Update src/streamlit_app.py
Browse files- src/streamlit_app.py +42 -20
src/streamlit_app.py
CHANGED
|
@@ -2,47 +2,69 @@ import streamlit as st
|
|
| 2 |
import os
|
| 3 |
from rag_engine import ProjectRAGEngine
|
| 4 |
|
| 5 |
-
|
|
|
|
| 6 |
st.title("📂 Industrial Project Report Analyzer")
|
| 7 |
|
| 8 |
-
#
|
|
|
|
| 9 |
api_key = os.environ.get("OPENAI_API_KEY")
|
| 10 |
if not api_key:
|
| 11 |
api_key = st.sidebar.text_input("Enter OpenAI API Key", type="password")
|
| 12 |
|
| 13 |
if not api_key:
|
| 14 |
-
st.warning("Please provide an OpenAI API Key to
|
| 15 |
st.stop()
|
| 16 |
|
|
|
|
| 17 |
if "engine" not in st.session_state:
|
| 18 |
st.session_state.engine = ProjectRAGEngine(api_key)
|
| 19 |
|
| 20 |
-
# Document
|
| 21 |
-
uploaded_files = st.sidebar.file_uploader(
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
|
| 23 |
if uploaded_files:
|
|
|
|
| 24 |
if "processed_files" not in st.session_state or len(st.session_state.processed_files) != len(uploaded_files):
|
| 25 |
-
with st.spinner("
|
| 26 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
paths = []
|
| 28 |
for f in uploaded_files:
|
| 29 |
-
|
| 30 |
-
with open(
|
| 31 |
-
|
|
|
|
|
|
|
|
|
|
| 32 |
st.session_state.engine.process_documents(paths)
|
| 33 |
st.session_state.processed_files = [f.name for f in uploaded_files]
|
| 34 |
-
st.success("
|
| 35 |
|
| 36 |
-
#
|
| 37 |
query = st.chat_input("Ex: 'Compare the budgets of these projects'")
|
|
|
|
| 38 |
if query:
|
| 39 |
-
with st.chat_message("user"):
|
|
|
|
|
|
|
| 40 |
with st.chat_message("assistant"):
|
|
|
|
| 41 |
answer, sources = st.session_state.engine.get_answer(query)
|
| 42 |
-
st.
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
import os
|
| 3 |
from rag_engine import ProjectRAGEngine
|
| 4 |
|
| 5 |
+
# Page configuration to ensure wide display for citations [cite: 31]
|
| 6 |
+
st.set_page_config(page_title="IIR Project Analyzer", layout="wide")
|
| 7 |
st.title("📂 Industrial Project Report Analyzer")
|
| 8 |
|
| 9 |
+
# 1. API Key Security Logic [cite: 21]
|
| 10 |
+
# Checks Hugging Face Secrets first, then falls back to user input
|
| 11 |
api_key = os.environ.get("OPENAI_API_KEY")
|
| 12 |
if not api_key:
|
| 13 |
api_key = st.sidebar.text_input("Enter OpenAI API Key", type="password")
|
| 14 |
|
| 15 |
if not api_key:
|
| 16 |
+
st.warning("Please provide an OpenAI API Key in the sidebar to proceed.")
|
| 17 |
st.stop()
|
| 18 |
|
| 19 |
+
# 2. Session Persistence [cite: 8, 44]
|
| 20 |
if "engine" not in st.session_state:
|
| 21 |
st.session_state.engine = ProjectRAGEngine(api_key)
|
| 22 |
|
| 23 |
+
# 3. Document Management Interface [cite: 6, 29]
|
| 24 |
+
uploaded_files = st.sidebar.file_uploader(
|
| 25 |
+
"Upload Project PDFs",
|
| 26 |
+
type="pdf",
|
| 27 |
+
accept_multiple_files=True
|
| 28 |
+
)
|
| 29 |
|
| 30 |
if uploaded_files:
|
| 31 |
+
# Check if files are already processed to prevent redundant API calls [cite: 8]
|
| 32 |
if "processed_files" not in st.session_state or len(st.session_state.processed_files) != len(uploaded_files):
|
| 33 |
+
with st.spinner("Analyzing project reports..."):
|
| 34 |
+
# Ensure the temp directory has write permissions to avoid 403/Permission errors
|
| 35 |
+
temp_dir = "temp"
|
| 36 |
+
if not os.path.exists(temp_dir):
|
| 37 |
+
os.makedirs(temp_dir, mode=0o777)
|
| 38 |
+
|
| 39 |
paths = []
|
| 40 |
for f in uploaded_files:
|
| 41 |
+
path = os.path.join(temp_dir, f.name)
|
| 42 |
+
with open(path, "wb") as out:
|
| 43 |
+
out.write(f.getbuffer())
|
| 44 |
+
paths.append(path)
|
| 45 |
+
|
| 46 |
+
# Simultaneous querying setup [cite: 7]
|
| 47 |
st.session_state.engine.process_documents(paths)
|
| 48 |
st.session_state.processed_files = [f.name for f in uploaded_files]
|
| 49 |
+
st.success("Reports indexed. Ready for queries.")
|
| 50 |
|
| 51 |
+
# 4. Question Answering Interface [cite: 30-31]
|
| 52 |
query = st.chat_input("Ex: 'Compare the budgets of these projects'")
|
| 53 |
+
|
| 54 |
if query:
|
| 55 |
+
with st.chat_message("user"):
|
| 56 |
+
st.write(query)
|
| 57 |
+
|
| 58 |
with st.chat_message("assistant"):
|
| 59 |
+
# Handle cases with no documents [cite: 25]
|
| 60 |
answer, sources = st.session_state.engine.get_answer(query)
|
| 61 |
+
st.markdown(f"### Response\n{answer}") # cite: 16
|
| 62 |
+
|
| 63 |
+
# Mandatory Source Attribution Display [cite: 11, 17, 31]
|
| 64 |
+
if sources:
|
| 65 |
+
with st.expander("View Source Attribution & Quotes"): # cite: 18
|
| 66 |
+
for idx, s in enumerate(sources):
|
| 67 |
+
doc_name = os.path.basename(s['metadata']['source'])
|
| 68 |
+
page_num = s['metadata']['page'] + 1 # PDF index starts at 0
|
| 69 |
+
st.markdown(f"**Source {idx+1}:** {doc_name} (Page {page_num})")
|
| 70 |
+
st.caption(f"Verbatim Quote: \"{s['content'][:300]}...\"")
|