def add_entries(self, entries: [DatasetEntryVO]):
try:
entries = [vo.to_array() for vo in entries]
for batch in chunked(entries, 100):
DatasetEntryEntity.insert_many(batch).execute()
except IntegrityError as ex:
raise Exception(
"one or more files have already been loaded into this dataset")
def get_label(self, entity_id: int):
dse = DatasetEntryEntity.alias()
lbl = LabelEntity.alias()
query = (dse.select(lbl.name).join(
lbl, on=(dse.label == lbl.id)).where(dse.id == entity_id))
result = list(query.dicts().execute())
return result[0]["name"] if len(result) > 0 else None
def fetch_entries_for_classification(self, ds_id):
en: DatasetEntryEntity = DatasetEntryEntity.alias("en")
lbl: LabelEntity = LabelEntity.alias("lbl")
query = (en.select(en.file_path,
lbl.name).join(lbl,
JOIN.INNER,
on=en.label == lbl.id).where(
en.dataset == ds_id
and en.label.is_null(False)))
cursor = query.dicts().execute()
return list(cursor)
def fetch_labels(self, dataset_id: int = None):
e = DatasetEntryEntity.alias("e")
l = LabelEntity.alias("l")
query = (e.select(
e.file_path.alias("image"), l.name.alias("label")).join(
l, on=(e.label == l.id)).where((e.dataset_id == dataset_id)
& (e.label.is_null(False))))
cursor = query.dicts().execute()
result = []
for row in cursor:
result.append(row)
return result
def find_by_path(self, ds_id, image_path):
query = (DatasetEntryEntity.select().where(
(DatasetEntryEntity.dataset == ds_id)
& (DatasetEntryEntity.file_path.endswith(
os.path.split(image_path)[1]))))
# print(query)
cursor = query.dicts().execute()
result = list(cursor)
vo = DatasetEntryVO()
if len(result) > 0:
row = result[0]
for k, v in row.items():
setattr(vo, k, v)
return vo
return None
def fetch_all_with_size(self):
ds = DatasetEntity.alias("ds")
m = DatasetEntryEntity.alias("m")
query = (ds.select(ds.id, ds.name, ds.data_type,
fn.SUM(m.file_size).alias("size"),
fn.COUNT(ds.id).alias("count")).join(
m, JOIN.LEFT_OUTER,
on=ds.id == m.dataset_id).group_by(ds.id))
query_results = list(query.dicts().execute())
result = []
for ds in query_results:
vo = DatasetVO()
result.append(vo)
for k, v in ds.items():
setattr(vo, k, v)
return result
def fetch_all_by_dataset(self, dataset_id: int = None):
ann = AnnotationEntity.alias("a")
ds_entry = DatasetEntryEntity.alias("i")
lbl = LabelEntity.alias("l")
query = (ann.select(ds_entry.file_path.alias("image"),
ann.kind.alias("annot_kind"),
ann.points.alias("annot_points"),
lbl.name.alias("label_name"),
lbl.color.alias("label_color")).join(
ds_entry, on=(ann.entry == ds_entry.id)).join(
lbl,
on=(ann.label == lbl.id),
join_type=JOIN.LEFT_OUTER).where(
ds_entry.dataset == dataset_id))
cursor = query.dicts().execute()
result = []
for row in cursor:
result.append(row)
return result
def delete_entry(self, id):
return DatasetEntryEntity.delete_by_id(id)
def tag_entries(self, entries: [DatasetEntryVO], label: LabelVO):
ids = [vo.id for vo in entries]
for chunk in MiscUtilities.chunk(ids, 100):
rows = (DatasetEntryEntity.update(label=label.id).where(
DatasetEntryEntity.id.in_(list(chunk))).execute())
def delete(self, id: int):
result = LabelEntity.delete_by_id(id)
DatasetEntryEntity.update(label=None).where(DatasetEntryEntity.label == id).execute()
return result