tiptap-rusty-parser 0.6.2

//! Core data model: [`Node`] and [`Mark`].
//!
//! Mirrors Tiptap's `JSONContent` interface. Schema-agnostic: any node/mark
//! `type` is accepted and unknown JSON fields are preserved in `extra` for
//! faithful roundtrip.

use serde::de::{self, DeserializeSeed, Deserializer, MapAccess, SeqAccess, Visitor};
use serde::ser::{SerializeMap, Serializer};
use serde::{Deserialize, Serialize};
use serde_json::{Map, Value};
use std::fmt;
use std::marker::PhantomData;

/// A single Tiptap/ProseMirror node.
///
/// All structural fields are optional to faithfully distinguish "missing" from
/// "empty" on roundtrip. Unknown top-level keys land in [`Node::extra`].
///
/// `attrs` and `extra` are boxed so the common attr-less, extra-less node stays
/// small (one pointer each); `extra: None` and `Some(empty)` compare equal and
/// serialize identically.
///
/// Serde: serialized as a JSON object with keys in the order `type`, `attrs`,
/// `content`, `marks`, `text` (each only when `Some`), then `extra` in its own
/// order. On deserialize, a `null` known field reads as `None`, a repeated
/// known field is a `duplicate field` error, and a repeated unknown key keeps
/// its first position with the last value.
#[derive(Debug, Clone, Default)]
pub struct Node {
    /// Node type, e.g. `"doc"`, `"paragraph"`, `"text"`.
    pub node_type: Option<String>,

    /// Node attributes.
    pub attrs: Option<Box<Map<String, Value>>>,

    /// Child nodes.
    pub content: Option<Vec<Node>>,

    /// Marks applied to this node (typically text nodes).
    pub marks: Option<Vec<Mark>>,

    /// Text payload (text nodes only).
    pub text: Option<String>,

    /// Any unknown/extra top-level fields, preserved verbatim (`None` = none).
    pub extra: Option<Box<Map<String, Value>>>,
}

/// A mark applied to a node (e.g. `bold`, `italic`, `link`).
///
/// Serde: `type` (required, string), then `attrs` (when `Some`), then `extra`;
/// same `null`/duplicate rules as [`Node`].
#[derive(Debug, Clone)]
pub struct Mark {
    /// Mark type, e.g. `"bold"`.
    pub mark_type: String,

    /// Mark attributes (e.g. `href` for a link).
    pub attrs: Option<Box<Map<String, Value>>>,

    /// Any unknown/extra fields, preserved verbatim (`None` = none).
    pub extra: Option<Box<Map<String, Value>>>,
}

impl Mark {
    /// Create a mark of `mark_type` with no attrs.
    ///
    /// ```
    /// use tiptap_rusty_parser::Mark;
    /// let m = Mark::new("bold");
    /// assert_eq!(m.mark_type, "bold");
    /// ```
    #[inline]
    pub fn new(mark_type: impl Into<String>) -> Self {
        Mark {
            mark_type: mark_type.into(),
            attrs: None,
            extra: None,
        }
    }

    /// Builder: set one attr.
    #[inline]
    pub fn attr(mut self, key: impl Into<String>, value: impl Into<Value>) -> Self {
        self.attrs
            .get_or_insert_with(Default::default)
            .insert(key.into(), value.into());
        self
    }

    /// Attrs map, if any.
    #[inline]
    pub fn attrs(&self) -> Option<&Map<String, Value>> {
        self.attrs.as_deref()
    }

    /// Mutable attrs map, creating it if absent.
    #[inline]
    pub fn attrs_mut(&mut self) -> &mut Map<String, Value> {
        self.attrs.get_or_insert_with(Default::default)
    }

    /// Unknown/extra fields, if any (may be `Some(empty)`).
    #[inline]
    pub fn extra(&self) -> Option<&Map<String, Value>> {
        self.extra.as_deref()
    }

    /// Mutable extra map, creating it if absent.
    #[inline]
    pub fn extra_mut(&mut self) -> &mut Map<String, Value> {
        self.extra.get_or_insert_with(Default::default)
    }
}

impl Node {
    /// Attrs map, if any.
    #[inline]
    pub fn attrs(&self) -> Option<&Map<String, Value>> {
        self.attrs.as_deref()
    }

    /// Unknown/extra fields, if any (may be `Some(empty)`).
    #[inline]
    pub fn extra(&self) -> Option<&Map<String, Value>> {
        self.extra.as_deref()
    }

