From 0c2bf11efd39dcb0afe02b5dfdb23af8ce163c8e Mon Sep 17 00:00:00 2001
From: Yoann Schneider <yschneider@teklia.com>
Date: Mon, 26 Sep 2022 10:15:39 +0000
Subject: [PATCH] Add worker_run_id on publication fields and in cache
---
arkindex_worker/cache.py | 5 ++
arkindex_worker/worker/classification.py | 30 +++++++---
arkindex_worker/worker/element.py | 3 +
arkindex_worker/worker/entity.py | 3 +
arkindex_worker/worker/metadata.py | 1 +
arkindex_worker/worker/transcription.py | 7 +++
tests/test_cache.py | 10 ++--
.../test_classifications.py | 57 ++++++++++++++++++-
tests/test_elements_worker/test_elements.py | 12 ++++
tests/test_elements_worker/test_entities.py | 14 +++++
tests/test_elements_worker/test_metadata.py | 5 +-
.../test_transcriptions.py | 37 ++++++++++++
12 files changed, 167 insertions(+), 17 deletions(-)
diff --git a/arkindex_worker/cache.py b/arkindex_worker/cache.py
index 4baa14e5..fe7a4d62 100644
--- a/arkindex_worker/cache.py
+++ b/arkindex_worker/cache.py
@@ -95,6 +95,7 @@ class CachedElement(Model):
mirrored = BooleanField(default=False)
initial = BooleanField(default=False)
worker_version_id = UUIDField(null=True)
+ worker_run_id = UUIDField(null=True)
confidence = FloatField(null=True)
class Meta:
@@ -173,6 +174,7 @@ class CachedTranscription(Model):
confidence = FloatField()
orientation = CharField(max_length=50)
worker_version_id = UUIDField(null=True)
+ worker_run_id = UUIDField(null=True)
class Meta:
database = db
@@ -190,6 +192,7 @@ class CachedClassification(Model):
confidence = FloatField()
state = CharField(max_length=10)
worker_version_id = UUIDField(null=True)
+ worker_run_id = UUIDField(null=True)
class Meta:
database = db
@@ -207,6 +210,7 @@ class CachedEntity(Model):
validated = BooleanField(default=False)
metas = JSONField(null=True)
worker_version_id = UUIDField(null=True)
+ worker_run_id = UUIDField(null=True)
class Meta:
database = db
@@ -225,6 +229,7 @@ class CachedTranscriptionEntity(Model):
offset = IntegerField(constraints=[Check("offset >= 0")])
length = IntegerField(constraints=[Check("length > 0")])
worker_version_id = UUIDField(null=True)
+ worker_run_id = UUIDField(null=True)
confidence = FloatField(null=True)
class Meta:
diff --git a/arkindex_worker/worker/classification.py b/arkindex_worker/worker/classification.py
index 53229776..4d9c0b7f 100644
--- a/arkindex_worker/worker/classification.py
+++ b/arkindex_worker/worker/classification.py
@@ -105,6 +105,7 @@ class ClassificationMixin(object):
"element": str(element.id),
"ml_class": self.get_ml_class_id(ml_class),
"worker_version": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
"confidence": confidence,
"high_confidence": high_confidence,
},
@@ -121,6 +122,7 @@ class ClassificationMixin(object):
"confidence": created["confidence"],
"state": created["state"],
"worker_version_id": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
}
]
CachedClassification.insert_many(to_insert).execute()
@@ -130,15 +132,23 @@ class ClassificationMixin(object):
)
except ErrorResponse as e:
# Detect already existing classification
- if (
- e.status_code == 400
- and "non_field_errors" in e.content
- and "The fields element, worker_version, ml_class must make a unique set."
- in e.content["non_field_errors"]
- ):
- logger.warning(
- f"This worker version has already set {ml_class} on element {element.id}"
- )
+ if e.status_code == 400 and "non_field_errors" in e.content:
+ if (
+ "The fields element, worker_version, ml_class must make a unique set."
+ in e.content["non_field_errors"]
+ ):
+ logger.warning(
+ f"This worker version has already set {ml_class} on element {element.id}"
+ )
+ elif (
+ "The fields element, worker_run, ml_class must make a unique set."
