devup-editor-markdown 1.0.1

use devup_editor_core::{Block, BlockId, Document, IdGenerator, Mark, TextSpan};
use pulldown_cmark::{Event, HeadingLevel, Options, Parser, Tag, TagEnd};
use serde_json::{Map, Value};

/// Parse a Markdown string into a [`Document`].
///
/// Supported: paragraphs, headings (h1-h3), lists, todo items (`- [ ]`),
/// blockquotes, fenced code blocks, bold/italic/code/strike marks.
pub fn parse_markdown(src: &str, id_gen: &mut dyn IdGenerator) -> Document {
    let mut doc = Document::new();
    let mut builder = DocBuilder::new(&mut doc, id_gen);

    let mut opts = Options::empty();
    opts.insert(Options::ENABLE_TASKLISTS);
    opts.insert(Options::ENABLE_STRIKETHROUGH);

    let parser = Parser::new_ext(src, opts);
    for event in parser {
        builder.handle(event);
    }
    builder.finish();
    doc
}

/// Active mark stack while walking inline events.
#[derive(Default)]
struct InlineState {
    bold: u32,
    italic: u32,
    code: u32,
    strike: u32,
    html_marks: Vec<Vec<Mark>>,
    spans: Vec<TextSpan>,
}

impl InlineState {
    fn current_marks(&self) -> Vec<Mark> {
        let mut marks = Vec::new();
        if self.bold > 0 {
            marks.push(Mark::new("bold"));
        }
        if self.italic > 0 {
            marks.push(Mark::new("italic"));
        }
        if self.code > 0 {
            marks.push(Mark::new("code"));
        }
        if self.strike > 0 {
            marks.push(Mark::new("strike"));
        }
        for html_marks in &self.html_marks {
            marks.extend(html_marks.clone());
        }
        marks
    }

    fn push_text(&mut self, text: &str) {
        if text.is_empty() {
            return;
        }
        let marks = self.current_marks();
        // Merge with previous span if same marks
        if let Some(last) = self.spans.last_mut() {
            if last.marks == marks {
                last.text.push_str(text);
                return;
            }
        }
        self.spans.push(TextSpan {
            text: text.to_string(),
            marks,
        });
    }

    fn take(&mut self) -> Vec<TextSpan> {
        std::mem::take(&mut self.spans)
    }
}

/// Tracks which kind of block we're currently building.
enum CurrentBlock {
    None,
    Paragraph,
    Heading,
    Code {
        language: Option<String>,
    },
    Quote,
    /// list item: (ordered?, indent level, todo state)
    ListItem {
        ordered: bool,
        indent: usize,
        todo: Option<bool>,
    },
}

struct DocBuilder<'a> {
    doc: &'a mut Document,
    id_gen: &'a mut dyn IdGenerator,
    inline: InlineState,
    current: CurrentBlock,
    /// Nesting depth of `<list>` tags from pulldown-cmark.
    list_depth: usize,
    /// Whether each open list is ordered (LIFO stack).
    list_ordered_stack: Vec<bool>,
}

impl<'a> DocBuilder<'a> {
    fn new(doc: &'a mut Document, id_gen: &'a mut dyn IdGenerator) -> Self {
        Self {
            doc,
            id_gen,
            inline: InlineState::default(),
            current: CurrentBlock::None,
            list_depth: 0,
            list_ordered_stack: Vec::new(),
        }
    }

    fn handle(&mut self, event: Event<'_>) {
        match event {
            Event::Start(tag) => self.start(tag),
            Event::End(tag) => self.end(tag),
            Event::Text(t) => self.inline.push_text(&t),
            Event::Html(html) | Event::InlineHtml(html) => self.handle_inline_html(&html),
            Event::Code(t) => {
                self.inline.code += 1;
                self.inline.push_text(&t);
                self.inline.code -= 1;
            }
            Event::SoftBreak | Event::HardBreak => self.inline.push_text("\n"),
            Event::TaskListMarker(checked) => {
                if let CurrentBlock::ListItem { todo, .. } = &mut self.current {
                    *todo = Some(checked);
                }
            }
            _ => {}
        }
    }

    fn start(&mut self, tag: Tag<'_>) {
        match tag {
            Tag::Paragraph => {
                if matches!(self.current, CurrentBlock::None) {
                    self.current = CurrentBlock::Paragraph;
                }
            }
            Tag::Heading { .. } => {
                self.current = CurrentBlock::Heading;
            }
            Tag::CodeBlock(kind) => {
                let language = match kind {
                    pulldown_cmark::CodeBlockKind::Fenced(lang) if !lang.is_empty() => {
                        Some(lang.into_string())
                    }
                    _ => None,
                };
                self.current = CurrentBlock::Code { language };
            }
            Tag::BlockQuote(_) => {
                self.current = CurrentBlock::Quote;
            }
            Tag::List(start) => {
                self.list_depth += 1;
                self.list_ordered_stack.push(start.is_some());
            }
            Tag::Item => {
                let ordered = *self.list_ordered_stack.last().unwrap_or(&false);
                self.current = CurrentBlock::ListItem {
                    ordered,
                    indent: self.list_depth.saturating_sub(1),
                    todo: None,
                };
            }
            Tag::Strong => self.inline.bold += 1,
            Tag::Emphasis => self.inline.italic += 1,
            Tag::Strikethrough => self.inline.strike += 1,
            _ => {}
        }
    }

