Metadata-Version: 2.4
Name: ape-langchain
Version: 0.1.1
Summary: LangChain integration layer for the Ape AI-first programming language
Author-email: David Van Aelst <david@skyrah.be>
License: MIT
Project-URL: Homepage, https://github.com/Quynah/ape-langchain
Project-URL: Repository, https://github.com/Quynah/ape-langchain
Project-URL: Documentation, https://github.com/Quynah/ape-langchain#readme
Project-URL: Issues, https://github.com/Quynah/ape-langchain/issues
Keywords: ape,langchain,ai,llm,deterministic,agent
Classifier: Development Status :: 3 - Alpha
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Topic :: Software Development :: Libraries :: Python Modules
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.11
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: ape-lang>=0.2.0
Requires-Dist: langchain>=0.1.0
Requires-Dist: typing-extensions>=4.0.0
Provides-Extra: dev
Requires-Dist: pytest>=7.0.0; extra == "dev"
Requires-Dist: pytest-cov>=4.0.0; extra == "dev"
Requires-Dist: black>=23.0.0; extra == "dev"
Requires-Dist: mypy>=1.0.0; extra == "dev"
Dynamic: license-file

# ape-langchain

**LangChain integration layer for the Ape AI-first programming language**

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
[![Python 3.11+](https://img.shields.io/badge/python-3.11+-blue.svg)](https://www.python.org/downloads/)

## What is this?

`ape-langchain` bridges the [Ape programming language](https://github.com/Quynah/Ape) with [LangChain](https://github.com/langchain-ai/langchain), enabling you to:

- **Use Ape tasks as LangChain tools** with deterministic execution
- **Eliminate AI hallucination** in tool invocation through strict validation
- **Build reliable AI agents** with formally defined, constraint-validated workflows

Ape is a deterministic AI-first language designed to remove ambiguity from AI-generated code. `ape-langchain` makes Ape tasks seamlessly callable from LangChain agents.

## Installation

```bash
pip install ape-langchain
```

This will automatically install `ape-lang` and `langchain` as dependencies.

## Quick Start

### 1. Write an Ape Task

Create `pricing.ape`:

```ape
task calculate_total_price
  inputs:
    base_price: String
    tax_rate: String
  outputs:
    total: String
  constraints:
    - base_price must be numeric
    - tax_rate must be between 0 and 1
  steps:
    - Parse base_price as decimal
    - Multiply by (1 + tax_rate)
    - Return as string
```

### 2. Use it in LangChain

```python
from ape_langchain import ApeTool
from langchain.agents import initialize_agent, AgentType
from langchain.llms import OpenAI

# Wrap Ape task as LangChain tool
pricing_tool = ApeTool(
    ape_file="pricing.ape",
    function="calculate_total_price",
    description="Calculate total price with tax"
)

# Create agent with deterministic tool
llm = OpenAI(temperature=0)
agent = initialize_agent(
    tools=[pricing_tool],
    llm=llm,
    agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION,
    verbose=True
)

# Execute - AI decides when to call, but execution is deterministic
result = agent.run("What's the total for $99.99 with 21% tax?")
print(result)
```

### 3. Or Use ApeChain Directly

For simpler workflows without full agent orchestration:

```python
from ape_langchain import ApeChain

chain = ApeChain.from_ape_file("pricing.ape", function="calculate_total_price")

result = chain.run(base_price="99.99", tax_rate="0.21")
print(result)  # Deterministic, validated output
```

## Why Use This?

### Problem: AI Tool Invocation is Unreliable

```python
# Traditional approach - AI can hallucinate parameters, miss constraints
def calculate_price(base, tax):
    return base * (1 + tax)  # What if base is negative? What if tax > 1?
```

### Solution: Ape + LangChain

```python
# Ape approach - constraints enforced, no ambiguity
pricing_tool = ApeTool("pricing.ape", "calculate_total_price")
# - Input validation automatic
# - Constraints checked before execution
# - No hallucination possible
```

## API Overview

### `ApeTool`

LangChain `BaseTool` adapter for Ape tasks.

```python
from ape_langchain import ApeTool

tool = ApeTool(
    ape_file="path/to/module.ape",
    function="task_name",
    description="Optional description for LLM"
)

# Use in LangChain agents
agent = initialize_agent(tools=[tool], llm=llm, ...)
```

### `ApeChain`

High-level wrapper for direct execution.

```python
from ape_langchain import ApeChain

chain = ApeChain.from_ape_file("module.ape", function="task_name")
result = chain.run(param1="value", param2="value")
```

### `ApeTask`

Low-level task representation with JSON Schema export.

```python
from ape_langchain import ApeTask
from ape import compile

module = compile("module.ape")
task = ApeTask.from_module(module, "task_name")

# Get JSON Schema for OpenAI/Anthropic function calling
schema = task.to_json_schema()
```

## Features

- ✅ **Deterministic execution** - Ape constraints prevent invalid inputs
- ✅ **Type-safe** - Full type checking before execution
- ✅ **LangChain native** - Works with all LangChain agent types
- ✅ **JSON Schema export** - Compatible with OpenAI/Anthropic function calling
- ✅ **Error handling** - Clear exception types for debugging

## Requirements

- Python 3.11+
- ape-lang >=0.2.0
- langchain >=0.1.0

## Development Status

**v0.1.0** - Initial release

Currently supports:
- Basic Ape task wrapping
- LangChain tool integration
- Simple type mapping (all parameters as strings for now)

Roadmap:
- Advanced type system (int, float, list, dict)
- Async execution support
- Multi-task orchestration
- Native Ape control flow support

## Contributing

Contributions welcome! See the [main Ape repository](https://github.com/Quynah/Ape) for guidelines.

## License

MIT License - see LICENSE file for details.

## Links

- [Ape Language](https://github.com/Quynah/Ape)
- [LangChain](https://github.com/langchain-ai/langchain)
- [Documentation](https://github.com/Quynah/Ape/tree/main/docs)

---

**Built with Ape 🦍 - Because AI should execute deterministically, not approximately.**