+ in e.content["non_field_errors"]
+ ):
+ logger.warning(
+ f"This worker run has already set {ml_class} on element {element.id}"
+ )
+ else:
+ raise
return
# Propagate any other API error
@@ -202,6 +212,7 @@ class ClassificationMixin(object):
body={
"parent": str(element.id),
"worker_version": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
"classifications": classifications,
},
)["classifications"]
@@ -220,6 +231,7 @@ class ClassificationMixin(object):
"confidence": created_cl["confidence"],
"state": created_cl["state"],
"worker_version_id": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
}
for created_cl in created_cls
]
diff --git a/arkindex_worker/worker/element.py b/arkindex_worker/worker/element.py
index 0a30e8da..4caac206 100644
--- a/arkindex_worker/worker/element.py
+++ b/arkindex_worker/worker/element.py
@@ -139,6 +139,7 @@ class ElementMixin(object):
"polygon": polygon,
"parent": element.id,
"worker_version": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
"confidence": confidence,
},
)
@@ -234,6 +235,7 @@ class ElementMixin(object):
id=parent.id,
body={
"worker_version": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
"elements": elements,
},
)
@@ -266,6 +268,7 @@ class ElementMixin(object):
"image_id": image_id,
"polygon": element["polygon"],
"worker_version_id": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
"confidence": element.get("confidence"),
}
for idx, element in enumerate(elements)
diff --git a/arkindex_worker/worker/entity.py b/arkindex_worker/worker/entity.py
index 0ccc7ddc..aa2ab829 100644
--- a/arkindex_worker/worker/entity.py
+++ b/arkindex_worker/worker/entity.py
@@ -71,6 +71,7 @@ class EntityMixin(object):
"validated": validated,
"corpus": self.corpus_id,
"worker_version": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
},
)
self.report.add_entity(element.id, entity["id"], type.value, name)
@@ -141,6 +142,7 @@ class EntityMixin(object):
"length": length,
"offset": offset,
"worker_version_id": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
}
if confidence is not None:
body["confidence"] = confidence
@@ -161,6 +163,7 @@ class EntityMixin(object):
offset=offset,
length=length,
worker_version_id=self.worker_version_id,
+ worker_run_id=self.worker_run_id,
confidence=confidence,
)
except IntegrityError as e:
diff --git a/arkindex_worker/worker/metadata.py b/arkindex_worker/worker/metadata.py
index 635ed629..8d0e8f5f 100644
--- a/arkindex_worker/worker/metadata.py
+++ b/arkindex_worker/worker/metadata.py
@@ -102,6 +102,7 @@ class MetaDataMixin(object):
"value": value,
"entity_id": entity,
"worker_version": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
},
)
self.report.add_metadata(element.id, metadata["id"], type.value, name)
diff --git a/arkindex_worker/worker/transcription.py b/arkindex_worker/worker/transcription.py
index 77f9287b..e454e20d 100644
--- a/arkindex_worker/worker/transcription.py
+++ b/arkindex_worker/worker/transcription.py
@@ -83,6 +83,7 @@ class TranscriptionMixin(object):
body={
"text": text,
"worker_version": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
"confidence": confidence,
"orientation": orientation.value,
},
@@ -101,6 +102,7 @@ class TranscriptionMixin(object):
"confidence": created["confidence"],
"orientation": created["orientation"],
"worker_version_id": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
}
]
CachedTranscription.insert_many(to_insert).execute()
@@ -171,6 +173,7 @@ class TranscriptionMixin(object):
"CreateTranscriptions",
body={
"worker_version": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
"transcriptions": transcriptions_payload,
},
)["transcriptions"]
@@ -189,6 +192,7 @@ class TranscriptionMixin(object):
"confidence": created_tr["confidence"],
"orientation": created_tr["orientation"],
"worker_version_id": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
}
for created_tr in created_trs
]
@@ -284,6 +288,7 @@ class TranscriptionMixin(object):
body={
"element_type": sub_element_type,
"worker_version": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
"transcriptions": transcriptions_payload,
"return_elements": True,
},
@@ -321,6 +326,7 @@ class TranscriptionMixin(object):
"image_id": element.image_id,
