Spaces:
Paused
Paused
Update helper_functions.py
Browse files- helper_functions.py +10 -3
helper_functions.py
CHANGED
|
@@ -1,14 +1,21 @@
|
|
| 1 |
from langchain_community.document_loaders import PyMuPDFLoader, TextLoader
|
| 2 |
from langchain_community.vectorstores import Qdrant
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
|
| 4 |
-
def process_file(file):
|
| 5 |
documents = []
|
| 6 |
if file.path.endswith(".pdf"):
|
| 7 |
-
loader = PyMuPDFLoader(
|
| 8 |
docs = loader.load()
|
| 9 |
documents.extend(docs)
|
| 10 |
else:
|
| 11 |
-
loader = TextLoader(
|
| 12 |
docs = loader.load()
|
| 13 |
documents.extend(docs)
|
| 14 |
return documents
|
|
|
|
| 1 |
from langchain_community.document_loaders import PyMuPDFLoader, TextLoader
|
| 2 |
from langchain_community.vectorstores import Qdrant
|
| 3 |
+
import os
|
| 4 |
+
|
| 5 |
+
def process_file(uploaded_file):
|
| 6 |
+
# save the file temporarily
|
| 7 |
+
temp_file = "./temp.pdf"
|
| 8 |
+
with open(temp_file, "wb") as file:
|
| 9 |
+
file.write(uploaded_file.getvalue())
|
| 10 |
+
file_name = uploaded_file.name
|
| 11 |
|
|
|
|
| 12 |
documents = []
|
| 13 |
if file.path.endswith(".pdf"):
|
| 14 |
+
loader = PyMuPDFLoader(temp_file)
|
| 15 |
docs = loader.load()
|
| 16 |
documents.extend(docs)
|
| 17 |
else:
|
| 18 |
+
loader = TextLoader(tmp_location)
|
| 19 |
docs = loader.load()
|
| 20 |
documents.extend(docs)
|
| 21 |
return documents
|