-entities', kwargs={'pk': str(self.page.id)}))
+ self.element.corpus = self.private_corpus
+ self.element.save()
+ response = self.client.get(reverse('api:element-entities', kwargs={'pk': str(self.element.id)}))
self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
@patch('arkindex.documents.api.entities.ESEntity')
@@ -765,7 +772,7 @@ class TestEntitiesAPI(FixtureAPITestCase):
def test_list_element_links(self):
link = EntityLink.objects.create(parent=self.entity, child=self.entity_bis, role=self.role)
- response = self.client.get(reverse('api:element-links', kwargs={'pk': str(self.page.id)}))
+ response = self.client.get(reverse('api:element-links', kwargs={'pk': str(self.element.id)}))
self.assertEqual(response.status_code, status.HTTP_200_OK)
data = response.json()
self.assertListEqual(
diff --git a/arkindex/documents/tests/test_ml_results.py b/arkindex/documents/tests/test_ml_results.py
index e166cac4f377319adccb8214eb3facb0ef887b95..fa8ddd48c8c181407c65653e0179c291339f4ca0 100644
--- a/arkindex/documents/tests/test_ml_results.py
+++ b/arkindex/documents/tests/test_ml_results.py
@@ -49,7 +49,7 @@ class TestMLResults(FixtureTestCase):
def test_element_stats(self):
self.client.force_login(self.superuser)
- self.assertEqual(self.page.transcriptions.count(), 4)
+ self.assertEqual(self.page.transcriptions.count(), 1)
self.assertEqual(self.page.classifications.count(), 1)
self.assertEqual(Entity.objects.filter(transcriptions__element=self.page).count(), 1)
self.assertEqual(Entity.objects.filter(metadatas__element=self.page).count(), 1)
@@ -66,7 +66,7 @@ class TestMLResults(FixtureTestCase):
'revision': '4.2',
'internal': False,
'classifications_count': 1,
- 'transcriptions_count': 4,
+ 'transcriptions_count': 1,
'entities_count': 2,
}
])
diff --git a/arkindex/documents/tests/test_search.py b/arkindex/documents/tests/test_search.py
index d30b8bf5105eed5d09e086e229c0ea1a4656ef11..26839b442a56988da09828f9745555876d87f860 100644
--- a/arkindex/documents/tests/test_search.py
+++ b/arkindex/documents/tests/test_search.py
@@ -98,7 +98,7 @@ class TestSearchAPI(FixtureAPITestCase):
def test_element_transcription_search(self):
elt = Element.objects.get(name="Volume 1, page 1r")
- ts = Transcription.objects.filter(text="PARIS", zone__image__path='img1')
+ ts = Transcription.objects.filter(text="PARIS", element__zone__image__path='img1')
self.es_mock.count.return_value = {'count': 1}
self.es_mock.search.return_value = self.build_es_response(
@@ -190,7 +190,7 @@ class TestSearchAPI(FixtureAPITestCase):
def test_iiif_transcription_search(self):
# Filter to only get transcriptions from volume 1
unfiltered = Transcription.objects.filter(text="PARIS")
- expected = Transcription.objects.filter(text="PARIS", zone__image__path__in=['img1', 'img2', 'img3'])
+ expected = Transcription.objects.filter(text="PARIS", element__zone__image__path__in=['img1', 'img2', 'img3'])
vol = Element.objects.get(name='Volume 1')
self.es_mock.count.return_value = {'count': len(unfiltered)}
diff --git a/arkindex/documents/tests/test_transcriptions.py b/arkindex/documents/tests/test_transcriptions.py
index 2024377959a673cca1a94bcb54853db34cbe6cd0..e1d08bc7e2a6801ca55e0b7a2f7d94e4535438c2 100644
--- a/arkindex/documents/tests/test_transcriptions.py
+++ b/arkindex/documents/tests/test_transcriptions.py
@@ -1,7 +1,6 @@
from django.urls import reverse
