实现多工程数据存储支持

This commit is contained in:
wanyaokun
2024-08-13 13:10:52 +08:00
parent 3a81a83033
commit 7e58a1a223
9 changed files with 97 additions and 51 deletions
+2 -5
View File
@@ -20,7 +20,6 @@ class FileLoaderConfig(BaseModel):
raise ValueError(f"Directory '{v}' does not exist")
return v
def llama_parse_parser():
if os.getenv("LLAMA_CLOUD_API_KEY") is None:
raise ValueError(
@@ -35,7 +34,6 @@ def llama_parse_parser():
)
return parser
def llama_parse_extractor() -> Dict[str, LlamaParse]:
from llama_parse.utils import SUPPORTED_FILE_TYPES
@@ -45,8 +43,7 @@ def llama_parse_extractor() -> Dict[str, LlamaParse]:
def llama_local_extractor() -> Dict[str, BaseReader]:
return {"json" : JSONReader}
def get_file_documents(config: FileLoaderConfig):
def get_file_documents(config: FileLoaderConfig, childPath: str):
from llama_index.core.readers import SimpleDirectoryReader
try:
@@ -63,7 +60,7 @@ def get_file_documents(config: FileLoaderConfig):
file_extractor = llama_local_extractor()
reader = SimpleDirectoryReader(
config.data_dir,
os.path.join(config.data_dir,childPath.replace('_','\\')),
recursive=True,
filename_as_id=True,
raise_on_error=True,