"polygon": transcription["polygon"],
"worker_version_id": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
}
)
@@ -336,6 +342,7 @@ class TranscriptionMixin(object):
"orientation", TextOrientation.HorizontalLeftToRight
).value,
"worker_version_id": self.worker_version_id,
+ "worker_run_id": self.worker_run_id,
}
)
diff --git a/tests/test_cache.py b/tests/test_cache.py
index 6cdcd984..41dd82f1 100644
--- a/tests/test_cache.py
+++ b/tests/test_cache.py
@@ -58,12 +58,12 @@ def test_create_tables(tmp_path):
init_cache_db(db_path)
create_tables()
- expected_schema = """CREATE TABLE "classifications" ("id" TEXT NOT NULL PRIMARY KEY, "element_id" TEXT NOT NULL, "class_name" TEXT NOT NULL, "confidence" REAL NOT NULL, "state" VARCHAR(10) NOT NULL, "worker_version_id" TEXT, FOREIGN KEY ("element_id") REFERENCES "elements" ("id"))
-CREATE TABLE "elements" ("id" TEXT NOT NULL PRIMARY KEY, "parent_id" TEXT, "type" VARCHAR(50) NOT NULL, "image_id" TEXT, "polygon" text, "rotation_angle" INTEGER NOT NULL, "mirrored" INTEGER NOT NULL, "initial" INTEGER NOT NULL, "worker_version_id" TEXT, "confidence" REAL, FOREIGN KEY ("image_id") REFERENCES "images" ("id"))
-CREATE TABLE "entities" ("id" TEXT NOT NULL PRIMARY KEY, "type" VARCHAR(50) NOT NULL, "name" TEXT NOT NULL, "validated" INTEGER NOT NULL, "metas" text, "worker_version_id" TEXT)
+ expected_schema = """CREATE TABLE "classifications" ("id" TEXT NOT NULL PRIMARY KEY, "element_id" TEXT NOT NULL, "class_name" TEXT NOT NULL, "confidence" REAL NOT NULL, "state" VARCHAR(10) NOT NULL, "worker_version_id" TEXT, "worker_run_id" TEXT, FOREIGN KEY ("element_id") REFERENCES "elements" ("id"))
+CREATE TABLE "elements" ("id" TEXT NOT NULL PRIMARY KEY, "parent_id" TEXT, "type" VARCHAR(50) NOT NULL, "image_id" TEXT, "polygon" text, "rotation_angle" INTEGER NOT NULL, "mirrored" INTEGER NOT NULL, "initial" INTEGER NOT NULL, "worker_version_id" TEXT, "worker_run_id" TEXT, "confidence" REAL, FOREIGN KEY ("image_id") REFERENCES "images" ("id"))
+CREATE TABLE "entities" ("id" TEXT NOT NULL PRIMARY KEY, "type" VARCHAR(50) NOT NULL, "name" TEXT NOT NULL, "validated" INTEGER NOT NULL, "metas" text, "worker_version_id" TEXT, "worker_run_id" TEXT)
CREATE TABLE "images" ("id" TEXT NOT NULL PRIMARY KEY, "width" INTEGER NOT NULL, "height" INTEGER NOT NULL, "url" TEXT NOT NULL)
-CREATE TABLE "transcription_entities" ("transcription_id" TEXT NOT NULL, "entity_id" TEXT NOT NULL, "offset" INTEGER NOT NULL CHECK (offset >= 0), "length" INTEGER NOT NULL CHECK (length > 0), "worker_version_id" TEXT, "confidence" REAL, PRIMARY KEY ("transcription_id", "entity_id"), FOREIGN KEY ("transcription_id") REFERENCES "transcriptions" ("id"), FOREIGN KEY ("entity_id") REFERENCES "entities" ("id"))
-CREATE TABLE "transcriptions" ("id" TEXT NOT NULL PRIMARY KEY, "element_id" TEXT NOT NULL, "text" TEXT NOT NULL, "confidence" REAL NOT NULL, "orientation" VARCHAR(50) NOT NULL, "worker_version_id" TEXT, FOREIGN KEY ("element_id") REFERENCES "elements" ("id"))"""
+CREATE TABLE "transcription_entities" ("transcription_id" TEXT NOT NULL, "entity_id" TEXT NOT NULL, "offset" INTEGER NOT NULL CHECK (offset >= 0), "length" INTEGER NOT NULL CHECK (length > 0), "worker_version_id" TEXT, "worker_run_id" TEXT, "confidence" REAL, PRIMARY KEY ("transcription_id", "entity_id"), FOREIGN KEY ("transcription_id") REFERENCES "transcriptions" ("id"), FOREIGN KEY ("entity_id") REFERENCES "entities" ("id"))
+CREATE TABLE "transcriptions" ("id" TEXT NOT NULL PRIMARY KEY, "element_id" TEXT NOT NULL, "text" TEXT NOT NULL, "confidence" REAL NOT NULL, "orientation" VARCHAR(50) NOT NULL, "worker_version_id" TEXT, "worker_run_id" TEXT, FOREIGN KEY ("element_id") REFERENCES "elements" ("id"))"""
actual_schema = "\n".join(
[
diff --git a/tests/test_elements_worker/test_classifications.py b/tests/test_elements_worker/test_classifications.py
index 32f9d876..5f2be88e 100644
--- a/tests/test_elements_worker/test_classifications.py
+++ b/tests/test_elements_worker/test_classifications.py
@@ -250,6 +250,7 @@ def test_create_classification_wrong_ml_class(mock_elements_worker, responses):
"element": "12341234-1234-1234-1234-123412341234",
