Update app.py
Browse files
app.py
CHANGED
@@ -7,9 +7,7 @@ from getpass import getpass
|
|
7 |
|
8 |
openai_api_key = os.getenv('OPENAI_API_KEY')
|
9 |
|
10 |
-
|
11 |
-
# Configure LlamaIndex with OpenAI LLM and Embeddings
|
12 |
-
# -------------------------------------------------------
|
13 |
from llama_index.llms.openai import OpenAI
|
14 |
from llama_index.embeddings.openai import OpenAIEmbedding
|
15 |
from llama_index.core import Settings
|
@@ -17,14 +15,12 @@ from llama_index.core import Settings
|
|
17 |
Settings.llm = OpenAI(model="gpt-3.5-turbo", temperature=0.4)
|
18 |
Settings.embed_model = OpenAIEmbedding(model="text-embedding-ada-002")
|
19 |
|
20 |
-
|
21 |
-
# Import document readers, index, vector store, memory, etc.
|
22 |
-
# -------------------------------------------------------
|
23 |
from llama_index.core import SimpleDirectoryReader, VectorStoreIndex, StorageContext
|
24 |
from llama_index.vector_stores.qdrant import QdrantVectorStore
|
25 |
from llama_index.core.memory import ChatMemoryBuffer
|
26 |
|
27 |
-
|
28 |
chat_engine = None
|
29 |
index = None
|
30 |
query_engine = None
|
@@ -33,32 +29,22 @@ client = None
|
|
33 |
vector_store = None
|
34 |
storage_context = None
|
35 |
|
36 |
-
|
37 |
-
# Function to process uploaded files and build the index.
|
38 |
-
# -------------------------------------------------------
|
39 |
def process_upload(files):
|
40 |
-
"""
|
41 |
-
Accepts a list of uploaded file paths, saves them to a local folder,
|
42 |
-
loads them as documents, and builds the vector index and chat engine.
|
43 |
-
"""
|
44 |
upload_dir = "uploaded_files"
|
45 |
if not os.path.exists(upload_dir):
|
46 |
os.makedirs(upload_dir)
|
47 |
else:
|
48 |
-
# Clear any existing files in the folder.
|
49 |
for f in os.listdir(upload_dir):
|
50 |
os.remove(os.path.join(upload_dir, f))
|
51 |
|
52 |
-
# 'files' is a list of file paths (Gradio's File component with type="file")
|
53 |
for file_path in files:
|
54 |
file_name = os.path.basename(file_path)
|
55 |
dest = os.path.join(upload_dir, file_name)
|
56 |
shutil.copy(file_path, dest)
|
57 |
|
58 |
-
# Load documents from the saved folder.
|
59 |
documents = SimpleDirectoryReader(upload_dir).load_data()
|
60 |
|
61 |
-
# Build the index and chat engine using Qdrant as the vector store.
|
62 |
global client, vector_store, storage_context, index, query_engine, memory, chat_engine
|
63 |
client = qdrant_client.QdrantClient(location=":memory:")
|
64 |
|
@@ -88,12 +74,9 @@ def process_upload(files):
|
|
88 |
|
89 |
return "Documents uploaded and index built successfully!"
|
90 |
|
91 |
-
|
92 |
-
# Chat function that uses the built chat engine.
|
93 |
-
# -------------------------------------------------------
|
94 |
def chat_with_ai(user_input, chat_history):
|
95 |
global chat_engine
|
96 |
-
# Check if the chat engine is initialized.
|
97 |
if chat_engine is None:
|
98 |
return chat_history, "Please upload documents first."
|
99 |
|
@@ -101,7 +84,6 @@ def chat_with_ai(user_input, chat_history):
|
|
101 |
references = response.source_nodes
|
102 |
ref, pages = [], []
|
103 |
|
104 |
-
# Extract file names from the source nodes (if available)
|
105 |
for node in references:
|
106 |
file_name = node.metadata.get('file_name')
|
107 |
if file_name and file_name not in ref:
|
@@ -114,23 +96,17 @@ def chat_with_ai(user_input, chat_history):
|
|
114 |
chat_history.append((user_input, str(response)))
|
115 |
return chat_history, ""
|
116 |
|
117 |
-
|
118 |
-
# Function to clear the chat history.
|
119 |
-
# -------------------------------------------------------
|
120 |
def clear_history():
|
121 |
return [], ""
