create dataset (#2074)

### What problem does this PR solve?

You can use sdk to create a dataset

### Type of change

- [x] New Feature

---------

Co-authored-by: root <root@xwg>
This commit is contained in:
LiuHua
2024-08-23 18:38:20 +08:00
committed by GitHub
parent 6228b1bd53
commit e14e0ec695
7 changed files with 125 additions and 26 deletions

View File

View File

@@ -0,0 +1,30 @@
class Base(object):
def __init__(self, rag, res_dict):
self.rag = rag
for k, v in res_dict.items():
if isinstance(v, dict):
self.__dict__[k] = Base(rag, v)
else:
self.__dict__[k] = v
def to_json(self):
pr = {}
for name in dir(self):
value = getattr(self, name)
if not name.startswith('__') and not callable(value) and name != "rag":
if isinstance(value, Base):
pr[name] = value.to_json()
else:
pr[name] = value
return pr
def post(self, path, param):
res = self.rag.post(path,param)
return res
def get(self, path, params=''):
res = self.rag.get(path,params)
return res

View File

@@ -0,0 +1,33 @@
from .base import Base
class DataSet(Base):
class ParseConfig(Base):
def __init__(self, rag, res_dict):
self.chunk_token_count = 128
self.layout_recognize = True
self.delimiter = '\n!?。;!?'
self.task_page_size = 12
super().__init__(rag, res_dict)
def __init__(self, rag, res_dict):
self.id = ""
self.name = ""
self.avatar = ""
self.tenant_id = None
self.description = ""
self.language = "English"
self.embedding_model = ""
self.permission = "me"
self.document_count = 0
self.chunk_count = 0
self.parse_method = 0
self.parser_config = None
super().__init__(rag, res_dict)
def delete(self):
try:
self.post("/rm", {"kb_id": self.id})
return True
except Exception:
return False