"ml_class": "new-ml-class-1234",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"confidence": 0.42,
"high_confidence": True,
},
@@ -401,6 +402,7 @@ def test_create_classification(responses, mock_elements_worker):
"element": "12341234-1234-1234-1234-123412341234",
"ml_class": "0000",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"confidence": 0.42,
"high_confidence": True,
}
@@ -426,6 +428,7 @@ def test_create_classification_with_cache(responses, mock_elements_worker_with_c
"element": "12341234-1234-1234-1234-123412341234",
"ml_class": "0000",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"confidence": 0.42,
"high_confidence": True,
"state": "pending",
@@ -450,6 +453,7 @@ def test_create_classification_with_cache(responses, mock_elements_worker_with_c
"element": "12341234-1234-1234-1234-123412341234",
"ml_class": "0000",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"confidence": 0.42,
"high_confidence": True,
}
@@ -468,11 +472,14 @@ def test_create_classification_with_cache(responses, mock_elements_worker_with_c
confidence=0.42,
state="pending",
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
)
]
-def test_create_classification_duplicate(responses, mock_elements_worker):
+def test_create_classification_duplicate_worker_version(
+ responses, mock_elements_worker
+):
mock_elements_worker.classes = {
"11111111-1111-1111-1111-111111111111": {"a_class": "0000"}
}
@@ -506,6 +513,50 @@ def test_create_classification_duplicate(responses, mock_elements_worker):
"element": "12341234-1234-1234-1234-123412341234",
"ml_class": "0000",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
+ "confidence": 0.42,
+ "high_confidence": True,
+ }
+
+ # Classification has NOT been created
+ assert mock_elements_worker.report.report_data["elements"] == {}
+
+
+def test_create_classification_duplicate_worker_run(responses, mock_elements_worker):
+ mock_elements_worker.classes = {
+ "11111111-1111-1111-1111-111111111111": {"a_class": "0000"}
+ }
+ elt = Element({"id": "12341234-1234-1234-1234-123412341234"})
+ responses.add(
+ responses.POST,
+ "http://testserver/api/v1/classifications/",
+ status=400,
+ json={
+ "non_field_errors": [
+ "The fields element, worker_run, ml_class must make a unique set."
+ ]
+ },
+ )
+
+ mock_elements_worker.create_classification(
+ element=elt,
+ ml_class="a_class",
+ confidence=0.42,
+ high_confidence=True,
+ )
+
+ assert len(responses.calls) == len(BASE_API_CALLS) + 1
+ assert [
+ (call.request.method, call.request.url) for call in responses.calls
+ ] == BASE_API_CALLS + [
+ ("POST", "http://testserver/api/v1/classifications/"),
+ ]
+
+ assert json.loads(responses.calls[-1].request.body) == {
+ "element": "12341234-1234-1234-1234-123412341234",
+ "ml_class": "0000",
+ "worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"confidence": 0.42,
"high_confidence": True,
}
@@ -827,6 +878,7 @@ def test_create_classifications(responses, mock_elements_worker_with_cache):
json={
"parent": str(elt.id),
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"classifications": [
{
"id": "00000000-0000-0000-0000-000000000000",
@@ -860,6 +912,7 @@ def test_create_classifications(responses, mock_elements_worker_with_cache):
assert json.loads(responses.calls[-1].request.body) == {
"parent": str(elt.id),
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"classifications": classes,
}
@@ -872,6 +925,7 @@ def test_create_classifications(responses, mock_elements_worker_with_cache):
confidence=0.75,
state="pending",
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
),
CachedClassification(
id=UUID("11111111-1111-1111-1111-111111111111"),
@@ -880,5 +934,6 @@ def test_create_classifications(responses, mock_elements_worker_with_cache):
confidence=0.25,
state="pending",
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
),
]
diff --git a/tests/test_elements_worker/test_elements.py b/tests/test_elements_worker/test_elements.py
index 46a60630..2d894feb 100644
--- a/tests/test_elements_worker/test_elements.py
+++ b/tests/test_elements_worker/test_elements.py
@@ -574,6 +574,7 @@ def test_create_sub_element(responses, mock_elements_worker, slim_output):
"polygon": [[1, 1], [2, 2], [2, 1], [1, 2]],
