[wip] Modal RAG example

codegen-examples/examples/langchain_agent/README.md

@@ -0,0 +1,82 @@
+# Codegen LangChain Agent Example
+
+<p align="center">
+  <a href="https://docs.codegen.com/tutorials/build-code-agent">
+    <img src="https://i.imgur.com/6RF9W0z.jpeg" />
+  </a>
+</p>
+
+<h2 align="center">
+  Build an intelligent code agent with LangChain and Codegen
+</h2>
+
+<div align="center">
+
+[![Documentation](https://img.shields.io/badge/Docs-docs.codegen.com-purple?style=flat-square)](https://docs.codegen.com/tutorials/build-code-agent)
+[![License](https://img.shields.io/badge/Code%20License-Apache%202.0-gray?&color=gray)](https://github.com/codegen-sh/codegen-sdk/tree/develop?tab=Apache-2.0-1-ov-file)
+
+</div>
+
+This example demonstrates how to build an intelligent code agent using Codegen's LangChain integration. The agent can analyze and manipulate codebases using natural language commands.
+
+## Quick Start
+
+```python
+from codegen import Codebase
+from codegen.extensions.langchain import create_codebase_agent
+
+# Initialize codebase
+codebase = Codebase.from_repo("fastapi/fastapi")
+
+# Create the agent
+agent = create_codebase_agent(codebase=codebase, model_name="gpt-4", verbose=True)
+
+# Ask the agent to analyze code
+result = agent.invoke({"input": "What are the dependencies of the FastAPI class?", "config": {"configurable": {"session_id": "demo"}}})
+print(result["output"])
+```
+
+## Installation
+
+```bash
+# Install dependencies
+pip install modal-client codegen langchain langchain-openai
+
+# Run the example
+python run.py
+```
+
+## Available Tools
+
+The agent comes with several built-in tools for code operations:
+
+- `ViewFileTool`: View file contents and metadata
+- `ListDirectoryTool`: List directory contents
+- `SearchTool`: Search code using regex
+- `EditFileTool`: Edit file contents
+- `CreateFileTool`: Create new files
+- `DeleteFileTool`: Delete files
+- `RenameFileTool`: Rename files and update imports
+- `MoveSymbolTool`: Move functions/classes between files
+- `RevealSymbolTool`: Analyze symbol dependencies
+- `SemanticEditTool`: Make semantic code edits
+- `CommitTool`: Commit changes to disk
+
+## Example Operations
+
+The agent can perform various code analysis and manipulation tasks:
+
+```python
+# Analyze dependencies
+agent.invoke({"input": "What are the dependencies of the reveal_symbol function?", "config": {"configurable": {"session_id": "demo"}}})
+
+# Find usage patterns
+agent.invoke({"input": "Show me examples of dependency injection in the codebase", "config": {"configurable": {"session_id": "demo"}}})
+
+# Move code
+agent.invoke({"input": "Move the validate_email function to validation_utils.py", "config": {"configurable": {"session_id": "demo"}}})
+```
+
+## Learn More
+
+- [Full Tutorial](https://docs.codegen.com/tutorials/build-code-agent)

codegen-examples/examples/langchain_agent/run.py

@@ -0,0 +1,107 @@
+"""Demo implementation of an agent with Codegen tools."""
+
+from codegen import Codebase
+from codegen.extensions.langchain.tools import (
+    CommitTool,
+    CreateFileTool,
+    DeleteFileTool,
+    EditFileTool,
+    ListDirectoryTool,
+    MoveSymbolTool,
+    RenameFileTool,
+    RevealSymbolTool,
+    SearchTool,
+    SemanticEditTool,
+    ViewFileTool,
+)
+from codegen.sdk.enums import ProgrammingLanguage
+from langchain import hub
+from langchain.agents import AgentExecutor
+from langchain.agents.openai_functions_agent.base import OpenAIFunctionsAgent
+from langchain_core.chat_history import ChatMessageHistory
+from langchain_core.runnables.history import RunnableWithMessageHistory
+from langchain_openai import ChatOpenAI
+
+
+def create_codebase_agent(
+    codebase: Codebase,
+    model_name: str = "gpt-4o",
+    temperature: float = 0,
+    verbose: bool = True,
+) -> RunnableWithMessageHistory:
+    """Create an agent with all codebase tools.
+
+    Args:
+        codebase: The codebase to operate on
+        model_name: Name of the model to use (default: gpt-4)
+        temperature: Model temperature (default: 0)
+        verbose: Whether to print agent's thought process (default: True)
+
+    Returns:
+        Initialized agent with message history
+    """
+    # Initialize language model
+    llm = ChatOpenAI(
+        model_name=model_name,
+        temperature=temperature,
+    )
+
+    # Get all codebase tools
+    tools = [
+        ViewFileTool(codebase),
+        ListDirectoryTool(codebase),
+        SearchTool(codebase),
+        EditFileTool(codebase),
+        CreateFileTool(codebase),
+        DeleteFileTool(codebase),
+        RenameFileTool(codebase),
+        MoveSymbolTool(codebase),
+        RevealSymbolTool(codebase),
+        SemanticEditTool(codebase),
+        CommitTool(codebase),
+    ]
+
+    # Get the prompt to use
+    prompt = hub.pull("hwchase17/openai-functions-agent")
+
+    # Create the agent
+    agent = OpenAIFunctionsAgent(
+        llm=llm,
+        tools=tools,
+        prompt=prompt,
+    )
+
+    # Create the agent executor
+    agent_executor = AgentExecutor(
+        agent=agent,
+        tools=tools,
+        verbose=verbose,
+    )
+
+    # Create message history handler
+    message_history = ChatMessageHistory()
+
+    # Wrap with message history
+    return RunnableWithMessageHistory(
+        agent_executor,
+        lambda session_id: message_history,
+        input_messages_key="input",
+        history_messages_key="chat_history",
+    )
+
+
+if __name__ == "__main__":
+    # Initialize codebase
+    print("Initializing codebase...")
+    codebase = Codebase.from_repo("fastapi/fastapi", programming_language=ProgrammingLanguage.PYTHON)
+
+    # Create agent with history
+    print("Creating agent...")
+    agent = create_codebase_agent(codebase)
+
+    print("\nAsking agent to analyze symbol relationships...")
+    result = agent.invoke(
+        {"input": "What are the dependencies of the reveal_symbol function?"},
+        config={"configurable": {"session_id": "demo"}},
+    )
+    print("Messages:", result["messages"])

