56 lines
1.7 KiB
Python
56 lines
1.7 KiB
Python
import json
|
|
|
|
|
|
def test_camembert_load_is_idempotent_and_reuses_process_session(tmp_path, monkeypatch):
|
|
import camembert_ner_manager as module
|
|
|
|
model_dir = tmp_path / "camembert-bio-deid" / "onnx"
|
|
model_dir.mkdir(parents=True)
|
|
(model_dir / "model.onnx").write_bytes(b"fake")
|
|
(model_dir / "config.json").write_text(
|
|
json.dumps({"id2label": {"0": "O", "1": "B-PER"}}),
|
|
encoding="utf-8",
|
|
)
|
|
(model_dir.parent / "VERSION.json").write_text(
|
|
json.dumps({"current_version": "v-test", "versions": {"v-test": {"f1": 1, "recall": 1}}}),
|
|
encoding="utf-8",
|
|
)
|
|
|
|
created_sessions = []
|
|
|
|
class FakeSessionOptions:
|
|
inter_op_num_threads = 0
|
|
intra_op_num_threads = 0
|
|
|
|
class FakeOrt:
|
|
SessionOptions = FakeSessionOptions
|
|
|
|
@staticmethod
|
|
def InferenceSession(path, sess_options=None, providers=None):
|
|
session = {"path": path, "providers": providers}
|
|
created_sessions.append(session)
|
|
return session
|
|
|
|
class FakeTokenizer:
|
|
@staticmethod
|
|
def from_pretrained(path):
|
|
return {"tokenizer_path": path}
|
|
|
|
monkeypatch.setattr(module, "_ORT_AVAILABLE", True)
|
|
monkeypatch.setattr(module, "_TOKENIZERS_AVAILABLE", True)
|
|
monkeypatch.setattr(module, "ort", FakeOrt)
|
|
monkeypatch.setattr(module, "AutoTokenizer", FakeTokenizer)
|
|
module._PROCESS_CACHE.clear()
|
|
|
|
first = module.CamembertNerManager(model_dir)
|
|
first.load()
|
|
first.load()
|
|
|
|
second = module.CamembertNerManager(model_dir)
|
|
second.load()
|
|
|
|
assert len(created_sessions) == 1
|
|
assert first.is_loaded()
|
|
assert second.is_loaded()
|
|
assert first._session is second._session
|