diff --git a/arkindex_worker/worker.py b/arkindex_worker/worker.py
index ee8b05692cfd0c36e2fe70d38c3f97480b1f98b2..d6381b752b78137af75bd30ae41ba8538ad48c67 100644
--- a/arkindex_worker/worker.py
+++ b/arkindex_worker/worker.py
@@ -5,11 +5,11 @@ import logging
import os
import sys
import uuid
+from enum import Enum
from apistar.exceptions import ErrorResponse
from arkindex import ArkindexClient, options_from_env
-from arkindex_common.enums import EntityType, TranscriptionType
from arkindex_worker import logger
from arkindex_worker.models import Element
from arkindex_worker.reporting import Reporter
@@ -73,6 +73,24 @@ class BaseWorker(object):
"""Override this method to implement your own process"""
+class TranscriptionType(Enum):
+ Page = "page"
+ Paragraph = "paragraph"
+ Line = "line"
+ Word = "word"
+ Character = "character"
+
+
+class EntityType(Enum):
+ Person = "person"
+ Location = "location"
+ Subject = "subject"
+ Organization = "organization"
+ Misc = "misc"
+ Number = "number"
+ Date = "date"
+
+
class ElementsWorker(BaseWorker):
def __init__(self, description="Arkindex Elements Worker"):
super().__init__(description)
diff --git a/tests/test_elements_worker.py b/tests/test_elements_worker.py
index b94312fbb2d64218d11a8384a9b6b5f4c41a3263..7f44936ff37a13984a46dfbc7dffe5bfe4c65e17 100644
--- a/tests/test_elements_worker.py
+++ b/tests/test_elements_worker.py
@@ -379,3 +379,402 @@ def test_create_sub_element(responses):
responses.calls[0].request.url
== "https://arkindex.teklia.com/api/v1/elements/create/"
)
+
+
+def test_create_transcription_wrong_element():
+ worker = ElementsWorker()
+ with pytest.raises(AssertionError) as e:
+ worker.create_transcription(
+ element=None, text="i am a line", type="line", score=0.42,
+ )
+ assert str(e.value) == "element shouldn't be null and should be of type Element"
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_transcription(
+ element="not element type", text="i am a line", type="line", score=0.42,
+ )
+ assert str(e.value) == "element shouldn't be null and should be of type Element"
+
+
+def test_create_transcription_wrong_type():
+ worker = ElementsWorker()
+ elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_transcription(
+ element=elt, text="i am a line", type=None, score=0.42,
+ )
+ assert str(e.value) == "type shouldn't be null and should be of type str"
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_transcription(
+ element=elt, text="i am a line", type=1234, score=0.42,
+ )
+ assert str(e.value) == "type shouldn't be null and should be of type str"
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_transcription(
+ element=elt,
+ text="i am a line",
+ type="not_a_transcription_type",
+ score=0.42,
+ )
+ assert str(e.value) == "type should be an allowed transcription type"
+
+
+def test_create_transcription_wrong_text():
+ worker = ElementsWorker()
+ elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_transcription(
+ element=elt, text=None, type="line", score=0.42,
+ )
+ assert str(e.value) == "text shouldn't be null and should be of type str"
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_transcription(
+ element=elt, text=1234, type="line", score=0.42,
+ )
+ assert str(e.value) == "text shouldn't be null and should be of type str"
+
+
+def test_create_transcription_wrong_score():
+ worker = ElementsWorker()
+ elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_transcription(
+ element=elt, text="i am a line", type="line", score=None,
+ )
+ assert (
+ str(e.value) == "score shouldn't be null and should be a float in [0..1] range"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_transcription(
+ element=elt, text="i am a line", type="line", score="wrong score",
+ )
+ assert (
+ str(e.value) == "score shouldn't be null and should be a float in [0..1] range"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_transcription(
+ element=elt, text="i am a line", type="line", score=0,
+ )
+ assert (
+ str(e.value) == "score shouldn't be null and should be a float in [0..1] range"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_transcription(
