external_knowledge_service.py 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294
  1. import json
  2. from copy import deepcopy
  3. from datetime import UTC, datetime
  4. from typing import Any, Optional, Union, cast
  5. import httpx
  6. import validators
  7. from constants import HIDDEN_VALUE
  8. from core.helper import ssrf_proxy
  9. from core.rag.entities.metadata_entities import MetadataCondition
  10. from extensions.ext_database import db
  11. from models.dataset import (
  12. Dataset,
  13. ExternalKnowledgeApis,
  14. ExternalKnowledgeBindings,
  15. )
  16. from services.entities.external_knowledge_entities.external_knowledge_entities import (
  17. Authorization,
  18. ExternalKnowledgeApiSetting,
  19. )
  20. from services.errors.dataset import DatasetNameDuplicateError
  21. class ExternalDatasetService:
  22. @staticmethod
  23. def get_external_knowledge_apis(page, per_page, tenant_id, search=None) -> tuple[list[ExternalKnowledgeApis], int]:
  24. query = ExternalKnowledgeApis.query.filter(ExternalKnowledgeApis.tenant_id == tenant_id).order_by(
  25. ExternalKnowledgeApis.created_at.desc()
  26. )
  27. if search:
  28. query = query.filter(ExternalKnowledgeApis.name.ilike(f"%{search}%"))
  29. external_knowledge_apis = query.paginate(page=page, per_page=per_page, max_per_page=100, error_out=False)
  30. return external_knowledge_apis.items, external_knowledge_apis.total
  31. @classmethod
  32. def validate_api_list(cls, api_settings: dict):
  33. if not api_settings:
  34. raise ValueError("api list is empty")
  35. if "endpoint" not in api_settings and not api_settings["endpoint"]:
  36. raise ValueError("endpoint is required")
  37. if "api_key" not in api_settings and not api_settings["api_key"]:
  38. raise ValueError("api_key is required")
  39. @staticmethod
  40. def create_external_knowledge_api(tenant_id: str, user_id: str, args: dict) -> ExternalKnowledgeApis:
  41. settings = args.get("settings")
  42. if settings is None:
  43. raise ValueError("settings is required")
  44. ExternalDatasetService.check_endpoint_and_api_key(settings)
  45. external_knowledge_api = ExternalKnowledgeApis(
  46. tenant_id=tenant_id,
  47. created_by=user_id,
  48. updated_by=user_id,
  49. name=args.get("name"),
  50. description=args.get("description", ""),
  51. settings=json.dumps(args.get("settings"), ensure_ascii=False),
  52. )
  53. db.session.add(external_knowledge_api)
  54. db.session.commit()
  55. return external_knowledge_api
  56. @staticmethod
  57. def check_endpoint_and_api_key(settings: dict):
  58. if "endpoint" not in settings or not settings["endpoint"]:
  59. raise ValueError("endpoint is required")
  60. if "api_key" not in settings or not settings["api_key"]:
  61. raise ValueError("api_key is required")
  62. endpoint = f"{settings['endpoint']}/retrieval"
  63. api_key = settings["api_key"]
  64. if not validators.url(endpoint, simple_host=True):
  65. if not endpoint.startswith("http://") and not endpoint.startswith("https://"):
  66. raise ValueError(f"invalid endpoint: {endpoint} must start with http:// or https://")
  67. else:
  68. raise ValueError(f"invalid endpoint: {endpoint}")
  69. try:
  70. response = httpx.post(endpoint, headers={"Authorization": f"Bearer {api_key}"})
  71. except Exception as e:
  72. raise ValueError(f"failed to connect to the endpoint: {endpoint}")
  73. if response.status_code == 502:
  74. raise ValueError(f"Bad Gateway: failed to connect to the endpoint: {endpoint}")
  75. if response.status_code == 404:
  76. raise ValueError(f"Not Found: failed to connect to the endpoint: {endpoint}")
  77. if response.status_code == 403:
  78. raise ValueError(f"Forbidden: Authorization failed with api_key: {api_key}")
  79. @staticmethod
  80. def get_external_knowledge_api(external_knowledge_api_id: str) -> ExternalKnowledgeApis:
  81. external_knowledge_api: Optional[ExternalKnowledgeApis] = ExternalKnowledgeApis.query.filter_by(
  82. id=external_knowledge_api_id
  83. ).first()
  84. if external_knowledge_api is None:
  85. raise ValueError("api template not found")
  86. return external_knowledge_api
  87. @staticmethod
  88. def update_external_knowledge_api(tenant_id, user_id, external_knowledge_api_id, args) -> ExternalKnowledgeApis:
  89. external_knowledge_api: Optional[ExternalKnowledgeApis] = ExternalKnowledgeApis.query.filter_by(
  90. id=external_knowledge_api_id, tenant_id=tenant_id
  91. ).first()
  92. if external_knowledge_api is None:
  93. raise ValueError("api template not found")
