Extract epub covers

crates/pile-value/src/extract/item/epub/epub_cover.rs

@@ -0,0 +1,76 @@
+use epub::doc::EpubDoc;
+use mime::Mime;
+use pile_io::SyncReadBridge;
+use std::sync::{Arc, OnceLock};
+use tracing::trace;
+
+use crate::{
+	extract::traits::ExtractState,
+	value::{Item, PileValue},
+};
+
+pub struct EpubCoverExtractor {
+	item: Item,
+	output: OnceLock<Option<(Mime, Vec<u8>)>>,
+}
+
+impl EpubCoverExtractor {
+	pub fn new(item: &Item) -> Self {
+		Self {
+			item: item.clone(),
+			output: OnceLock::new(),
+		}
+	}
+
+	async fn get_inner(&self) -> Result<Option<&(Mime, Vec<u8>)>, std::io::Error> {
+		if let Some(x) = self.output.get() {
+			return Ok(x.as_ref());
+		}
+
+		let reader = SyncReadBridge::new_current(self.item.read().await?);
+		let result = tokio::task::spawn_blocking(move || {
+			let mut doc = EpubDoc::from_reader(reader)?;
+			let cover_id = match doc.get_cover_id() {
+				Ok(id) => id,
+				Err(_) => return Ok::<_, anyhow::Error>(None),
+			};
+
+			let mime: Mime = doc
+				.resources
+				.get(&cover_id)
+				.and_then(|(_, mime_str)| mime_str.parse().ok())
+				.unwrap_or(mime::IMAGE_JPEG);
+
+			let bytes = doc.get_cover()?;
+			Ok(Some((mime, bytes)))
+		})
+		.await?;
+
+		let result = match result {
+			Ok(x) => x,
+			Err(error) => match error.downcast::<std::io::Error>() {
+				Ok(x) => return Err(x),
+				Err(error) => {
+					trace!(message = "Could not extract epub cover", ?error, key = ?self.item.key());
+					None
+				}
+			},
+		};
+
+		Ok(self.output.get_or_init(|| result).as_ref())
+	}
+
+	pub async fn get(&self, state: &ExtractState) -> Result<Option<PileValue>, std::io::Error> {
+		if !state.ignore_mime && self.item.mime().essence_str() != "application/epub+zip" {
+			return Ok(None);
+		}
+
+		Ok(self
+			.get_inner()
+			.await?
+			.map(|(mime, bytes)| PileValue::Blob {
+				mime: mime.clone(),
+				bytes: Arc::new(bytes.clone()),
+			}))
+	}
+}

crates/pile-value/src/extract/item/epub/mod.rs

@@ -1,6 +1,9 @@
 use pile_config::Label;
 use std::sync::Arc;
 
+mod epub_cover;
+pub use epub_cover::*;
+
 mod epub_meta;
 pub use epub_meta::*;
 
@@ -15,6 +18,7 @@ use crate::{
 pub struct EpubExtractor {
 	text: Arc<EpubTextExtractor>,
 	meta: Arc<EpubMetaExtractor>,
+	cover: Arc<EpubCoverExtractor>,
 }
 
 impl EpubExtractor {
@@ -22,6 +26,7 @@ impl EpubExtractor {
 		Self {
 			text: Arc::new(EpubTextExtractor::new(item)),
 			meta: Arc::new(EpubMetaExtractor::new(item)),
+			cover: Arc::new(EpubCoverExtractor::new(item)),
 		}
 	}
 }
@@ -43,6 +48,7 @@ impl ObjectExtractor for EpubExtractor {
 			)),
 
 			("meta", None) => Ok(Some(PileValue::ObjectExtractor(self.meta.clone()))),
+			("cover", None) => self.cover.get(state).await,
 			_ => Ok(None),
 		}
 	}
@@ -52,6 +58,7 @@ impl ObjectExtractor for EpubExtractor {
 		Ok(vec![
 			Label::new("text").unwrap(),
 			Label::new("meta").unwrap(),
+			Label::new("cover").unwrap(),
 		])
 	}
 
@@ -78,6 +85,18 @@ impl ObjectExtractor for EpubExtractor {
 				continue;
 			}
 
+			if k.as_str() == "cover" {
+				let summary = match &v {
+					PileValue::Blob { mime, bytes } => {
+						format!("<Blob ({}, {} bytes)>", mime, bytes.len())
+					}
+					PileValue::Null => "<null>".to_owned(),
+					_ => "<cover>".to_owned(),
+				};
+				map.insert(k.to_string(), serde_json::Value::String(summary));
+				continue;
+			}
+
 			map.insert(k.to_string(), Box::pin(v.to_json(state)).await?);
 		}