src/codegen/extensions/modal/api.py → codegen-examples/examples/modal_repo_analytics/api.py

@@ -1,9 +1,8 @@
 """Modal API endpoint for repository analysis."""
 
-import modal
-from pydantic import BaseModel
-
+import modal  # deptry: ignore
 from codegen import Codebase
+from pydantic import BaseModel
 
 # Create image with dependencies
 image = modal.Image.debian_slim(python_version="3.13").apt_install("git").pip_install("fastapi[standard]", "codegen>=0.5.30")

codegen-examples/examples/modal_repo_rag/README.md

@@ -0,0 +1,120 @@
+# Codegen RAG Q&A API
+
+<p align="center">
+  <a href="https://docs.codegen.com">
+    <img src="https://i.imgur.com/6RF9W0z.jpeg" />
+  </a>
+</p>
+
+<h2 align="center">
+  Answer questions about any GitHub repository using RAG
+</h2>
+
+<div align="center">
+
+[![Documentation](https://img.shields.io/badge/Docs-docs.codegen.com-purple?style=flat-square)](https://docs.codegen.com)
+[![License](https://img.shields.io/badge/Code%20License-Apache%202.0-gray?&color=gray)](https://github.com/codegen-sh/codegen-sdk/tree/develop?tab=Apache-2.0-1-ov-file)
+
+</div>
+
+This example demonstrates how to build a RAG-powered code Q&A API using Codegen's VectorIndex and Modal. The API can answer questions about any GitHub repository by:
+
+1. Creating embeddings for all files in the repository
+1. Finding the most relevant files for a given question
+1. Using GPT-4 to generate an answer based on the context
+
+## Quick Start
+
+1. Install dependencies:
+
+```bash
+pip install modal-client codegen openai
+```
+
+2. Create a Modal volume for storing indices:
+
+```bash
+modal volume create codegen-indices
+```
+
+3. Start the API server:
+
+```bash
+modal serve api.py
+```
+
+4. Test with curl:
+
+```bash
+curl -X POST "http://localhost:8000/answer_code_question" \
+  -H "Content-Type: application/json" \
+  -d '{
+    "repo_name": "fastapi/fastapi",
+    "query": "How does FastAPI handle dependency injection?"
+  }'
+```
+
+## API Reference
+
+### POST /answer_code_question
+
+Request body:
+
+```json
+{
+  "repo_name": "owner/repo",
+  "query": "Your question about the code"
+}
+```
+
+Response format:
+
+```json
+{
+  "status": "success",
+  "error": "",
+  "answer": "Detailed answer based on the code...",
+  "context": [
+    {
+      "filepath": "path/to/file.py",
+      "snippet": "Relevant code snippet..."
+    }
+  ]
+}
+```
+
+## How It Works
+
+1. The API uses Codegen to clone and analyze the repository
+1. It creates/loads a VectorIndex of all files using OpenAI's embeddings
+1. For each question:
+   - Finds the most semantically similar files
+   - Extracts relevant code snippets
+   - Uses GPT-4 to generate an answer based on the context
+
+## Development
+
+The API is built using:
+
+- Modal for serverless deployment
+- Codegen for repository analysis
+- OpenAI for embeddings and Q&A
+- FastAPI for the web endpoint
+
+To deploy changes:
+
+```bash
+modal deploy api.py
+```
+
+## Environment Variables
+
+Required environment variables:
+
+- `OPENAI_API_KEY`: Your OpenAI API key
+
+## Learn More
+
+- [Codegen Documentation](https://docs.codegen.com)
+- [Modal Documentation](https://modal.com/docs)
+- [VectorIndex Tutorial](https://docs.codegen.com/building-with-codegen/semantic-code-search)