data_source.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301
  1. import datetime
  2. import json
  3. from cachetools import TTLCache
  4. from flask import request, current_app
  5. from flask_login import current_user
  6. from core.login.login import login_required
  7. from flask_restful import Resource, marshal_with, fields, reqparse, marshal
  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 libs.helper import TimestampField
  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. integrate_icon_fields = {
  23. 'type': fields.String,
  24. 'url': fields.String,
  25. 'emoji': fields.String
  26. }
  27. integrate_page_fields = {
  28. 'page_name': fields.String,
  29. 'page_id': fields.String,
  30. 'page_icon': fields.Nested(integrate_icon_fields, allow_null=True),
  31. 'parent_id': fields.String,
  32. 'type': fields.String
  33. }
  34. integrate_workspace_fields = {
  35. 'workspace_name': fields.String,
  36. 'workspace_id': fields.String,
  37. 'workspace_icon': fields.String,
  38. 'pages': fields.List(fields.Nested(integrate_page_fields)),
  39. 'total': fields.Integer
  40. }
  41. integrate_fields = {
  42. 'id': fields.String,
  43. 'provider': fields.String,
  44. 'created_at': TimestampField,
  45. 'is_bound': fields.Boolean,
  46. 'disabled': fields.Boolean,
  47. 'link': fields.String,
  48. 'source_info': fields.Nested(integrate_workspace_fields)
  49. }
  50. integrate_list_fields = {
  51. 'data': fields.List(fields.Nested(integrate_fields)),
  52. }
  53. @setup_required
  54. @login_required
  55. @account_initialization_required
  56. @marshal_with(integrate_list_fields)
  57. def get(self):
  58. # get workspace data source integrates
  59. data_source_integrates = db.session.query(DataSourceBinding).filter(
  60. DataSourceBinding.tenant_id == current_user.current_tenant_id,
  61. DataSourceBinding.disabled == False
  62. ).all()
  63. base_url = request.url_root.rstrip('/')
  64. data_source_oauth_base_path = "/console/api/oauth/data-source"
  65. providers = ["notion"]
  66. integrate_data = []
  67. for provider in providers:
  68. # existing_integrate = next((ai for ai in data_source_integrates if ai.provider == provider), None)
  69. existing_integrates = filter(lambda item: item.provider == provider, data_source_integrates)
  70. if existing_integrates:
  71. for existing_integrate in list(existing_integrates):
  72. integrate_data.append({
  73. 'id': existing_integrate.id,
  74. 'provider': provider,
  75. 'created_at': existing_integrate.created_at,
  76. 'is_bound': True,
  77. 'disabled': existing_integrate.disabled,
  78. 'source_info': existing_integrate.source_info,
  79. 'link': f'{base_url}{data_source_oauth_base_path}/{provider}'
  80. })
  81. else:
  82. integrate_data.append({
  83. 'id': None,
  84. 'provider': provider,
  85. 'created_at': None,
  86. 'source_info': None,
  87. 'is_bound': False,
  88. 'disabled': None,
  89. 'link': f'{base_url}{data_source_oauth_base_path}/{provider}'
  90. })
  91. return {'data': integrate_data}, 200
  92. @setup_required
  93. @login_required
  94. @account_initialization_required
  95. def patch(self, binding_id, action):
  96. binding_id = str(binding_id)
  97. action = str(action)
  98. data_source_binding = DataSourceBinding.query.filter_by(
  99. id=binding_id
  100. ).first()
  101. if data_source_binding is None:
  102. raise NotFound('Data source binding not found.')
  103. # enable binding
  104. if action == 'enable':
  105. if data_source_binding.disabled:
  106. data_source_binding.disabled = False
  107. data_source_binding.updated_at = datetime.datetime.utcnow()
  108. db.session.add(data_source_binding)
  109. db.session.commit()
  110. else:
  111. raise ValueError('Data source is not disabled.')
  112. # disable binding
  113. if action == 'disable':
  114. if not data_source_binding.disabled:
  115. data_source_binding.disabled = True
  116. data_source_binding.updated_at = datetime.datetime.utcnow()
  117. db.session.add(data_source_binding)
  118. db.session.commit()
  119. else:
  120. raise ValueError('Data source is disabled.')
  121. return {'result': 'success'}, 200
  122. class DataSourceNotionListApi(Resource):
  123. integrate_icon_fields = {
  124. 'type': fields.String,
  125. 'url': fields.String,
  126. 'emoji': fields.String
  127. }
  128. integrate_page_fields = {
  129. 'page_name': fields.String,
  130. 'page_id': fields.String,
  131. 'page_icon': fields.Nested(integrate_icon_fields, allow_null=True),
  132. 'is_bound': fields.Boolean,
  133. 'parent_id': fields.String,
  134. 'type': fields.String
  135. }
  136. integrate_workspace_fields = {
  137. 'workspace_name': fields.String,
  138. 'workspace_id': fields.String,
  139. 'workspace_icon': fields.String,
  140. 'pages': fields.List(fields.Nested(integrate_page_fields))
  141. }
  142. integrate_notion_info_list_fields = {
  143. 'notion_info': fields.List(fields.Nested(integrate_workspace_fields)),
  144. }
  145. @setup_required
  146. @login_required
  147. @account_initialization_required
  148. @marshal_with(integrate_notion_info_list_fields)
  149. def get(self):
  150. dataset_id = request.args.get('dataset_id', default=None, type=str)
  151. exist_page_ids = []
  152. # import notion in the exist dataset
  153. if dataset_id:
  154. dataset = DatasetService.get_dataset(dataset_id)
  155. if not dataset:
  156. raise NotFound('Dataset not found.')