  94. if args.get("settings") and args.get("settings").get("api_key") == HIDDEN_VALUE:
  95. args.get("settings")["api_key"] = external_knowledge_api.settings_dict.get("api_key")
  96. external_knowledge_api.name = args.get("name")
  97. external_knowledge_api.description = args.get("description", "")
  98. external_knowledge_api.settings = json.dumps(args.get("settings"), ensure_ascii=False)
  99. external_knowledge_api.updated_by = user_id
  100. external_knowledge_api.updated_at = datetime.now(UTC).replace(tzinfo=None)
  101. db.session.commit()
  102. return external_knowledge_api
  103. @staticmethod
  104. def delete_external_knowledge_api(tenant_id: str, external_knowledge_api_id: str):
  105. external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
  106. id=external_knowledge_api_id, tenant_id=tenant_id
  107. ).first()
  108. if external_knowledge_api is None:
  109. raise ValueError("api template not found")
  110. db.session.delete(external_knowledge_api)
  111. db.session.commit()
  112. @staticmethod
  113. def external_knowledge_api_use_check(external_knowledge_api_id: str) -> tuple[bool, int]:
  114. count = ExternalKnowledgeBindings.query.filter_by(external_knowledge_api_id=external_knowledge_api_id).count()
  115. if count > 0:
  116. return True, count
  117. return False, 0
  118. @staticmethod
  119. def get_external_knowledge_binding_with_dataset_id(tenant_id: str, dataset_id: str) -> ExternalKnowledgeBindings:
  120. external_knowledge_binding: Optional[ExternalKnowledgeBindings] = ExternalKnowledgeBindings.query.filter_by(
  121. dataset_id=dataset_id, tenant_id=tenant_id
  122. ).first()
  123. if not external_knowledge_binding:
  124. raise ValueError("external knowledge binding not found")
  125. return external_knowledge_binding
  126. @staticmethod
  127. def document_create_args_validate(tenant_id: str, external_knowledge_api_id: str, process_parameter: dict):
  128. external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
  129. id=external_knowledge_api_id, tenant_id=tenant_id
  130. ).first()
  131. if external_knowledge_api is None:
  132. raise ValueError("api template not found")
  133. settings = json.loads(external_knowledge_api.settings)
  134. for setting in settings:
  135. custom_parameters = setting.get("document_process_setting")
  136. if custom_parameters:
  137. for parameter in custom_parameters:
  138. if parameter.get("required", False) and not process_parameter.get(parameter.get("name")):
  139. raise ValueError(f"{parameter.get('name')} is required")
  140. @staticmethod
  141. def process_external_api(
  142. settings: ExternalKnowledgeApiSetting, files: Union[None, dict[str, Any]]
  143. ) -> httpx.Response:
  144. """
  145. do http request depending on api bundle
  146. """
  147. kwargs = {
  148. "url": settings.url,
  149. "headers": settings.headers,
  150. "follow_redirects": True,
  151. }
  152. response: httpx.Response = getattr(ssrf_proxy, settings.request_method)(
  153. data=json.dumps(settings.params), files=files, **kwargs
  154. )
  155. return response
  156. @staticmethod
  157. def assembling_headers(authorization: Authorization, headers: Optional[dict] = None) -> dict[str, Any]:
  158. authorization = deepcopy(authorization)
  159. if headers:
  160. headers = deepcopy(headers)
  161. else:
  162. headers = {}
  163. if authorization.type == "api-key":
  164. if authorization.config is None:
  165. raise ValueError("authorization config is required")
  166. if authorization.config.api_key is None:
  167. raise ValueError("api_key is required")
  168. if not authorization.config.header:
  169. authorization.config.header = "Authorization"
  170. if authorization.config.type == "bearer":
  171. headers[authorization.config.header] = f"Bearer {authorization.config.api_key}"
  172. elif authorization.config.type == "basic":
  173. headers[authorization.config.header] = f"Basic {authorization.config.api_key}"
  174. elif authorization.config.type == "custom":
  175. headers[authorization.config.header] = authorization.config.api_key
  176. return headers
  177. @staticmethod
  178. def get_external_knowledge_api_settings(settings: dict) -> ExternalKnowledgeApiSetting:
  179. return ExternalKnowledgeApiSetting.parse_obj(settings)
  180. @staticmethod
  181. def create_external_dataset(tenant_id: str, user_id: str, args: dict) -> Dataset:
  182. # check if dataset name already exists
  183. if Dataset.query.filter_by(name=args.get("name"), tenant_id=tenant_id).first():
  184. raise DatasetNameDuplicateError(f"Dataset with name {args.get('name')} already exists.")
