external_knowledge_service.py 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289
  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 extensions.ext_database import db
  10. from models.dataset import (
  11. Dataset,
  12. ExternalKnowledgeApis,
  13. ExternalKnowledgeBindings,
  14. )
  15. from services.entities.external_knowledge_entities.external_knowledge_entities import (
  16. Authorization,
  17. ExternalKnowledgeApiSetting,
  18. )
  19. from services.errors.dataset import DatasetNameDuplicateError
  20. class ExternalDatasetService:
  21. @staticmethod
  22. def get_external_knowledge_apis(page, per_page, tenant_id, search=None) -> tuple[list[ExternalKnowledgeApis], int]:
  23. query = ExternalKnowledgeApis.query.filter(ExternalKnowledgeApis.tenant_id == tenant_id).order_by(
  24. ExternalKnowledgeApis.created_at.desc()
  25. )
  26. if search:
  27. query = query.filter(ExternalKnowledgeApis.name.ilike(f"%{search}%"))
  28. external_knowledge_apis = query.paginate(page=page, per_page=per_page, max_per_page=100, error_out=False)
  29. return external_knowledge_apis.items, external_knowledge_apis.total
  30. @classmethod
  31. def validate_api_list(cls, api_settings: dict):
  32. if not api_settings:
  33. raise ValueError("api list is empty")
  34. if "endpoint" not in api_settings and not api_settings["endpoint"]:
  35. raise ValueError("endpoint is required")
  36. if "api_key" not in api_settings and not api_settings["api_key"]:
  37. raise ValueError("api_key is required")
  38. @staticmethod
  39. def create_external_knowledge_api(tenant_id: str, user_id: str, args: dict) -> ExternalKnowledgeApis:
  40. settings = args.get("settings")
  41. if settings is None:
  42. raise ValueError("settings is required")
  43. ExternalDatasetService.check_endpoint_and_api_key(settings)
  44. external_knowledge_api = ExternalKnowledgeApis(
  45. tenant_id=tenant_id,
  46. created_by=user_id,
  47. updated_by=user_id,
  48. name=args.get("name"),
  49. description=args.get("description", ""),
  50. settings=json.dumps(args.get("settings"), ensure_ascii=False),
  51. )
  52. db.session.add(external_knowledge_api)
  53. db.session.commit()
  54. return external_knowledge_api
  55. @staticmethod
  56. def check_endpoint_and_api_key(settings: dict):
  57. if "endpoint" not in settings or not settings["endpoint"]:
  58. raise ValueError("endpoint is required")
  59. if "api_key" not in settings or not settings["api_key"]:
  60. raise ValueError("api_key is required")
  61. endpoint = f"{settings['endpoint']}/retrieval"
  62. api_key = settings["api_key"]
  63. if not validators.url(endpoint, simple_host=True):
  64. if not endpoint.startswith("http://") and not endpoint.startswith("https://"):
  65. raise ValueError(f"invalid endpoint: {endpoint} must start with http:// or https://")
  66. else:
  67. raise ValueError(f"invalid endpoint: {endpoint}")
  68. try:
  69. response = httpx.post(endpoint, headers={"Authorization": f"Bearer {api_key}"})
  70. except Exception as e:
  71. raise ValueError(f"failed to connect to the endpoint: {endpoint}")
  72. if response.status_code == 502:
  73. raise ValueError(f"Bad Gateway: failed to connect to the endpoint: {endpoint}")
  74. if response.status_code == 404:
  75. raise ValueError(f"Not Found: failed to connect to the endpoint: {endpoint}")
  76. if response.status_code == 403:
  77. raise ValueError(f"Forbidden: Authorization failed with api_key: {api_key}")
  78. @staticmethod
  79. def get_external_knowledge_api(external_knowledge_api_id: str) -> ExternalKnowledgeApis:
  80. external_knowledge_api: Optional[ExternalKnowledgeApis] = ExternalKnowledgeApis.query.filter_by(
  81. id=external_knowledge_api_id
  82. ).first()
  83. if external_knowledge_api is None:
  84. raise ValueError("api template not found")
  85. return external_knowledge_api