from rest_framework import status
from arkindex.project.tests import FixtureAPITestCase
-from django.contrib.gis.geos import LinearRing
from arkindex_common.enums import TranscriptionType
from arkindex_common.ml_tool import MLToolType
from arkindex.documents.models import Corpus, DataSource
@@ -52,116 +51,123 @@ class TestTranscriptions(FixtureAPITestCase):
self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
def test_list_element_transcriptions(self):
- self.page.transcriptions.all().delete()
- transcriptions = []
- for i in range(1, 11):
- zone, _ = self.page.zone.image.zones.get_or_create(
- polygon=LinearRing((0, 0), (0, i), (i, i), (i, 0), (0, 0))
- )
- # Create transcriptions on the page with their own zones
- transcriptions.append(self.page.transcriptions.create(
- source_id=self.src.id, type=TranscriptionType.Word, zone=zone,
- ))
+ tr1 = self.page.transcriptions.get()
+ tr2 = self.page.transcriptions.create(
+ type=TranscriptionType.Word,
+ text='something',
+ worker_version=self.worker_version,
+ score=0.369,
+ )
self.client.force_login(self.user)
- with self.assertNumQueries(10):
+ with self.assertNumQueries(11):
response = self.client.get(reverse('api:element-transcriptions', kwargs={'pk': str(self.page.id)}))
self.assertEqual(response.status_code, status.HTTP_200_OK)
- results = response.json()['results']
- self.assertEqual(len(results), 10)
- self.assertCountEqual(
- [(tr['id'], tr['element']) for tr in results],
- # Element should not be serialized in case recursive parameter is not set
- [(str(tr.id), None) for tr in transcriptions]
- )
+ self.assertCountEqual(response.json()['results'], [
+ {
+ 'id': str(tr1.id),
+ 'type': 'page',
+ 'text': 'Lorem ipsum dolor sit amet',
+ 'score': 1.0,
+ 'source': {
+ 'id': str(self.src.id),
+ 'type': 'recognizer',
+ 'slug': 'test',
+ 'name': 'Test Recognizer',
+ 'revision': '4.2',
+ 'internal': False,
+ },
+ 'worker_version_id': None,
+ 'element': None,
+ },
+ {
+ 'id': str(tr2.id),
+ 'type': 'word',
+ 'text': 'something',
+ 'score': 0.369,
+ 'source': None,
+ 'worker_version_id': str(self.worker_version.id),
+ 'element': None,
+ }
+ ])
def test_list_transcriptions_recursive(self):
- for i in range(1, 5):
- # Add 4 transcriptions on the page line
- self.line.transcriptions.create(worker_version=self.worker_version, type=TranscriptionType.Line, text=f'Text {i}')
- for i in range(1, 5):
- # Add 4 transcribed line children
- zone, _ = self.page.zone.image.zones.get_or_create(
- polygon=LinearRing((0, 0), (0, i + 1), (i + 1, i + 1), (i + 1, 0), (0, 0))
- )
- line = self.page.corpus.elements.create(zone=zone, type=self.line.type, name=f'Added line {i}')
- line.transcriptions.create(source_id=self.src.id, type=TranscriptionType.Line, text=f'Added text {i}')
- line.add_parent(self.page)
-
self.client.force_login(self.user)
- with self.assertNumQueries(14):
+ with self.assertNumQueries(13):
response = self.client.get(
reverse('api:element-transcriptions', kwargs={'pk': str(self.page.id)}),
data={'recursive': 'true'}
)
self.assertEqual(response.status_code, status.HTTP_200_OK)
- results = response.json()['results']
- self.assertEqual(len(results), 12)
-
- page_polygon = [[0, 0], [0, 1000], [1000, 1000], [1000, 0], [0, 0]]
- line_polygon = [[400, 400], [400, 500], [500, 500], [500, 400], [400, 400]]
self.assertCountEqual(
[
