textlane-html 0.6.0

use serde_json::{Map, Value, json};
use textlane_core::hast::{Node, PropertyValue};

use crate::helpers::property::find_property;
use crate::helpers::schema::HTML;
use crate::helpers::value::write_value;
use crate::models::options::ExtractToc;

/// The headings of a tree as @stefanprobst/rehype-extract-toc nests them: each
/// `{depth, value, id?, children?}`, under the closest heading of a lower rank before it.
pub fn extract_toc(node: &Node<'_>, options: &ExtractToc) -> Value {
    let skipped: Vec<(String, Option<&str>)> = (options.skip.iter())
        .map(|rule| match rule.split_once('=') {
            Some((name, value)) => (read_attribute(name), Some(value)),
            None => (read_attribute(rule), None),
        })
        .collect();
    let min = options.min_depth.unwrap_or(1);
    let max = options.max_depth.unwrap_or(6);
    let mut headings = Vec::new();
    let mut stack: Vec<(&Node<'_>, bool)> = vec![(node, false)];
    while let Some((node, inside)) = stack.pop() {
        let (children, inside) = match node {
            Node::Root(root) => (&root.children, inside),
            Node::Element(element) => {
                let left_out = inside
                    || element.properties.iter().any(|property| {
                        let attribute = read_attribute(property.name);
                        skipped.iter().any(|(name, value)| {
                            *name == attribute
                                && match value {
                                    Some(value) => write_value(&property.value) == *value,
                                    None => !matches!(
                                        property.value,
                                        PropertyValue::Null
                                            | PropertyValue::Boolean(false)
                                    ),
                                }
                        })
                    });
                let depth = read_rank(element.tag_name)
                    .filter(|depth| !left_out && (min..=max).contains(depth));
                if let Some(depth) = depth {
                    let mut heading = Map::new();
                    heading.insert("depth".into(), json!(depth));
                    heading.insert("value".into(), json!(read_text(node)));
                    if let Some(id) = read_id(element) {
                        heading.insert("id".into(), id);
                    }
                    headings.push((depth, heading));
                }
                let inside = inside
                    || (options.skip_inside.iter())
                        .any(|name| name.eq_ignore_ascii_case(element.tag_name));
                (&element.children, inside)
            }
            _ => continue,
        };
        stack.extend(children.iter().rev().map(|child| (child, inside)));
    }
    nest_headings(headings)
}

// `createTree`: a heading deeper than the one before opens it as a parent; a shallower one
// closes parents down to one shallower than itself
fn nest_headings(headings: Vec<(u8, Map<String, Value>)>) -> Value {
    let mut parents: Vec<(u8, Map<String, Value>, Vec<Value>)> =
        vec![(0, Map::new(), vec![])];
    let mut previous_depth = 0;
    for (depth, heading) in headings {
        if depth > previous_depth {
            // the previous heading becomes a parent once something nests in it
        } else {
            while parents.len() > 1 && parents.last().is_some_and(|(d, _, _)| *d >= depth)
            {
                close_parent(&mut parents);
            }
        }
        parents.push((depth, heading, vec![]));
        previous_depth = depth;
    }
    while parents.len() > 1 {
        close_parent(&mut parents);
    }
    let root = parents
        .pop()
        .map(|(_, _, children)| children)
        .unwrap_or_default();
    Value::Array(root)
}

// The innermost open heading joins its parent's children, with its own when it has any
fn close_parent(parents: &mut Vec<(u8, Map<String, Value>, Vec<Value>)>) {
    let Some((_, mut heading, children)) = parents.pop() else {
        return;
    };
    if !children.is_empty() {
        heading.insert("children".into(), Value::Array(children));
    }
    if let Some((_, _, siblings)) = parents.last_mut() {
        siblings.push(Value::Object(heading));
    }
}

// hast-util-heading-rank
fn read_rank(tag_name: &str) -> Option<u8> {
    match tag_name.to_ascii_lowercase().as_bytes() {
        [b'h', digit @ b'1'..=b'6'] => Some(digit - b'0'),
        _ => None,
    }
}

// hast-util-to-string: the text of every descendant, in order
fn read_text(node: &Node<'_>) -> String {
    let mut text = String::new();
    let mut stack = vec![node];
    while let Some(node) = stack.pop() {
        match node {
            Node::Text(node) => text.push_str(node.value),
            Node::Element(element) => stack.extend(element.children.iter().rev()),
            Node::Root(root) => stack.extend(root.children.iter().rev()),
            _ => {}
        }
    }
    text
}

fn read_id(element: &textlane_core::hast::Element<'_>) -> Option<Value> {
    let property = element
        .properties
        .iter()
        .find(|property| property.name == "id")?;
    match &property.value {
        PropertyValue::String(id) => Some(json!(id)),
        PropertyValue::Number(id) => Some(json!(id)),
        PropertyValue::Boolean(id) => Some(json!(id)),
        PropertyValue::List(_) | PropertyValue::Null => None,
    }
}

// The HTML attribute a property or attribute name stands for, `dataStep` and `data-step` alike
fn read_attribute(name: &str) -> String {
    find_property(&HTML, name).attribute.to_lowercase()
}