💾 Archived View for random-projects.net › blog › 2021-03-05-gemtext-parser.gemini captured on 2021-12-03 at 14:04:38. Gemini links have been rewritten to link to archived content

-=-=-=-=-=-=-

zig-gemtext - A new gemini text parser

Date: 05.03.2021

Intro

In the last two days i created a new library called zig-gemtext which provides both spec-compliant parsing as well as rendering gemini text into the following formats:

gemini text ("autoformatting")
rich text
html
markdown

The library is written in Zig, but exposes both a Zig and C api, so it can be used from pretty much any programming language or platform. In theory, the library should be compilable to wasm as well and be used from there.

The parser itself is a non-blocking streaming parser, so it's perfectly suitable for any asynchronous or stream processing, also combinable with the renderer.

You can get the source of the library here:

github.com/MasterQ32/zig-gemtext

Usage

Here's a small usage example of the library:

pub fn main() !void {
    var document = try gemtext.Document.parse(
      std.heap.page_allocator,
      std.io.getStdIn().reader(),
    );
    defer document.deinit();

    try gemtext.renderer.html(
      document.fragments.items, 
      std.io.getStdOut().writer(),
    );
}

Performance

I tested the library with a pretty naive benchmark and got those results:

148489113 Bytes (148 MB, 142 MiB) copied, 6,13719 s, 24,2 MB/s

1392348 lines in total
6,13719 seconds
226870 lines per second
140 MB source
24.2 MB/s

I think it can be faster, but it's good enough for now.

Source code of the benchmark