xsd_parser/quick_xml/

deserialize.rs

use std::fmt::Debug;
use std::marker::PhantomData;
use std::str::{from_utf8, FromStr};

use quick_xml::events::Event;

use super::{Error, ErrorKind, XmlReader, XmlReaderSync};

/// Trait that defines the [`Deserializer`] for a type.
pub trait WithDeserializer: Sized {
    /// The deserializer to use for this type.
    type Deserializer: for<'de> Deserializer<'de, Self>;
}

impl<X> WithDeserializer for X
where
    X: DeserializeBytes + Debug,
{
    type Deserializer = ContentDeserializer<X>;
}

/// Trait that defines a deserializer that can be used to construct a type from a
/// XML [`Event`]s.
pub trait Deserializer<'de, T>: Debug + Sized {
    /// Initializes a new deserializer from the passed `reader` and the initial `event`.
    ///
    /// # Errors
    ///
    /// Returns an [`Error`] if the initialization of the deserializer failed.
    fn init<R>(reader: &R, event: Event<'de>) -> DeserializerResult<'de, T, Self>
    where
        R: XmlReader;

    /// Processes the next XML [`Event`].
    ///
    /// # Errors
    ///
    /// Returns an [`Error`] if processing the event failed.
    fn next<R>(self, reader: &R, event: Event<'de>) -> DeserializerResult<'de, T, Self>
    where
        R: XmlReader;

    /// Force the deserializer to finish.
    ///
    /// # Errors
    ///
    /// Returns an [`Error`] if the deserializer could not finish.
    fn finish<R>(self, reader: &R) -> Result<T, Error>
    where
        R: XmlReader;
}

/// Result type returned by the [`Deserializer`] trait.
pub type DeserializerResult<'a, T, B> = Result<DeserializerOutput<'a, T, B>, Error>;

/// Type that is used to bundle the output of a [`Deserializer`] operation.
#[derive(Debug)]
pub struct DeserializerOutput<'a, T, B> {
    /// Contains the actual type constructed by the deserializer, once the deserializer has
    /// finished it's construction. If this is `None`, constructing the object has
    /// not been finished yet.
    pub data: Option<T>,

    /// Contains the deserializer after an operation on the deserializer has been executed.
    /// If this is `None` the deserializer is finished and can not be re-used.
    pub deserializer: Option<B>,

    /// Contains the processed event if it was not consumed by the deserializer.
    pub event: Option<Event<'a>>,

    /// Whether the deserializer allows other XML elements in the current state or not.
    /// If this is set to `true` and the `event` is not consumed, the event should
    /// be skipped. For [`Event::Start`] this would mean to skip the whole element
    /// until the corresponding [`Event::End`] is received.
    pub allow_any: bool,
}

/// Trait that could be implemented by types to support deserialization from XML
/// using the [`quick_xml`] crate.
pub trait DeserializeSync<'de, R>: Sized
where
    R: XmlReaderSync<'de>,
{
    /// Error that is returned by the `deserialize` method.
    type Error;

    /// Deserialize the type from the passed `reader`.
    ///
    /// # Errors
    ///
    /// Will return a suitable error if the operation failed.
    fn deserialize(reader: &mut R) -> Result<Self, Self::Error>;
}

impl<'de, R, X> DeserializeSync<'de, R> for X
where
    R: XmlReaderSync<'de>,
    X: WithDeserializer,
{
    type Error = Error;

    fn deserialize(reader: &mut R) -> Result<Self, Self::Error> {
        DeserializeHelper::new(reader).deserialize_sync()
    }
}

/// Trait that could be implemented by types to support asynchronous
/// deserialization from XML using the [`quick_xml`] crate.
#[cfg(feature = "async")]
pub trait DeserializeAsync<'de, R>: Sized
where
    R: super::XmlReaderAsync<'de>,
{
    /// Future that is returned by the [`deserialize_async`] method.
    type Future<'x>: std::future::Future<Output = Result<Self, Self::Error>>
    where
        R: 'x,
        'de: 'x;

    /// Error that is returned by the future generated by the [`deserialize_async`] method.
    type Error;

    /// Asynchronously deserializes the type from the passed `reader`.
    fn deserialize_async<'x>(reader: &'x mut R) -> Self::Future<'x>
    where
        'de: 'x;
}

#[cfg(feature = "async")]
impl<'de, R, X> DeserializeAsync<'de, R> for X
where
    R: super::XmlReaderAsync<'de>,
    X: WithDeserializer,
{
    type Future<'x>
        = std::pin::Pin<Box<dyn std::future::Future<Output = Result<Self, Self::Error>> + 'x>>
    where
        R: 'x,
        'de: 'x;

    type Error = Error;

    fn deserialize_async<'x>(reader: &'x mut R) -> Self::Future<'x>
    where
        'de: 'x,
    {
        Box::pin(async move { DeserializeHelper::new(reader).deserialize_async().await })
    }
}

/// Trait that could be implemented by types to support deserialization from
/// XML byte streams using the [`quick_xml`] crate.
///
/// This is usually implemented for simple types like numbers, strings or enums.
pub trait DeserializeBytes: Sized {
    /// Try to deserialize the type from bytes.
    ///
    /// This is used to deserialize the type from attributes or raw element
    /// content.
    ///
    /// # Errors
    ///
    /// Returns a suitable [`Error`] if the deserialization was not successful.
    fn deserialize_bytes<R: XmlReader>(reader: &R, bytes: &[u8]) -> Result<Self, Error>;
}

impl<X> DeserializeBytes for X
where
    X: FromStr,
    X::Err: std::error::Error + Send + Sync + 'static,
{
    fn deserialize_bytes<R: XmlReader>(reader: &R, bytes: &[u8]) -> Result<Self, Error> {
        let _reader = reader;
        let s = from_utf8(bytes).map_err(Error::from)?;