+ element=elt, text="i am a line", type="line", score=2.00,
+ )
+ assert (
+ str(e.value) == "score shouldn't be null and should be a float in [0..1] range"
+ )
+
+
+def test_create_transcription_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}/transcription/",
+ status=500,
+ )
+
+ with pytest.raises(ErrorResponse):
+ worker.create_transcription(
+ element=elt, text="i am a line", type="line", score=0.42,
+ )
+
+ assert len(responses.calls) == 1
+ assert (
+ responses.calls[0].request.url
+ == f"https://arkindex.teklia.com/api/v1/element/{elt.id}/transcription/"
+ )
+
+
+def test_create_transcription(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}/transcription/",
+ status=200,
+ )
+
+ worker.create_transcription(
+ element=elt, text="i am a line", type="line", score=0.42,
+ )
+
+ assert len(responses.calls) == 1
+ assert (
+ responses.calls[0].request.url
+ == f"https://arkindex.teklia.com/api/v1/element/{elt.id}/transcription/"
+ )
+
+
+def test_create_classification_wrong_element():
+ worker = ElementsWorker()
+ with pytest.raises(AssertionError) as e:
+ worker.create_classification(
+ element=None, ml_class="a_class", confidence=0.42, high_confidence=True,
+ )
+ assert str(e.value) == "element shouldn't be null and should be of type Element"
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_classification(
+ element="not element type",
+ ml_class="a_class",
+ confidence=0.42,
+ high_confidence=True,
+ )
+ assert str(e.value) == "element shouldn't be null and should be of type Element"
+
+
+def test_create_classification_wrong_ml_class():
+ worker = ElementsWorker()
+ elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_classification(
+ element=elt, ml_class=None, confidence=0.42, high_confidence=True,
+ )
+ assert str(e.value) == "ml_class shouldn't be null and should be of type str"
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_classification(
+ element=elt, ml_class=1234, confidence=0.42, high_confidence=True,
+ )
+ assert str(e.value) == "ml_class shouldn't be null and should be of type str"
+
+
+def test_create_classification_wrong_confidence():
+ worker = ElementsWorker()
+ elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_classification(
+ element=elt, ml_class="a_class", confidence=None, high_confidence=True,
+ )
+ assert (
+ str(e.value)
+ == "confidence shouldn't be null and should be a float in [0..1] range"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_classification(
+ element=elt,
+ ml_class="a_class",
+ confidence="wrong confidence",
+ high_confidence=True,
+ )
+ assert (
+ str(e.value)
+ == "confidence shouldn't be null and should be a float in [0..1] range"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_classification(
+ element=elt, ml_class="a_class", confidence=0, high_confidence=True,
+ )
+ assert (
+ str(e.value)
+ == "confidence shouldn't be null and should be a float in [0..1] range"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_classification(
+ element=elt, ml_class="a_class", confidence=2.00, high_confidence=True,
+ )
+ assert (
+ str(e.value)
+ == "confidence shouldn't be null and should be a float in [0..1] range"
+ )
+
+
+def test_create_classification_wrong_high_confidence():
+ worker = ElementsWorker()
+ elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_classification(
+ element=elt, ml_class="a_class", confidence=0.42, high_confidence=None,
+ )
+ assert (
+ str(e.value) == "high_confidence shouldn't be null and should be of type bool"
+ )
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_classification(
+ element=elt,
+ ml_class="a_class",
+ confidence=0.42,
+ high_confidence="wrong high_confidence",
+ )
+ assert (
+ str(e.value) == "high_confidence shouldn't be null and should be of type bool"
+ )
+
+
+def test_create_classification_api_error(responses):
+ worker = ElementsWorker()
+ worker.configure()
+ elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+ responses.add(
+ responses.POST,
+ "https://arkindex.teklia.com/api/v1/classifications/",