    fn end(&mut self, tag: TagEnd) {
        match tag {
            TagEnd::Paragraph => {
                if matches!(self.current, CurrentBlock::Paragraph) {
                    self.flush_block("paragraph", Map::new());
                }
            }
            TagEnd::Heading(level) => {
                if matches!(self.current, CurrentBlock::Heading) {
                    let mut props = Map::new();
                    props.insert("level".into(), Value::from(u64::from(heading_level(level))));
                    self.flush_block("heading", props);
                }
            }
            TagEnd::CodeBlock => {
                if let CurrentBlock::Code { language } = &self.current {
                    let mut props = Map::new();
                    if let Some(lang) = language.clone() {
                        props.insert("language".into(), Value::String(lang));
                    }
                    self.flush_block("code", props);
                }
            }
            TagEnd::BlockQuote(_) => {
                self.flush_block("quote", Map::new());
            }
            TagEnd::List(_) => {
                self.list_depth = self.list_depth.saturating_sub(1);
                self.list_ordered_stack.pop();
            }
            TagEnd::Item => {
                if let CurrentBlock::ListItem {
                    ordered,
                    indent,
                    todo,
                } = self.current
                {
                    let mut props = Map::new();
                    let block_type = if let Some(checked) = todo {
                        props.insert("checked".into(), Value::Bool(checked));
                        if indent > 0 {
                            props.insert(
                                "indent".into(),
                                Value::from(i64::try_from(indent).unwrap_or(i64::MAX)),
                            );
                        }
                        "todo"
                    } else {
                        let style = if ordered { "ordered" } else { "unordered" };
                        props.insert("style".into(), Value::String(style.into()));
                        if indent > 0 {
                            props.insert(
                                "indent".into(),
                                Value::from(i64::try_from(indent).unwrap_or(i64::MAX)),
                            );
                        }
                        "list"
                    };
                    self.flush_block(block_type, props);
                }
            }
            TagEnd::Strong => self.inline.bold = self.inline.bold.saturating_sub(1),
            TagEnd::Emphasis => self.inline.italic = self.inline.italic.saturating_sub(1),
            TagEnd::Strikethrough => self.inline.strike = self.inline.strike.saturating_sub(1),
            _ => {}
        }
    }

    fn flush_block(&mut self, ty: &str, props: Map<String, Value>) {
        let spans = self.inline.take();
        let id = BlockId::new(self.id_gen.next_id().to_string());
        let mut block = Block::new(id, ty);
        block.content = spans;
        block.props = props;
        self.doc.push_root_block(block);
        self.current = CurrentBlock::None;
    }

    fn finish(&mut self) {
        // If a block is still in progress (e.g. trailing paragraph), flush it.
        if !matches!(self.current, CurrentBlock::None) {
            self.flush_block("paragraph", Map::new());
        }
    }

    fn handle_inline_html(&mut self, html: &str) {
        match parse_html_span_marks(html) {
            HtmlSpanMarks::Push(marks) => self.inline.html_marks.push(marks),
            HtmlSpanMarks::Pop => {
                self.inline.html_marks.pop();
            }
            HtmlSpanMarks::Ignore => {}
        }
    }
}

enum HtmlSpanMarks {
    Push(Vec<Mark>),
    Pop,
    Ignore,
}

fn parse_html_span_marks(html: &str) -> HtmlSpanMarks {
    let trimmed = html.trim();
    let lower = trimmed.to_lowercase();
    if lower.starts_with("</span") {
        return HtmlSpanMarks::Pop;
    }
    if !lower.starts_with("<span") {
        return HtmlSpanMarks::Ignore;
    }

    let Some(style_attr) = extract_html_attr(trimmed, "style") else {
        return HtmlSpanMarks::Ignore;
    };

    let mut style_map = Map::new();
    for decl in style_attr.split(';') {
        let Some((raw_key, raw_value)) = decl.split_once(':') else {
            continue;
        };
        let key = raw_key.trim().to_ascii_lowercase();
        let value = raw_value.trim();
        if value.is_empty() {
            continue;
        }
        match key.as_str() {
            "color" => {
                style_map.insert("color".into(), Value::String(value.into()));
            }
            "background-color" => {
                style_map.insert("backgroundColor".into(), Value::String(value.into()));
            }
            _ => {}
        }
    }

    if style_map.is_empty() {
        return HtmlSpanMarks::Ignore;
    }

    let mut marks = Vec::new();
    if let Some(color) = style_map.get("color").cloned() {
        let mut attrs = Map::new();
        let mut style = Map::new();
        style.insert("color".into(), color);
        attrs.insert("style".into(), Value::Object(style));
        marks.push(Mark::with_attrs("color", attrs));
    }
    if let Some(background) = style_map.get("backgroundColor").cloned() {
        let mut attrs = Map::new();
        let mut style = Map::new();
        style.insert("backgroundColor".into(), background);
        attrs.insert("style".into(), Value::Object(style));
        marks.push(Mark::with_attrs("highlight", attrs));
    }