        X::from_str(s).map_err(Error::custom)
    }
}

/// Implements a [`Deserializer`] for any type that implements [`DeserializeBytes`].
#[derive(Debug)]
pub struct ContentDeserializer<T> {
    data: Vec<u8>,
    marker: PhantomData<T>,
}

impl<'de, T> Deserializer<'de, T> for ContentDeserializer<T>
where
    T: DeserializeBytes + Debug,
{
    fn init<R>(reader: &R, event: Event<'de>) -> DeserializerResult<'de, T, Self>
    where
        R: XmlReader,
    {
        match event {
            Event::Start(_) => Ok(DeserializerOutput {
                data: None,
                deserializer: Some(Self {
                    data: Vec::new(),
                    marker: PhantomData,
                }),
                event: None,
                allow_any: false,
            }),
            Event::Empty(_) => {
                let data = T::deserialize_bytes(reader, &[])?;

                Ok(DeserializerOutput {
                    data: Some(data),
                    deserializer: None,
                    event: None,
                    allow_any: false,
                })
            }
            event => Ok(DeserializerOutput {
                data: None,
                deserializer: None,
                event: Some(event),
                allow_any: false,
            }),
        }
    }

    fn next<R>(mut self, reader: &R, event: Event<'de>) -> DeserializerResult<'de, T, Self>
    where
        R: XmlReader,
    {
        match event {
            Event::Text(x) => {
                self.data.extend_from_slice(&x.into_inner());

                Ok(DeserializerOutput {
                    data: None,
                    deserializer: Some(self),
                    event: None,
                    allow_any: false,
                })
            }
            Event::End(_) => {
                let data = self.finish(reader)?;

                Ok(DeserializerOutput {
                    data: Some(data),
                    deserializer: None,
                    event: None,
                    allow_any: false,
                })
            }
            event => Ok(DeserializerOutput {
                data: None,
                deserializer: Some(self),
                event: Some(event),
                allow_any: false,
            }),
        }
    }

    fn finish<R>(self, reader: &R) -> Result<T, Error>
    where
        R: XmlReader,
    {
        T::deserialize_bytes(reader, self.data[..].trim_ascii())
    }
}

/* DeserializeHelper */

struct DeserializeHelper<'a, 'de, T, R>
where
    T: WithDeserializer,
{
    reader: &'a mut R,
    deserializer: Option<T::Deserializer>,
    skip_depth: Option<usize>,
    marker: PhantomData<&'de ()>,
}

impl<'a, 'de, T, R> DeserializeHelper<'a, 'de, T, R>
where
    T: WithDeserializer,
    R: XmlReader,
{
    fn new(reader: &'a mut R) -> Self {
        Self {
            reader,
            deserializer: None,
            skip_depth: None,
            marker: PhantomData,
        }
    }

    fn handle_event(&mut self, event: Event<'_>) -> Result<Option<T>, Error> {
        let ret = match self.deserializer.take() {
            None => T::Deserializer::init(self.reader, event),
            Some(b) => b.next(self.reader, event),
        };

        let DeserializerOutput {
            data,
            deserializer,
            event,
            allow_any,
        } = ret?;

        self.deserializer = deserializer;

        match event {
            None
            | Some(
                Event::Decl(_)
                | Event::Text(_)
                | Event::Comment(_)
                | Event::DocType(_)
                | Event::PI(_),
            ) => (),
            Some(event) if allow_any => {
                if matches!(event, Event::Start(_)) {
                    self.skip_depth = Some(1);
                }
            }
            Some(event) => return Err(ErrorKind::UnexpectedEvent(event.into_owned()).into()),
        }

        Ok(data)
    }

    fn handle_skip(&mut self, event: Event<'de>) -> Option<Event<'de>> {
        let Some(skip_depth) = self.skip_depth.as_mut() else {
            return Some(event);
        };

        match event {
            Event::Start(_) => *skip_depth += 1,
            Event::End(_) if *skip_depth == 1 => {
                self.skip_depth = None;

                return None;
            }
            Event::End(_) => *skip_depth -= 1,
            Event::Eof => return Some(Event::Eof),
            _ => (),
        }

        None
    }
}

impl<'de, T, R> DeserializeHelper<'_, 'de, T, R>
where
    T: WithDeserializer,
    R: XmlReaderSync<'de>,
{
    fn deserialize_sync(&mut self) -> Result<T, Error> {
        loop {
            let event = self.reader.read_event()?;

            if let Some(event) = self.handle_skip(event) {
                if let Some(data) = self
                    .handle_event(event)
                    .map_err(|error| self.reader.extend_error(error))?
                {
                    return Ok(data);
                }
            }
        }
    }
}
#[cfg(feature = "async")]
impl<'de, T, R> DeserializeHelper<'_, 'de, T, R>
where
    T: WithDeserializer,
    R: super::XmlReaderAsync<'de>,
{
    async fn deserialize_async(&mut self) -> Result<T, Error> {
        loop {
            let event = self.reader.read_event_async().await?;

            if let Some(event) = self.handle_skip(event) {
                if let Some(data) = self.handle_event(event)? {
                    return Ok(data);
                }
            }
        }
    }
}