"parent": "12341234-1234-1234-1234-123412341234",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"confidence": None,
}
if slim_output:
@@ -619,6 +620,7 @@ def test_create_sub_element_confidence(responses, mock_elements_worker):
"polygon": [[1, 1], [2, 2], [2, 1], [1, 2]],
"parent": "12341234-1234-1234-1234-123412341234",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"confidence": 0.42,
}
assert sub_element_id == "12345678-1234-1234-1234-123456789123"
@@ -994,6 +996,7 @@ def test_create_elements_cached_element(responses, mock_elements_worker_with_cac
}
],
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
}
assert created_ids == [{"id": "497f6eca-6276-4993-bfeb-53cbbbba6f08"}]
@@ -1007,6 +1010,7 @@ def test_create_elements_cached_element(responses, mock_elements_worker_with_cac
image_id="c0fec0fe-c0fe-c0fe-c0fe-c0fec0fec0fe",
polygon=[[1, 1], [2, 2], [2, 1], [1, 2]],
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
),
]
@@ -1061,6 +1065,7 @@ def test_create_elements(responses, mock_elements_worker_with_cache, tmp_path):
}
],
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
}
assert created_ids == [{"id": "497f6eca-6276-4993-bfeb-53cbbbba6f08"}]
@@ -1075,6 +1080,7 @@ def test_create_elements(responses, mock_elements_worker_with_cache, tmp_path):
image_id="c0fec0fe-c0fe-c0fe-c0fe-c0fec0fec0fe",
polygon=[[1, 1], [2, 2], [2, 1], [1, 2]],
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
confidence=None,
)
]
@@ -1134,6 +1140,7 @@ def test_create_elements_confidence(
}
],
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
}
assert created_ids == [{"id": "497f6eca-6276-4993-bfeb-53cbbbba6f08"}]
@@ -1148,6 +1155,7 @@ def test_create_elements_confidence(
image_id="c0fec0fe-c0fe-c0fe-c0fe-c0fec0fec0fe",
polygon=[[1, 1], [2, 2], [2, 1], [1, 2]],
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
confidence=0.42,
)
]
@@ -1392,6 +1400,7 @@ def test_list_element_children(responses, mock_elements_worker):
"best_classes": None,
"has_children": None,
"worker_version_id": None,
+ "worker_run_id": None,
},
{
"id": "1111",
@@ -1403,6 +1412,7 @@ def test_list_element_children(responses, mock_elements_worker):
"best_classes": None,
"has_children": None,
"worker_version_id": None,
+ "worker_run_id": None,
},
{
"id": "2222",
@@ -1414,6 +1424,7 @@ def test_list_element_children(responses, mock_elements_worker):
"best_classes": None,
"has_children": None,
"worker_version_id": None,
+ "worker_run_id": None,
},
]
responses.add(
@@ -1456,6 +1467,7 @@ def test_list_element_children_manual_worker_version(responses, mock_elements_wo
"best_classes": None,
"has_children": None,
"worker_version_id": None,
+ "worker_run_id": None,
}
]
responses.add(
diff --git a/tests/test_elements_worker/test_entities.py b/tests/test_elements_worker/test_entities.py
index 0df2fe4a..82012dcb 100644
--- a/tests/test_elements_worker/test_entities.py
+++ b/tests/test_elements_worker/test_entities.py
@@ -187,6 +187,7 @@ def test_create_entity(responses, mock_elements_worker):
"validated": None,
"corpus": "11111111-1111-1111-1111-111111111111",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
}
assert entity_id == "12345678-1234-1234-1234-123456789123"
@@ -220,6 +221,7 @@ def test_create_entity_with_cache(responses, mock_elements_worker_with_cache):
"validated": None,
"corpus": "11111111-1111-1111-1111-111111111111",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
}
assert entity_id == "12345678-1234-1234-1234-123456789123"
@@ -232,6 +234,7 @@ def test_create_entity_with_cache(responses, mock_elements_worker_with_cache):
validated=False,
metas={},
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
)
]
@@ -419,6 +422,7 @@ def test_create_transcription_entity_no_confidence(responses, mock_elements_work
"offset": 5,
"length": 10,
"worker_version_id": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
}
@@ -457,6 +461,7 @@ def test_create_transcription_entity_with_confidence(responses, mock_elements_wo
"offset": 5,
"length": 10,
"worker_version_id": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"confidence": 0.33,
}
@@ -496,6 +501,7 @@ def test_create_transcription_entity_confidence_none(responses, mock_elements_wo
"offset": 5,
"length": 10,