    if marks.is_empty() {
        HtmlSpanMarks::Ignore
    } else {
        HtmlSpanMarks::Push(marks)
    }
}

fn extract_html_attr(tag: &str, attr_name: &str) -> Option<String> {
    let lower = tag.to_ascii_lowercase();
    let needle = format!("{attr_name}=");
    let start = lower.find(&needle)? + needle.len();
    let rest = &tag[start..];
    let quote = rest.chars().next()?;
    if quote != '"' && quote != '\'' {
        return None;
    }
    let rest = &rest[1..];
    let end = rest.find(quote)?;
    Some(rest[..end].to_string())
}

fn heading_level(level: HeadingLevel) -> u8 {
    match level {
        HeadingLevel::H1 => 1,
        HeadingLevel::H2 => 2,
        HeadingLevel::H3 => 3,
        HeadingLevel::H4 => 4,
        HeadingLevel::H5 => 5,
        HeadingLevel::H6 => 6,
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use devup_editor_core::SequentialIdGenerator;

    fn parse(md: &str) -> Document {
        let mut id_gen = SequentialIdGenerator::new("test");
        parse_markdown(md, &mut id_gen)
    }

    #[test]
    fn parse_paragraph() {
        let doc = parse("hello world");
        assert_eq!(doc.root_block_count(), 1);
        let block = doc.get_block(&BlockId::new("test-1")).unwrap();
        assert_eq!(block.ty, "paragraph");
        assert_eq!(block.plain_text(), "hello world");
    }

    #[test]
    fn parse_heading() {
        let doc = parse("# Title");
        let block = doc.get_block(&BlockId::new("test-1")).unwrap();
        assert_eq!(block.ty, "heading");
        assert_eq!(
            block.props.get("level").and_then(serde_json::Value::as_u64),
            Some(1)
        );
    }

    #[test]
    fn parse_bold_italic() {
        let doc = parse("**bold** and *italic*");
        let block = doc.get_block(&BlockId::new("test-1")).unwrap();
        assert!(
            block
                .content
                .iter()
                .any(|s| s.marks.iter().any(|m| m.ty == "bold"))
        );
        assert!(
            block
                .content
                .iter()
                .any(|s| s.marks.iter().any(|m| m.ty == "italic"))
        );
    }

    #[test]
    fn parse_inline_html_color_and_highlight_marks() {
        let doc = parse("<span style=\"color:#ff0000;background-color:#fff000\">가나</span>");
        let block = doc.get_block(&BlockId::new("test-1")).unwrap();
        assert_eq!(block.content.len(), 1);
        let marks = &block.content[0].marks;
        assert!(marks.iter().any(|m| m.ty == "color"));
        assert!(marks.iter().any(|m| m.ty == "highlight"));
    }

    #[test]
    fn parse_todo() {
        let doc = parse("- [x] done\n- [ ] todo");
        assert_eq!(doc.root_block_count(), 2);
        let first = doc.get_block(&BlockId::new("test-1")).unwrap();
        assert_eq!(first.ty, "todo");
        assert_eq!(
            first
                .props
                .get("checked")
                .and_then(serde_json::Value::as_bool),
            Some(true)
        );
    }

    #[test]
    fn parse_nested_todo_preserves_indent() {
        let doc = parse("- [ ] parent\n  - [x] child");
        let block = doc
            .root_block_ids()
            .iter()
            .filter_map(|id| doc.get_block(id))
            .find(|block| {
                block.ty == "todo"
                    && block
                        .props
                        .get("indent")
                        .and_then(serde_json::Value::as_i64)
                        == Some(1)
                    && block
                        .props
                        .get("checked")
                        .and_then(serde_json::Value::as_bool)
                        == Some(true)
            })
            .unwrap();
        assert_eq!(block.ty, "todo");
        assert_eq!(
            block
                .props
                .get("indent")
                .and_then(serde_json::Value::as_i64),
            Some(1)
        );
        assert_eq!(
            block
                .props
                .get("checked")
                .and_then(serde_json::Value::as_bool),
            Some(true)
        );
    }

    #[test]
    fn parse_fenced_code_block() {
        let doc = parse("```rust\nfn main() {}\n```");
        let block = doc.get_block(&BlockId::new("test-1")).unwrap();
        assert_eq!(block.ty, "code");
        assert_eq!(
            block.props.get("language").and_then(|v| v.as_str()),
            Some("rust")
        );
        assert!(block.plain_text().contains("fn main"));
    }

    #[test]
    fn parse_code_block_without_language() {
        let doc = parse("```\nplain code\n```");
        let block = doc.get_block(&BlockId::new("test-1")).unwrap();
        assert_eq!(block.ty, "code");
        assert!(block.props.get("language").is_none());
    }
}