+ status=500,
+ )
+
+ with pytest.raises(ErrorResponse):
+ worker.create_classification(
+ element=elt, ml_class="a_class", confidence=0.42, high_confidence=True,
+ )
+
+ assert len(responses.calls) == 1
+ assert (
+ responses.calls[0].request.url
+ == "https://arkindex.teklia.com/api/v1/classifications/"
+ )
+
+
+def test_create_classification(responses):
+ worker = ElementsWorker()
+ worker.configure()
+ elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+ responses.add(
+ responses.POST,
+ "https://arkindex.teklia.com/api/v1/classifications/",
+ status=200,
+ )
+
+ worker.create_classification(
+ element=elt, ml_class="a_class", confidence=0.42, high_confidence=True,
+ )
+
+ assert len(responses.calls) == 1
+ assert (
+ responses.calls[0].request.url
+ == "https://arkindex.teklia.com/api/v1/classifications/"
+ )
+
+
+def test_create_entity_wrong_name():
+ worker = ElementsWorker()
+ with pytest.raises(AssertionError) as e:
+ worker.create_entity(
+ name=None, type="person", corpus="12341234-1234-1234-1234-123412341234",
+ )
+ assert str(e.value) == "name shouldn't be null and should be of type str"
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_entity(
+ name=1234, type="person", corpus="12341234-1234-1234-1234-123412341234",
+ )
+ assert str(e.value) == "name shouldn't be null and should be of type str"
+
+
+def test_create_entity_wrong_type():
+ worker = ElementsWorker()
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_entity(
+ name="Bob Bob", type=None, corpus="12341234-1234-1234-1234-123412341234",
+ )
+ assert str(e.value) == "type shouldn't be null and should be of type str"
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_entity(
+ name="Bob Bob", type=1234, corpus="12341234-1234-1234-1234-123412341234",
+ )
+ assert str(e.value) == "type shouldn't be null and should be of type str"
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_entity(
+ name="Bob Bob",
+ type="not_an_entity_type",
+ corpus="12341234-1234-1234-1234-123412341234",
+ )
+ assert str(e.value) == "type should be an allowed entity type"
+
+
+def test_create_entity_wrong_corpus():
+ worker = ElementsWorker()
+ with pytest.raises(AssertionError) as e:
+ worker.create_entity(
+ name="Bob Bob", type="person", corpus=None,
+ )
+ assert str(e.value) == "corpus shouldn't be null and should be of type str"
+
+ with pytest.raises(AssertionError) as e:
+ worker.create_entity(
+ name="Bob Bob", type="person", corpus=1234,
+ )
+ assert str(e.value) == "corpus shouldn't be null and should be of type str"
+
+
+def test_create_entity_wrong_metas():
+ worker = ElementsWorker()
+ with pytest.raises(AssertionError) as e:
+ worker.create_entity(
+ name="Bob Bob",
+ type="person",
+ corpus="12341234-1234-1234-1234-123412341234",
+ metas="wrong metas",
+ )
+ assert str(e.value) == "metas should be of type dict"
+
+
+def test_create_entity_wrong_validated():
+ worker = ElementsWorker()
+ with pytest.raises(AssertionError) as e:
+ worker.create_entity(
+ name="Bob Bob",
+ type="person",
+ corpus="12341234-1234-1234-1234-123412341234",
+ validated="wrong validated",
+ )
+ assert str(e.value) == "validated should be of type bool"
+
+
+def test_create_entity_api_error(responses):
+ worker = ElementsWorker()
+ worker.configure()
+ responses.add(
+ responses.POST, "https://arkindex.teklia.com/api/v1/entity/", status=500,
+ )
+
+ with pytest.raises(ErrorResponse):
+ worker.create_entity(
+ name="Bob Bob",
+ type="person",
+ corpus="12341234-1234-1234-1234-123412341234",
+ )
+
+ assert len(responses.calls) == 1
+ assert (
+ responses.calls[0].request.url == "https://arkindex.teklia.com/api/v1/entity/"
+ )
+
+
+def test_create_entity(responses):
+ worker = ElementsWorker()
+ worker.configure()
+ responses.add(
+ responses.POST, "https://arkindex.teklia.com/api/v1/entity/", status=200,
+ )
+
+ worker.create_entity(
+ name="Bob Bob", type="person", corpus="12341234-1234-1234-1234-123412341234",
+ )
+
+ assert len(responses.calls) == 1
+ assert (
+ responses.calls[0].request.url == "https://arkindex.teklia.com/api/v1/entity/"
+ )