data_source.py 9.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248
  1. import datetime
  2. import json
  3. from cachetools import TTLCache
  4. from flask import request
  5. from flask_login import current_user
  6. from libs.login import login_required
  7. from flask_restful import Resource, marshal_with, reqparse
  8. from werkzeug.exceptions import NotFound
  9. from controllers.console import api
  10. from controllers.console.setup import setup_required
  11. from controllers.console.wraps import account_initialization_required
  12. from core.data_loader.loader.notion import NotionLoader
  13. from core.indexing_runner import IndexingRunner
  14. from extensions.ext_database import db
  15. from fields.data_source_fields import integrate_notion_info_list_fields, integrate_list_fields
  16. from models.dataset import Document
  17. from models.source import DataSourceBinding
  18. from services.dataset_service import DatasetService, DocumentService
  19. from tasks.document_indexing_sync_task import document_indexing_sync_task
  20. cache = TTLCache(maxsize=None, ttl=30)
  21. class DataSourceApi(Resource):
  22. @setup_required
  23. @login_required
  24. @account_initialization_required
  25. @marshal_with(integrate_list_fields)
  26. def get(self):
  27. # get workspace data source integrates
  28. data_source_integrates = db.session.query(DataSourceBinding).filter(
  29. DataSourceBinding.tenant_id == current_user.current_tenant_id,
  30. DataSourceBinding.disabled == False
  31. ).all()
  32. base_url = request.url_root.rstrip('/')
  33. data_source_oauth_base_path = "/console/api/oauth/data-source"
  34. providers = ["notion"]
  35. integrate_data = []
  36. for provider in providers:
  37. # existing_integrate = next((ai for ai in data_source_integrates if ai.provider == provider), None)
  38. existing_integrates = filter(lambda item: item.provider == provider, data_source_integrates)
  39. if existing_integrates:
  40. for existing_integrate in list(existing_integrates):
  41. integrate_data.append({
  42. 'id': existing_integrate.id,
  43. 'provider': provider,
  44. 'created_at': existing_integrate.created_at,
  45. 'is_bound': True,
  46. 'disabled': existing_integrate.disabled,
  47. 'source_info': existing_integrate.source_info,
  48. 'link': f'{base_url}{data_source_oauth_base_path}/{provider}'
  49. })
  50. else:
  51. integrate_data.append({
  52. 'id': None,
  53. 'provider': provider,
  54. 'created_at': None,
  55. 'source_info': None,
  56. 'is_bound': False,
  57. 'disabled': None,
  58. 'link': f'{base_url}{data_source_oauth_base_path}/{provider}'
  59. })
  60. return {'data': integrate_data}, 200
  61. @setup_required
  62. @login_required
  63. @account_initialization_required
  64. def patch(self, binding_id, action):
  65. binding_id = str(binding_id)
  66. action = str(action)
  67. data_source_binding = DataSourceBinding.query.filter_by(
  68. id=binding_id
  69. ).first()
  70. if data_source_binding is None:
  71. raise NotFound('Data source binding not found.')
  72. # enable binding
  73. if action == 'enable':
  74. if data_source_binding.disabled:
  75. data_source_binding.disabled = False
  76. data_source_binding.updated_at = datetime.datetime.utcnow()
  77. db.session.add(data_source_binding)
  78. db.session.commit()
  79. else:
  80. raise ValueError('Data source is not disabled.')
  81. # disable binding
  82. if action == 'disable':
  83. if not data_source_binding.disabled:
  84. data_source_binding.disabled = True
  85. data_source_binding.updated_at = datetime.datetime.utcnow()
  86. db.session.add(data_source_binding)
  87. db.session.commit()
  88. else:
  89. raise ValueError('Data source is disabled.')
  90. return {'result': 'success'}, 200
  91. class DataSourceNotionListApi(Resource):
  92. @setup_required
  93. @login_required
  94. @account_initialization_required
  95. @marshal_with(integrate_notion_info_list_fields)
  96. def get(self):
  97. dataset_id = request.args.get('dataset_id', default=None, type=str)
  98. exist_page_ids = []
  99. # import notion in the exist dataset
  100. if dataset_id:
  101. dataset = DatasetService.get_dataset(dataset_id)
  102. if not dataset:
  103. raise NotFound('Dataset not found.')
  104. if dataset.data_source_type != 'notion_import':
  105. raise ValueError('Dataset is not notion type.')