  185. external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
  186. id=args.get("external_knowledge_api_id"), tenant_id=tenant_id
  187. ).first()
  188. if external_knowledge_api is None:
  189. raise ValueError("api template not found")
  190. dataset = Dataset(
  191. tenant_id=tenant_id,
  192. name=args.get("name"),
  193. description=args.get("description", ""),
  194. provider="external",
  195. retrieval_model=args.get("external_retrieval_model"),
  196. created_by=user_id,
  197. )
  198. db.session.add(dataset)
  199. db.session.flush()
  200. external_knowledge_binding = ExternalKnowledgeBindings(
  201. tenant_id=tenant_id,
  202. dataset_id=dataset.id,
  203. external_knowledge_api_id=args.get("external_knowledge_api_id"),
  204. external_knowledge_id=args.get("external_knowledge_id"),
  205. created_by=user_id,
  206. )
  207. db.session.add(external_knowledge_binding)
  208. db.session.commit()
  209. return dataset
  210. @staticmethod
  211. def fetch_external_knowledge_retrieval(
  212. tenant_id: str,
  213. dataset_id: str,
  214. query: str,
  215. external_retrieval_parameters: dict,
  216. metadata_condition: Optional[MetadataCondition] = None,
  217. ) -> list:
  218. external_knowledge_binding = ExternalKnowledgeBindings.query.filter_by(
  219. dataset_id=dataset_id, tenant_id=tenant_id
  220. ).first()
  221. if not external_knowledge_binding:
  222. raise ValueError("external knowledge binding not found")
  223. external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
  224. id=external_knowledge_binding.external_knowledge_api_id
  225. ).first()
  226. if not external_knowledge_api:
  227. raise ValueError("external api template not found")
  228. settings = json.loads(external_knowledge_api.settings)
  229. headers = {"Content-Type": "application/json"}
  230. if settings.get("api_key"):
  231. headers["Authorization"] = f"Bearer {settings.get('api_key')}"
  232. score_threshold_enabled = external_retrieval_parameters.get("score_threshold_enabled") or False
  233. score_threshold = external_retrieval_parameters.get("score_threshold", 0.0) if score_threshold_enabled else 0.0
  234. request_params = {
  235. "retrieval_setting": {
  236. "top_k": external_retrieval_parameters.get("top_k"),
  237. "score_threshold": score_threshold,
  238. },
  239. "query": query,
  240. "knowledge_id": external_knowledge_binding.external_knowledge_id,
  241. "metadata_condition": metadata_condition.model_dump() if metadata_condition else None,
  242. }
  243. response = ExternalDatasetService.process_external_api(
  244. ExternalKnowledgeApiSetting(
  245. url=f"{settings.get('endpoint')}/retrieval",
  246. request_method="post",
  247. headers=headers,
  248. params=request_params,
  249. ),
  250. None,
  251. )
  252. if response.status_code == 200:
  253. return cast(list[Any], response.json().get("records", []))
  254. return []