From 95c31fab8c0919bdccd8327be9d29abdd3f0749b Mon Sep 17 00:00:00 2001
From: Eva Bardou <ebardou@teklia.com>
Date: Tue, 25 Aug 2020 11:47:36 +0000
Subject: [PATCH] Add a helper to reach CreateElementTranscriptions API
endpoint
---
arkindex_worker/worker.py | 65 +++++
tests/test_elements_worker.py | 519 ++++++++++++++++++++++++++++++++++
2 files changed, 584 insertions(+)
diff --git a/arkindex_worker/worker.py b/arkindex_worker/worker.py
index 3fc01b31..8f10b07e 100644
--- a/arkindex_worker/worker.py
+++ b/arkindex_worker/worker.py
@@ -321,3 +321,68 @@ class ElementsWorker(BaseWorker):
self.report.add_entity(element.id, entity["id"], type.value, name)
return entity["id"]
+
+ def create_element_transcriptions(
+ self, element, sub_element_type, transcription_type, transcriptions
+ ):
+ """
+ Create multiple sub elements with their transcriptions on the given element through API
+ """
+ assert element and isinstance(
+ element, Element
+ ), "element shouldn't be null and should be of type Element"
+ assert sub_element_type and isinstance(
+ sub_element_type, str
+ ), "sub_element_type shouldn't be null and should be of type str"
+ assert transcription_type and isinstance(
+ transcription_type, TranscriptionType
+ ), "transcription_type shouldn't be null and should be of type TranscriptionType"
+ assert transcriptions and isinstance(
+ transcriptions, list
+ ), "transcriptions shouldn't be null and should be of type list"
+
+ for index, transcription in enumerate(transcriptions):
+ text = transcription.get("text")
+ assert text and isinstance(
+ text, str
+ ), f"Transcription at index {index} in transcriptions: text shouldn't be null and should be of type str"
+
+ score = transcription.get("score")
+ assert (
+ score and isinstance(score, float) and 0 <= score <= 1
+ ), f"Transcription at index {index} in transcriptions: score shouldn't be null and should be a float in [0..1] range"
+
+ polygon = transcription.get("polygon")
+ assert polygon and isinstance(
+ polygon, list
+ ), f"Transcription at index {index} in transcriptions: polygon shouldn't be null and should be of type list"
+ assert (
+ len(polygon) >= 3
+ ), f"Transcription at index {index} in transcriptions: polygon should have at least three points"
+ assert all(
+ isinstance(point, list) and len(point) == 2 for point in polygon
+ ), f"Transcription at index {index} in transcriptions: polygon points should be lists of two items"
+ assert all(
+ isinstance(coord, (int, float)) for point in polygon for coord in point
+ ), f"Transcription at index {index} in transcriptions: polygon points should be lists of two numbers"
+
+ annotations = self.api_client.request(
+ "CreateElementTranscriptions",
+ id=element.id,
+ body={
+ "element_type": sub_element_type,
+ "transcription_type": transcription_type.value,
+ "worker_version": self.worker_version_id,
+ "transcriptions": transcriptions,
+ "return_elements": True,
+ },
+ )
+ for annotation in annotations:
+ if annotation["created"]:
+ logger.debug(
+ f"A sub_element of {element.id} with type {sub_element_type} was created during transcriptions bulk creation"
+ )
+ self.report.add_element(element.id, sub_element_type)
+ self.report.add_transcription(annotation["id"], transcription_type.value)
+
+ return annotations
diff --git a/tests/test_elements_worker.py b/tests/test_elements_worker.py
index 42fccaea..22d1aaf5 100644
--- a/tests/test_elements_worker.py
+++ b/tests/test_elements_worker.py
@@ -12,6 +12,24 @@ from apistar.exceptions import ErrorResponse
from arkindex_worker.models import Element
from arkindex_worker.worker import ElementsWorker, EntityType, TranscriptionType
+TRANSCRIPTIONS_SAMPLE = [
+ {
+ "polygon": [[100, 150], [700, 150], [700, 200], [100, 200]],
+ "score": 0.5,
+ "text": "The",
+ },
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "first",
+ },
+ {
+ "polygon": [[1000, 300], [1200, 300], [1200, 500], [1000, 500]],
+ "score": 0.9,
+ "text": "line",
+ },
+]
+
def test_cli_default(monkeypatch):
_, path = tempfile.mkstemp()
@@ -877,3 +895,504 @@ def test_create_entity(responses):
