Flows.rs

"Τὰ πάντα ῥεῖ καὶ οὐδὲν μένει" — Heraclitus

Building blocks for flow-based programming (FBP) in Rust. This collection of crates implements a growing shrink-wrap inventory of standard, reusable dataflow blocks for common use cases.

Tip

🚧 We are building in public. This is presently under heavy construction.

[Features] | [Prerequisites] | [Installation] | [Examples] | [Reference] | [Development]

✨ Features

• Provides components for flow-based programming (FBP) based on Tokio.
• Enables dataflow systems through reusable components called blocks.
• Built on the dataflow primitives provided by the Async-Flow project.
• Supports opting out of any feature using comprehensive feature flags.
• Adheres to the Rust API Guidelines in its naming conventions.
• Cuts red tape: 100% free and unencumbered public domain software.

🛠️ Prerequisites

• Rust 1.85+ (2024 edition)

⬇️ Installation

Installation via Cargo

cargo add flows

👉 Examples

Importing the Library

use flows::{audio, hash, image, io, json, math, rand, text};

Implementing Blocks

Implementing a split_string block

use flows::{Inputs, Outputs, Result};

/// A block that splits input strings based on a delimiter.
async fn split_string(delim: &str, mut ins: Inputs<String>, outs: Outputs<String>) -> Result {
    while let Some(input) = ins.recv().await? {
        for output in input.split(delim) {
            outs.send(output.into()).await?;
        }
    }
    Ok(())
}

Implementing an add_ints block

use flows::{Inputs, Outputs, Result};

/// A block that outputs the sums of input numbers.
async fn add_ints(mut lhs: Inputs<i64>, mut rhs: Inputs<i64>, sums: Outputs<i64>) -> Result {
    loop {
        let (a, b) = tokio::try_join!(lhs.recv(), rhs.recv())?;
        match (a, b) {
            (Some(a), Some(b)) => sums.send(a + b).await?,
            _ => break,
        }
    }
    Ok(())
}

📚 Reference

docs.rs/flows

Packages

Package	Summary	Crate	Documentation
flows	Flow-based programming (FBP).
flows-arrow	Flow-based data processing with Apache Arrow.
flows-audio	Flow-based audio processing.
flows-datafusion	Flow-based query processing with Apache DataFusion.
flows-derive	Derive macros for flow-based programming (FBP).
flows-hash	Flow-based cryptographic hashing.
flows-http	Flow-based HTTP requests & responses.
flows-image	Flow-based image processing.
flows-io	Flow-based I/O readers & writers.
flows-json	Flow-based JSON encoding & decoding.
flows-math	Flow-based mathematical operations.
flows-rand	Flow-based random number generation.
flows-text	Flow-based text processing.

Integrations

TBD

Glossary

• System: A collection of blocks that are connected together. Systems are the top-level entities in dataflow programs.

• Block: An encapsulated system component that processes messages. Blocks are the autonomous units of computation in a system.

• Port: A named connection point on a block that sends or receives messages. Ports are the only interfaces through which blocks communicate with each other.

• Message: A unit of data that flows between blocks in a system, from port to port. Any Rust type that implements the Send + Sync + 'static traits can be used as a message.

👨‍💻 Development

git clone https://github.com/artob/flows.rs.git

---