cecilia-uu
commited on
Commit
·
a009daf
1
Parent(s):
090b2e7
delete_dataset method and tests created (#1186)
Browse files### What problem does this PR solve?
This PR have completed both HTTP API and Python SDK for
'delete_dataset". In addition, there are tests for it.
### Type of change
- [x] New Feature (non-breaking change which adds functionality)
- api/apps/dataset_api.py +45 -5
- sdk/python/ragflow/ragflow.py +17 -2
- sdk/python/test/test_dataset.py +15 -0
api/apps/dataset_api.py
CHANGED
|
@@ -135,9 +135,52 @@ def list_datasets():
|
|
| 135 |
|
| 136 |
@manager.route('/<dataset_id>', methods=['DELETE'])
|
| 137 |
@login_required
|
|
|
|
| 138 |
def remove_dataset(dataset_id):
|
| 139 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 140 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 141 |
|
| 142 |
@manager.route('/<dataset_id>', methods=['PUT'])
|
| 143 |
@login_required
|
|
@@ -146,10 +189,7 @@ def update_dataset(dataset_id):
|
|
| 146 |
return construct_json_result(code=RetCode.DATA_ERROR, message=f"attempt to update dataset: {dataset_id}")
|
| 147 |
|
| 148 |
|
| 149 |
-
|
| 150 |
-
@login_required
|
| 151 |
-
def get_dataset(dataset_id):
|
| 152 |
-
return construct_json_result(code=RetCode.DATA_ERROR, message=f"attempt to get detail of dataset: {dataset_id}")
|
| 153 |
|
| 154 |
|
| 155 |
|
|
|
|
| 135 |
|
| 136 |
@manager.route('/<dataset_id>', methods=['DELETE'])
|
| 137 |
@login_required
|
| 138 |
+
@validate_request("dataset_id")
|
| 139 |
def remove_dataset(dataset_id):
|
| 140 |
+
req = request.json
|
| 141 |
+
try:
|
| 142 |
+
kbs = KnowledgebaseService.query(
|
| 143 |
+
created_by=current_user.id, id=req["dataset_id"])
|
| 144 |
+
if not kbs:
|
| 145 |
+
return construct_json_result(
|
| 146 |
+
data=False, message=f'Only owner of knowledgebase authorized for this operation.',
|
| 147 |
+
code=RetCode.OPERATING_ERROR)
|
| 148 |
+
|
| 149 |
+
for doc in DocumentService.query(kb_id=req["dataset_id"]):
|
| 150 |
+
if not DocumentService.remove_document(doc, kbs[0].tenant_id):
|
| 151 |
+
return construct_json_result(
|
| 152 |
+
message="Database error (Document removal)!")
|
| 153 |
+
f2d = File2DocumentService.get_by_document_id(doc.id)
|
| 154 |
+
FileService.filter_delete([File.source_type == FileSource.KNOWLEDGEBASE, File.id == f2d[0].file_id])
|
| 155 |
+
File2DocumentService.delete_by_document_id(doc.id)
|
| 156 |
+
|
| 157 |
+
if not KnowledgebaseService.delete_by_id(req["dataset_id"]):
|
| 158 |
+
return construct_json_result(
|
| 159 |
+
message="Database error (Knowledgebase removal)!")
|
| 160 |
+
return construct_json_result(code=RetCode.DATA_ERROR, message=f"attempt to remove dataset: {dataset_id}")
|
| 161 |
+
except Exception as e:
|
| 162 |
+
return construct_error_response(e)
|
| 163 |
|
| 164 |
+
# ------------------------------ get details of a dataset ----------------------------------------
|
| 165 |
+
@manager.route('/<dataset_id>', methods=['GET'])
|
| 166 |
+
@login_required
|
| 167 |
+
@validate_request("dataset_id")
|
| 168 |
+
def get_dataset():
|
| 169 |
+
dataset_id = request.args["dataset_id"]
|
| 170 |
+
try:
|
| 171 |
+
dataset = KnowledgebaseService.get_detail(dataset_id)
|
| 172 |
+
if not dataset:
|
| 173 |
+
return construct_json_result(
|
| 174 |
+
message="Can't find this knowledgebase!")
