dc09rs/src/markdown.rs

use std::{
    io::{BufWriter, Write},
    path::Path,
};

use pulldown_cmark::{CowStr, Options, Parser};

pub fn compile_markdown(
    src: impl AsRef<Path>,
    html: impl AsRef<Path>,
    gmi: impl AsRef<Path>,
) -> std::io::Result<()> {
    let src_text = std::fs::read_to_string(src)?;

    let mut html = create_file(html)?;
    let mut gmi = create_file(gmi)?;

    let mut state = State::Start;
    let mut ol = false; // ordered list
    let mut counter: u64 = 0;
    let mut links: Vec<GmiLink<'_>> = vec![];

    for event in Parser::new_ext(&src_text, Options::all()) {
        use pulldown_cmark::Event::*;

        dbg!(&event);

        match event {
            Start(tag) => {
                use pulldown_cmark::{CodeBlockKind, HeadingLevel, Tag::*};

                match tag {
                    Paragraph => {
                        html.write_all(b"<p>")?;
                        write_paragraph_start(&mut gmi, state)?;
                    }

                    Heading { level, id, .. } => {
                        if let Some(id) = id {
                            html.write_fmt(format_args!("<{} id=\"{}\">", level, id))?;
                        } else {
                            html.write_fmt(format_args!("<{}>", level))?;
                        }

                        write_paragraph_start(&mut gmi, state)?;
                        let hashes = match level {
                            HeadingLevel::H1 => "# ",
                            HeadingLevel::H2 => "## ",
                            _ => "### ",
                        };
                        gmi.write_all(hashes.as_bytes())?;
                        state = State::Paragraph;
                    }

                    BlockQuote(_kind) => {
                        html.write_all(b"<blockquote>")?;

                        write_paragraph_start(&mut gmi, state)?;
                        gmi.write_all(b"> ")?;
                        state = State::Quote;
                    }

                    CodeBlock(CodeBlockKind::Fenced(lang)) => {
                        // TODO: highlighting with syntect
                        html.write_all(b"<pre><code>")?;

                        write_paragraph_start(&mut gmi, state)?;
                        gmi.write_fmt(format_args!("```{}\r\n", lang))?;
                    }

                    CodeBlock(CodeBlockKind::Indented) => {
                        html.write_all(b"<pre><code>")?;

                        write_paragraph_start(&mut gmi, state)?;
                        gmi.write_all(b"```")?;
                    }

                    List(None) => {
                        html.write_all(b"<ul>")?;
                        write_paragraph_start(&mut gmi, state)?;
                        state = State::Start;
                    }

                    List(Some(counter_start)) => {
                        counter = counter_start;
                        ol = true;

                        if counter_start == 1 {
                            html.write_all(b"<ol>")?;
                        } else {
                            html.write_fmt(format_args!("<ol start=\"{}\">", counter_start))?;
                        }

                        write_paragraph_start(&mut gmi, state)?;
                        state = State::Start;
                    }

                    Item => {
                        html.write_all(b"<li>")?;

                        if state != State::Start {
                            gmi.write_all(b"\r\n")?;
                            state = State::Paragraph;
                        }
                        if ol {
                            gmi.write_fmt(format_args!("{}. ", counter))?;
                            counter += 1;
                        } else {
                            gmi.write_all(b"* ")?;
                        }
                    }

                    Strong => {
                        html.write_all(b"<b>")?;
                        write_inline(&mut gmi, state, "**", &mut links)?;
                    }

                    Emphasis => {
                        html.write_all(b"<i>")?;
                        write_inline(&mut gmi, state, "*", &mut links)?;
                    }

                    Strikethrough => {
                        html.write_all(b"<s>")?;
                        write_inline(&mut gmi, state, "~", &mut links)?;
                    }

                    Link { dest_url, .. } => {
                        html.write_fmt(format_args!("<a href=\"{}\">", dest_url))?;
                        links.push(GmiLink {
                            title: String::new(),
                            url: dest_url,
                        });
                        state = State::Link;
                    }

                    Image { dest_url, .. } => {
                        links.push(GmiLink {
                            title: String::new(),
                            url: dest_url,
                        });
                        state = State::Link;
                    }

                    Table(_align) => {
                        html.write_all(b"<table>")?;

                        write_paragraph_start(&mut gmi, state)?;
                        // gmi ??
                    }

                    TableHead => {
                        html.write_all(b"<thead><tr>")?;
                        state = State::TableHead;
                    }

                    TableRow => {
                        html.write_all(b"<tr>")?;
                    }

                    TableCell => {
                        if state == State::TableHead {
                            html.write_all(b"<th>")?;
                        } else {
                            html.write_all(b"<td>")?;
                        }
                    }

                    MetadataBlock(_) => {
                        state = State::Metadata(state != State::Start);
                    }

                    other => {
                        eprintln!("Unsupported tag: {:?}", other);
                    }
                }
            }

            End(tag) => {
                use pulldown_cmark::TagEnd::*;

                match tag {
                    Paragraph => {
                        html.write_all(b"</p>")?;

