Extractor rewrite

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

@@ -0,0 +1,95 @@
+use epub::doc::EpubDoc;
+use pile_config::Label;
+use std::{
+	collections::HashMap,
+	sync::{Arc, OnceLock},
+};
+use tracing::trace;
+
+use crate::{
+	extract::traits::ObjectExtractor,
+	value::{Item, PileValue, SyncReadBridge},
+};
+
+pub struct EpubMetaExtractor {
+	item: Item,
+	output: OnceLock<HashMap<Label, PileValue>>,
+}
+
+impl EpubMetaExtractor {
+	pub fn new(item: &Item) -> Self {
+		Self {
+			item: item.clone(),
+			output: OnceLock::new(),
+		}
+	}
+
+	async fn get_inner(&self) -> Result<&HashMap<Label, PileValue>, std::io::Error> {
+		if let Some(x) = self.output.get() {
+			return Ok(x);
+		}
+
+		let key = self.item.key();
+		let ext = key.as_str().rsplit('.').next();
+		if !matches!(ext, Some("epub")) {
+			return Ok(self.output.get_or_init(HashMap::new));
+		}
+
+		let reader = SyncReadBridge::new_current(self.item.read().await?);
+		let raw_meta = tokio::task::spawn_blocking(move || {
+			let doc = EpubDoc::from_reader(reader)
+				.map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidData, e.to_string()))?;
+
+			let fields: &[&'static str] = &[
+				"title",
+				"creator",
+				"description",
+				"language",
+				"publisher",
+				"date",
+				"subject",
+				"identifier",
+			];
+
+			let meta: Vec<(&'static str, Option<String>)> =
+				fields.iter().map(|&key| (key, doc.mdata(key))).collect();
+
+			Ok::<_, std::io::Error>(meta)
+		})
+		.await
+		.map_err(std::io::Error::other)?;
+
+		let raw_meta = match raw_meta {
+			Ok(x) => x,
+			Err(error) => {
+				trace!(message = "Could not process epub", ?error, key = ?self.item.key());
+				return Ok(self.output.get_or_init(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(Arc::new(s.into())),
+				None => PileValue::Null,
+			};
+			output.insert(label, value);
+		}
+
+		return Ok(self.output.get_or_init(|| output));
+	}
+}
+
+#[async_trait::async_trait]
+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> {
+		Ok(self.get_inner().await?.keys().cloned().collect())
+	}
+}

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

@@ -0,0 +1,105 @@
+use epub::doc::EpubDoc;
+use pile_config::Label;
+use std::{
+	collections::HashMap,
+	sync::{Arc, OnceLock},
+};
+use tracing::debug;
+
+use crate::{
+	extract::traits::ObjectExtractor,
+	value::{Item, PileValue, SyncReadBridge},
+};
+
+pub struct EpubTextExtractor {
+	item: Item,
+	output: OnceLock<HashMap<Label, PileValue>>,
+}
+
+impl EpubTextExtractor {
+	pub fn new(item: &Item) -> Self {
+		Self {
+			item: item.clone(),
+			output: OnceLock::new(),
+		}
+	}
+
+	async fn get_inner(&self) -> Result<&HashMap<Label, PileValue>, std::io::Error> {
+		if let Some(x) = self.output.get() {
+			return Ok(x);
+		}
+
+		let key = self.item.key();
+		let ext = key.as_str().rsplit('.').next();
+		if !matches!(ext, Some("epub")) {
+			return Ok(self.output.get_or_init(HashMap::new));
+		}
+
+		let reader = SyncReadBridge::new_current(self.item.read().await?);
+		let raw_text = tokio::task::spawn_blocking(move || {
+			let mut doc = EpubDoc::from_reader(reader)
+				.map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidData, e.to_string()))?;
+
+			let mut text_parts: Vec<String> = Vec::new();
+
+			loop {
+				if let Ok(content) = doc.get_current_str() {
+					text_parts.push(strip_html(&content));
+				}
+				if doc.go_next().is_err() {
+					break;
+				}
+			}
+
+			Ok::<_, std::io::Error>(text_parts.join(" "))
+		})
+		.await
+		.map_err(std::io::Error::other)?;
+
+		let raw_text = match raw_text {
+			Ok(x) => x,
+			Err(error) => {
+				debug!(message = "Could not process epub", ?error, key = ?self.item.key());
+				return Ok(self.output.get_or_init(HashMap::new));
+			}
+		};
+
+		#[expect(clippy::unwrap_used)]
+		let output = HashMap::from([(
+			Label::new("text").unwrap(),
+			PileValue::String(Arc::new(raw_text.into())),
+		)]);
+
+		let _ = self.output.set(output);
+		#[expect(clippy::unwrap_used)]
+		return Ok(self.output.get().unwrap());
+	}
+}
+
+/// Strip HTML/XHTML tags from a string, leaving only text nodes.
+fn strip_html(html: &str) -> String {
+	let mut result = String::with_capacity(html.len());
+	let mut in_tag = false;
+
+	for c in html.chars() {
+		match c {
+			'<' => in_tag = true,
+			'>' => in_tag = false,
+			_ if !in_tag => result.push(c),
+			_ => {}
+		}
+	}
+
+	result
+}
+
+#[async_trait::async_trait]
+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> {
+		Ok(self.get_inner().await?.keys().cloned().collect())
+	}
+}