|
122 |
|
123 |
-
|
124 |
-
# Build the Gradio interface.
|
125 |
-
# -------------------------------------------------------
|
126 |
def gradio_interface():
|
127 |
with gr.Blocks() as demo:
|
128 |
gr.Markdown("# Chat Interface for LlamaIndex with File Upload")
|
129 |
|
130 |
-
# Use Tabs to separate the file upload and chat interfaces.
|
131 |
with gr.Tab("Upload Documents"):
|
132 |
gr.Markdown("Upload PDF, Excel, CSV, DOC/DOCX, or TXT files below:")
|
133 |
-
# The file upload widget: we specify allowed file types.
|
134 |
file_upload = gr.File(
|
135 |
label="Upload Files",
|
136 |
file_count="multiple",
|
@@ -158,5 +134,5 @@ def gradio_interface():
|
|
158 |
|
159 |
return demo
|
160 |
|
161 |
-
|
162 |
gradio_interface().launch(debug=True)
|
|
|
7 |
|
8 |
openai_api_key = os.getenv('OPENAI_API_KEY')
|
9 |
|
10 |
+
|
|
|
|
|
11 |
from llama_index.llms.openai import OpenAI
|
12 |
from llama_index.embeddings.openai import OpenAIEmbedding
|
13 |
from llama_index.core import Settings
|
|
|
15 |
Settings.llm = OpenAI(model="gpt-3.5-turbo", temperature=0.4)
|
16 |
Settings.embed_model = OpenAIEmbedding(model="text-embedding-ada-002")
|
17 |
|
18 |
+
|
|
|
|
|
19 |
from llama_index.core import SimpleDirectoryReader, VectorStoreIndex, StorageContext
|
20 |
from llama_index.vector_stores.qdrant import QdrantVectorStore
|
21 |
from llama_index.core.memory import ChatMemoryBuffer
|
22 |
|
23 |
+
|
24 |
chat_engine = None
|
25 |
index = None
|
26 |
query_engine = None
|
|
|
29 |
vector_store = None
|
30 |
storage_context = None
|
31 |
|
32 |
+
|
|
|
|
|
33 |
def process_upload(files):
|
|
|
|
|
|
|
|
|
34 |
upload_dir = "uploaded_files"
|
35 |
if not os.path.exists(upload_dir):
|
36 |
os.makedirs(upload_dir)
|
37 |
else:
|
|
|
38 |
for f in os.listdir(upload_dir):
|
39 |
os.remove(os.path.join(upload_dir, f))
|
40 |
|
|
|
41 |
for file_path in files:
|
42 |
file_name = os.path.basename(file_path)
|
43 |
dest = os.path.join(upload_dir, file_name)
|
44 |
shutil.copy(file_path, dest)
|
45 |
|
|
|
46 |
documents = SimpleDirectoryReader(upload_dir).load_data()
|
47 |
|
|
|
48 |
global client, vector_store, storage_context, index, query_engine, memory, chat_engine
|
49 |
client = qdrant_client.QdrantClient(location=":memory:")
|
50 |
|
|
|
74 |
|
75 |
return "Documents uploaded and index built successfully!"
|
76 |
|
77 |
+
|
|
|
|
|
78 |
def chat_with_ai(user_input, chat_history):
|
79 |
global chat_engine
|
|
|
80 |
if chat_engine is None:
|
81 |
return chat_history, "Please upload documents first."
|
82 |
|
|
|
84 |
references = response.source_nodes
|
85 |
ref, pages = [], []
|
86 |
|
|
|
87 |
for node in references:
|
88 |
file_name = node.metadata.get('file_name')
|
89 |
if file_name and file_name not in ref:
|
|
|
96 |
chat_history.append((user_input, str(response)))
|
97 |
return chat_history, ""
|
98 |
|
99 |
+
|
|
|
|
|
100 |
def clear_history():
|
101 |
return [], ""
|
102 |
|
103 |
+
|
|
|
|
|
104 |
def gradio_interface():
|
105 |
with gr.Blocks() as demo:
|
106 |
gr.Markdown("# Chat Interface for LlamaIndex with File Upload")
|
107 |
|
|
|
108 |
with gr.Tab("Upload Documents"):
|
109 |
gr.Markdown("Upload PDF, Excel, CSV, DOC/DOCX, or TXT files below:")
|
|
|
110 |
file_upload = gr.File(
|
111 |
label="Upload Files",
|
112 |
file_count="multiple",
|
|
|
134 |
|
135 |
return demo
|
136 |
|
137 |
+
|
138 |
gradio_interface().launch(debug=True)
|