perf: improve XML deserialization speed

2 files changed, 334 insertions, 0 deletions

diff --git a/src/deserialization/buffer_deserializer.rs b/src/deserialization/buffer_deserializer.rs
new file mode 100644
index 0000000..652e1ff
--- /dev/null
+++ b/src/deserialization/buffer_deserializer.rs
@@ -0,0 +1,210 @@
+use std::any::type_name;
+use std::error::Error;
+use std::io::BufRead;
+
+use quick_xml::events::{BytesStart, Event};
+use quick_xml::Reader;
+
+use crate::deserialization::{
+    Deserialize,
+    Deserializer,
+    DeserializerError,
+    IgnoreEnd,
+    WrappedDeserializeError,
+};
+
+macro_rules! read_event {
+    ($self: ident) => {{
+        let event = if let Some(leftover_event) = $self.leftover_event.take() {
+            leftover_event
+        } else {
+            $self.reader.read_event_into(&mut $self.buf)?.into_owned()
+        };
+
+        if let Event::Eof = &event {
+            return Err(DeserializerError::UnexpectedEndOfFile);
+        }
+
+        event
+    }};
+}
+
+pub struct BufferDeserializer<Source>
+{
+    reader: Reader<Source>,
+    leftover_event: Option<Event<'static>>,
+    buf: Vec<u8>,
+}
+
+impl<Source> BufferDeserializer<Source>
+where
+    Source: BufRead,
+{
+    pub fn new(source: Source) -> Self
+    {
+        let mut reader = Reader::from_reader(source);
+
+        reader.trim_text(true);
+        reader.expand_empty_elements(true);
+
+        Self {
+            reader,
+            leftover_event: None,
+            buf: Vec::new(),
+        }
+    }
+}
+
+impl<Source> Deserializer for BufferDeserializer<Source>
+where
+    Source: BufRead,
+{
+    fn de_tag<De: Deserialize>(
+        &mut self,
+        tag_name: &str,
+        ignore_end: IgnoreEnd,
+    ) -> Result<De, DeserializerError>
+    {
+        self.de_tag_with(tag_name, ignore_end, De::deserialize)
+    }
+
+    fn de_tag_with<Output, Err, DeserializeFn>(
+        &mut self,
+        tag_name: &str,
+        ignore_end: IgnoreEnd,
+        deserialize: DeserializeFn,
+    ) -> Result<Output, DeserializerError>
+    where
+        Err: Error + Send + Sync + 'static,
+        DeserializeFn: FnOnce(&BytesStart, &mut Self) -> Result<Output, Err>,
+    {
+        let deserialized = match read_event!(self) {
+            Event::Start(start) if start.name().as_ref() == tag_name.as_bytes() => {
+                deserialize(&start, self).map_err(|err| {
+                    DeserializerError::DeserializeFailed(
+                        type_name::<Output>(),
+                        WrappedDeserializeError::new(err),
+                    )
+                })?
+            }
+            event => {
+                self.leftover_event = Some(event.clone().into_owned());
+
+                return Err(DeserializerError::UnexpectedEvent {
+                    expected_event_name: format!("start({tag_name})"),
+                    found_event: event,
+                });
+            }
+        };
+
+        if let IgnoreEnd::No = ignore_end {
+            self.read_end_event(tag_name)?;
+        }
+
+        Ok(deserialized)
+    }
+
+    fn de_tag_list<De: Deserialize>(
+        &mut self,
+        tag_name: &str,
+    ) -> Result<Vec<De>, DeserializerError>
+    {
+        let mut deserialized_items = Vec::new();
+
+        loop {
+            let start = match read_event!(self) {
+                Event::Start(start) if start.name().as_ref() == tag_name.as_bytes() => {
+                    start
+                }
+                Event::Comment(_) => {
+                    continue;
+                }
+                event => {
+                    self.leftover_event = Some(event.into_owned());
+                    break;
+                }
+            };
+
+            let deserialized = De::deserialize(&start, self).map_err(|err| {
+                DeserializerError::DeserializeFailed(
+                    type_name::<De>(),
+                    WrappedDeserializeError::new(err),
+                )
+            })?;
+
+            self.read_end_event(tag_name)?;
+
+            deserialized_items.push(deserialized);
+        }
+
+        Ok(deserialized_items)
+    }
+
+    fn de_text(&mut self) -> Result<String, DeserializerError>
+    {
+        let text = match read_event!(self) {
+            Event::Text(text) => Ok(text),
+            event => {
+                self.leftover_event = Some(event.clone().into_owned());
+
+                Err(DeserializerError::UnexpectedEvent {
+                    expected_event_name: "text".to_string(),
+                    found_event: event,
+                })
+            }
+        }?