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

@@ -0,0 +1,46 @@
+use pile_config::Label;
+use std::sync::Arc;
+
+mod epub_meta;
+pub use epub_meta::*;
+
+mod epub_text;
+pub use epub_text::*;
+
+use crate::{
+	extract::traits::ObjectExtractor,
+	value::{Item, PileValue},
+};
+
+pub struct EpubExtractor {
+	text: Arc<EpubTextExtractor>,
+	meta: Arc<EpubMetaExtractor>,
+}
+
+impl EpubExtractor {
+	pub fn new(item: &Item) -> Self {
+		Self {
+			text: Arc::new(EpubTextExtractor::new(item)),
+			meta: Arc::new(EpubMetaExtractor::new(item)),
+		}
+	}
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for EpubExtractor {
+	async fn field(&self, name: &pile_config::Label) -> Result<Option<PileValue>, std::io::Error> {
+		match name.as_str() {
+			"text" => self.text.field(name).await,
+			"meta" => Ok(Some(PileValue::ObjectExtractor(self.meta.clone()))),
+			_ => Ok(None),
+		}
+	}
+
+	#[expect(clippy::unwrap_used)]
+	async fn fields(&self) -> Result<Vec<Label>, std::io::Error> {
+		Ok(vec![
+			Label::new("text").unwrap(),
+			Label::new("meta").unwrap(),
+		])
+	}
+}

crates/pile-value/src/extract/item/exif.rs

@@ -0,0 +1,96 @@
+use pile_config::Label;
+use std::{
+	collections::HashMap,
+	io::BufReader,
+	sync::{Arc, OnceLock},
+};
+use tracing::trace;
+
+use crate::{
+	extract::traits::ObjectExtractor,
+	value::{Item, PileValue, SyncReadBridge},
+};
+
+pub struct ExifExtractor {
+	item: Item,
+	output: OnceLock<HashMap<Label, PileValue>>,
+}
+
+impl ExifExtractor {
+	pub fn new(item: &Item) -> Self {
+		Self {
+			item: item.clone(),
+			output: OnceLock::new(),
+		}
+	}
+
+	async fn get_inner(&self) -> Result<&HashMap<Label, PileValue>, std::io::Error> {
+		if let Some(x) = self.output.get() {
+			return Ok(x);
+		}
+
+		let reader = SyncReadBridge::new_current(self.item.read().await?);
+		let raw_fields = tokio::task::spawn_blocking(move || {
+			let mut br = BufReader::new(reader);
+			let exif = exif::Reader::new()
+				.read_from_container(&mut br)
+				.map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidData, e.to_string()))?;
+
+			let fields: Vec<(String, String)> = exif
+				.fields()
+				.map(|f| {
+					(
+						f.tag.to_string(),
+						f.display_value().with_unit(&exif).to_string(),
+					)
+				})
+				.collect();
+
+			Ok::<_, std::io::Error>(fields)
+		})
+		.await
+		.map_err(std::io::Error::other)?;
+
+		let raw_fields = match raw_fields {
+			Ok(x) => x,
+			Err(error) => {
+				trace!(message = "Could not process exif", ?error, key = ?self.item.key());
+				return Ok(self.output.get_or_init(HashMap::new));
+			}
+		};
+
+		let mut output: HashMap<Label, PileValue> = HashMap::new();
+
+		for (tag_name, value) in raw_fields {
+			let Some(label) = tag_to_label(&tag_name) else {
+				continue;
+			};
+			// First occurrence wins (PRIMARY IFD comes before THUMBNAIL)
+			output
+				.entry(label)
+				.or_insert_with(|| PileValue::String(Arc::new(value.into())));
+		}
+
+		return Ok(self.output.get_or_init(|| output));
+	}
+}
+
+fn tag_to_label(tag: &str) -> Option<Label> {
+	let sanitized: String = tag
+		.chars()
+		.map(|c| if c == ' ' { '_' } else { c })
+		.filter(|c| Label::VALID_CHARS.contains(*c))
+		.collect();
+	Label::new(sanitized)
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for ExifExtractor {
+	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> {
+		Ok(self.get_inner().await?.keys().cloned().collect())
+	}
+}