|
| 175 |
+
return construct_json_result(code=RetCode.DATA_ERROR, message=f"attempt to get detail of dataset: {dataset_id}")
|
| 176 |
+
except Exception as e:
|
| 177 |
+
return construct_json_result(e)
|
| 178 |
+
|
| 179 |
+
# ------------------------------ update a dataset --------------------------------------------
|
| 180 |
+
@manager.route('/<dataset_id>', methods=['GET'])
|
| 181 |
+
@login_required
|
| 182 |
+
def get_dataset(dataset_id):
|
| 183 |
+
return construct_json_result(code=RetCode.DATA_ERROR, message=f"attempt to get detail of dataset: {dataset_id}")
|
| 184 |
|
| 185 |
@manager.route('/<dataset_id>', methods=['PUT'])
|
| 186 |
@login_required
|
|
|
|
| 189 |
return construct_json_result(code=RetCode.DATA_ERROR, message=f"attempt to update dataset: {dataset_id}")
|
| 190 |
|
| 191 |
|
| 192 |
+
|
|
|
|
|
|
|
|
|
|
| 193 |
|
| 194 |
|
| 195 |
|
sdk/python/ragflow/ragflow.py
CHANGED
|
@@ -39,8 +39,23 @@ class RAGFlow:
|
|
| 39 |
result_dict = json.loads(res.text)
|
| 40 |
return result_dict
|
| 41 |
|
| 42 |
-
def delete_dataset(self, dataset_name
|
| 43 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 44 |
|
| 45 |
def list_dataset(self, offset=0, count=-1, orderby="create_time", desc=True):
|
| 46 |
params = {
|
|
|
|
| 39 |
result_dict = json.loads(res.text)
|
| 40 |
return result_dict
|
| 41 |
|
| 42 |
+
def delete_dataset(self, dataset_name):
|
| 43 |
+
dataset_id = self.find_dataset_id_by_name(dataset_name)
|
| 44 |
+
if not dataset_id:
|
| 45 |
+
return {"success": False, "message": "Dataset not found."}
|
| 46 |
+
|
| 47 |
+
res = requests.delete(f"{self.dataset_url}/{dataset_id}", headers=self.authorization_header)
|
| 48 |
+
if res.status_code == 200:
|
| 49 |
+
return {"success": True, "message": "Dataset deleted successfully!"}
|
| 50 |
+
else:
|
| 51 |
+
return {"success": False, "message": f"Other status code: {res.status_code}"}
|
| 52 |
+
|
| 53 |
+
def find_dataset_id_by_name(self, dataset_name):
|
| 54 |
+
res = requests.get(self.dataset_url, headers=self.authorization_header)
|
| 55 |
+
for dataset in res.json()['data']:
|
| 56 |
+
if dataset['name'] == dataset_name:
|
| 57 |
+
return dataset['id']
|
| 58 |
+
return None
|
| 59 |
|
| 60 |
def list_dataset(self, offset=0, count=-1, orderby="create_time", desc=True):
|
| 61 |
params = {
|
sdk/python/test/test_dataset.py
CHANGED
|
@@ -101,6 +101,21 @@ class TestDataset(TestSdk):
|
|
| 101 |
_, res = response
|
| 102 |
assert "IndexError" in res['message']
|
| 103 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 104 |
|
| 105 |
|
| 106 |
|
|
|
|
| 101 |
_, res = response
|
| 102 |
assert "IndexError" in res['message']
|
| 103 |
|
| 104 |
+
def test_delete_one_dataset_with_success(self):
|
| 105 |
+
# get the real name of the created dataset
|
| 106 |
+
ragflow = RAGFlow(API_KEY, HOST_ADDRESS)
|
| 107 |
+
res = ragflow.create_dataset("kb0")
|
| 108 |
+
real_dataset_name = res['data']['dataset_name']
|
| 109 |
+
print("name", real_dataset_name)
|
| 110 |
+
# delete this dataset
|
| 111 |
+
result = ragflow.delete_dataset(real_dataset_name)
|
| 112 |
+
print(result)
|
| 113 |
+
assert result["success"] is True
|
| 114 |
+
|
| 115 |
+
def test_delete_dataset_with_not_existing_dataset(self):
|
| 116 |
+
ragflow = RAGFlow(API_KEY, HOST_ADDRESS)
|
| 117 |
+
res = ragflow.delete_dataset("weird_dataset")
|
| 118 |
+
assert res["success"] is False
|
| 119 |
|
| 120 |
|
| 121 |
|