my-project-public

repository

loading code, commits, and activity

repositories

loading repo index

my-project-public — gitlawb

#1	import hashlib
#2
#3	import pytest
#4
#5	from embedchain.loaders.json import JSONLoader
#6
#7
#8	def test_load_data(mocker):
#9	content = "temp.json"
#10
#11	mock_document = {
#12	"doc_id": hashlib.sha256((content + ", ".join(["content1", "content2"])).encode()).hexdigest(),
#13	"data": [
#14	{"content": "content1", "meta_data": {"url": content}},
#15	{"content": "content2", "meta_data": {"url": content}},
#16	],
#17	}
#18
#19	mocker.patch("embedchain.loaders.json.JSONLoader.load_data", return_value=mock_document)
#20
#21	json_loader = JSONLoader()
#22
#23	result = json_loader.load_data(content)
#24
#25	assert "doc_id" in result
#26	assert "data" in result
#27
#28	expected_data = [
#29	{"content": "content1", "meta_data": {"url": content}},
#30	{"content": "content2", "meta_data": {"url": content}},
#31	]
#32
#33	assert result["data"] == expected_data
#34
#35	expected_doc_id = hashlib.sha256((content + ", ".join(["content1", "content2"])).encode()).hexdigest()
#36	assert result["doc_id"] == expected_doc_id
#37
#38
#39	def test_load_data_url(mocker):
#40	content = "https://example.com/posts.json"
#41
#42	mocker.patch("os.path.isfile", return_value=False)
#43	mocker.patch(
#44	"embedchain.loaders.json.JSONReader.load_data",
#45	return_value=[
#46	{
#47	"text": "content1",
#48	},
#49	{
#50	"text": "content2",
#51	},
#52	],
#53	)
#54
#55	mock_response = mocker.Mock()
#56	mock_response.status_code = 200
#57	mock_response.json.return_value = {"document1": "content1", "document2": "content2"}
#58
#59	mocker.patch("requests.get", return_value=mock_response)
#60
#61	result = JSONLoader.load_data(content)
#62
#63	assert "doc_id" in result
#64	assert "data" in result
#65
#66	expected_data = [
#67	{"content": "content1", "meta_data": {"url": content}},
#68	{"content": "content2", "meta_data": {"url": content}},
#69	]
#70
#71	assert result["data"] == expected_data
#72
#73	expected_doc_id = hashlib.sha256((content + ", ".join(["content1", "content2"])).encode()).hexdigest()
#74	assert result["doc_id"] == expected_doc_id
#75
#76
#77	def test_load_data_invalid_string_content(mocker):
#78	mocker.patch("os.path.isfile", return_value=False)
#79	mocker.patch("requests.get")
#80
#81	content = "123: 345}"
#82
#83	with pytest.raises(ValueError, match="Invalid content to load json data from"):
#84	JSONLoader.load_data(content)
#85
#86
#87	def test_load_data_invalid_url(mocker):
#88	mocker.patch("os.path.isfile", return_value=False)
#89
#90	mock_response = mocker.Mock()
#91	mock_response.status_code = 404
#92	mocker.patch("requests.get", return_value=mock_response)
#93
#94	content = "http://invalid-url.com/"
#95
#96	with pytest.raises(ValueError, match=f"Invalid content to load json data from: {content}"):
#97	JSONLoader.load_data(content)
#98
#99
#100	def test_load_data_from_json_string(mocker):
#101	content = '{"foo": "bar"}'
#102
#103	content_url_str = hashlib.sha256((content).encode("utf-8")).hexdigest()
#104
#105	mocker.patch("os.path.isfile", return_value=False)
#106	mocker.patch(
#107	"embedchain.loaders.json.JSONReader.load_data",
#108	return_value=[
#109	{
#110	"text": "content1",
#111	},
#112	{
#113	"text": "content2",
#114	},
#115	],
#116	)
#117
#118	result = JSONLoader.load_data(content)
#119
#120	assert "doc_id" in result
#121	assert "data" in result
#122
#123	expected_data = [
#124	{"content": "content1", "meta_data": {"url": content_url_str}},
#125	{"content": "content2", "meta_data": {"url": content_url_str}},
#126	]
#127
#128	assert result["data"] == expected_data
#129
#130	expected_doc_id = hashlib.sha256((content_url_str + ", ".join(["content1", "content2"])).encode()).hexdigest()
#131	assert result["doc_id"] == expected_doc_id
#132

z6Mkq5mY3JWtxoxUobWcfNHm7AkRubgSWEZTkBVqZXJviFZ5/my-project-public