data_source.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269
  1. import datetime
  2. import json
  3. from flask import request
  4. from flask_login import current_user
  5. from flask_restful import Resource, marshal_with, reqparse
  6. from werkzeug.exceptions import NotFound
  7. from controllers.console import api
  8. from controllers.console.setup import setup_required
  9. from controllers.console.wraps import account_initialization_required
  10. from core.indexing_runner import IndexingRunner
  11. from core.rag.extractor.entity.extract_setting import ExtractSetting
  12. from core.rag.extractor.notion_extractor import NotionExtractor
  13. from extensions.ext_database import db
  14. from fields.data_source_fields import integrate_list_fields, integrate_notion_info_list_fields
  15. from libs.login import login_required
  16. from models import DataSourceOauthBinding, Document
  17. from services.dataset_service import DatasetService, DocumentService
  18. from tasks.document_indexing_sync_task import document_indexing_sync_task
  19. class DataSourceApi(Resource):
  20. @setup_required
  21. @login_required
  22. @account_initialization_required
  23. @marshal_with(integrate_list_fields)
  24. def get(self):
  25. # get workspace data source integrates
  26. data_source_integrates = (
  27. db.session.query(DataSourceOauthBinding)
  28. .filter(
  29. DataSourceOauthBinding.tenant_id == current_user.current_tenant_id,
  30. DataSourceOauthBinding.disabled == False,
  31. )
  32. .all()
  33. )
  34. base_url = request.url_root.rstrip("/")
  35. data_source_oauth_base_path = "/console/api/oauth/data-source"
  36. providers = ["notion"]
  37. integrate_data = []
  38. for provider in providers:
  39. # existing_integrate = next((ai for ai in data_source_integrates if ai.provider == provider), None)
  40. existing_integrates = filter(lambda item: item.provider == provider, data_source_integrates)
  41. if existing_integrates:
  42. for existing_integrate in list(existing_integrates):
  43. integrate_data.append(
  44. {
  45. "id": existing_integrate.id,
  46. "provider": provider,
  47. "created_at": existing_integrate.created_at,
  48. "is_bound": True,
  49. "disabled": existing_integrate.disabled,
  50. "source_info": existing_integrate.source_info,
  51. "link": f"{base_url}{data_source_oauth_base_path}/{provider}",
  52. }
  53. )
  54. else:
  55. integrate_data.append(
  56. {
  57. "id": None,
  58. "provider": provider,
  59. "created_at": None,
  60. "source_info": None,
  61. "is_bound": False,
  62. "disabled": None,
  63. "link": f"{base_url}{data_source_oauth_base_path}/{provider}",
  64. }
  65. )
  66. return {"data": integrate_data}, 200
  67. @setup_required
  68. @login_required
  69. @account_initialization_required
  70. def patch(self, binding_id, action):
  71. binding_id = str(binding_id)
  72. action = str(action)
  73. data_source_binding = DataSourceOauthBinding.query.filter_by(id=binding_id).first()
  74. if data_source_binding is None:
  75. raise NotFound("Data source binding not found.")
  76. # enable binding
  77. if action == "enable":
  78. if data_source_binding.disabled:
  79. data_source_binding.disabled = False
  80. data_source_binding.updated_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
  81. db.session.add(data_source_binding)
  82. db.session.commit()
  83. else:
  84. raise ValueError("Data source is not disabled.")
  85. # disable binding
  86. if action == "disable":
  87. if not data_source_binding.disabled:
  88. data_source_binding.disabled = True
  89. data_source_binding.updated_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
  90. db.session.add(data_source_binding)
  91. db.session.commit()
  92. else:
  93. raise ValueError("Data source is disabled.")
  94. return {"result": "success"}, 200
  95. class DataSourceNotionListApi(Resource):
  96. @setup_required
  97. @login_required
  98. @account_initialization_required
  99. @marshal_with(integrate_notion_info_list_fields)
  100. def get(self):
  101. dataset_id = request.args.get("dataset_id", default=None, type=str)
  102. exist_page_ids = []
  103. # import notion in the exist dataset
  104. if dataset_id:
  105. dataset = DatasetService.get_dataset(dataset_id)
  106. if not dataset:
  107. raise NotFound("Dataset not found.")
  108. if dataset.data_source_type != "notion_import":
  109. raise ValueError("Dataset is not notion type.")
