qdrant_test_ui.py

import streamlit as st
import os
import tempfile
from qdrant_client import QdrantClient, models
from langchain.vectorstores import Qdrant
from langchain.embeddings import FakeEmbeddings
from pathlib import Path
import shutil

st.set_page_config(
    page_title="FormPilot - Qdrant Dependency Test",
    page_icon="🧪",
    layout="wide",
)

st.title("🧪 FormPilot - Qdrant Dependency Test UI")

# Create a sidebar with instructions
with st.sidebar:
    st.header("Test Steps")
    st.info("""
    This app helps verify your Qdrant setup for FormPilot.
    
    1. Check package versions
    2. Test basic Qdrant functionality
    3. Test LangChain integration
    4. Test form ingestion process simulation
    """)
    
    st.markdown("---")
    st.subheader("Package Versions")
    
    try:
        import pkg_resources
        
        packages = [
            'streamlit', 'pandas', 'qdrant-client', 'langchain', 
            'langchain-openai', 'openai', 'tqdm', 'python-dotenv'
        ]
        
        for package in packages:
            try:
                version = pkg_resources.get_distribution(package).version
                st.success(f"{package}: {version}")
            except pkg_resources.DistributionNotFound:
                st.error(f"{package}: Not installed")
    except ImportError:
        st.error("Could not import pkg_resources")

# Main content area
st.header("1. Create Test Environment")

# Use temporary directory or let user specify path
use_temp_dir = st.checkbox("Use temporary directory for tests", value=True)

if use_temp_dir:
    temp_dir = tempfile.mkdtemp()
    qdrant_path = os.path.join(temp_dir, "qdrant_test_data")
    st.info(f"Using temporary directory: {qdrant_path}")
else:
    qdrant_path = st.text_input("Enter path for Qdrant data:", value="./qdrant_test_data")
    os.makedirs(qdrant_path, exist_ok=True)

# Create a test collection
st.header("2. Test Basic Qdrant Functionality")

test_collection_name = "test_collection"

if st.button("Create Test Collection"):
    with st.spinner("Creating collection..."):
        try:
            client = QdrantClient(path=qdrant_path)
            
            try:
                # Check if collection exists using get_collection API
                client.get_collection(collection_name=test_collection_name)
                st.info(f"Collection '{test_collection_name}' already exists")
            except Exception:
                # Collection doesn't exist, create it
                client.create_collection(
                    collection_name=test_collection_name,
                    vectors_config=models.VectorParams(size=2, distance=models.Distance.COSINE),
                )
                st.success(f"Created collection '{test_collection_name}'")
            
            # Insert test vectors
            vectors = [[1.0, 0.0], [0.0, 1.0]]
            payloads = [{"text": "test1"}, {"text": "test2"}]
            ids = [1, 2]
            
            client.upload_collection(
                collection_name=test_collection_name,
                vectors=vectors,
                payload=payloads,
                ids=ids
            )
            
            # Check count
            count = client.count(test_collection_name).count
            st.success(f"Number of vectors in collection: {count}")
            
            # Search for similar vectors
            search_result = client.search(
                collection_name=test_collection_name,
                query_vector=[1.0, 0.0],
                limit=1
            )
            
            st.json({"Search Result": [{"id": res.id, "score": res.score, "payload": res.payload} for res in search_result]})
        except Exception as e:
            st.error(f"Error: {e}")

# Test LangChain integration
st.header("3. Test LangChain Integration")

langchain_collection = "langchain_test"

if st.button("Test LangChain + Qdrant"):
    with st.spinner("Testing LangChain integration..."):
        try:
            # Create a fake embeddings model for testing
            class TestEmbeddings(FakeEmbeddings):
                def embed_documents(self, texts):
                    # Return 1536-dim vectors (like OpenAI)
                    return [[1.0] * 1536 for _ in texts]
                
                def embed_query(self, text):
                    # Return 1536-dim vector
                    return [1.0] * 1536
            
            embeddings = TestEmbeddings(size=1536)
            
            # Create a new directory for this test
            langchain_path = os.path.join(Path(qdrant_path).parent, "langchain_test")
            os.makedirs(langchain_path, exist_ok=True)
            st.info(f"LangChain test path: {langchain_path}")
            