crates/pile-value/src/extract/item/flac.rs

@@ -0,0 +1,162 @@
+use mime::Mime;
+use pile_config::Label;
+use pile_flac::{FlacBlock, FlacReader};
+use std::{
+	collections::HashMap,
+	io::BufReader,
+	sync::{Arc, OnceLock},
+};
+
+use crate::{
+	extract::traits::{ListExtractor, ObjectExtractor},
+	value::{Item, PileValue, SyncReadBridge},
+};
+
+pub struct FlacImagesExtractor {
+	item: Item,
+}
+
+impl FlacImagesExtractor {
+	pub fn new(item: &Item) -> Self {
+		Self { item: item.clone() }
+	}
+
+	async fn get_images(&self) -> Result<Vec<PileValue>, std::io::Error> {
+		let reader = SyncReadBridge::new_current(self.item.read().await?);
+		let raw_images = tokio::task::spawn_blocking(move || {
+			let reader = FlacReader::new(BufReader::new(reader));
+			let mut images: Vec<(Mime, Vec<u8>)> = Vec::new();
+			for block in reader {
+				match block.map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidData, e))? {
+					FlacBlock::Picture(picture) => {
+						images.push((picture.mime, picture.img_data));
+					}
+					FlacBlock::AudioFrame(_) => break,
+					_ => {}
+				}
+			}
+			Ok::<_, std::io::Error>(images)
+		})
+		.await
+		.map_err(std::io::Error::other)??;
+
+		Ok(raw_images
+			.into_iter()
+			.map(|(mime, data)| PileValue::Blob {
+				mime,
+				bytes: Arc::new(data),
+			})
+			.collect())
+	}
+}
+
+#[async_trait::async_trait]
+impl ListExtractor for FlacImagesExtractor {
+	async fn get<'a>(&'a self, idx: usize) -> Result<Option<PileValue>, std::io::Error> {
+		Ok(self.get_images().await?.into_iter().nth(idx))
+	}
+
+	async fn len(&self) -> Result<usize, std::io::Error> {
+		Ok(self.get_images().await?.len())
+	}
+}
+
+pub struct FlacExtractor {
+	item: Item,
+	output: OnceLock<HashMap<Label, PileValue>>,
+	images: Option<PileValue>,
+}
+
+impl FlacExtractor {
+	pub fn new(item: &Item) -> Self {
+		let is_flac = match item {
+			Item::File { path, .. } => path.to_str().unwrap_or_default().ends_with(".flac"),
+			Item::S3 { key, .. } => key.ends_with(".flac"),
+		};
+
+		let images =
+			is_flac.then(|| PileValue::ListExtractor(Arc::new(FlacImagesExtractor::new(item))));
+
+		Self {
+			item: item.clone(),
+			output: OnceLock::new(),
+			images,
+		}
+	}
+
+	async fn get_inner(&self) -> Result<&HashMap<Label, PileValue>, std::io::Error> {
+		if let Some(x) = self.output.get() {
+			return Ok(x);
+		}
+
+		let key = match &self.item {
+			Item::File { path, .. } => path.to_str().unwrap_or_default().to_owned(),
+			Item::S3 { key, .. } => key.to_string(),
+		};
+
+		if !key.ends_with(".flac") {
+			let _ = self.output.set(HashMap::new());
+			#[expect(clippy::unwrap_used)]
+			return Ok(self.output.get().unwrap());
+		}
+
+		let reader = SyncReadBridge::new_current(self.item.read().await?);
+		let raw_tags = tokio::task::spawn_blocking(move || {
+			let reader = FlacReader::new(BufReader::new(reader));
+			let mut tags: Vec<(String, String)> = Vec::new();
+			for block in reader {
+				match block.map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidData, e))? {
+					FlacBlock::VorbisComment(comment) => {
+						for (k, v) in comment.comment.comments {
+							tags.push((k.to_string().to_lowercase(), v.into()));
+						}
+					}
+					FlacBlock::AudioFrame(_) => break,
+					_ => {}
+				}
+			}
+			Ok::<_, std::io::Error>(tags)
+		})
+		.await
+		.map_err(std::io::Error::other)??;
+
+		let mut output: HashMap<Label, Vec<PileValue>> = HashMap::new();
+		for (k, v) in raw_tags {
+			if let Some(label) = Label::new(k) {
+				output
+					.entry(label)
+					.or_default()
+					.push(PileValue::String(Arc::new(v.into())));
+			}
+		}
+		let output: HashMap<Label, PileValue> = output
+			.into_iter()
+			.map(|(k, v)| (k, PileValue::Array(Arc::new(v))))
+			.collect();
+
+		let _ = self.output.set(output);
+		#[expect(clippy::unwrap_used)]
+		return Ok(self.output.get().unwrap());
+	}
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for FlacExtractor {
+	async fn field(&self, name: &Label) -> Result<Option<PileValue>, std::io::Error> {
+		if name.as_str() == "images"
+			&& let Some(ref images) = self.images
+		{
+			return Ok(Some(images.clone()));
+		}
+		Ok(self.get_inner().await?.get(name).cloned())
+	}
+
+	async fn fields(&self) -> Result<Vec<Label>, std::io::Error> {
+		let mut fields = self.get_inner().await?.keys().cloned().collect::<Vec<_>>();
+		if self.images.is_some() {
+			#[expect(clippy::unwrap_used)]
+			fields.push(Label::new("images").unwrap());
+		}
+		Ok(fields)
+	}
+}

