created app.py

app.py

@@ -0,0 +1,65 @@
+import streamlit as st
+import os
+from app.pdf_to_image import convert_pdf_to_images
+from app.rag import query_rag
+from app.embed_and_store import embed_all_images
+
+base_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+pdf_dir = os.path.join(base_dir, "data", "pdfs")
+image_dir = os.path.join(base_dir, "data", "images")
+
+os.makedirs(pdf_dir, exist_ok=True)
+os.makedirs(image_dir, exist_ok=True)
+
+st.title("🧠 Image-based PDF RAG")
+
+if "processed_files" not in st.session_state:
+    st.session_state.processed_files = set()
+if "image_paths" not in st.session_state:
+    st.session_state.image_paths = []
+
+uploaded_file = st.file_uploader("Upload a PDF file", type=["pdf"])
+if uploaded_file:
+    pdf_path = os.path.join(pdf_dir, uploaded_file.name)
+
+    if uploaded_file.name not in st.session_state.processed_files:
+        with st.spinner("Processing PDF and creating embeddings..."):
+            with open(pdf_path, "wb") as f:
+                f.write(uploaded_file.read())
+
+            image_paths = convert_pdf_to_images(pdf_path, image_dir)
+
+            for image_path in image_paths:
+                embed_all_images(image_path)
+
+            st.session_state.processed_files.add(uploaded_file.name)
+            st.session_state.image_paths = image_paths
+
+            st.success(
+                f"PDF processed! {len(image_paths)} pages converted and embedded for search."
+            )
+    else:
+        st.success("PDF already processed. Ready to query.")
+        image_paths = st.session_state.image_paths
+else:
+    image_paths = []
+
+# Query UI
+if image_paths:
+    st.subheader("Query PDF Content")
+    page_options = [f"Page {i+1}" for i in range(len(image_paths))]
+    selected_page = st.selectbox("Select a page to query:", page_options)
+
+    if page_options:
+        page_idx = page_options.index(selected_page)
+        selected_image = image_paths[page_idx]
+
+        with st.expander("📄 View selected page"):
+            st.image(selected_image, caption=f"Page {page_idx+1}")
+
+        query = st.text_input("🔍 Ask a question about this page:")
+        if query and st.button("Search"):
+            with st.spinner("Searching with Gemini..."):
+                result = query_rag(selected_image, query)
+                st.markdown("### 💬 Answer")
+                st.write(result)

app/embed_and_store.py

@@ -1,8 +1,8 @@
 # app/embed_and_store.py
 
 import os
-from embed import get_image_embedding
-from chroma_utils import add_embedding
+from app.embed import get_image_embedding
+from app.chroma_utils import add_embedding
 
 # Create absolute paths for data directories
 base_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))

app/rag.py

@@ -1,7 +1,4 @@
-from embed import get_image_embedding
-from chroma_utils import search_embedding
-# app/rag.py
-from gemini_vision import ask_image_question
+from app.gemini_vision import ask_image_question
 
 def query_rag(image_path: str, question: str) -> str:
     return ask_image_question(image_path, question)

app/ui.py

@@ -1,65 +1,65 @@
-import streamlit as st
-import os
-from pdf_to_image import convert_pdf_to_images
-from rag import query_rag
-from embed_and_store import embed_all_images
+# import streamlit as st
+# import os
+# from pdf_to_image import convert_pdf_to_images
+# from rag import query_rag
+# from embed_and_store import embed_all_images
 
-base_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
-pdf_dir = os.path.join(base_dir, "data", "pdfs")
-image_dir = os.path.join(base_dir, "data", "images")
+# base_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+# pdf_dir = os.path.join(base_dir, "data", "pdfs")
+# image_dir = os.path.join(base_dir, "data", "images")
 
-os.makedirs(pdf_dir, exist_ok=True)
-os.makedirs(image_dir, exist_ok=True)
+# os.makedirs(pdf_dir, exist_ok=True)
+# os.makedirs(image_dir, exist_ok=True)
 
