Owned items, static values

crates/pile-dataset/src/extract/epub/epub_meta.rs

@@ -1,24 +1,27 @@
 use epub::doc::EpubDoc;
 use pile_config::Label;
-use std::{collections::HashMap, sync::OnceLock};
+use std::{
+	collections::HashMap,
+	sync::{Arc, OnceLock},
+};
 use tracing::trace;
 
 use crate::{Item, PileValue, SyncReadBridge, extract::ObjectExtractor};
 
-pub struct EpubMetaExtractor<'a> {
-	item: &'a Item,
-	output: OnceLock<HashMap<Label, PileValue<'a>>>,
+pub struct EpubMetaExtractor {
+	item: Item,
+	output: OnceLock<HashMap<Label, PileValue>>,
 }
 
-impl<'a> EpubMetaExtractor<'a> {
-	pub fn new(item: &'a Item) -> Self {
+impl EpubMetaExtractor {
+	pub fn new(item: &Item) -> Self {
 		Self {
-			item,
+			item: item.clone(),
 			output: OnceLock::new(),
 		}
 	}
 
-	async fn get_inner(&self) -> Result<&HashMap<Label, PileValue<'a>>, std::io::Error> {
+	async fn get_inner(&self) -> Result<&HashMap<Label, PileValue>, std::io::Error> {
 		if let Some(x) = self.output.get() {
 			return Ok(x);
 		}
@@ -61,13 +64,13 @@ impl<'a> EpubMetaExtractor<'a> {
 			}
 		};
 
-		let mut output: HashMap<Label, PileValue<'a>> = HashMap::new();
+		let mut output: HashMap<Label, PileValue> = HashMap::new();
 
 		#[expect(clippy::unwrap_used)]
 		for (key, val) in raw_meta {
 			let label = Label::new(key).unwrap();
 			let value = match val {
-				Some(s) => PileValue::String(s.into()),
+				Some(s) => PileValue::String(Arc::new(s.into())),
 				None => PileValue::Null,
 			};
 			output.insert(label, value);
@@ -78,12 +81,9 @@ impl<'a> EpubMetaExtractor<'a> {
 }
 
 #[async_trait::async_trait]
-impl ObjectExtractor for EpubMetaExtractor<'_> {
-	async fn field<'a>(
-		&'a self,
-		name: &Label,
-	) -> Result<Option<&'a PileValue<'a>>, std::io::Error> {
-		Ok(self.get_inner().await?.get(name))
+impl ObjectExtractor for EpubMetaExtractor {
+	async fn field(&self, name: &Label) -> Result<Option<PileValue>, std::io::Error> {
+		Ok(self.get_inner().await?.get(name).cloned())
 	}
 
 	async fn fields(&self) -> Result<Vec<Label>, std::io::Error> {

crates/pile-dataset/src/extract/epub/epub_text.rs

@@ -1,24 +1,27 @@
 use epub::doc::EpubDoc;
 use pile_config::Label;
-use std::{collections::HashMap, sync::OnceLock};
+use std::{
+	collections::HashMap,
+	sync::{Arc, OnceLock},
+};
 use tracing::debug;
 
 use crate::{Item, PileValue, SyncReadBridge, extract::ObjectExtractor};
 
-pub struct EpubTextExtractor<'a> {
-	item: &'a Item,
-	output: OnceLock<HashMap<Label, PileValue<'a>>>,
+pub struct EpubTextExtractor {
+	item: Item,
+	output: OnceLock<HashMap<Label, PileValue>>,
 }
 
-impl<'a> EpubTextExtractor<'a> {
-	pub fn new(item: &'a Item) -> Self {
+impl EpubTextExtractor {
+	pub fn new(item: &Item) -> Self {
 		Self {
-			item,
+			item: item.clone(),
 			output: OnceLock::new(),
 		}
 	}
 
-	async fn get_inner(&self) -> Result<&HashMap<Label, PileValue<'a>>, std::io::Error> {
+	async fn get_inner(&self) -> Result<&HashMap<Label, PileValue>, std::io::Error> {
 		if let Some(x) = self.output.get() {
 			return Ok(x);
 		}
@@ -61,7 +64,7 @@ impl<'a> EpubTextExtractor<'a> {
 		#[expect(clippy::unwrap_used)]
 		let output = HashMap::from([(
 			Label::new("text").unwrap(),
-			PileValue::String(raw_text.into()),
+			PileValue::String(Arc::new(raw_text.into())),
 		)]);
 
 		let _ = self.output.set(output);
@@ -88,12 +91,9 @@ fn strip_html(html: &str) -> String {
 }
 
 #[async_trait::async_trait]
-impl ObjectExtractor for EpubTextExtractor<'_> {
-	async fn field<'a>(
-		&'a self,
-		name: &Label,
-	) -> Result<Option<&'a PileValue<'a>>, std::io::Error> {
-		Ok(self.get_inner().await?.get(name))
+impl ObjectExtractor for EpubTextExtractor {
+	async fn field(&self, name: &Label) -> Result<Option<PileValue>, std::io::Error> {
+		Ok(self.get_inner().await?.get(name).cloned())
 	}
 
 	async fn fields(&self) -> Result<Vec<Label>, std::io::Error> {

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

@@ -12,13 +12,13 @@ use crate::{
 	extract::{MapExtractor, ObjectExtractor},
 };
 
-pub struct EpubExtractor<'a> {
-	inner: MapExtractor<'a>,
+pub struct EpubExtractor {
+	inner: MapExtractor,
 }
 
-impl<'a> EpubExtractor<'a> {
+impl EpubExtractor {
 	#[expect(clippy::unwrap_used)]
-	pub fn new(item: &'a Item) -> Self {
+	pub fn new(item: &Item) -> Self {
 		let inner = MapExtractor {
 			inner: HashMap::from([
 				(
@@ -37,19 +37,8 @@ impl<'a> EpubExtractor<'a> {
 }
 
 #[async_trait::async_trait]
-impl ObjectExtractor for EpubExtractor<'_> {
-	async fn field<'a>(
-		&'a self,
-		name: &pile_config::Label,
-	) -> Result<Option<&'a PileValue<'a>>, std::io::Error> {
-		#[expect(clippy::unwrap_used)]
-		if name.as_str() == "text" {
-			match self.inner.inner.get(name).unwrap() {
-				PileValue::ObjectExtractor(x) => return x.field(name).await,
-				_ => unreachable!(),
-			};
-		}
-
+impl ObjectExtractor for EpubExtractor {
+	async fn field(&self, name: &pile_config::Label) -> Result<Option<PileValue>, std::io::Error> {
 		self.inner.field(name).await
 	}