crates/pile-value/src/extract/item/fs.rs

@@ -0,0 +1,80 @@
+use pile_config::Label;
+use std::{
+	collections::HashMap,
+	path::Component,
+	sync::{Arc, OnceLock},
+};
+
+use crate::{
+	extract::traits::ObjectExtractor,
+	value::{Item, PileValue},
+};
+
+pub struct FsExtractor {
+	item: Item,
+	output: OnceLock<HashMap<Label, PileValue>>,
+}
+
+impl FsExtractor {
+	pub fn new(item: &Item) -> Self {
+		Self {
+			item: item.clone(),
+			output: OnceLock::new(),
+		}
+	}
+
+	fn get_inner(&self) -> Result<&HashMap<Label, PileValue>, std::io::Error> {
+		if let Some(x) = self.output.get() {
+			return Ok(x);
+		}
+
+		let Item::File { path, .. } = &self.item else {
+			return Ok(self.output.get_or_init(HashMap::new));
+		};
+
+		#[expect(clippy::unwrap_used)]
+		let output = HashMap::from([
+			(
+				Label::new("extension").unwrap(),
+				path.extension()
+					.and_then(|x| x.to_str())
+					.map(|x| PileValue::String(Arc::new(x.into())))
+					.unwrap_or(PileValue::Null),
+			),
+			(
+				Label::new("path").unwrap(),
+				path.to_str()
+					.map(|x| PileValue::String(Arc::new(x.into())))
+					.unwrap_or(PileValue::Null),
+			),
+			(
+				Label::new("segments").unwrap(),
+				path.components()
+					.map(|x| match x {
+						Component::CurDir => Some(".".to_owned()),
+						Component::Normal(x) => x.to_str().map(|x| x.to_owned()),
+						Component::ParentDir => Some("..".to_owned()),
+						Component::RootDir => Some("/".to_owned()),
+						Component::Prefix(x) => x.as_os_str().to_str().map(|x| x.to_owned()),
+					})
+					.map(|x| x.map(|x| PileValue::String(Arc::new(x.into()))))
+					.collect::<Option<Vec<_>>>()
+					.map(|v| PileValue::Array(Arc::new(v)))
+					.unwrap_or(PileValue::Null),
+			),
+		]);
+
+		return Ok(self.output.get_or_init(|| output));
+	}
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for FsExtractor {
+	async fn field(&self, name: &Label) -> Result<Option<PileValue>, std::io::Error> {
+		Ok(self.get_inner()?.get(name).cloned())
+	}
+
+	async fn fields(&self) -> Result<Vec<Label>, std::io::Error> {
+		Ok(self.get_inner()?.keys().cloned().collect())
+	}
+}