"worker_version_id": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
}
@@ -513,12 +519,14 @@ def test_create_transcription_entity_with_cache(
confidence=0.42,
orientation=TextOrientation.HorizontalLeftToRight,
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
)
CachedEntity.create(
id=UUID("11111111-1111-1111-1111-111111111111"),
type="person",
name="Bob Bob",
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
)
responses.add(
@@ -553,6 +561,7 @@ def test_create_transcription_entity_with_cache(
"offset": 5,
"length": 10,
"worker_version_id": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
}
# Check that created transcription entity was properly stored in SQLite cache
@@ -563,6 +572,7 @@ def test_create_transcription_entity_with_cache(
offset=5,
length=10,
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
)
]
@@ -581,12 +591,14 @@ def test_create_transcription_entity_with_confidence_with_cache(
confidence=0.42,
orientation=TextOrientation.HorizontalLeftToRight,
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
)
CachedEntity.create(
id=UUID("11111111-1111-1111-1111-111111111111"),
type="person",
name="Bob Bob",
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
)
responses.add(
@@ -623,6 +635,7 @@ def test_create_transcription_entity_with_confidence_with_cache(
"offset": 5,
"length": 10,
"worker_version_id": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"confidence": 0.77,
}
@@ -634,6 +647,7 @@ def test_create_transcription_entity_with_confidence_with_cache(
offset=5,
length=10,
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
confidence=0.77,
)
]
diff --git a/tests/test_elements_worker/test_metadata.py b/tests/test_elements_worker/test_metadata.py
index feb19c6d..40e2b423 100644
--- a/tests/test_elements_worker/test_metadata.py
+++ b/tests/test_elements_worker/test_metadata.py
@@ -194,6 +194,7 @@ def test_create_metadata(responses, mock_elements_worker):
"value": "La Turbine, Grenoble 38000",
"entity_id": None,
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
}
assert metadata_id == "12345678-1234-1234-1234-123456789123"
@@ -221,8 +222,8 @@ def test_create_metadatas(responses, mock_elements_worker, metadatas):
"http://testserver/api/v1/element/12341234-1234-1234-1234-123412341234/metadata/bulk/",
status=201,
json={
- "worker_version": mock_elements_worker.worker_version_id,
- "worker_run_id": mock_elements_worker.worker_run_id,
+ "worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"metadata_list": [
{
"id": "fake_metadata_id",
diff --git a/tests/test_elements_worker/test_transcriptions.py b/tests/test_elements_worker/test_transcriptions.py
index 0411db22..655df6e4 100644
--- a/tests/test_elements_worker/test_transcriptions.py
+++ b/tests/test_elements_worker/test_transcriptions.py
@@ -134,6 +134,7 @@ def test_create_transcription_default_orientation(responses, mock_elements_worke
"text": "Animula vagula blandula",
"confidence": 0.42,
"worker_version_id": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
},
)
mock_elements_worker.create_transcription(
@@ -144,6 +145,7 @@ def test_create_transcription_default_orientation(responses, mock_elements_worke
assert json.loads(responses.calls[-1].request.body) == {
"text": "Animula vagula blandula",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"confidence": 0.42,
"orientation": "horizontal-lr",
}
@@ -160,6 +162,7 @@ def test_create_transcription_orientation(responses, mock_elements_worker):
"text": "Animula vagula blandula",
"confidence": 0.42,
"worker_version_id": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
},
)
mock_elements_worker.create_transcription(
@@ -171,6 +174,7 @@ def test_create_transcription_orientation(responses, mock_elements_worker):
assert json.loads(responses.calls[-1].request.body) == {
"text": "Animula vagula blandula",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"confidence": 0.42,
"orientation": "vertical-lr",
}
@@ -230,6 +234,7 @@ def test_create_transcription(responses, mock_elements_worker):
"text": "i am a line",
"confidence": 0.42,
"worker_version_id": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
},
)
@@ -249,6 +254,7 @@ def test_create_transcription(responses, mock_elements_worker):
