Extractor refactor, S3 support

crates/pile-dataset/src/extract/mod.rs

@@ -1,5 +1,5 @@
 use pile_config::Label;
-use std::{collections::HashMap, rc::Rc};
+use std::{collections::HashMap, sync::Arc};
 
 mod flac;
 pub use flac::*;
@@ -13,59 +13,73 @@ pub use fs::*;
 mod pdf;
 pub use pdf::*;
 
-mod sidecar;
-pub use sidecar::*;
+mod toml;
+pub use toml::*;
 
 mod map;
 pub use map::*;
 
+mod sidecar;
+pub use sidecar::*;
+
+use crate::Item;
+
 /// An attachment that extracts metadata from an [Item].
 ///
 /// Metadata is exposed as an immutable map of {label: value},
 /// much like a json object.
-pub trait Extractor<I: crate::Item> {
+#[async_trait::async_trait]
+pub trait Extractor: Send + Sync {
 	/// Get the field at `name` from `item`.
 	/// - returns `None` if `name` is not a valid field
 	/// - returns `Some(Null)` if `name` is not available
-	fn field<'a>(
+	async fn field<'a>(
 		&'a self,
 		name: &pile_config::Label,
-	) -> Result<Option<&'a crate::PileValue<'a, I>>, std::io::Error>;
+	) -> Result<Option<&'a crate::PileValue<'a>>, std::io::Error>;
 
 	/// Return all fields in this extractor.
 	/// `Self::field` must return [Some] for all these keys
 	/// and [None] for all others.
-	fn fields(&self) -> Result<Vec<Label>, std::io::Error>;
+	async fn fields(&self) -> Result<Vec<Label>, std::io::Error>;
 }
 
-pub struct MetaExtractor<'a, I: crate::Item> {
-	inner: MapExtractor<'a, I>,
+pub struct MetaExtractor<'a> {
+	inner: MapExtractor<'a>,
 }
 
-impl<'a> MetaExtractor<'a, crate::FileItem> {
+//
+// MARK: file
+//
+
+impl<'a> MetaExtractor<'a> {
 	#[expect(clippy::unwrap_used)]
-	pub fn new(item: &'a crate::FileItem) -> Self {
+	pub fn new(item: &'a Item) -> Self {
 		let inner = MapExtractor {
 			inner: HashMap::from([
 				(
 					Label::new("flac").unwrap(),
-					crate::PileValue::Extractor(Rc::new(FlacExtractor::new(item))),
+					crate::PileValue::Extractor(Arc::new(FlacExtractor::new(item))),
 				),
 				(
 					Label::new("id3").unwrap(),
-					crate::PileValue::Extractor(Rc::new(Id3Extractor::new(item))),
+					crate::PileValue::Extractor(Arc::new(Id3Extractor::new(item))),
 				),
 				(
 					Label::new("fs").unwrap(),
-					crate::PileValue::Extractor(Rc::new(FsExtractor::new(item))),
+					crate::PileValue::Extractor(Arc::new(FsExtractor::new(item))),
 				),
 				(
 					Label::new("pdf").unwrap(),
-					crate::PileValue::Extractor(Rc::new(PdfExtractor::new(item))),
+					crate::PileValue::Extractor(Arc::new(PdfExtractor::new(item))),
+				),
+				(
+					Label::new("toml").unwrap(),
+					crate::PileValue::Extractor(Arc::new(TomlExtractor::new(item))),
 				),
 				(
 					Label::new("sidecar").unwrap(),
-					crate::PileValue::Extractor(Rc::new(SidecarExtractor::new(item))),
+					crate::PileValue::Extractor(Arc::new(SidecarExtractor::new(item))),
 				),
 			]),
 		};
@@ -74,16 +88,17 @@ impl<'a> MetaExtractor<'a, crate::FileItem> {
 	}
 }
 
-impl Extractor<crate::FileItem> for MetaExtractor<'_, crate::FileItem> {
-	fn field<'a>(
+#[async_trait::async_trait]
+impl Extractor for MetaExtractor<'_> {
+	async fn field<'a>(
 		&'a self,
 		name: &pile_config::Label,
-	) -> Result<Option<&'a crate::PileValue<'a, crate::FileItem>>, std::io::Error> {
-		self.inner.field(name)
+	) -> Result<Option<&'a crate::PileValue<'a>>, std::io::Error> {
+		self.inner.field(name).await
 	}
 
 	#[expect(clippy::unwrap_used)]
-	fn fields(&self) -> Result<Vec<Label>, std::io::Error> {
+	async fn fields(&self) -> Result<Vec<Label>, std::io::Error> {
 		return Ok(vec![
 			Label::new("flac").unwrap(),
 			Label::new("id3").unwrap(),
@@ -92,4 +107,4 @@ impl Extractor<crate::FileItem> for MetaExtractor<'_, crate::FileItem> {
 			Label::new("sidecar").unwrap(),
 		]);
 	}
-}
+}