my-project-public

repository

loading code, commits, and activity

repositories

loading repo index

my-project-public — gitlawb

#1	import hashlib
#2	import os
#3	import sys
#4	from unittest.mock import mock_open, patch
#5
#6	import pytest
#7
#8	if sys.version_info > (3, 10): # as `match` statement was introduced in python 3.10
#9	from deepgram import PrerecordedOptions
#10
#11	from embedchain.loaders.audio import AudioLoader
#12
#13
#14	@pytest.fixture
#15	def setup_audio_loader(mocker):
#16	mock_dropbox = mocker.patch("deepgram.DeepgramClient")
#17	mock_dbx = mocker.MagicMock()
#18	mock_dropbox.return_value = mock_dbx
#19
#20	os.environ["DEEPGRAM_API_KEY"] = "test_key"
#21	loader = AudioLoader()
#22	loader.client = mock_dbx
#23
#24	yield loader, mock_dbx
#25
#26	if "DEEPGRAM_API_KEY" in os.environ:
#27	del os.environ["DEEPGRAM_API_KEY"]
#28
#29
#30	@pytest.mark.skipif(
#31	sys.version_info < (3, 10), reason="Test skipped for Python 3.9 or lower"
#32	) # as `match` statement was introduced in python 3.10
#33	def test_initialization(setup_audio_loader):
#34	"""Test initialization of AudioLoader."""
#35	loader, _ = setup_audio_loader
#36	assert loader is not None
#37
#38
#39	@pytest.mark.skipif(
#40	sys.version_info < (3, 10), reason="Test skipped for Python 3.9 or lower"
#41	) # as `match` statement was introduced in python 3.10
#42	def test_load_data_from_url(setup_audio_loader):
#43	loader, mock_dbx = setup_audio_loader
#44	url = "https://example.com/audio.mp3"
#45	expected_content = "This is a test audio transcript."
#46
#47	mock_response = {"results": {"channels": [{"alternatives": [{"transcript": expected_content}]}]}}
#48	mock_dbx.listen.prerecorded.v.return_value.transcribe_url.return_value = mock_response
#49
#50	result = loader.load_data(url)
#51
#52	doc_id = hashlib.sha256((expected_content + url).encode()).hexdigest()
#53	expected_result = {
#54	"doc_id": doc_id,
#55	"data": [
#56	{
#57	"content": expected_content,
#58	"meta_data": {"url": url},
#59	}
#60	],
#61	}
#62
#63	assert result == expected_result
#64	mock_dbx.listen.prerecorded.v.assert_called_once_with("1")
#65	mock_dbx.listen.prerecorded.v.return_value.transcribe_url.assert_called_once_with(
#66	{"url": url}, PrerecordedOptions(model="nova-2", smart_format=True)
#67	)
#68
#69
#70	@pytest.mark.skipif(
#71	sys.version_info < (3, 10), reason="Test skipped for Python 3.9 or lower"
#72	) # as `match` statement was introduced in python 3.10
#73	def test_load_data_from_file(setup_audio_loader):
#74	loader, mock_dbx = setup_audio_loader
#75	file_path = "local_audio.mp3"
#76	expected_content = "This is a test audio transcript."
#77
#78	mock_response = {"results": {"channels": [{"alternatives": [{"transcript": expected_content}]}]}}
#79	mock_dbx.listen.prerecorded.v.return_value.transcribe_file.return_value = mock_response
#80
#81	# Mock the file reading functionality
#82	with patch("builtins.open", mock_open(read_data=b"some data")) as mock_file:
#83	result = loader.load_data(file_path)
#84
#85	doc_id = hashlib.sha256((expected_content + file_path).encode()).hexdigest()
#86	expected_result = {
#87	"doc_id": doc_id,
#88	"data": [
#89	{
#90	"content": expected_content,
#91	"meta_data": {"url": file_path},
#92	}
#93	],
#94	}
#95
#96	assert result == expected_result
#97	mock_dbx.listen.prerecorded.v.assert_called_once_with("1")
#98	mock_dbx.listen.prerecorded.v.return_value.transcribe_file.assert_called_once_with(
#99	{"buffer": mock_file.return_value}, PrerecordedOptions(model="nova-2", smart_format=True)
#100	)
#101

z6Mkq5mY3JWtxoxUobWcfNHm7AkRubgSWEZTkBVqZXJviFZ5/my-project-public