- (data['element']['type'], data['worker_version_id'], data['element']['zone']['polygon'], data['text'])
- for data in results
+ (data['element']['type'], data['element']['zone']['polygon'], data['text'])
+ for data in response.json()['results']
], [
- ('page', None, page_polygon, 'PARIS'),
- ('page', None, page_polygon, 'ROY'),
- ('page', None, page_polygon, 'Lorem ipsum dolor sit amet'),
- ('page', None, page_polygon, 'DATUM'),
- ('text_line', str(self.worker_version.id), line_polygon, 'Text 1'),
- ('text_line', str(self.worker_version.id), line_polygon, 'Text 2'),
- ('text_line', str(self.worker_version.id), line_polygon, 'Text 3'),
- ('text_line', str(self.worker_version.id), line_polygon, 'Text 4'),
- ('text_line', None, [[0, 0], [0, 2], [2, 2], [2, 0], [0, 0]], 'Added text 1'),
- ('text_line', None, [[0, 0], [0, 3], [3, 3], [3, 0], [0, 0]], 'Added text 2'),
- ('text_line', None, [[0, 0], [0, 4], [4, 4], [4, 0], [0, 0]], 'Added text 3'),
- ('text_line', None, [[0, 0], [0, 5], [5, 5], [5, 0], [0, 0]], 'Added text 4')
+ ('page', [[0, 0], [0, 1000], [1000, 1000], [1000, 0], [0, 0]], 'Lorem ipsum dolor sit amet'),
+ ('word', [[100, 100], [100, 200], [200, 200], [200, 100], [100, 100]], 'PARIS'),
+ ('word', [[400, 400], [400, 500], [500, 500], [500, 400], [400, 400]], 'ROY'),
+ ('word', [[700, 700], [700, 800], [800, 800], [800, 700], [700, 700]], 'DATUM'),
]
)
def test_list_transcriptions_recursive_filtered(self):
- for i in range(1, 5):
- # Add 4 transcriptions on the page line
- self.line.transcriptions.create(source_id=self.src.id, type=TranscriptionType.Line, text=f'Text {i}')
-
self.client.force_login(self.user)
- with self.assertNumQueries(12):
+ with self.assertNumQueries(13):
response = self.client.get(
reverse('api:element-transcriptions', kwargs={'pk': str(self.page.id)}),
- data={'recursive': 'true', 'type': 'line'}
+ data={'recursive': 'true', 'type': 'word'}
)
self.assertEqual(response.status_code, status.HTTP_200_OK)
- results = response.json()['results']
- self.assertEqual(len(results), 4)
- for tr in results:
- self.assertEqual(tr.get('type'), 'line')
+ self.assertCountEqual(
+ [
+ (data['element']['type'], data['element']['zone']['polygon'], data['text'])
+ for data in response.json()['results']
+ ], [
+ ('word', [[100, 100], [100, 200], [200, 200], [200, 100], [100, 100]], 'PARIS'),
+ ('word', [[400, 400], [400, 500], [500, 500], [500, 400], [400, 400]], 'ROY'),
+ ('word', [[700, 700], [700, 800], [800, 800], [800, 700], [700, 700]], 'DATUM'),
+ ]
+ )
def test_list_worker_version_transcriptions(self):
-
- for i in range(1, 5):
- # Add 4 transcriptions on the page line with a specific worker_version
- self.line.transcriptions.create(
- type=TranscriptionType.Line,
- text=f'Text {i}',
- worker_version=self.worker_version
- )
+ worker_transcription = self.page.transcriptions.create(
+ type=TranscriptionType.Word,
+ text='something',
+ worker_version=self.worker_version,
+ score=0.369,
+ )
self.client.force_login(self.user)
- with self.assertNumQueries(12):
+ with self.assertNumQueries(13):
response = self.client.get(
reverse('api:element-transcriptions', kwargs={'pk': str(self.page.id)}),
data={'recursive': 'true', 'worker_version': str(self.worker_version.id)}
)
self.assertEqual(response.status_code, status.HTTP_200_OK)
- results = response.json()['results']
- self.assertEqual(len(results), 4)