crates/pile-value/src/extract/item/id3.rs

@@ -0,0 +1,135 @@
+use id3::Tag;
+use pile_config::Label;
+use std::{
+	borrow::Cow,
+	collections::HashMap,
+	io::BufReader,
+	sync::{Arc, OnceLock},
+};
+
+use crate::{
+	extract::traits::ObjectExtractor,
+	value::{Item, PileValue, SyncReadBridge},
+};
+
+pub struct Id3Extractor {
+	item: Item,
+	output: OnceLock<HashMap<Label, PileValue>>,
+}
+
+impl Id3Extractor {
+	pub fn new(item: &Item) -> Self {
+		Self {
+			item: item.clone(),
+			output: OnceLock::new(),
+		}
+	}
+
+	async fn get_inner(&self) -> Result<&HashMap<Label, PileValue>, std::io::Error> {
+		if let Some(x) = self.output.get() {
+			return Ok(x);
+		}
+
+		let key = self.item.key();
+		let ext = key.as_str().rsplit('.').next();
+		if !matches!(ext, Some("mp3") | Some("aiff") | Some("aif") | Some("wav")) {
+			return Ok(self.output.get_or_init(HashMap::new));
+		}
+
+		let reader = SyncReadBridge::new_current(self.item.read().await?);
+		let tag = match tokio::task::spawn_blocking(move || Tag::read_from2(BufReader::new(reader)))
+			.await
+		{
+			Ok(Ok(tag)) => tag,
+
+			Ok(Err(id3::Error {
+				kind: id3::ErrorKind::NoTag,
+				..
+			})) => {
+				return Ok(self.output.get_or_init(HashMap::new));
+			}
+
+			Ok(Err(id3::Error {
+				kind: id3::ErrorKind::Io(e),
+				..
+			})) => return Err(e),
+
+			Ok(Err(e)) => return Err(std::io::Error::new(std::io::ErrorKind::InvalidData, e)),
+			Err(e) => return Err(e.into()),
+		};
+
+		let mut output: HashMap<Label, Vec<PileValue>> = HashMap::new();
+		for frame in tag.frames() {
+			if let Some(texts) = frame.content().text_values() {
+				let name = frame_id_to_field(frame.id());
+				if let Some(key) = Label::new(name) {
+					for text in texts {
+						output
+							.entry(key.clone())
+							.or_default()
+							.push(PileValue::String(Arc::new(text.into())));
+					}
+				}
+			}
+		}
+
+		let output = output
+			.into_iter()
+			.map(|(k, v)| (k, PileValue::Array(Arc::new(v))))
+			.collect();
+
+		return Ok(self.output.get_or_init(|| output));
+	}
+}
+
+/// Map an ID3 frame ID to the equivalent Vorbis Comment field name.
+/// Falls back to the lowercased frame ID if no mapping exists.
+fn frame_id_to_field(id: &str) -> Cow<'static, str> {
+	match id {
+		// spell:off
+		"TIT2" => Cow::Borrowed("title"),
+		"TIT1" => Cow::Borrowed("grouping"),
+		"TIT3" => Cow::Borrowed("subtitle"),
+		"TPE1" => Cow::Borrowed("artist"),
+		"TPE2" => Cow::Borrowed("albumartist"),
+		"TPE3" => Cow::Borrowed("conductor"),
+		"TOPE" => Cow::Borrowed("originalartist"),
+		"TALB" => Cow::Borrowed("album"),
+		"TOAL" => Cow::Borrowed("originalalbum"),
+		"TRCK" => Cow::Borrowed("tracknumber"),
+		"TPOS" => Cow::Borrowed("discnumber"),
+		"TSST" => Cow::Borrowed("discsubtitle"),
+		"TDRC" | "TYER" => Cow::Borrowed("date"),
+		"TDOR" | "TORY" => Cow::Borrowed("originaldate"),
+		"TCON" => Cow::Borrowed("genre"),
+		"TCOM" => Cow::Borrowed("composer"),
+		"TEXT" => Cow::Borrowed("lyricist"),
+		"TPUB" => Cow::Borrowed("label"),
+		"TSRC" => Cow::Borrowed("isrc"),
+		"TBPM" => Cow::Borrowed("bpm"),
+		"TLAN" => Cow::Borrowed("language"),
+		"TMED" => Cow::Borrowed("media"),
+		"TMOO" => Cow::Borrowed("mood"),
+		"TCOP" => Cow::Borrowed("copyright"),
+		"TENC" => Cow::Borrowed("encodedby"),
+		"TSSE" => Cow::Borrowed("encodersettings"),
+		"TSOA" => Cow::Borrowed("albumsort"),
+		"TSOP" => Cow::Borrowed("artistsort"),
+		"TSOT" => Cow::Borrowed("titlesort"),
+		"MVNM" => Cow::Borrowed("movement"),
+		"MVIN" => Cow::Borrowed("movementnumber"),
+		_ => Cow::Owned(id.to_lowercase()),
+		// spell:on
+	}
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for Id3Extractor {
+	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> {
+		Ok(self.get_inner().await?.keys().cloned().collect())
+	}
+}