+        .unescape()?;
+
+        Ok(text.to_string())
+    }
+
+    fn skip_to_tag_start(&mut self, tag_name: &str) -> Result<(), DeserializerError>
+    {
+        loop {
+            match read_event!(self) {
+                Event::Start(start) if start.name().as_ref() == tag_name.as_bytes() => {
+                    self.leftover_event = Some(Event::Start(start).into_owned());
+
+                    break;
+                }
+                _ => {}
+            }
+        }
+
+        Ok(())
+    }
+
+    fn skip_to_tag_end(&mut self, tag_name: &str) -> Result<(), DeserializerError>
+    {
+        loop {
+            match read_event!(self) {
+                Event::End(end) if end.name().as_ref() == tag_name.as_bytes() => {
+                    self.leftover_event = Some(Event::End(end).into_owned());
+
+                    return Ok(());
+                }
+                _ => {}
+            }
+        }
+    }
+}
+
+impl<Source> BufferDeserializer<Source>
+where
+    Source: BufRead,
+{
+    fn read_end_event(&mut self, tag_name: &str) -> Result<(), DeserializerError>
+    {
+        let event = read_event!(self);
+
+        if matches!(&event, Event::End(end) if end.name().as_ref() == tag_name.as_bytes())
+        {
+            return Ok(());
+        }
+
+        Err(DeserializerError::UnexpectedEvent {
+            expected_event_name: "end".to_string(),
+            found_event: event.into_owned(),
+        })
+    }
+}
diff --git a/src/deserialization/mod.rs b/src/deserialization/mod.rs
new file mode 100644
index 0000000..fa25e4b
--- /dev/null
+++ b/src/deserialization/mod.rs
@@ -0,0 +1,124 @@
+use std::error::Error;
+use std::ops::Deref;
+
+use quick_xml::events::{BytesStart, Event};
+
+pub mod buffer_deserializer;
+
+pub trait Deserialize: Sized
+{
+    type Error: Error + Send + Sync + 'static;
+
+    fn deserialize<TDeserializer: Deserializer>(
+        start: &BytesStart,
+        deserializer: &mut TDeserializer,
+    ) -> Result<Self, Self::Error>;
+}
+
+pub trait Deserializer
+{
+    fn de_tag<De: Deserialize>(
+        &mut self,
+        tag_name: &str,
+        ignore_end: IgnoreEnd,
+    ) -> Result<De, DeserializerError>;
+
+    fn de_tag_with<Output, Err, DeserializeFn>(
+        &mut self,
+        tag_name: &str,
+        ignore_end: IgnoreEnd,
+        deserialize: DeserializeFn,
+    ) -> Result<Output, DeserializerError>
+    where
+        Err: Error + Send + Sync + 'static,
+        DeserializeFn: FnOnce(&BytesStart, &mut Self) -> Result<Output, Err>;
+
+    fn de_tag_list<De: Deserialize>(
+        &mut self,
+        tag_name: &str,
+    ) -> Result<Vec<De>, DeserializerError>;
+
+    fn de_text(&mut self) -> Result<String, DeserializerError>;
+
+    fn skip_to_tag_start(&mut self, tag_name: &str) -> Result<(), DeserializerError>;
+
+    fn skip_to_tag_end(&mut self, tag_name: &str) -> Result<(), DeserializerError>;
+}
+
+pub enum IgnoreEnd
+{
+    Yes,
+    No,
+}
+
+/// Function pointer type passable to [`Deserializer::de_tag_with`].
+pub type DeserializeWithFn<Output, Err, Deserializer> =
+    fn(&BytesStart, &mut Deserializer) -> Result<Output, Err>;
+
+#[derive(Debug, thiserror::Error)]
+pub enum DeserializerError
+{
+    #[error("Failed to read")]
+    ReadFailed(#[from] quick_xml::Error),
+
+    #[error("Failed to deserialize {0}")]
+    DeserializeFailed(&'static str, #[source] WrappedDeserializeError),
+
+    #[error("Expected {expected_event_name} event. Found {found_event:?}")]
+    UnexpectedEvent
+    {
+        expected_event_name: String,
+        found_event: Event<'static>,
+    },
+
+    #[error("Unexpected end of file")]
+    UnexpectedEndOfFile,
+}
+
+#[derive(Debug, thiserror::Error)]
+#[error(transparent)]
+pub struct WrappedDeserializeError(Box<dyn Error + Send + Sync>);
+
+impl WrappedDeserializeError
+{
+    fn new<Err: Error + Send + Sync + 'static>(err: Err) -> Self
+    {
+        Self(Box::new(err))
+    }
+}
+
+impl Deref for WrappedDeserializeError
+{
+    type Target = dyn Error;
+
+    fn deref(&self) -> &Self::Target
+    {
+        self.0.as_ref()
+    }
+}
+
+pub trait ResultExt<Value>
+{
+    fn try_event(self) -> Result<Option<Value>, DeserializerError>;
+}
+
+impl<Value> ResultExt<Value> for Result<Value, DeserializerError>
+{
+    fn try_event(self) -> Result<Option<Value>, DeserializerError>
+    {
+        self.map_or_else(
+            |err| {
+                if let DeserializerError::UnexpectedEvent {
+                    expected_event_name: _,
+                    found_event: _,
+                } = err
+                {
+                    return Ok(None);
+                }
+
+                Err(err)
+            },
+            |value| Ok(Some(value)),
+        )
+    }
+}