mylloon.fr/src/misc/markdown.rs

use crate::misc::date::Date;
use comrak::nodes::{AstNode, NodeValue};
use comrak::{
    format_html, parse_document, Arena, ComrakExtensionOptions, ComrakOptions, ComrakParseOptions,
    ComrakRenderOptions, ListStyleType,
};
use ramhorns::Content;
use serde::Deserialize;
use std::fs;

#[derive(Default, Deserialize, Content)]
pub struct FileMetadata {
    pub title: Option<String>,
    pub link: Option<String>,
    pub date: Option<Date>,
}

#[derive(Content)]
pub struct Metadata {
    pub info: FileMetadata,
    pub mermaid: bool,
    pub syntax_highlight: bool,
}

#[derive(Content)]
pub struct File {
    pub metadata: Metadata,
    pub content: String,
}

/// Options used for parser and compiler MD --> HTML
pub fn get_options() -> ComrakOptions {
    ComrakOptions {
        extension: ComrakExtensionOptions {
            strikethrough: true,
            tagfilter: true,
            table: true,
            autolink: true,
            tasklist: true,
            superscript: true,
            header_ids: Some(String::new()),
            footnotes: false, // true?
            description_lists: true,
            front_matter_delimiter: Some("---".to_owned()),
        },
        parse: ComrakParseOptions {
            smart: false, // true one day?
            default_info_string: None,
            relaxed_tasklist_matching: true,
        },
        render: ComrakRenderOptions {
            hardbreaks: false, // could be true? change by metadata could be good for compatibility
            github_pre_lang: false,
            full_info_string: true,
            width: 0, // 0 mean disabled?
            unsafe_: true,
            escape: false, // may change in the future?
            list_style: ListStyleType::Dash,
            sourcepos: false,
        },
    }
}

pub fn read(raw_text: &str) -> File {
    let arena = Arena::new();

    // LaTeX conversion to MathML
    let text = latex2mathml::replace(raw_text).unwrap_or_default();

    let options = get_options();
    let root = parse_document(&arena, &text, &options);

    // Find metadata
    let metadata = get_metadata(root);

    // Convert to HTML
    let mut html = vec![];
    format_html(root, &options, &mut html).unwrap();

    let mermaid_name = "mermaid";

    File {
        metadata: Metadata {
            info: metadata,
            mermaid: check_mermaid(root, mermaid_name.to_owned()),
            syntax_highlight: check_code(root, &[mermaid_name.to_owned()]),
        },
        content: String::from_utf8(html).unwrap(),
    }
}

/// Read markdown file
pub fn read_file(filename: &str) -> Option<File> {
    match fs::read_to_string(filename) {
        Ok(text) => Some(read(&text)),
        _ => None,
    }
}

/// Fetch metadata from AST
pub fn get_metadata<'a>(root: &'a AstNode<'a>) -> FileMetadata {
    match root
        .children()
        .find_map(|node| match &node.data.borrow().value {
            NodeValue::FrontMatter(text) => {
                // '-' correspond to `front_matter_delimiter`
                serde_yaml::from_str(text.trim_matches(&['-', '\n'] as &[_])).unwrap_or_default()
            }
            _ => None,
        }) {
        Some(data) => data,
        None => FileMetadata::default(),
    }
}

/// Recursively check whether mermaid diagrams are in the AST
fn check_mermaid<'a>(root: &'a AstNode<'a>, mermaid_str: String) -> bool {
    root.children().any(|node| match &node.data.borrow().value {
        // Check if code of block define a mermaid diagram
        NodeValue::CodeBlock(code_block) => code_block.info == mermaid_str,
        _ => false,
    })
}

/// Recursively check if code is in the AST
fn check_code<'a>(root: &'a AstNode<'a>, blacklist: &[String]) -> bool {
    root.children().any(|node| match &node.data.borrow().value {
        // Detect code in paragraph
        /* NodeValue::Paragraph => match &node.children().next() {
            Some(child) => matches!(child.data.borrow().value, NodeValue::Code(_)),
            None => false,
        }, */
        // Detect blocks of code where the lang isn't in the blacklist
        NodeValue::CodeBlock(code_block) => !blacklist.contains(&code_block.info),
        _ => false,
    })
}