mlem/src/index.rs

use derive_more::Constructor;
use serde::Serialize;
use serde_json::value::Value;
use std::collections::HashMap;
use tera::{Context, Tera};

#[derive(Constructor, Debug, Serialize)]
pub struct BlogPost {
    pub title: String,
    pub human_date: String,
    pub sort_date: i64,  // Sort date = unix timestamp
    pub content: String, // Unformatted Content of blog post can be used to display a preview or reading time estimate. Will be html when assigned but later turned into raw text
    pub output_file_name: String,
}

pub fn generate(mut blog_posts: Vec<BlogPost>, template_dir: &String) -> String {
    for post in &mut blog_posts {
        post.content = get_unformatted_text(post.content.clone());
    }

    let mut tera = Tera::new(&format!("{}/*", template_dir)).unwrap();
    tera.autoescape_on(vec![]);
    tera.register_filter("truncate", truncate);
    let mut context = Context::new();
    context.insert("blog_posts", &blog_posts);
    tera.render("index.html", &context).unwrap()
}

fn get_unformatted_text(html: String) -> String {
    let frag = scraper::Html::parse_fragment(&html);
    let mut unformatted_text = String::new();
    for node in frag.tree {
        if let scraper::node::Node::Text(text) = node {
            unformatted_text.push_str(&text);
        }
    }
    unformatted_text
}

fn truncate(value: &Value, args: &HashMap<String, Value>) -> Result<Value, tera::Error> {
    let mut value = value.as_str().unwrap().to_string();
    let new_len: usize = args.get("len").unwrap().as_str().unwrap().parse().unwrap();
    value.truncate(new_len);
    Ok(Value::String(value.to_string()))
}

#[cfg(test)]
mod tests {
    use crate::index::*;
    #[test]
    fn truncate_shortens() {
        let mut args: HashMap<String, Value> = HashMap::new();
        args.insert("len".to_string(), Value::String("4".to_string()));
        let truncated_string = truncate(&Value::String("Meow Nya".to_string()), &args).unwrap();
        assert_eq!(truncated_string.as_str().unwrap(), "Meow");
    }

    #[test]
    fn unformat_html() {
        let html_src = "<p>Meow nyaaa<em> UwU</em></p>".to_string();
        let unformatted_text = get_unformatted_text(html_src);
        assert_eq!(unformatted_text, "Meow nyaaa UwU");
    }
}