flowo

A Pythonic DSL for Flowgorithm (.fprg) that can simplify AI-assisted work by generating valid visual programs from Python code.

This thing is basically a port and redesign of the Rust crate fgrs.

AI & Vibecoding

Flowgorithm's native XML format is a nightmare for LLMs (like Gemini, Claude, or ChatGPT) to generate correctly. flowo provides a clean, type-safe Python DSL that is significantly easier for AI coding assistants to write.

By using flowo, you can "vibecode" your Flowgorithm assignments by prompting for Python code and letting the library handle the strict XML schema, reserved keywords, and variable declarations automatically.

Features

• Context Manager API: Natural Pythonic syntax for nested structures (with flow.if_():, with flow.for_():).
• Strict Validation: Prevents invalid Flowgorithm variable names and reserved keyword usage.
• Variable Tracking: Ensures variables are declared before use.
• Intrinsic Functions: Helper functions for all Flowgorithm built-ins (Abs, Sin, Len, ToInteger, etc.).
• Zero Dependencies: Built entirely on the Python Standard Library.

Installation

pip install flowo

Quick Start

from flowo import Flow, Type

flow = Flow("HelloWorld.fprg")

with flow.function("Main"):
    flow.declare("name", Type.STRING)
    flow.output('"What is your name?"')
    flow.input("name")
    flow.output('"Hello, " & name')

flow.to_fprg()

Advanced Usage

flowo supports all Flowgorithm features (that I've used during my uni study) including arrays, multi-argument declarations, and nested loops:

from flowo import Flow, Type, Sin, PI

flow = Flow()
with flow.function("Main"):
    flow.declare("i, n", Type.INTEGER)
    flow.declare("val", Type.REAL)
    
    flow.assign("n", "10")
    with flow.for_("i", "0", "n - 1"):
        flow.assign("val", Sin(f"(i / n) * 2 * {PI}"))
        flow.output("val")

flow.to_fprg("SineWave.fprg")

License

MIT