-st.title("🧠 Image-based PDF RAG")
+# st.title("🧠 Image-based PDF RAG")
 
-if "processed_files" not in st.session_state:
-    st.session_state.processed_files = set()
-if "image_paths" not in st.session_state:
-    st.session_state.image_paths = []
+# if "processed_files" not in st.session_state:
+#     st.session_state.processed_files = set()
+# if "image_paths" not in st.session_state:
+#     st.session_state.image_paths = []
 
-uploaded_file = st.file_uploader("Upload a PDF file", type=["pdf"])
-if uploaded_file:
-    pdf_path = os.path.join(pdf_dir, uploaded_file.name)
+# uploaded_file = st.file_uploader("Upload a PDF file", type=["pdf"])
+# if uploaded_file:
+#     pdf_path = os.path.join(pdf_dir, uploaded_file.name)
 
-    if uploaded_file.name not in st.session_state.processed_files:
-        with st.spinner("Processing PDF and creating embeddings..."):
-            with open(pdf_path, "wb") as f:
-                f.write(uploaded_file.read())
+#     if uploaded_file.name not in st.session_state.processed_files:
+#         with st.spinner("Processing PDF and creating embeddings..."):
+#             with open(pdf_path, "wb") as f:
+#                 f.write(uploaded_file.read())
 
-            image_paths = convert_pdf_to_images(pdf_path, image_dir)
+#             image_paths = convert_pdf_to_images(pdf_path, image_dir)
 
-            for image_path in image_paths:
-                embed_all_images(image_path)
+#             for image_path in image_paths:
+#                 embed_all_images(image_path)
 
-            st.session_state.processed_files.add(uploaded_file.name)
-            st.session_state.image_paths = image_paths
+#             st.session_state.processed_files.add(uploaded_file.name)
+#             st.session_state.image_paths = image_paths
 
-            st.success(
-                f"PDF processed! {len(image_paths)} pages converted and embedded for search."
-            )
-    else:
-        st.success("PDF already processed. Ready to query.")
-        image_paths = st.session_state.image_paths
-else:
-    image_paths = []
+#             st.success(
+#                 f"PDF processed! {len(image_paths)} pages converted and embedded for search."
+#             )
+#     else:
+#         st.success("PDF already processed. Ready to query.")
+#         image_paths = st.session_state.image_paths
+# else:
+#     image_paths = []
 
-# Query UI
-if image_paths:
-    st.subheader("Query PDF Content")
-    page_options = [f"Page {i+1}" for i in range(len(image_paths))]
-    selected_page = st.selectbox("Select a page to query:", page_options)
+# # Query UI
+# if image_paths:
+#     st.subheader("Query PDF Content")
+#     page_options = [f"Page {i+1}" for i in range(len(image_paths))]
+#     selected_page = st.selectbox("Select a page to query:", page_options)
 
-    if page_options:
-        page_idx = page_options.index(selected_page)
-        selected_image = image_paths[page_idx]
+#     if page_options:
+#         page_idx = page_options.index(selected_page)
+#         selected_image = image_paths[page_idx]
 
-        with st.expander("📄 View selected page"):
-            st.image(selected_image, caption=f"Page {page_idx+1}")
+#         with st.expander("📄 View selected page"):
+#             st.image(selected_image, caption=f"Page {page_idx+1}")
 
-        query = st.text_input("🔍 Ask a question about this page:")
-        if query and st.button("Search"):
-            with st.spinner("Searching with Gemini..."):
-                result = query_rag(selected_image, query)
-                st.markdown("### 💬 Answer")
-                st.write(result)
+#         query = st.text_input("🔍 Ask a question about this page:")
+#         if query and st.button("Search"):
+#             with st.spinner("Searching with Gemini..."):
+#                 result = query_rag(selected_image, query)
+#                 st.markdown("### 💬 Answer")
+#                 st.write(result)