Many field paths

crates/pile-dataset/src/serve/field.rs

@@ -1,6 +1,6 @@
 use axum::{
 	Json,
-	extract::{Query, State},
+	extract::{Query, RawQuery, State},
 	http::{StatusCode, header},
 	response::{IntoResponse, Response},
 };
@@ -17,19 +17,19 @@ use crate::Datasets;
 pub struct FieldQuery {
 	source: String,
 	key: String,
-	path: String,
 	#[serde(default)]
 	download: bool,
 }
 
-/// Extract a specific field from an item's metadata
+/// Extract a specific field from an item's metadata.
+/// Multiple `path` parameters may be provided; the first non-null result is returned.
 #[utoipa::path(
 	get,
 	path = "/field",
 	params(
 		("source" = String, Query, description = "Source label"),
 		("key" = String, Query, description = "Item key"),
-		("path" = String, Query, description = "Object path (e.g. $.flac.title)"),
+		("path" = String, Query, description = "Object path (e.g. $.flac.title); repeat for fallbacks"),
 	),
 	responses(
 		(status = 200, description = "Field value as JSON"),
@@ -41,43 +41,73 @@ pub struct FieldQuery {
 pub async fn get_field(
 	State(state): State<Arc<Datasets>>,
 	Query(params): Query<FieldQuery>,
+	RawQuery(raw_query): RawQuery,
 ) -> Response {
 	let start = Instant::now();
-	debug!(
-		message = "Serving /field",
-		source = params.source,
-		key = params.key,
-		path = params.path,
-	);
 
 	let label = match Label::try_from(params.source.clone()) {
 		Ok(l) => l,
 		Err(e) => return (StatusCode::BAD_REQUEST, format!("{e:?}")).into_response(),
 	};
 
-	let path: ObjectPath = match params.path.parse() {
-		Ok(p) => p,
-		Err(e) => return (StatusCode::BAD_REQUEST, format!("{e:?}")).into_response(),
+	// Collect all `path` query params in order (supports repeated ?path=...&path=...)
+	let raw = raw_query.as_deref().unwrap_or("");
+	let paths: Vec<ObjectPath> = {
+		let mut result = Vec::new();
+		for part in raw.split('&') {
+			if let Some((k, v)) = part.split_once('=')
+				&& k == "path"
+			{
+				match v.parse::<ObjectPath>() {
+					Ok(p) => result.push(p),
+					Err(e) => {
+						return (StatusCode::BAD_REQUEST, format!("{e:?}")).into_response();
+					}
+				}
+			}
+		}
+		result
 	};
 
+	if paths.is_empty() {
+		return (StatusCode::BAD_REQUEST, "Missing `path` query parameter").into_response();
+	}
+
+	debug!(
+		message = "Serving /field",
+		source = params.source,
+		key = params.key,
+		paths = paths.len(),
+	);
+
 	let Some(item) = state.get(&label, &params.key).await else {
 		return StatusCode::NOT_FOUND.into_response();
 	};
 
-	let state = ExtractState { ignore_mime: false };
-
+	let extract_state = ExtractState { ignore_mime: false };
 	let item = PileValue::Item(item);
-	let value = match item.query(&state, &path).await {
-		Ok(Some(v)) => v,
-		Ok(None) => return StatusCode::NOT_FOUND.into_response(),
-		Err(e) => return (StatusCode::INTERNAL_SERVER_ERROR, format!("{e:?}")).into_response(),
+
+	// Try each path in order, returning the first non-null result
+	let mut value = None;
+	for path in &paths {
+		match item.query(&extract_state, path).await {
+			Ok(Some(PileValue::Null)) | Ok(None) => continue,
+			Ok(Some(v)) => {
+				value = Some(v);
+				break;
+			}
+			Err(e) => return (StatusCode::INTERNAL_SERVER_ERROR, format!("{e:?}")).into_response(),
+		}
+	}
+
+	let Some(value) = value else {
+		return StatusCode::NOT_FOUND.into_response();
 	};
 
 	debug!(
 		message = "Served /field",
 		source = params.source,
 		key = params.key,
-		path = params.path,
 		time_ms = start.elapsed().as_millis()
 	);
 
@@ -106,7 +136,7 @@ pub async fn get_field(
 			bytes.as_ref().clone(),
 		)
 			.into_response(),
-		_ => match value.to_json(&state).await {
+		_ => match value.to_json(&extract_state).await {
 			Ok(json) => (
 				StatusCode::OK,
 				[(header::CONTENT_DISPOSITION, disposition.to_owned())],

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?);
 		}