  86. @staticmethod
  87. def update_external_knowledge_api(tenant_id, user_id, external_knowledge_api_id, args) -> ExternalKnowledgeApis:
  88. external_knowledge_api: Optional[ExternalKnowledgeApis] = ExternalKnowledgeApis.query.filter_by(
  89. id=external_knowledge_api_id, tenant_id=tenant_id
  90. ).first()
  91. if external_knowledge_api is None:
  92. raise ValueError("api template not found")
  93. if args.get("settings") and args.get("settings").get("api_key") == HIDDEN_VALUE:
  94. args.get("settings")["api_key"] = external_knowledge_api.settings_dict.get("api_key")
  95. external_knowledge_api.name = args.get("name")
  96. external_knowledge_api.description = args.get("description", "")
  97. external_knowledge_api.settings = json.dumps(args.get("settings"), ensure_ascii=False)
  98. external_knowledge_api.updated_by = user_id
  99. external_knowledge_api.updated_at = datetime.now(UTC).replace(tzinfo=None)
  100. db.session.commit()
  101. return external_knowledge_api
  102. @staticmethod
  103. def delete_external_knowledge_api(tenant_id: str, external_knowledge_api_id: str):
  104. external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
  105. id=external_knowledge_api_id, tenant_id=tenant_id
  106. ).first()
  107. if external_knowledge_api is None:
  108. raise ValueError("api template not found")
  109. db.session.delete(external_knowledge_api)
  110. db.session.commit()
  111. @staticmethod
  112. def external_knowledge_api_use_check(external_knowledge_api_id: str) -> tuple[bool, int]:
  113. count = ExternalKnowledgeBindings.query.filter_by(external_knowledge_api_id=external_knowledge_api_id).count()
  114. if count > 0:
  115. return True, count
  116. return False, 0
  117. @staticmethod
  118. def get_external_knowledge_binding_with_dataset_id(tenant_id: str, dataset_id: str) -> ExternalKnowledgeBindings:
  119. external_knowledge_binding: Optional[ExternalKnowledgeBindings] = ExternalKnowledgeBindings.query.filter_by(
  120. dataset_id=dataset_id, tenant_id=tenant_id
  121. ).first()
  122. if not external_knowledge_binding:
  123. raise ValueError("external knowledge binding not found")
  124. return external_knowledge_binding
  125. @staticmethod
  126. def document_create_args_validate(tenant_id: str, external_knowledge_api_id: str, process_parameter: dict):
  127. external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
  128. id=external_knowledge_api_id, tenant_id=tenant_id
  129. ).first()
  130. if external_knowledge_api is None:
  131. raise ValueError("api template not found")
  132. settings = json.loads(external_knowledge_api.settings)
  133. for setting in settings:
  134. custom_parameters = setting.get("document_process_setting")
  135. if custom_parameters:
  136. for parameter in custom_parameters:
  137. if parameter.get("required", False) and not process_parameter.get(parameter.get("name")):
  138. raise ValueError(f'{parameter.get("name")} is required')
  139. @staticmethod
  140. def process_external_api(
  141. settings: ExternalKnowledgeApiSetting, files: Union[None, dict[str, Any]]
  142. ) -> httpx.Response:
  143. """
  144. do http request depending on api bundle
  145. """
  146. kwargs = {
  147. "url": settings.url,
  148. "headers": settings.headers,
  149. "follow_redirects": True,
  150. }
  151. response: httpx.Response = getattr(ssrf_proxy, settings.request_method)(
  152. data=json.dumps(settings.params), files=files, **kwargs
  153. )
  154. return response
  155. @staticmethod
  156. def assembling_headers(authorization: Authorization, headers: Optional[dict] = None) -> dict[str, Any]:
  157. authorization = deepcopy(authorization)
  158. if headers:
  159. headers = deepcopy(headers)
  160. else:
  161. headers = {}
  162. if authorization.type == "api-key":
  163. if authorization.config is None:
  164. raise ValueError("authorization config is required")
