riveter

A command-line utility written in Rust to "rivet" together the structure and content of a project directory. It generates a combined overview suitable for providing context to Large Language Models (LLMs) or for documentation purposes.

Motivation

When working with LLMs, providing sufficient context about a codebase or project structure is crucial for getting relevant and accurate responses. Manually copying directory trees and file contents is tedious and error-prone. This tool automates the process, creating a single output (either text or XML) that represents the project's layout and the content of its readable files.

Features

• Directory Tree: Generates a visual representation of the directory structure.
• File Content Concatenation: Reads and includes the content of text files.
• Output Formats: Supports plain text (text) and structured XML (xml) output.
• .gitignore Integration: Automatically respects .gitignore rules found within the project to exclude irrelevant files/directories (can be disabled).
• Custom Exclusions: Allows specifying additional ignore patterns (using gitignore syntax) via the command line.
• Depth Control: Limits the directory traversal depth.
• Graceful File Handling: Skips binary files or files with non-UTF8 content, issuing warnings instead of crashing.
• Cross-Platform: Built with Rust, runs on Linux, macOS, and Windows.

Installation

Prerequisites

• Rust programming language and Cargo (latest stable version recommended).

From Pre-Built Binaries

You can use cargo-binstall to install a pre-built binary for your platform.

cargo binstall riverter

From Source

1. Clone the repository:

   git clone https://github.com/d4mr/riveter
   cd riveter
   

2. Build:

   cargo build --release
   

   The executable will be located at target/release/riveter.
3. (Optional) Install globally using Cargo: From within the cloned directory:

   cargo install --path .
   

   This will install the binary into your Cargo bin directory (usually ~/.cargo/bin/), making it available in your PATH.

Usage

The tool is run from the command line.

riveter [OPTIONS]

By default, it processes the current directory (.) and outputs in text format to standard output.

[!TIP] You can pipe the output to pbcopy on macOS to copy it to your clipboard

riveter -f xml | pbcopy

Options

Here's the output from riveter --help:

Processes a directory, creating a structure view (text tree or XML) and concatenating readable file contents. Uses gitignore-style patterns for exclusion and optionally respects .gitignore files. Defaults to the current directory if none is specified.

Usage: riveter [OPTIONS]

Options:
  -d, --directory <DIRECTORY>
          The directory to process

          [default: .]

  -f, --format <FORMAT>
          Output format

          [default: text]
          [possible values: text, xml]

  -m, --max-depth <DEPTH>
          Maximum depth to traverse (0 means no limit)

          [default: 0]

  -x, --exclude [<PATTERN>...]
          Gitignore-style patterns to exclude files/directories. Applied *after* gitignore rules

      --respect-gitignore
          Respect .gitignore files found in the directory structure

  -h, --help
          Print help (see a summary with '-h')

  -V, --version
          Print version

Key Arguments:

• -d, --directory <DIR>: Specify the target directory to process.
• -f, --format <FORMAT>: Choose the output format (text or xml).
• -m, --max-depth <N>: Limit directory traversal to N levels deep (0 = unlimited).
• -x, --exclude <PATTERN>: Provide one or more gitignore-style patterns to exclude specific files or directories (e.g., -x "*.log" -x "build/"). This is applied in addition to .gitignore rules.
• --respect-gitignore=false: Disable the automatic use of .gitignore files.

Examples

1. Process the current directory (default settings):

   riveter
   

   (Outputs text format, respects .gitignore, unlimited depth)

2. Process a specific project directory and output as XML:

   riveter -d ../my-other-project -f xml
   

3. Process the current directory, excluding log files and the dist folder:

   riveter -x "*.log" -x "dist/"
   

4. Process the current directory, but only show the first 2 levels:

   riveter -m 2
   

5. Process the current directory, withour respecting any .gitignore files:

   riveter --respect-gitignore=false
   

6. Save the XML output to a file:

   riveter -f xml > project_context.xml
   

Output Formats

Text Format (-f text)

Provides a human-readable output:

--- Directory Tree ---
my_project/
  src/
    main.rs
    lib.rs
  tests/
    integration_test.rs
  .gitignore
  Cargo.toml
  README.md

--- File Contents ---
========================================
File: .gitignore
========================================
/target
Cargo.lock

========================================
File: Cargo.toml
========================================
[package]
name = "my_project"
# ... rest of Cargo.toml content ...

========================================
File: README.md
========================================
# My Project
This is a sample project.
# ... rest of README.md content ...

========================================
File: src/lib.rs
========================================
// Library code here
pub fn add(left: usize, right: usize) -> usize {
    left + right
}

========================================
File: src/main.rs
========================================
// Main application code here
fn main() {
    println!("Hello, world!");
}

========================================
File: tests/integration_test.rs
========================================
// Integration tests here
#[test]
fn it_works() {
    assert_eq!(2 + 2, 4);
}

XML Format (-f xml)

Provides a structured XML output, suitable for programmatic parsing:

<?xml version="1.0" encoding="UTF-8"?>
<projectContext rootPath="/absolute/path/to/my_project">
  <tree>
    <dir name="my_project">
      <file name=".gitignore"/>
      <file name="Cargo.toml"/>
      <file name="README.md"/>
      <dir name="src">
        <file name="lib.rs"/>
        <file name="main.rs"/>
      </dir>
      <dir name="tests">
        <file name="integration_test.rs"/>
      </dir>
    </dir>
  </tree>
  <fileContents>
    <file path=".gitignore"><![CDATA[/target
Cargo.lock
]]></file>
    <file path="Cargo.toml"><![CDATA[[package]
name = "my_project"
# ... rest of Cargo.toml content ...
]]></file>
    <file path="README.md"><![CDATA[# My Project
This is a sample project.
# ... rest of README.md content ...
]]></file>
    <file path="src/lib.rs"><![CDATA[// Library code here
pub fn add(left: usize, right: usize) -> usize {
    left + right
}
]]></file>
    <file path="src/main.rs"><![CDATA[// Main application code here
fn main() {
    println!("Hello, world!");
}
]]></file>
    <file path="tests/integration_test.rs"><![CDATA[// Integration tests here
#[test]
fn it_works() {
    assert_eq!(2 + 2, 4);
}
]]></file>
  </fileContents>
</projectContext>

Note: File content is wrapped in <![CDATA[...]]> sections to handle special characters correctly.

License

Licensed under Apache License, Version 2.0 (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)

Contributing

Contributions are welcome! Please feel free to submit issues or pull requests.

Acknowledgements

This tool relies on several excellent Rust crates, including:

• clap for command-line argument parsing.
• ignore for directory traversal and .gitignore handling.
• xmlwriter for generating XML output.