Plans on doing the other way (editting the front matter)

[SeaDve]

Very useful library btw

[yuchanns]

More detail in your description please?

[SeaDve]

Something like extracting the front matter, editing it, then writing back to the file

[yuchanns]

Got it. I think add an api such as compose will be a nice idea. Or else do you have any better advice?

For example:

use gray_matter::Matter;
use gray_matter::engine::YAML;
use serde::Deserialize;

const INPUT: &str = r#"---
title: gray-matter-rs
tags:
  - gray-matter
  - rust
---
Some excerpt
---
Other stuff
"#;

fn main() {
	#[derive(Deserialize, Debug)]
    struct FrontMatter {
        title: String,
        tags: Vec<String>
    }
	let mut result_with_struct = matter.parse_with_struct::<FrontMatter>(INPUT).unwrap();
	result_with_struct.data.title = "gray-matter-rs-modified".to_owned();
	let output = matter.compose(result_with_struct);
	println!("{:?}", output)
	// ---
	// title: gray-matter-rs-modified
	// tags:
	//  - gray-matter
	//  - rust
	// ---
	// Some excerpt
	// ---
	// Other stuff

}

[SeaDve]

i think that's a good api. I dont have any better idea

[kmaasrud]

We could implement the Display trait on ParsedEntity and ParsedEntityStruct, which would allow result_with_struct.to_string() to output the full input, but with changes reflected.

[yuchanns]

We could implement the Display trait on ParsedEntity and ParsedEntityStruct, which would allow result_with_struct.to_string() to output the full input, but with changes reflected.

That's great!

[kmaasrud]

I've set a new requirement for the Engine trait:

https://github.com/kmaasrud/gray-matter-rs/blob/1183ce888bc5473a524629da3b6ac9e7193ac4e4/src/engine.rs#L23

This requirement means every engine will need to define how the front matter they've parsed shall be output, e.g.:

https://github.com/kmaasrud/gray-matter-rs/blob/1183ce888bc5473a524629da3b6ac9e7193ac4e4/src/engine/toml.rs#L20-L23

This is in turn is used in the fmt function for ParsedEntity:

https://github.com/kmaasrud/gray-matter-rs/blob/1183ce888bc5473a524629da3b6ac9e7193ac4e4/src/entity.rs#L58-L66

I was not able to implement Display on ParsedEntityStruct, since we would need to define a trait limitation on what structs can be converted into a Pod, which quickly becomes very complex. JSON and TOML are up and running, and I'm gonna finish YAML soon.

[yuchanns]

I was not able to implement Display on ParsedEntityStruct, since we would need to define a trait limitation on what structs can be converted into a Pod, which quickly becomes very complex. JSON and TOML are up and running, and I'm gonna finish YAML soon.

Since now we can deserialized Pod into data implementing serde::de::DeserializeOwned in parse_with_struct, how about add an extra limitation trait serde::ser::Serialize? In that way we can easily implement Display on ParsedEntityStruct without making things complex, just serialize structs into Pod. Then we can smoothly replace all serd trait by ourselves once traits Deserialize and Serialize of Pod implemented.