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

@@ -0,0 +1,99 @@
+mod flac;
+use std::{collections::HashMap, sync::Arc};
+
+pub use flac::*;
+
+mod id3;
+pub use id3::*;
+
+mod fs;
+pub use fs::*;
+
+mod epub;
+pub use epub::*;
+
+mod exif;
+pub use exif::*;
+
+mod pdf;
+pub use pdf::*;
+
+mod toml;
+use pile_config::Label;
+pub use toml::*;
+
+mod sidecar;
+pub use sidecar::*;
+
+use crate::{
+	extract::{misc::MapExtractor, traits::ObjectExtractor},
+	value::{Item, PileValue},
+};
+
+pub struct ItemExtractor {
+	inner: MapExtractor,
+}
+
+impl ItemExtractor {
+	#[expect(clippy::unwrap_used)]
+	pub fn new(item: &Item) -> Self {
+		let inner = MapExtractor {
+			inner: HashMap::from([
+				(
+					Label::new("flac").unwrap(),
+					PileValue::ObjectExtractor(Arc::new(FlacExtractor::new(item))),
+				),
+				(
+					Label::new("id3").unwrap(),
+					PileValue::ObjectExtractor(Arc::new(Id3Extractor::new(item))),
+				),
+				(
+					Label::new("fs").unwrap(),
+					PileValue::ObjectExtractor(Arc::new(FsExtractor::new(item))),
+				),
+				(
+					Label::new("epub").unwrap(),
+					PileValue::ObjectExtractor(Arc::new(EpubExtractor::new(item))),
+				),
+				(
+					Label::new("exif").unwrap(),
+					PileValue::ObjectExtractor(Arc::new(ExifExtractor::new(item))),
+				),
+				(
+					Label::new("pdf").unwrap(),
+					PileValue::ObjectExtractor(Arc::new(PdfExtractor::new(item))),
+				),
+				(
+					Label::new("toml").unwrap(),
+					PileValue::ObjectExtractor(Arc::new(TomlExtractor::new(item))),
+				),
+				(
+					Label::new("sidecar").unwrap(),
+					PileValue::ObjectExtractor(Arc::new(SidecarExtractor::new(item))),
+				),
+			]),
+		};
+
+		Self { inner }
+	}
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for ItemExtractor {
+	async fn field(&self, name: &pile_config::Label) -> Result<Option<PileValue>, std::io::Error> {
+		self.inner.field(name).await
+	}
+
+	#[expect(clippy::unwrap_used)]
+	async fn fields(&self) -> Result<Vec<Label>, std::io::Error> {
+		return Ok(vec![
+			Label::new("flac").unwrap(),
+			Label::new("id3").unwrap(),
+			Label::new("fs").unwrap(),
+			Label::new("epub").unwrap(),
+			Label::new("exif").unwrap(),
+			Label::new("pdf").unwrap(),
+			Label::new("sidecar").unwrap(),
+		]);
+	}
+}