  165. if authorization.config.api_key is None:
  166. raise ValueError("api_key is required")
  167. if not authorization.config.header:
  168. authorization.config.header = "Authorization"
  169. if authorization.config.type == "bearer":
  170. headers[authorization.config.header] = f"Bearer {authorization.config.api_key}"
  171. elif authorization.config.type == "basic":
  172. headers[authorization.config.header] = f"Basic {authorization.config.api_key}"
  173. elif authorization.config.type == "custom":
  174. headers[authorization.config.header] = authorization.config.api_key
  175. return headers
  176. @staticmethod
  177. def get_external_knowledge_api_settings(settings: dict) -> ExternalKnowledgeApiSetting:
  178. return ExternalKnowledgeApiSetting.parse_obj(settings)
  179. @staticmethod
  180. def create_external_dataset(tenant_id: str, user_id: str, args: dict) -> Dataset:
  181. # check if dataset name already exists
  182. if Dataset.query.filter_by(name=args.get("name"), tenant_id=tenant_id).first():
  183. raise DatasetNameDuplicateError(f"Dataset with name {args.get('name')} already exists.")
  184. external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
  185. id=args.get("external_knowledge_api_id"), tenant_id=tenant_id
  186. ).first()
  187. if external_knowledge_api is None:
  188. raise ValueError("api template not found")
  189. dataset = Dataset(
  190. tenant_id=tenant_id,
  191. name=args.get("name"),
  192. description=args.get("description", ""),
  193. provider="external",
  194. retrieval_model=args.get("external_retrieval_model"),
  195. created_by=user_id,
  196. )
  197. db.session.add(dataset)
  198. db.session.flush()
  199. external_knowledge_binding = ExternalKnowledgeBindings(
  200. tenant_id=tenant_id,
  201. dataset_id=dataset.id,
  202. external_knowledge_api_id=args.get("external_knowledge_api_id"),
  203. external_knowledge_id=args.get("external_knowledge_id"),
  204. created_by=user_id,
  205. )
  206. db.session.add(external_knowledge_binding)
  207. db.session.commit()
  208. return dataset
  209. @staticmethod
  210. def fetch_external_knowledge_retrieval(
  211. tenant_id: str, dataset_id: str, query: str, external_retrieval_parameters: dict
  212. ) -> list:
  213. external_knowledge_binding = ExternalKnowledgeBindings.query.filter_by(
  214. dataset_id=dataset_id, tenant_id=tenant_id
  215. ).first()
  216. if not external_knowledge_binding:
  217. raise ValueError("external knowledge binding not found")
  218. external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
  219. id=external_knowledge_binding.external_knowledge_api_id
  220. ).first()
  221. if not external_knowledge_api:
  222. raise ValueError("external api template not found")
  223. settings = json.loads(external_knowledge_api.settings)
  224. headers = {"Content-Type": "application/json"}
  225. if settings.get("api_key"):
  226. headers["Authorization"] = f"Bearer {settings.get('api_key')}"
  227. score_threshold_enabled = external_retrieval_parameters.get("score_threshold_enabled") or False
  228. score_threshold = external_retrieval_parameters.get("score_threshold", 0.0) if score_threshold_enabled else 0.0
  229. request_params = {
  230. "retrieval_setting": {
  231. "top_k": external_retrieval_parameters.get("top_k"),
  232. "score_threshold": score_threshold,
  233. },
  234. "query": query,
  235. "knowledge_id": external_knowledge_binding.external_knowledge_id,
  236. }
  237. response = ExternalDatasetService.process_external_api(
  238. ExternalKnowledgeApiSetting(
  239. url=f"{settings.get('endpoint')}/retrieval",
  240. request_method="post",
  241. headers=headers,
  242. params=request_params,
  243. ),
  244. None,
  245. )
  246. if response.status_code == 200:
  247. return cast(list[Any], response.json().get("records", []))
  248. return []