Spaces:
Sleeping
Sleeping
Update app.py
Browse filesmoved metadata and faiss indexing to flask backend
app.py
CHANGED
@@ -32,45 +32,19 @@ embedding_dim = 768 # Adjust according to model
|
|
32 |
def store_document_data(PDF_FILE, METADATA_FILE, INDEX_FILE):
|
33 |
print(" Storing document...")
|
34 |
|
35 |
-
|
36 |
# Extract text from the PDF
|
37 |
text = extract_text_from_pdf(PDF_FILE)
|
38 |
if not text:
|
39 |
return "Could not extract any text from the PDF."
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
print("
|
44 |
-
|
45 |
-
|
46 |
else:
|
47 |
-
|
48 |
-
metadata = {}
|
49 |
-
|
50 |
-
if INDEX_FILE:
|
51 |
-
# extract Faiss
|
52 |
-
print("index_file recieved")
|
53 |
-
index = faiss.read_index(INDEX_FILE)
|
54 |
-
else:
|
55 |
-
print(" No FAISS index found. Creating a new one.")
|
56 |
-
index = faiss.IndexFlatL2(embedding_dim) # Empty FAISS index
|
57 |
-
|
58 |
-
# Generate and store embedding
|
59 |
-
embedding = embedding_model.encode([text]).astype(np.float32)
|
60 |
-
index.add(embedding) # Add to FAISS index
|
61 |
-
print(" Embeddings generated")
|
62 |
-
|
63 |
-
# Get FAISS index for the new document
|
64 |
-
doc_index = index.ntotal - 1
|
65 |
-
|
66 |
-
# Update metadata with FAISS index
|
67 |
-
metadata[str(doc_index)] = PDF_FILE
|
68 |
-
print(" Saved Metadata")
|
69 |
-
|
70 |
-
return json.dumps({
|
71 |
-
"metadata": metadata,
|
72 |
-
"index": index
|
73 |
-
})
|
74 |
|
75 |
def retrieve_document(query):
|
76 |
print(f"Retrieving document based on:\n{query}")
|
|
|
32 |
def store_document_data(PDF_FILE, METADATA_FILE, INDEX_FILE):
|
33 |
print(" Storing document...")
|
34 |
|
35 |
+
if PDF_FILE:
|
36 |
# Extract text from the PDF
|
37 |
text = extract_text_from_pdf(PDF_FILE)
|
38 |
if not text:
|
39 |
return "Could not extract any text from the PDF."
|
40 |
+
|
41 |
+
# Generate and return embedding
|
42 |
+
embedding = embedding_model.encode([text]).astype(np.float32)
|
43 |
+
print("Embeddings generated")
|
44 |
+
|
45 |
+
return embedding
|
46 |
else:
|
47 |
+
return "No PDF file provided."
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
48 |
|
49 |
def retrieve_document(query):
|
50 |
print(f"Retrieving document based on:\n{query}")
|