  110. documents = Document.query.filter_by(
  111. dataset_id=dataset_id,
  112. tenant_id=current_user.current_tenant_id,
  113. data_source_type="notion_import",
  114. enabled=True,
  115. ).all()
  116. if documents:
  117. for document in documents:
  118. data_source_info = json.loads(document.data_source_info)
  119. exist_page_ids.append(data_source_info["notion_page_id"])
  120. # get all authorized pages
  121. data_source_bindings = DataSourceOauthBinding.query.filter_by(
  122. tenant_id=current_user.current_tenant_id, provider="notion", disabled=False
  123. ).all()
  124. if not data_source_bindings:
  125. return {"notion_info": []}, 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 {"notion_info": pre_import_info_list}, 200
  144. class DataSourceNotionApi(Resource):
  145. @setup_required
  146. @login_required
  147. @account_initialization_required
  148. def get(self, workspace_id, page_id, page_type):
  149. workspace_id = str(workspace_id)
  150. page_id = str(page_id)
  151. data_source_binding = DataSourceOauthBinding.query.filter(
  152. db.and_(
  153. DataSourceOauthBinding.tenant_id == current_user.current_tenant_id,
  154. DataSourceOauthBinding.provider == "notion",
  155. DataSourceOauthBinding.disabled == False,
  156. DataSourceOauthBinding.source_info["workspace_id"] == f'"{workspace_id}"',
  157. )
  158. ).first()
  159. if not data_source_binding:
  160. raise NotFound("Data source binding not found.")
  161. extractor = NotionExtractor(
  162. notion_workspace_id=workspace_id,
  163. notion_obj_id=page_id,
  164. notion_page_type=page_type,
  165. notion_access_token=data_source_binding.access_token,
  166. tenant_id=current_user.current_tenant_id,
  167. )
  168. text_docs = extractor.extract()
  169. return {"content": "\n".join([doc.page_content for doc in text_docs])}, 200
  170. @setup_required
  171. @login_required
  172. @account_initialization_required
  173. def post(self):
  174. parser = reqparse.RequestParser()
  175. parser.add_argument("notion_info_list", type=list, required=True, nullable=True, location="json")
  176. parser.add_argument("process_rule", type=dict, required=True, nullable=True, location="json")
  177. parser.add_argument("doc_form", type=str, default="text_model", required=False, nullable=False, location="json")
  178. parser.add_argument(
  179. "doc_language", type=str, default="English", required=False, nullable=False, location="json"
  180. )
  181. args = parser.parse_args()
  182. # validate args
  183. DocumentService.estimate_args_validate(args)
  184. notion_info_list = args["notion_info_list"]
  185. extract_settings = []
  186. for notion_info in notion_info_list:
  187. workspace_id = notion_info["workspace_id"]
  188. for page in notion_info["pages"]:
  189. extract_setting = ExtractSetting(
  190. datasource_type="notion_import",
  191. notion_info={
  192. "notion_workspace_id": workspace_id,
  193. "notion_obj_id": page["page_id"],
  194. "notion_page_type": page["type"],
  195. "tenant_id": current_user.current_tenant_id,
  196. },
  197. document_model=args["doc_form"],
  198. )
  199. extract_settings.append(extract_setting)
  200. indexing_runner = IndexingRunner()
  201. response = indexing_runner.indexing_estimate(
  202. current_user.current_tenant_id,
  203. extract_settings,
  204. args["process_rule"],
  205. args["doc_form"],
  206. args["doc_language"],
  207. )
  208. return response, 200
  209. class DataSourceNotionDatasetSyncApi(Resource):
  210. @setup_required
  211. @login_required
  212. @account_initialization_required
  213. def get(self, dataset_id):
  214. dataset_id_str = str(dataset_id)
  215. dataset = DatasetService.get_dataset(dataset_id_str)
  216. if dataset is None:
  217. raise NotFound("Dataset not found.")
  218. documents = DocumentService.get_document_by_dataset_id(dataset_id_str)
  219. for document in documents:
  220. document_indexing_sync_task.delay(dataset_id_str, document.id)
  221. return 200
  222. class DataSourceNotionDocumentSyncApi(Resource):
  223. @setup_required
  224. @login_required
  225. @account_initialization_required
  226. def get(self, dataset_id, document_id):
  227. dataset_id_str = str(dataset_id)
  228. document_id_str = str(document_id)
  229. dataset = DatasetService.get_dataset(dataset_id_str)
  230. if dataset is None:
  231. raise NotFound("Dataset not found.")
  232. document = DocumentService.get_document(dataset_id_str, document_id_str)
  233. if document is None:
  234. raise NotFound("Document not found.")
  235. document_indexing_sync_task.delay(dataset_id_str, document_id_str)
  236. return 200
  237. api.add_resource(DataSourceApi, "/data-source/integrates", "/data-source/integrates/<uuid:binding_id>/<string:action>")
  238. api.add_resource(DataSourceNotionListApi, "/notion/pre-import/pages")
  239. api.add_resource(
  240. DataSourceNotionApi,
  241. "/notion/workspaces/<uuid:workspace_id>/pages/<uuid:page_id>/<string:page_type>/preview",
  242. "/datasets/notion-indexing-estimate",
  243. )
  244. api.add_resource(DataSourceNotionDatasetSyncApi, "/datasets/<uuid:dataset_id>/notion/sync")
  245. api.add_resource(
  246. DataSourceNotionDocumentSyncApi, "/datasets/<uuid:dataset_id>/documents/<uuid:document_id>/notion/sync"
  247. )