Spaces:
Paused
Paused
Create helper_functions.py
Browse files- helper_functions.py +25 -0
helper_functions.py
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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.endswith(".pdf"):
|
| 7 |
+
loader = PyMuPDFLoader(file)
|
| 8 |
+
docs = loader.load()
|
| 9 |
+
documents.extend(docs)
|
| 10 |
+
else:
|
| 11 |
+
loader = TextLoader(file)
|
| 12 |
+
docs = loader.load()
|
| 13 |
+
documents.extend(docs)
|
| 14 |
+
return documents
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
def add_to_qdrant(documents, embeddings, qdrant_client, collection_name):
|
| 18 |
+
|
| 19 |
+
Qdrant.from_documents(
|
| 20 |
+
documents,
|
| 21 |
+
embeddings,
|
| 22 |
+
url=qdrant_client.url,
|
| 23 |
+
prefer_grpc=True,
|
| 24 |
+
collection_name=collection_name,
|
| 25 |
+
)
|