                        if !links.is_empty() {
                            gmi.write_all(b"\r\n")?;
                            for (i, link) in links.iter().enumerate() {
                                gmi.write_fmt(format_args!(
                                    "\r\n=> {} [{}]: {}", // => https://... [1]: example
                                    link.url, i, link.title,
                                ))?;
                            }
                            links.clear();
                        }
                    }
                    Heading(level) => html.write_fmt(format_args!("<{}>", level))?,
                    BlockQuote(_) => html.write_all(b"</blockquote>")?,
                    CodeBlock => {
                        html.write_all(b"</code></pre>")?;
                        gmi.write_all(b"```")?;
                    }
                    List(ordered) => {
                        if ordered {
                            html.write_all(b"</ol>")?;
                            ol = false;
                        } else {
                            html.write_all(b"</ul>")?;
                        }
                    }
                    Item => html.write_all(b"</li>")?,
                    Strong => {
                        html.write_all(b"</b>")?;
                        write_inline(&mut gmi, state, "**", &mut links)?;
                    }
                    Emphasis => {
                        html.write_all(b"</i>")?;
                        write_inline(&mut gmi, state, "*", &mut links)?;
                    }
                    Strikethrough => {
                        html.write_all(b"</s>")?;
                        write_inline(&mut gmi, state, "~", &mut links)?;
                    }
                    Link | Image => {
                        gmi.write_fmt(format_args!(
                            "{}[{}]", // example[1] ...\r\n => https://... [1]: example
                            links.last().unwrap().title,
                            links.len(),
                        ))?;
                    }
                    TableHead => html.write_all(b"</tr></thead><tbody>")?,
                    TableRow => html.write_all(b"</tr>")?,
                    TableCell => {
                        if state == State::TableHead {
                            html.write_all(b"</th>")?
                        } else {
                            html.write_all(b"</td>")?
                        }
                    }
                    Table => html.write_all(b"</tbody></table>")?,
                    _ => {}
                }

                match state {
                    State::Metadata(false) => {
                        state = State::Start;
                    }
                    _ => {
                        state = State::Paragraph;
                    }
                }
            }

            Text(text) => {
                match state {
                    State::Metadata(_) => {
                        // TODO: parse yaml
                        continue;
                    }
                    _ => {}
                }

                html.write_all(text.as_bytes())?;
                write_inline(&mut gmi, state, &text, &mut links)?;
            }

            Code(code) => {
                html.write_all(b"<code>")?;
                html.write_all(code.as_bytes())?;
                html.write_all(b"</code>")?;

                write_inline(&mut gmi, state, "`", &mut links)?;
                write_inline(&mut gmi, state, &code, &mut links)?;
                write_inline(&mut gmi, state, "`", &mut links)?;
            }

            SoftBreak => {
                html.write_all(b" ")?;
                write_inline(&mut gmi, state, " ", &mut links)?;
            }

            HardBreak => {
                html.write_all(b"<br>")?;

                gmi.write_all(b"\r\n")?;
                if state == State::Quote {
                    gmi.write_all(b"> ")?;
                }
            }

            Rule => {
                html.write_all(b"<hr>")?;

                write_paragraph_start(&mut gmi, state)?;
                gmi.write_all(b"---")?;
            }

            TaskListMarker(done) => {
                if done {
                    html.write_all(b"<input type=checkbox checked disabled>")?;
                    write_inline(&mut gmi, state, "[x] ", &mut links)?;
                } else {
                    html.write_all(b"<input type=checkbox disabled>")?;
                    write_inline(&mut gmi, state, "[ ] ", &mut links)?;
                }
            }

            other => {
                eprintln!("Unsupported event: {:?}", other);
            }
        }
    }

    html.flush()?;
    gmi.flush()?;

    Ok(())
}

#[inline(always)]
fn create_file(path: impl AsRef<Path>) -> std::io::Result<BufWriter<std::fs::File>> {
    let f = std::fs::OpenOptions::new()
        .create(true)
        .write(true)
        .open(path)?;
    Ok(BufWriter::new(f))
}

#[inline(always)]
fn write_paragraph_start(gmi: &mut impl std::io::Write, state: State) -> std::io::Result<()> {
    if state != State::Start && state != State::Quote {
        gmi.write_all(b"\r\n\r\n")?;
    }
    Ok(())
}

#[inline(always)]
fn write_inline(
    gmi: &mut impl std::io::Write,
    state: State,
    text: &str,
    links: &mut Vec<GmiLink>,
) -> std::io::Result<()> {
    if state == State::Link {
        links.last_mut().unwrap().title.push_str(text);
    } else {
        gmi.write_all(text.as_bytes())?;
    }
    Ok(())
}

#[derive(Debug, Clone, Copy, PartialEq, Eq)]
enum State {
    Start,
    Paragraph,
    Link,
    Quote,
    TableHead,
    Metadata(bool),
}

struct GmiLink<'link> {
    title: String,
    url: CowStr<'link>,
}