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

@@ -0,0 +1,61 @@
+use pile_config::Label;
+use std::sync::Arc;
+
+#[cfg(feature = "pdfium")]
+mod pdf_pages;
+#[cfg(feature = "pdfium")]
+pub use pdf_pages::*;
+
+mod pdf_meta;
+pub use pdf_meta::*;
+
+mod pdf_text;
+pub use pdf_text::*;
+
+use crate::{
+	extract::traits::ObjectExtractor,
+	value::{Item, PileValue},
+};
+
+pub struct PdfExtractor {
+	text: Arc<PdfTextExtractor>,
+	meta: Arc<PdfMetaExtractor>,
+	#[cfg(feature = "pdfium")]
+	pages: Arc<PdfPagesExtractor>,
+}
+
+impl PdfExtractor {
+	pub fn new(item: &Item) -> Self {
+		Self {
+			text: Arc::new(PdfTextExtractor::new(item)),
+			meta: Arc::new(PdfMetaExtractor::new(item)),
+			#[cfg(feature = "pdfium")]
+			pages: Arc::new(PdfPagesExtractor::new(item)),
+		}
+	}
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for PdfExtractor {
+	async fn field(&self, name: &pile_config::Label) -> Result<Option<PileValue>, std::io::Error> {
+		match name.as_str() {
+			"text" => self.text.field(name).await,
+			"meta" => Ok(Some(PileValue::ObjectExtractor(self.meta.clone()))),
+			#[cfg(feature = "pdfium")]
+			"pages" => Ok(Some(PileValue::ListExtractor(self.pages.clone()))),
+			_ => Ok(None),
+		}
+	}
+
+	#[expect(clippy::unwrap_used)]
+	async fn fields(&self) -> Result<Vec<Label>, std::io::Error> {
+		Ok(vec![
+			Label::new("text").unwrap(),
+			Label::new("meta").unwrap(),
+			#[cfg(feature = "pdfium")]
+			Label::new("cover").unwrap(),
+			#[cfg(feature = "pdfium")]
+			Label::new("pages").unwrap(),
+		])
+	}
+}