assert json.loads(responses.calls[-1].request.body) == {
"text": "i am a line",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"confidence": 0.42,
"orientation": "horizontal-lr",
}
@@ -267,6 +273,7 @@ def test_create_transcription_with_cache(responses, mock_elements_worker_with_ca
"confidence": 0.42,
"orientation": "horizontal-lr",
"worker_version_id": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
},
)
@@ -286,6 +293,7 @@ def test_create_transcription_with_cache(responses, mock_elements_worker_with_ca
assert json.loads(responses.calls[-1].request.body) == {
"text": "i am a line",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"orientation": "horizontal-lr",
"confidence": 0.42,
}
@@ -299,6 +307,7 @@ def test_create_transcription_with_cache(responses, mock_elements_worker_with_ca
confidence=0.42,
orientation=TextOrientation.HorizontalLeftToRight,
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
)
]
@@ -317,6 +326,7 @@ def test_create_transcription_orientation_with_cache(
"confidence": 0.42,
"orientation": "vertical-lr",
"worker_version_id": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
},
)
mock_elements_worker_with_cache.create_transcription(
@@ -328,6 +338,7 @@ def test_create_transcription_orientation_with_cache(
assert json.loads(responses.calls[-1].request.body) == {
"text": "Animula vagula blandula",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"orientation": "vertical-lr",
"confidence": 0.42,
}
@@ -345,12 +356,14 @@ def test_create_transcription_orientation_with_cache(
"mirrored": False,
"initial": False,
"worker_version_id": None,
+ "worker_run_id": None,
"confidence": None,
},
"text": "Animula vagula blandula",
"confidence": 0.42,
"orientation": TextOrientation.VerticalLeftToRight.value,
"worker_version_id": UUID("12341234-1234-1234-1234-123412341234"),
+ "worker_run_id": UUID("56785678-5678-5678-5678-567856785678"),
}
]
@@ -663,6 +676,7 @@ def test_create_transcriptions(responses, mock_elements_worker_with_cache):
status=200,
json={
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"transcriptions": [
{
"id": "00000000-0000-0000-0000-000000000000",
@@ -695,6 +709,7 @@ def test_create_transcriptions(responses, mock_elements_worker_with_cache):
assert json.loads(responses.calls[-1].request.body) == {
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"transcriptions": [
{
"element_id": "11111111-1111-1111-1111-111111111111",
@@ -720,6 +735,7 @@ def test_create_transcriptions(responses, mock_elements_worker_with_cache):
confidence=0.75,
orientation=TextOrientation.HorizontalLeftToRight,
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
),
CachedTranscription(
id=UUID("11111111-1111-1111-1111-111111111111"),
@@ -728,6 +744,7 @@ def test_create_transcriptions(responses, mock_elements_worker_with_cache):
confidence=0.42,
orientation=TextOrientation.HorizontalLeftToRight,
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
),
]
@@ -755,6 +772,7 @@ def test_create_transcriptions_orientation(responses, mock_elements_worker_with_
status=200,
json={
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"transcriptions": [
{
"id": "00000000-0000-0000-0000-000000000000",
@@ -780,6 +798,7 @@ def test_create_transcriptions_orientation(responses, mock_elements_worker_with_
assert json.loads(responses.calls[-1].request.body) == {
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"transcriptions": [
{
"element_id": "11111111-1111-1111-1111-111111111111",
@@ -810,12 +829,14 @@ def test_create_transcriptions_orientation(responses, mock_elements_worker_with_
"mirrored": False,
"initial": False,
"worker_version_id": None,
+ "worker_run_id": None,
"confidence": None,
},
"text": "Animula vagula blandula",
"confidence": 0.12,
"orientation": TextOrientation.HorizontalRightToLeft.value,
"worker_version_id": UUID("12341234-1234-1234-1234-123412341234"),
+ "worker_run_id": UUID("56785678-5678-5678-5678-567856785678"),
},
{
"id": UUID("11111111-1111-1111-1111-111111111111"),
@@ -829,12 +850,14 @@ def test_create_transcriptions_orientation(responses, mock_elements_worker_with_
"mirrored": False,
"initial": False,
"worker_version_id": None,
+ "worker_run_id": None,
"confidence": None,
},
"text": "Hospes comesque corporis",