    /// Mutable extra map, creating it if absent.
    #[inline]
    pub fn extra_mut(&mut self) -> &mut Map<String, Value> {
        self.extra.get_or_insert_with(Default::default)
    }

    /// Extra field `key`.
    #[inline]
    pub fn extra_field(&self, key: &str) -> Option<&Value> {
        self.extra.as_ref()?.get(key)
    }

    /// Remove extra field `key`, dropping the map once empty.
    pub fn remove_extra(&mut self, key: &str) -> Option<Value> {
        let m = self.extra.as_mut()?;
        let prev = m.remove(key);
        if m.is_empty() {
            self.extra = None;
        }
        prev
    }
}

/// `None` and `Some(empty)` are the same extra set.
#[inline]
pub(crate) fn extra_eq(
    a: &Option<Box<Map<String, Value>>>,
    b: &Option<Box<Map<String, Value>>>,
) -> bool {
    match (a, b) {
        (None, None) => true,
        (Some(x), Some(y)) => x == y,
        (Some(m), None) | (None, Some(m)) => m.is_empty(),
    }
}

impl PartialEq for Node {
    #[inline]
    fn eq(&self, o: &Node) -> bool {
        self.node_type == o.node_type
            && self.text == o.text
            && self.attrs == o.attrs
            && self.marks == o.marks
            && extra_eq(&self.extra, &o.extra)
            && self.content == o.content
    }
}

impl PartialEq for Mark {
    #[inline]
    fn eq(&self, o: &Mark) -> bool {
        self.mark_type == o.mark_type && self.attrs == o.attrs && extra_eq(&self.extra, &o.extra)
    }
}

// Hand-written serde: one pass over the map, known keys straight into typed
// fields, the rest into `extra`. Avoids `#[serde(flatten)]`, which buffers
// every unknown entry through serde's private `Content` tree.

/// Object key: a known field (matched without allocating) or an unknown key.
enum Key {
    Type,
    Attrs,
    Content,
    Marks,
    Text,
    Other(String),
}

impl Key {
    #[inline]
    fn known(s: &str) -> Option<Key> {
        Some(match s {
            "type" => Key::Type,
            "attrs" => Key::Attrs,
            "content" => Key::Content,
            "marks" => Key::Marks,
            "text" => Key::Text,
            _ => return None,
        })
    }
}

struct KeyVisitor;

impl<'de> Visitor<'de> for KeyVisitor {
    type Value = Key;

    fn expecting(&self, f: &mut fmt::Formatter) -> fmt::Result {
        f.write_str("field identifier")
    }

    #[inline]
    fn visit_str<E: de::Error>(self, v: &str) -> Result<Key, E> {
        Ok(Key::known(v).unwrap_or_else(|| Key::Other(v.to_owned())))
    }

    #[inline]
    fn visit_string<E: de::Error>(self, v: String) -> Result<Key, E> {
        Ok(Key::known(&v).unwrap_or(Key::Other(v)))
    }

    fn visit_bytes<E: de::Error>(self, v: &[u8]) -> Result<Key, E> {
        match std::str::from_utf8(v) {
            Ok(s) => self.visit_str(s),
            Err(_) => Err(E::invalid_value(de::Unexpected::Bytes(v), &self)),
        }
    }
}

impl<'de> Deserialize<'de> for Key {
    #[inline]
    fn deserialize<D: Deserializer<'de>>(d: D) -> Result<Key, D::Error> {
        d.deserialize_identifier(KeyVisitor)
    }
}

/// Error on a repeated known field (matches the derive's behaviour).
#[inline]
fn once<E: de::Error>(seen: &mut u8, bit: u8, name: &'static str) -> Result<(), E> {
    if *seen & bit != 0 {
        return Err(E::duplicate_field(name));
    }
    *seen |= bit;
    Ok(())
}

/// Scratch stacks shared by one deserialize call. Children and marks collect
/// here, then move into exactly sized `Vec`s: serde's seq visitor grows by
/// doubling (JSON gives no length), which left ~2x the needed capacity.
#[derive(Default)]
struct Scratch {
    nodes: Vec<Node>,
    marks: Vec<Mark>,
}

impl<'de> Deserialize<'de> for Node {
    #[inline]
    fn deserialize<D: Deserializer<'de>>(d: D) -> Result<Node, D::Error> {
        NodeSeed(&mut Scratch::default()).deserialize(d)
    }
}

struct NodeSeed<'a>(&'a mut Scratch);

impl<'de> DeserializeSeed<'de> for NodeSeed<'_> {
    type Value = Node;