"worker_version": "12341234-1234-1234-1234-123412341234",
}
assert entity_id == "12345678-1234-1234-1234-123456789123"
+
+
+def test_create_element_transcriptions_wrong_element():
+ worker = ElementsWorker()
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=None,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=TRANSCRIPTIONS_SAMPLE,
+ )
+ assert str(e.value) == "element shouldn't be null and should be of type Element"
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element="not element type",
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=TRANSCRIPTIONS_SAMPLE,
+ )
+ assert str(e.value) == "element shouldn't be null and should be of type Element"
+
+
+def test_create_element_transcriptions_wrong_sub_element_type():
+ worker = ElementsWorker()
+ elt = Element({"zone": None})
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type=None,
+ transcription_type=TranscriptionType.Word,
+ transcriptions=TRANSCRIPTIONS_SAMPLE,
+ )
+ assert (
+ str(e.value) == "sub_element_type shouldn't be null and should be of type str"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type=1234,
+ transcription_type=TranscriptionType.Word,
+ transcriptions=TRANSCRIPTIONS_SAMPLE,
+ )
+ assert (
+ str(e.value) == "sub_element_type shouldn't be null and should be of type str"
+ )
+
+
+def test_create_element_transcriptions_wrong_transcription_type():
+ worker = ElementsWorker()
+ elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=None,
+ transcriptions=TRANSCRIPTIONS_SAMPLE,
+ )
+ assert (
+ str(e.value)
+ == "transcription_type shouldn't be null and should be of type TranscriptionType"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=1234,
+ transcriptions=TRANSCRIPTIONS_SAMPLE,
+ )
+ assert (
+ str(e.value)
+ == "transcription_type shouldn't be null and should be of type TranscriptionType"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type="not_a_transcription_type",
+ transcriptions=TRANSCRIPTIONS_SAMPLE,
+ )
+ assert (
+ str(e.value)
+ == "transcription_type shouldn't be null and should be of type TranscriptionType"
+ )
+
+
+def test_create_element_transcriptions_wrong_transcriptions():
+ worker = ElementsWorker()
+ elt = Element({"zone": None})
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=None,
+ )
+ assert str(e.value) == "transcriptions shouldn't be null and should be of type list"
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=1234,
+ )
+ assert str(e.value) == "transcriptions shouldn't be null and should be of type list"
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {
+ "polygon": [[100, 150], [700, 150], [700, 200], [100, 200]],
+ "score": 0.5,
+ },
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: text shouldn't be null and should be of type str"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {
+ "polygon": [[100, 150], [700, 150], [700, 200], [100, 200]],
+ "score": 0.5,
+ "text": None,
+ },
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: text shouldn't be null and should be of type str"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {
+ "polygon": [[100, 150], [700, 150], [700, 200], [100, 200]],
+ "score": 0.5,
+ "text": 1234,
+ },
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: text shouldn't be null and should be of type str"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {
+ "polygon": [[100, 150], [700, 150], [700, 200], [100, 200]],
+ "text": "word",
+ },
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: score shouldn't be null and should be a float in [0..1] range"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {
+ "polygon": [[100, 150], [700, 150], [700, 200], [100, 200]],
+ "score": None,
+ "text": "word",
+ },
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: score shouldn't be null and should be a float in [0..1] range"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {
+ "polygon": [[100, 150], [700, 150], [700, 200], [100, 200]],
+ "score": "a wrong score",
+ "text": "word",
+ },
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: score shouldn't be null and should be a float in [0..1] range"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {
