the-alchemists-of-arland / gray-matter-rs Goto Github PK

As mentioned in #14, there are likely a lot of things that can be done to improve the performance of our parser - as I was not very proficient in Rust when initially writing it. This issue can be used for tracking advances done in that regard.

You seem to be using Rust 1.49 in both the main.yml and the publish.yml workflow. See:

It seems I don't have direct push access to the main branch (needs review), so I can't change this. @yuchanns, could you update the workflow to the latest Rust version? The latest stable is 1.55, and my pending PR needs at least 1.52

While investigating #3, I realised that we don't really have a need for the polyglot datatype Pod. Since we nonetheless coerce it into a serde_json::Value upon deserializing, Pod can easily be replaced by Value - essentially skipping a conversion step.

This would increase performance (probably not by much, but every nanosecond counts) and drastically reduce code complexity.

Changing this would introduce an API change, so I don't see the rush to do it, but I'm leaving this issue here for whenever we deem it fit to implement.

Currently, it will panic at src/matter.rs:115:34 when you have something like

---
---
---

My suggested solution is to only parse the first two --- then include the succeeding to content

---
key: "Value"
---
    This is my content

gets parsed to

ParsedEntity {
    content: "This is my content"
}

Maybe introducing a configuration field for Matter to allow to set whether whitespace gets trimmed or not. Can be true by default for backwards compatibility.

A simple change to this line is not sufficient enough as the content then contains a leading newline:

\n    This is my content

Very useful library btw