  157. if dataset.data_source_type != 'notion_import':
  158. raise ValueError('Dataset is not notion type.')
  159. documents = Document.query.filter_by(
  160. dataset_id=dataset_id,
  161. tenant_id=current_user.current_tenant_id,
  162. data_source_type='notion_import',
  163. enabled=True
  164. ).all()
  165. if documents:
  166. for document in documents:
  167. data_source_info = json.loads(document.data_source_info)
  168. exist_page_ids.append(data_source_info['notion_page_id'])
  169. # get all authorized pages
  170. data_source_bindings = DataSourceBinding.query.filter_by(
  171. tenant_id=current_user.current_tenant_id,
  172. provider='notion',
  173. disabled=False
  174. ).all()
  175. if not data_source_bindings:
  176. return {
  177. 'notion_info': []
  178. }, 200
  179. pre_import_info_list = []
  180. for data_source_binding in data_source_bindings:
  181. source_info = data_source_binding.source_info
  182. pages = source_info['pages']
  183. # Filter out already bound pages
  184. for page in pages:
  185. if page['page_id'] in exist_page_ids:
  186. page['is_bound'] = True
  187. else:
  188. page['is_bound'] = False
  189. pre_import_info = {
  190. 'workspace_name': source_info['workspace_name'],
  191. 'workspace_icon': source_info['workspace_icon'],
  192. 'workspace_id': source_info['workspace_id'],
  193. 'pages': pages,
  194. }
  195. pre_import_info_list.append(pre_import_info)
  196. return {
  197. 'notion_info': pre_import_info_list
  198. }, 200
  199. class DataSourceNotionApi(Resource):
  200. @setup_required
  201. @login_required
  202. @account_initialization_required
  203. def get(self, workspace_id, page_id, page_type):
  204. workspace_id = str(workspace_id)
  205. page_id = str(page_id)
  206. data_source_binding = DataSourceBinding.query.filter(
  207. db.and_(
  208. DataSourceBinding.tenant_id == current_user.current_tenant_id,
  209. DataSourceBinding.provider == 'notion',
  210. DataSourceBinding.disabled == False,
  211. DataSourceBinding.source_info['workspace_id'] == f'"{workspace_id}"'
  212. )
  213. ).first()
  214. if not data_source_binding:
  215. raise NotFound('Data source binding not found.')
  216. loader = NotionLoader(
  217. notion_access_token=data_source_binding.access_token,
  218. notion_workspace_id=workspace_id,
  219. notion_obj_id=page_id,
  220. notion_page_type=page_type
  221. )
  222. text_docs = loader.load()
  223. return {
  224. 'content': "\n".join([doc.page_content for doc in text_docs])
  225. }, 200
  226. @setup_required
  227. @login_required
  228. @account_initialization_required
  229. def post(self):
  230. parser = reqparse.RequestParser()
  231. parser.add_argument('notion_info_list', type=list, required=True, nullable=True, location='json')
  232. parser.add_argument('process_rule', type=dict, required=True, nullable=True, location='json')
  233. args = parser.parse_args()
  234. # validate args
  235. DocumentService.estimate_args_validate(args)
  236. indexing_runner = IndexingRunner()
  237. response = indexing_runner.notion_indexing_estimate(current_user.current_tenant_id, args['notion_info_list'], args['process_rule'])
  238. return response, 200
  239. class DataSourceNotionDatasetSyncApi(Resource):
  240. @setup_required
  241. @login_required
  242. @account_initialization_required
  243. def get(self, dataset_id):
  244. dataset_id_str = str(dataset_id)
  245. dataset = DatasetService.get_dataset(dataset_id_str)
  246. if dataset is None:
  247. raise NotFound("Dataset not found.")
  248. documents = DocumentService.get_document_by_dataset_id(dataset_id_str)
  249. for document in documents:
  250. document_indexing_sync_task.delay(dataset_id_str, document.id)
  251. return 200
  252. class DataSourceNotionDocumentSyncApi(Resource):
  253. @setup_required
  254. @login_required
  255. @account_initialization_required
  256. def get(self, dataset_id, document_id):
  257. dataset_id_str = str(dataset_id)
  258. document_id_str = str(document_id)
  259. dataset = DatasetService.get_dataset(dataset_id_str)
  260. if dataset is None:
  261. raise NotFound("Dataset not found.")
  262. document = DocumentService.get_document(dataset_id_str, document_id_str)
  263. if document is None:
  264. raise NotFound("Document not found.")
  265. document_indexing_sync_task.delay(dataset_id_str, document_id_str)
  266. return 200
  267. api.add_resource(DataSourceApi, '/data-source/integrates', '/data-source/integrates/<uuid:binding_id>/<string:action>')
  268. api.add_resource(DataSourceNotionListApi, '/notion/pre-import/pages')
  269. api.add_resource(DataSourceNotionApi,
  270. '/notion/workspaces/<uuid:workspace_id>/pages/<uuid:page_id>/<string:page_type>/preview',
  271. '/datasets/notion-indexing-estimate')
  272. api.add_resource(DataSourceNotionDatasetSyncApi, '/datasets/<uuid:dataset_id>/notion/sync')
  273. api.add_resource(DataSourceNotionDocumentSyncApi, '/datasets/<uuid:dataset_id>/documents/<uuid:document_id>/notion/sync')