- for tr in results:
- self.assertEqual(tr.get('worker_version_id'), str(self.worker_version.id))
+ self.assertListEqual(response.json()['results'], [
+ {
+ 'id': str(worker_transcription.id),
+ 'type': 'word',
+ 'text': 'something',
+ 'score': 0.369,
+ 'source': None,
+ 'worker_version_id': str(self.worker_version.id),
+ 'element': {
+ 'id': str(self.page.id),
+ 'name': 'Volume 1, page 1r',
+ 'type': 'page',
+ 'zone': {
+ 'id': str(self.page.zone_id),
+ 'polygon': [[0, 0], [0, 1000], [1000, 1000], [1000, 0], [0, 0]],
+ },
+ },
+ }
+ ])
diff --git a/arkindex/images/tests/test_image_elements.py b/arkindex/images/tests/test_image_elements.py
index 9b5a138cba8f8d548218590a4f046b499cf60441..1aeb63f8276a26a93dbf2c96ab50a1d707aa8987 100644
--- a/arkindex/images/tests/test_image_elements.py
+++ b/arkindex/images/tests/test_image_elements.py
@@ -16,7 +16,7 @@ class TestImageElements(FixtureTestCase):
data = response.json()
self.assertListEqual(
[element['name'] for element in data['results']],
- ['Volume 1, page 1r', 'Surface A', 'Surface B', 'Text line'],
+ ['Volume 1, page 1r', 'Surface A', 'Surface B', 'Text line', 'DATUM', 'PARIS', 'ROY'],
)
def test_image_elements_type_filter(self):
diff --git a/arkindex/project/elastic.py b/arkindex/project/elastic.py
index 8b8430c84ba0fe83cfa87a41f73272152c8b850f..b92db343c83548843061d91ca4c15936936069d9 100644
--- a/arkindex/project/elastic.py
+++ b/arkindex/project/elastic.py
@@ -107,22 +107,7 @@ class ESElement(Document):
@classmethod
def from_model(cls, instance):
- from arkindex.documents.models import Element, Transcription
-
- if instance.zone:
- transcriptions = instance.transcriptions.all()
- else:
- children = Element.objects \
- .get_descending(instance.id) \
- .filter(zone__isnull=False) \
- .prefetch_related('zone')
- transcriptions = chain(*[
- Transcription.objects.filter(
- zone__image_id=elt.zone.image_id,
- zone__polygon__within=elt.zone.polygon
- )
- for elt in children
- ])
+ from arkindex.documents.models import Element
interpreted_dates = chain(*[md.get_dates() for md in instance.metadatas.all()])
date_range = {
@@ -142,13 +127,16 @@ class ESElement(Document):
element.name
for element in Element.objects.get_ascending(instance.id)
],
+ # Filter using Python here as metadatas are prefetched entirely
+ # and applying .filter would make an unnecessary SQL query
references=[
md.value.lower()
- for md in instance.metadatas.filter(type=MetaType.Reference)
+ for md in instance.metadatas.all()
+ if md.type == MetaType.Reference
],
transcriptions=list(map(
ESTranscriptionInnerDoc.from_model,
- transcriptions,
+ instance.transcriptions.all(),
)),
date_range=date_range,
)
diff --git a/arkindex/project/tests/test_elastic.py b/arkindex/project/tests/test_elastic.py
index 9db3b5a55b13e5b070a30217cd2032ecfd77451d..9f38e1d749192e63b1ad9727f6676f05111c23ff 100644
--- a/arkindex/project/tests/test_elastic.py
+++ b/arkindex/project/tests/test_elastic.py
@@ -61,7 +61,6 @@ class TestESDocuments(FixtureAPITestCase):
surface.transcriptions.create(
type=TranscriptionType.Word,
text='invisible transcription',
- zone=page.zone.image.zones.create(polygon=[(0, 0), (0, 1), (1, 1), (1, 0), (0, 0)]),
source=DataSource.objects.get(slug='test', type=MLToolType.Recognizer),
)
texts = [tr['text'] for tr in ESElement.from_model(page).to_dict()['transcriptions']]