"confidence": 0.21,
"orientation": TextOrientation.VerticalLeftToRight.value,
"worker_version_id": UUID("12341234-1234-1234-1234-123412341234"),
+ "worker_run_id": UUID("56785678-5678-5678-5678-567856785678"),
},
]
@@ -1307,6 +1330,7 @@ def test_create_element_transcriptions(responses, mock_elements_worker):
assert json.loads(responses.calls[-1].request.body) == {
"element_type": "page",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"transcriptions": [
{
"polygon": [[100, 150], [700, 150], [700, 200], [100, 200]],
@@ -1392,6 +1416,7 @@ def test_create_element_transcriptions_with_cache(
assert json.loads(responses.calls[-1].request.body) == {
"element_type": "page",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"transcriptions": [
{
"polygon": [[100, 150], [700, 150], [700, 200], [100, 200]],
@@ -1440,6 +1465,7 @@ def test_create_element_transcriptions_with_cache(
type="page",
polygon="[[100, 150], [700, 150], [700, 200], [100, 200]]",
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
),
CachedElement(
id=UUID("22222222-2222-2222-2222-222222222222"),
@@ -1447,6 +1473,7 @@ def test_create_element_transcriptions_with_cache(
type="page",
polygon="[[0, 0], [2000, 0], [2000, 3000], [0, 3000]]",
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
),
]
assert list(CachedTranscription.select()) == [
@@ -1457,6 +1484,7 @@ def test_create_element_transcriptions_with_cache(
confidence=0.5,
orientation=TextOrientation.HorizontalLeftToRight.value,
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
),
CachedTranscription(
id=UUID("67896789-6789-6789-6789-678967896789"),
@@ -1465,6 +1493,7 @@ def test_create_element_transcriptions_with_cache(
confidence=0.75,
orientation=TextOrientation.HorizontalLeftToRight.value,
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
),
CachedTranscription(
id=UUID("78907890-7890-7890-7890-789078907890"),
@@ -1473,6 +1502,7 @@ def test_create_element_transcriptions_with_cache(
confidence=0.9,
orientation=TextOrientation.HorizontalLeftToRight.value,
worker_version_id=UUID("12341234-1234-1234-1234-123412341234"),
+ worker_run_id=UUID("56785678-5678-5678-5678-567856785678"),
),
]
@@ -1534,6 +1564,7 @@ def test_create_transcriptions_orientation_with_cache(
assert json.loads(responses.calls[-1].request.body) == {
"element_type": "page",
"worker_version": "12341234-1234-1234-1234-123412341234",
+ "worker_run_id": "56785678-5678-5678-5678-567856785678",
"transcriptions": [
{
"polygon": [[100, 150], [700, 150], [700, 200], [100, 200]],
@@ -1588,12 +1619,14 @@ def test_create_transcriptions_orientation_with_cache(
"mirrored": False,
"initial": False,
"worker_version_id": UUID("12341234-1234-1234-1234-123412341234"),
+ "worker_run_id": UUID("56785678-5678-5678-5678-567856785678"),
"confidence": None,
},
"text": "Animula vagula blandula",
"confidence": 0.5,
"orientation": TextOrientation.HorizontalLeftToRight.value,
"worker_version_id": UUID("12341234-1234-1234-1234-123412341234"),
+ "worker_run_id": UUID("56785678-5678-5678-5678-567856785678"),
},
{
"id": UUID("67896789-6789-6789-6789-678967896789"),
@@ -1607,12 +1640,14 @@ def test_create_transcriptions_orientation_with_cache(
"mirrored": False,
"initial": False,
"worker_version_id": UUID("12341234-1234-1234-1234-123412341234"),
+ "worker_run_id": UUID("56785678-5678-5678-5678-567856785678"),
"confidence": None,
},
"text": "Hospes comesque corporis",
"confidence": 0.75,
"orientation": TextOrientation.VerticalLeftToRight.value,
"worker_version_id": UUID("12341234-1234-1234-1234-123412341234"),
+ "worker_run_id": UUID("56785678-5678-5678-5678-567856785678"),
},
{
"id": UUID("78907890-7890-7890-7890-789078907890"),
@@ -1626,12 +1661,14 @@ def test_create_transcriptions_orientation_with_cache(
"mirrored": False,
"initial": False,
"worker_version_id": UUID("12341234-1234-1234-1234-123412341234"),
+ "worker_run_id": UUID("56785678-5678-5678-5678-567856785678"),
"confidence": None,
},
"text": "Quae nunc abibis in loca",
"confidence": 0.9,
"orientation": TextOrientation.HorizontalRightToLeft.value,
"worker_version_id": UUID("12341234-1234-1234-1234-123412341234"),
+ "worker_run_id": UUID("56785678-5678-5678-5678-567856785678"),
},
]
--
GitLab