    #[inline]
    fn deserialize<D: Deserializer<'de>>(self, d: D) -> Result<Node, D::Error> {
        d.deserialize_map(self)
    }
}

impl<'de> Visitor<'de> for NodeSeed<'_> {
    type Value = Node;

    fn expecting(&self, f: &mut fmt::Formatter) -> fmt::Result {
        f.write_str("struct Node")
    }

    fn visit_map<A: MapAccess<'de>>(self, mut map: A) -> Result<Node, A::Error> {
        let mut n = Node::default();
        let mut seen = 0u8;
        while let Some(key) = map.next_key::<Key>()? {
            match key {
                Key::Type => {
                    once(&mut seen, 1, "type")?;
                    n.node_type = map.next_value()?;
                }
                Key::Attrs => {
                    once(&mut seen, 2, "attrs")?;
                    n.attrs = map.next_value()?;
                }
                Key::Content => {
                    once(&mut seen, 4, "content")?;
                    n.content = map.next_value_seed(OptVec::<Node>(self.0, PhantomData))?;
                }
                Key::Marks => {
                    once(&mut seen, 8, "marks")?;
                    n.marks = map.next_value_seed(OptVec::<Mark>(self.0, PhantomData))?;
                }
                Key::Text => {
                    once(&mut seen, 16, "text")?;
                    n.text = map.next_value()?;
                }
                Key::Other(k) => {
                    let v: Value = map.next_value()?;
                    n.extra_mut().insert(k, v);
                }
            }
        }
        Ok(n)
    }
}

/// Element type collected through a [`Scratch`] stack.
trait Stacked: Sized {
    fn stack(s: &mut Scratch) -> &mut Vec<Self>;
    fn next<'de, A: SeqAccess<'de>>(seq: &mut A, s: &mut Scratch)
        -> Result<Option<Self>, A::Error>;
}

impl Stacked for Node {
    #[inline]
    fn stack(s: &mut Scratch) -> &mut Vec<Node> {
        &mut s.nodes
    }

    #[inline]
    fn next<'de, A: SeqAccess<'de>>(
        seq: &mut A,
        s: &mut Scratch,
    ) -> Result<Option<Node>, A::Error> {
        seq.next_element_seed(NodeSeed(s))
    }
}

impl Stacked for Mark {
    #[inline]
    fn stack(s: &mut Scratch) -> &mut Vec<Mark> {
        &mut s.marks
    }

    #[inline]
    fn next<'de, A: SeqAccess<'de>>(
        seq: &mut A,
        _: &mut Scratch,
    ) -> Result<Option<Mark>, A::Error> {
        seq.next_element()
    }
}

/// `Option<Vec<T>>` with the same `null`/error behaviour as serde's impls.
struct OptVec<'a, T>(&'a mut Scratch, PhantomData<T>);

impl<'de, T: Stacked> DeserializeSeed<'de> for OptVec<'_, T> {
    type Value = Option<Vec<T>>;

    #[inline]
    fn deserialize<D: Deserializer<'de>>(self, d: D) -> Result<Self::Value, D::Error> {
        d.deserialize_option(self)
    }
}

impl<'de, T: Stacked> Visitor<'de> for OptVec<'_, T> {
    type Value = Option<Vec<T>>;

    fn expecting(&self, f: &mut fmt::Formatter) -> fmt::Result {
        f.write_str("option")
    }

    #[inline]
    fn visit_none<E: de::Error>(self) -> Result<Self::Value, E> {
        Ok(None)
    }

    #[inline]
    fn visit_unit<E: de::Error>(self) -> Result<Self::Value, E> {
        Ok(None)
    }

    #[inline]
    fn visit_some<D: Deserializer<'de>>(self, d: D) -> Result<Self::Value, D::Error> {
        d.deserialize_seq(SeqVec(self.0, PhantomData)).map(Some)
    }
}