+ "polygon": [[100, 150], [700, 150], [700, 200], [100, 200]],
+ "score": 0,
+ "text": "word",
+ },
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: score shouldn't be null and should be a float in [0..1] range"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {
+ "polygon": [[100, 150], [700, 150], [700, 200], [100, 200]],
+ "score": 2.00,
+ "text": "word",
+ },
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: score shouldn't be null and should be a float in [0..1] range"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {"score": 0.5, "text": "word"},
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: polygon shouldn't be null and should be of type list"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {"polygon": None, "score": 0.5, "text": "word"},
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: polygon shouldn't be null and should be of type list"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {"polygon": "not a polygon", "score": 0.5, "text": "word"},
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: polygon shouldn't be null and should be of type list"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {"polygon": [[1, 1], [2, 2]], "score": 0.5, "text": "word"},
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: polygon should have at least three points"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {
+ "polygon": [[1, 1, 1], [2, 2, 1], [2, 1, 1], [1, 2, 1]],
+ "score": 0.5,
+ "text": "word",
+ },
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: polygon points should be lists of two items"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {"polygon": [[1], [2], [2], [1]], "score": 0.5, "text": "word"},
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: polygon points should be lists of two items"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=[
+ {
+ "polygon": [[0, 0], [2000, 0], [2000, 3000], [0, 3000]],
+ "score": 0.75,
+ "text": "The",
+ },
+ {
+ "polygon": [["not a coord", 1], [2, 2], [2, 1], [1, 2]],
+ "score": 0.5,
+ "text": "word",
+ },
+ ],
+ )
+ assert (
+ str(e.value)
+ == "Transcription at index 1 in transcriptions: polygon points should be lists of two numbers"
+ )
+
+
+def test_create_element_transcriptions_api_error(responses):
+ worker = ElementsWorker()
+ worker.configure()
+ elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+ responses.add(
+ responses.POST,
+ f"https://arkindex.teklia.com/api/v1/element/{elt.id}/transcriptions/bulk/",
+ status=500,
+ )
+
+ with pytest.raises(ErrorResponse):
+ worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=TRANSCRIPTIONS_SAMPLE,
+ )
+
+ assert len(responses.calls) == 1
+ assert (
+ responses.calls[0].request.url
+ == f"https://arkindex.teklia.com/api/v1/element/{elt.id}/transcriptions/bulk/"
+ )
+
+
+def test_create_element_transcriptions(responses):
+ worker = ElementsWorker()
+ worker.configure()
+ elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+ responses.add(
+ responses.POST,
+ f"https://arkindex.teklia.com/api/v1/element/{elt.id}/transcriptions/bulk/",
+ status=200,
+ json=[
+ {"id": "word1_1_1", "created": False},
+ {"id": "word1_1_2", "created": False},
+ {"id": "word1_1_3", "created": False},
+ ],
+ )
+
+ annotations = worker.create_element_transcriptions(
+ element=elt,
+ sub_element_type="page",
+ transcription_type=TranscriptionType.Word,
+ transcriptions=TRANSCRIPTIONS_SAMPLE,
+ )
+
+ assert len(responses.calls) == 1
+ assert (
+ responses.calls[0].request.url
+ == f"https://arkindex.teklia.com/api/v1/element/{elt.id}/transcriptions/bulk/"
+ )
+ assert json.loads(responses.calls[0].request.body) == {
+ "element_type": "page",
+ "transcription_type": "word",
+ "worker_version": "12341234-1234-1234-1234-123412341234",
+ "transcriptions": TRANSCRIPTIONS_SAMPLE,
+ "return_elements": True,
+ }
+ assert annotations == [
+ {"id": "word1_1_1", "created": False},
+ {"id": "word1_1_2", "created": False},
+ {"id": "word1_1_3", "created": False},
+ ]
--
GitLab