            # Initialize Qdrant with LangChain
            try:
                from langchain.schema.document import Document
                docs = [Document(page_content="This is a test document", metadata={"source": "test"})]
                
                vectordb = Qdrant.from_documents(
                    documents=docs,
                    embedding=embeddings,
                    path=langchain_path,
                    collection_name=langchain_collection,
                )
                st.success("Successfully created Qdrant vector store with LangChain")
                
                # Test search
                results = vectordb.similarity_search("test query")
                st.json({"Search Results": [{"content": doc.page_content, "metadata": doc.metadata} for doc in results]})
                
            except Exception as e:
                st.error(f"LangChain integration error: {e}")
                st.error("This may indicate version incompatibility between LangChain and Qdrant")
                
        except Exception as e:
            st.error(f"Error: {e}")

# Simulate form ingestion
st.header("4. Test Form Ingestion Simulation")

if st.button("Simulate Form Ingestion"):
    with st.spinner("Simulating form ingestion process..."):
        try:
            # Create path for this test
            ingest_path = os.path.join(Path(qdrant_path).parent, "ingest_test")
            os.makedirs(ingest_path, exist_ok=True)
            
            # Create collection for ingestion test
            collection_name = "formpilot_test"
            client = QdrantClient(path=ingest_path)
            
            try:
                # Check if collection exists
                client.get_collection(collection_name=collection_name)
                st.info(f"Collection '{collection_name}' already exists")
            except Exception:
                # Collection doesn't exist, create it
                client.create_collection(
                    collection_name=collection_name,
                    vectors_config=models.VectorParams(size=1536, distance=models.Distance.COSINE),
                )
                st.success(f"Created collection '{collection_name}'")
            
            # Simulate adding vectors similar to ingest process
            test_vectors = [
                [0.1] * 1536,  # Simplified vector
                [0.2] * 1536
            ]
            
            test_payloads = [
                {"text": "I-485 form instructions page 1", "source": "I-485-instr.pdf:page-1", "form": "I-485"},
                {"text": "I-485 form instructions page 2", "source": "I-485-instr.pdf:page-2", "form": "I-485"}
            ]
            
            test_ids = [1, 2]
            
            client.upload_collection(
                collection_name=collection_name,
                vectors=test_vectors,
                payload=test_payloads,
                ids=test_ids,
                batch_size=64
            )
            
            # Verify count
            count = client.count(collection_name).count
            st.success(f"Number of vectors in test collection: {count}")
            
            # Retrieve a point to confirm structure
            point = client.retrieve(collection_name, ids=[1])
            if point:
                st.json({"Retrieved Point": {"id": point[0].id, "payload": point[0].payload}})
            else:
                st.error("Could not retrieve point")
                
        except Exception as e:
            st.error(f"Error: {e}")

# Cleanup
st.header("5. Cleanup")

if st.button("Clean Up Test Directories"):
    with st.spinner("Cleaning up..."):
        try:
            if use_temp_dir:
                shutil.rmtree(Path(qdrant_path).parent)
                st.success(f"Removed temp directory: {Path(qdrant_path).parent}")
            else:
                if os.path.exists(qdrant_path):
                    shutil.rmtree(qdrant_path)
                    st.success(f"Removed directory: {qdrant_path}")
                
                langchain_path = os.path.join(Path(qdrant_path).parent, "langchain_test")
                if os.path.exists(langchain_path):
                    shutil.rmtree(langchain_path)
                    st.success(f"Removed directory: {langchain_path}")
                
                ingest_path = os.path.join(Path(qdrant_path).parent, "ingest_test")
                if os.path.exists(ingest_path):
                    shutil.rmtree(ingest_path)
                    st.success(f"Removed directory: {ingest_path}")
        except Exception as e:
            st.error(f"Error during cleanup: {e}")

# Summary
st.header("6. Summary")

st.markdown("""
If all tests passed successfully, your Qdrant and LangChain dependencies are correctly configured for FormPilot.

### Common Issues:

1. **Version Incompatibilities**: Ensure you're using compatible versions of Qdrant client and LangChain.
2. **API Changes**: The LangChain API has changed significantly in recent versions. Your code may need updates.
3. **Missing Dependencies**: Make sure all required packages are installed.

### Recommendations:

Based on your code analysis, consider using these pinned versions:
```
qdrant-client==1.7.3
langchain==0.1.12
langchain-openai==0.1.0
```

Check the notebook for a more detailed dependency test if you need it.
""")