Spaces:
Sleeping
Sleeping
Tuchuanhuhuhu
commited on
Commit
·
08b7713
1
Parent(s):
12eb16f
bugfix: 如果文件解析失败,不影响后续文件
Browse files- modules/index_func.py +4 -2
modules/index_func.py
CHANGED
@@ -23,6 +23,7 @@ def get_documents(file_src):
|
|
23 |
filename = os.path.basename(filepath)
|
24 |
file_type = os.path.splitext(filename)[1]
|
25 |
logging.info(f"loading file: {filename}")
|
|
|
26 |
try:
|
27 |
if file_type == ".pdf":
|
28 |
logging.debug("Loading PDF...")
|
@@ -72,8 +73,9 @@ def get_documents(file_src):
|
|
72 |
logging.error(f"Error loading file: {filename}")
|
73 |
traceback.print_exc()
|
74 |
|
75 |
-
texts
|
76 |
-
|
|
|
77 |
logging.debug("Documents loaded.")
|
78 |
return documents
|
79 |
|
|
|
23 |
filename = os.path.basename(filepath)
|
24 |
file_type = os.path.splitext(filename)[1]
|
25 |
logging.info(f"loading file: {filename}")
|
26 |
+
texts = None
|
27 |
try:
|
28 |
if file_type == ".pdf":
|
29 |
logging.debug("Loading PDF...")
|
|
|
73 |
logging.error(f"Error loading file: {filename}")
|
74 |
traceback.print_exc()
|
75 |
|
76 |
+
if texts is not None:
|
77 |
+
texts = text_splitter.split_documents(texts)
|
78 |
+
documents.extend(texts)
|
79 |
logging.debug("Documents loaded.")
|
80 |
return documents
|
81 |
|