Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -64,26 +64,27 @@ def langchain_document_loader():
|
|
64 |
Files can be in txt, pdf, CSV or docx format.
|
65 |
"""
|
66 |
current_dir = os.getcwd()
|
67 |
-
TMP_DIR = current_dir
|
68 |
|
69 |
documents = []
|
70 |
|
|
|
71 |
txt_loader = DirectoryLoader(
|
72 |
TMP_DIR.as_posix(), glob="**/*.txt", loader_cls=TextLoader, show_progress=True
|
73 |
)
|
74 |
documents.extend(txt_loader.load())
|
75 |
-
|
76 |
pdf_loader = DirectoryLoader(
|
77 |
TMP_DIR.as_posix(), glob="*.pdf", loader_cls=PyPDFLoader, show_progress=True
|
78 |
)
|
79 |
documents.extend(pdf_loader.load())
|
80 |
-
|
81 |
csv_loader = DirectoryLoader(
|
82 |
TMP_DIR.as_posix(), glob="**/*.csv", loader_cls=CSVLoader, show_progress=True,
|
83 |
loader_kwargs={"encoding":"utf8"}
|
84 |
)
|
85 |
documents.extend(csv_loader.load())
|
86 |
-
|
87 |
doc_loader = DirectoryLoader(
|
88 |
TMP_DIR.as_posix(),
|
89 |
glob="**/*.docx",
|
|
|
64 |
Files can be in txt, pdf, CSV or docx format.
|
65 |
"""
|
66 |
current_dir = os.getcwd()
|
67 |
+
TMP_DIR = PureWindowsPath(current_dir)
|
68 |
|
69 |
documents = []
|
70 |
|
71 |
+
"""
|
72 |
txt_loader = DirectoryLoader(
|
73 |
TMP_DIR.as_posix(), glob="**/*.txt", loader_cls=TextLoader, show_progress=True
|
74 |
)
|
75 |
documents.extend(txt_loader.load())
|
76 |
+
"""
|
77 |
pdf_loader = DirectoryLoader(
|
78 |
TMP_DIR.as_posix(), glob="*.pdf", loader_cls=PyPDFLoader, show_progress=True
|
79 |
)
|
80 |
documents.extend(pdf_loader.load())
|
81 |
+
"""
|
82 |
csv_loader = DirectoryLoader(
|
83 |
TMP_DIR.as_posix(), glob="**/*.csv", loader_cls=CSVLoader, show_progress=True,
|
84 |
loader_kwargs={"encoding":"utf8"}
|
85 |
)
|
86 |
documents.extend(csv_loader.load())
|
87 |
+
"""
|
88 |
doc_loader = DirectoryLoader(
|
89 |
TMP_DIR.as_posix(),
|
90 |
glob="**/*.docx",
|