/// Seq visitor: push onto the scratch stack, then one exact-capacity move.
struct SeqVec<'a, T>(&'a mut Scratch, PhantomData<T>);

impl<'de, T: Stacked> Visitor<'de> for SeqVec<'_, T> {
    type Value = Vec<T>;

    fn expecting(&self, f: &mut fmt::Formatter) -> fmt::Result {
        f.write_str("a sequence")
    }

    fn visit_seq<A: SeqAccess<'de>>(self, mut seq: A) -> Result<Vec<T>, A::Error> {
        let s = self.0;
        let base = T::stack(s).len();
        while let Some(x) = T::next(&mut seq, s)? {
            T::stack(s).push(x);
        }
        let st = T::stack(s);
        let mut v = Vec::with_capacity(st.len() - base);
        v.extend(st.drain(base..));
        Ok(v)
    }
}

impl Serialize for Node {
    fn serialize<S: Serializer>(&self, s: S) -> Result<S::Ok, S::Error> {
        let len = self.node_type.is_some() as usize
            + self.attrs.is_some() as usize
            + self.content.is_some() as usize
            + self.marks.is_some() as usize
            + self.text.is_some() as usize
            + self.extra.as_ref().map_or(0, |e| e.len());
        let mut m = s.serialize_map(Some(len))?;
        if let Some(t) = &self.node_type {
            m.serialize_entry("type", t)?;
        }
        if let Some(a) = &self.attrs {
            m.serialize_entry("attrs", a)?;
        }
        if let Some(c) = &self.content {
            m.serialize_entry("content", c)?;
        }
        if let Some(mk) = &self.marks {
            m.serialize_entry("marks", mk)?;
        }
        if let Some(t) = &self.text {
            m.serialize_entry("text", t)?;
        }
        for (k, v) in self.extra.iter().flat_map(|e| e.iter()) {
            m.serialize_entry(k, v)?;
        }
        m.end()
    }
}

impl<'de> Deserialize<'de> for Mark {
    fn deserialize<D: Deserializer<'de>>(d: D) -> Result<Mark, D::Error> {
        struct V;
        impl<'de> Visitor<'de> for V {
            type Value = Mark;

            fn expecting(&self, f: &mut fmt::Formatter) -> fmt::Result {
                f.write_str("struct Mark")
            }

            fn visit_map<A: MapAccess<'de>>(self, mut map: A) -> Result<Mark, A::Error> {
                let mut mark_type: Option<String> = None;
                let mut attrs = None;
                let mut extra: Option<Box<Map<String, Value>>> = None;
                let mut seen = 0u8;
                while let Some(key) = map.next_key::<Key>()? {
                    match key {
                        Key::Type => {
                            once(&mut seen, 1, "type")?;
                            mark_type = Some(map.next_value()?);
                        }
                        Key::Attrs => {
                            once(&mut seen, 2, "attrs")?;
                            attrs = map.next_value()?;
                        }
                        Key::Content => extra_entry(&mut map, &mut extra, "content")?,
                        Key::Marks => extra_entry(&mut map, &mut extra, "marks")?,
                        Key::Text => extra_entry(&mut map, &mut extra, "text")?,
                        Key::Other(k) => {
                            let v: Value = map.next_value()?;
                            extra.get_or_insert_with(Default::default).insert(k, v);
                        }
                    }
                }
                let mark_type = mark_type.ok_or_else(|| de::Error::missing_field("type"))?;
                Ok(Mark {
                    mark_type,
                    attrs,
                    extra,
                })
            }
        }
        d.deserialize_map(V)
    }
}

/// `Mark` has no `content`/`marks`/`text` fields: those keys are unknown there.
#[inline]
fn extra_entry<'de, A: MapAccess<'de>>(
    map: &mut A,
    extra: &mut Option<Box<Map<String, Value>>>,
    key: &str,
) -> Result<(), A::Error> {
    let v: Value = map.next_value()?;
    extra
        .get_or_insert_with(Default::default)
        .insert(key.to_owned(), v);
    Ok(())
}

impl Serialize for Mark {
    fn serialize<S: Serializer>(&self, s: S) -> Result<S::Ok, S::Error> {
        let len = 1 + self.attrs.is_some() as usize + self.extra.as_ref().map_or(0, |e| e.len());
        let mut m = s.serialize_map(Some(len))?;
        m.serialize_entry("type", &self.mark_type)?;
        if let Some(a) = &self.attrs {
            m.serialize_entry("attrs", a)?;
        }
        for (k, v) in self.extra.iter().flat_map(|e| e.iter()) {
            m.serialize_entry(k, v)?;
        }
        m.end()
    }
}