streamlit deployment

.gitattributes

@@ -0,0 +1,3 @@
+vectorstore/* filter=lfs diff=lfs merge=lfs -text
+vectorstore/*.sqlite filter=lfs diff=lfs merge=lfs -text
+vectorstore/**/*.bin filter=lfs diff=lfs merge=lfs -text

src/paper_query/data/loaders.py

@@ -10,13 +10,24 @@
 from paper_query.llm import setup_model
 
 
-def pypdf_loader(file_path: str) -> Document:
+def pypdf_loader(file_path: str, interpret_images: bool = False, **image_kwargs) -> Document:
+    """Function to load a PDF file, optionally interpreting images."""
+    if interpret_images and "model" not in image_kwargs:
+        raise ValueError("When interpret_images is True, 'model' must be provided in image_kwargs.")
+
+    if interpret_images:
+        return _pypdf_loader_w_images(file_path, **image_kwargs)
+    else:
+        return _pypdf_loader(file_path)
+
+
+def _pypdf_loader(file_path: str) -> Document:
     """Function to load text from a PDF file."""
     logger.debug("Loading PDF file using PyPDFLoader")
     return PyPDFLoader(file_path, mode="single").load()[0]
 
 
-def pypdf_loader_w_images(
+def _pypdf_loader_w_images(
     file_path: str, model: str, provider: str, max_tokens: int = 1024
 ) -> Document:
     """Function to load text from a PDF file with images."""

src/paper_query/llm/models.py

@@ -1,10 +1,11 @@
 import os
 
 from langchain.chat_models import init_chat_model
+from langchain_core.language_models.chat_models import BaseChatModel
 from loguru import logger
 
 
-def setup_model(model_name: str, model_provider: str, **kwargs):
+def setup_model(model_name: str, model_provider: str, **kwargs) -> BaseChatModel:
     """Initialize the chat model."""
     logger.info(f"Initializing {model_name} model from {model_provider}")
     if model_provider == "openai":

src/paper_query/ui/strain_relief_app.py

@@ -30,7 +30,7 @@ def strain_relief_chatbot():
     """Chatbot for the StrainRelief paper."""
     initialize_session_state()
 
-    st.title("The StrainRelief Chatbot")
+    st.title("StrainReliefChat")
     chat_tab, about_tab = st.tabs(["Chat", "About"])
 
     st.sidebar.title("API Configuration")
@@ -46,12 +46,14 @@ def strain_relief_chatbot():
     # Display current model
     st.sidebar.markdown(f"Using **{st.session_state.model_name}** model.")
 
-    st.session_state.chatbot = HybridQueryChatbot(
-        model_name=st.session_state.model_name.lower(),
-        model_provider="openai",
-        paper_path=str(assets_dir / "strainrelief_preprint.pdf"),
-        references_dir=str(assets_dir / "references"),
-    )
+    # Only instantiate chatbot once and store in session state
+    if st.session_state.chatbot is None:
+        st.session_state.chatbot = HybridQueryChatbot(
+            model_name=st.session_state.model_name.lower(),
+            model_provider="openai",
+            paper_path=str(assets_dir / "strainrelief_preprint.pdf"),
+            references_dir=str(assets_dir / "references"),
+        )
 
     with chat_tab:
         if "messages" not in st.session_state:

test/data/test_loaders.py

@@ -5,15 +5,14 @@
 from paper_query.data.loaders import (
     code_loader,
     pypdf_loader,
-    pypdf_loader_w_images,
     references_loader,
 )
 
 
 def test_pypdf_loader(test_assets_dir):
     """Test the pypdf_loader function."""
     path = test_assets_dir / "example_pdf.pdf"
-    doc = pypdf_loader(path)
+    doc = pypdf_loader(path, interpret_images=False)
     assert isinstance(doc, Document)
 
 
@@ -22,7 +21,7 @@ def test_pypdf_loader_w_images(test_assets_dir):
     """Test the pypdf_loader_w_images function."""
     path = test_assets_dir / "example_pdf.pdf"
     # TODO: change to free model
-    doc = pypdf_loader_w_images(path, "gpt-4.1-nano", "openai")
+    doc = pypdf_loader(path, interpret_images=True, model="gpt-4.1-nano", provider="openai")
     assert isinstance(doc, Document)