  106. documents = Document.query.filter_by(
  107. dataset_id=dataset_id,
  108. tenant_id=current_user.current_tenant_id,
  109. data_source_type='notion_import',
  110. enabled=True
  111. ).all()
  112. if documents:
  113. for document in documents:
  114. data_source_info = json.loads(document.data_source_info)
  115. exist_page_ids.append(data_source_info['notion_page_id'])
  116. # get all authorized pages
  117. data_source_bindings = DataSourceBinding.query.filter_by(
  118. tenant_id=current_user.current_tenant_id,
  119. provider='notion',
  120. disabled=False
  121. ).all()
  122. if not data_source_bindings:
  123. return {
  124. 'notion_info': []
  125. }, 200
  126. pre_import_info_list = []
  127. for data_source_binding in data_source_bindings:
  128. source_info = data_source_binding.source_info
  129. pages = source_info['pages']
  130. # Filter out already bound pages
  131. for page in pages:
  132. if page['page_id'] in exist_page_ids:
  133. page['is_bound'] = True
  134. else:
  135. page['is_bound'] = False
  136. pre_import_info = {
  137. 'workspace_name': source_info['workspace_name'],
  138. 'workspace_icon': source_info['workspace_icon'],
  139. 'workspace_id': source_info['workspace_id'],
  140. 'pages': pages,
  141. }
  142. pre_import_info_list.append(pre_import_info)
  143. return {
  144. 'notion_info': pre_import_info_list
  145. }, 200
  146. class DataSourceNotionApi(Resource):
  147. @setup_required
  148. @login_required
  149. @account_initialization_required
  150. def get(self, workspace_id, page_id, page_type):
  151. workspace_id = str(workspace_id)
  152. page_id = str(page_id)
  153. data_source_binding = DataSourceBinding.query.filter(
  154. db.and_(
  155. DataSourceBinding.tenant_id == current_user.current_tenant_id,
  156. DataSourceBinding.provider == 'notion',
  157. DataSourceBinding.disabled == False,
  158. DataSourceBinding.source_info['workspace_id'] == f'"{workspace_id}"'
  159. )
  160. ).first()
  161. if not data_source_binding:
  162. raise NotFound('Data source binding not found.')
  163. loader = NotionLoader(
  164. notion_access_token=data_source_binding.access_token,
  165. notion_workspace_id=workspace_id,
  166. notion_obj_id=page_id,
  167. notion_page_type=page_type
  168. )
  169. text_docs = loader.load()
  170. return {
  171. 'content': "\n".join([doc.page_content for doc in text_docs])
  172. }, 200
  173. @setup_required
  174. @login_required
  175. @account_initialization_required
  176. def post(self):
  177. parser = reqparse.RequestParser()
  178. parser.add_argument('notion_info_list', type=list, required=True, nullable=True, location='json')
  179. parser.add_argument('process_rule', type=dict, required=True, nullable=True, location='json')
  180. args = parser.parse_args()
  181. # validate args
  182. DocumentService.estimate_args_validate(args)
  183. indexing_runner = IndexingRunner()
  184. response = indexing_runner.notion_indexing_estimate(current_user.current_tenant_id, args['notion_info_list'], args['process_rule'])
  185. return response, 200
  186. class DataSourceNotionDatasetSyncApi(Resource):
  187. @setup_required
  188. @login_required
  189. @account_initialization_required
  190. def get(self, dataset_id):
  191. dataset_id_str = str(dataset_id)
  192. dataset = DatasetService.get_dataset(dataset_id_str)
  193. if dataset is None:
  194. raise NotFound("Dataset not found.")
  195. documents = DocumentService.get_document_by_dataset_id(dataset_id_str)
  196. for document in documents:
  197. document_indexing_sync_task.delay(dataset_id_str, document.id)
  198. return 200
  199. class DataSourceNotionDocumentSyncApi(Resource):
  200. @setup_required
  201. @login_required
  202. @account_initialization_required
  203. def get(self, dataset_id, document_id):
  204. dataset_id_str = str(dataset_id)
  205. document_id_str = str(document_id)
  206. dataset = DatasetService.get_dataset(dataset_id_str)
  207. if dataset is None:
  208. raise NotFound("Dataset not found.")
  209. document = DocumentService.get_document(dataset_id_str, document_id_str)
  210. if document is None:
  211. raise NotFound("Document not found.")
  212. document_indexing_sync_task.delay(dataset_id_str, document_id_str)
  213. return 200
  214. api.add_resource(DataSourceApi, '/data-source/integrates', '/data-source/integrates/<uuid:binding_id>/<string:action>')
  215. api.add_resource(DataSourceNotionListApi, '/notion/pre-import/pages')
  216. api.add_resource(DataSourceNotionApi,
  217. '/notion/workspaces/<uuid:workspace_id>/pages/<uuid:page_id>/<string:page_type>/preview',
  218. '/datasets/notion-indexing-estimate')
  219. api.add_resource(DataSourceNotionDatasetSyncApi, '/datasets/<uuid:dataset_id>/notion/sync')
  220. api.add_resource(DataSourceNotionDocumentSyncApi, '/datasets/<uuid:dataset_id>/documents/<uuid:document_id>/notion/sync')