crates/pile-value/src/extract/item/pdf/pdf_meta.rs

@@ -0,0 +1,132 @@
+use pdf::file::FileOptions;
+use pdf::primitive::{Date, TimeRel};
+use pile_config::Label;
+use std::{
+	collections::HashMap,
+	io::BufReader,
+	sync::{Arc, OnceLock},
+};
+use tracing::trace;
+
+use crate::{
+	extract::traits::ObjectExtractor,
+	value::{Item, PileValue, SyncReadBridge},
+};
+
+pub struct PdfMetaExtractor {
+	item: Item,
+	output: OnceLock<HashMap<Label, PileValue>>,
+}
+
+impl PdfMetaExtractor {
+	pub fn new(item: &Item) -> Self {
+		Self {
+			item: item.clone(),
+			output: OnceLock::new(),
+		}
+	}
+
+	async fn get_inner(&self) -> Result<&HashMap<Label, PileValue>, std::io::Error> {
+		if let Some(x) = self.output.get() {
+			return Ok(x);
+		}
+
+		let reader = SyncReadBridge::new_current(self.item.read().await?);
+		let raw_meta = tokio::task::spawn_blocking(move || {
+			let mut bytes = Vec::new();
+			std::io::Read::read_to_end(&mut BufReader::new(reader), &mut bytes)?;
+
+			let file = match FileOptions::cached().load(bytes) {
+				Ok(x) => x,
+				Err(pdf::PdfError::Io { source }) => return Err(source),
+				Err(error) => {
+					return Err(std::io::Error::new(
+						std::io::ErrorKind::InvalidData,
+						error.to_string(),
+					));
+				}
+			};
+
+			let page_count = file.num_pages();
+
+			let mut meta: Vec<(&'static str, Option<String>)> = Vec::new();
+
+			if let Some(info) = &file.trailer.info_dict {
+				use pdf::primitive::PdfString;
+				let fields: &[(&'static str, Option<&PdfString>)] = &[
+					("title", info.title.as_ref()),
+					("author", info.author.as_ref()),
+					("subject", info.subject.as_ref()),
+					("keywords", info.keywords.as_ref()),
+					("creator", info.creator.as_ref()),
+					("producer", info.producer.as_ref()),
+				];
+
+				for (key, val) in fields {
+					meta.push((key, val.map(|s| s.to_string_lossy())));
+				}
+
+				meta.push((
+					"creation_date",
+					info.creation_date.as_ref().map(format_date),
+				));
+				meta.push(("mod_date", info.mod_date.as_ref().map(format_date)));
+			}
+
+			Ok::<_, std::io::Error>((page_count, meta))
+		})
+		.await
+		.map_err(std::io::Error::other)?;
+
+		let (page_count, raw_meta) = match raw_meta {
+			Ok(x) => x,
+			Err(error) => {
+				trace!(message = "Could not process pdf", ?error, key = ?self.item.key());
+				return Ok(self.output.get_or_init(HashMap::new));
+			}
+		};
+
+		let mut output: HashMap<Label, PileValue> = HashMap::new();
+
+		#[expect(clippy::unwrap_used)]
+		output.insert(
+			Label::new("pages").unwrap(),
+			PileValue::U64(page_count as u64),
+		);
+
+		#[expect(clippy::unwrap_used)]
+		for (key, val) in raw_meta {
+			let label = Label::new(key).unwrap();
+			let value = match val {
+				Some(s) => PileValue::String(Arc::new(s.into())),
+				None => PileValue::Null,
+			};
+			output.insert(label, value);
+		}
+
+		return Ok(self.output.get_or_init(|| output));
+	}
+}
+
+fn format_date(d: &Date) -> String {
+	let tz = match d.rel {
+		TimeRel::Universal => "Z".to_owned(),
+		TimeRel::Later => format!("+{:02}:{:02}", d.tz_hour, d.tz_minute),
+		TimeRel::Earlier => format!("-{:02}:{:02}", d.tz_hour, d.tz_minute),
+	};
+	format!(
+		"{:04}-{:02}-{:02}T{:02}:{:02}:{:02}{}",
+		d.year, d.month, d.day, d.hour, d.minute, d.second, tz
+	)
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for PdfMetaExtractor {
+	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> {
+		Ok(self.get_inner().await?.keys().cloned().collect())
+	}
+}

crates/pile-value/src/extract/item/pdf/pdf_pages.rs

@@ -0,0 +1,107 @@
+use image::ImageFormat;
+use pdfium_render::prelude::*;
+use std::{
+	io::{BufReader, Cursor},
+	sync::Arc,
+};
+use tracing::trace;
+
+use crate::{
+	extract::traits::ListExtractor,
+	value::{Item, PileValue, SyncReadBridge},
+};
+
+pub struct PdfPagesExtractor {
+	item: Item,
+}
+
+impl PdfPagesExtractor {
+	pub fn new(item: &Item) -> Self {
+		Self { item: item.clone() }
+	}
+
+	async fn get_bytes(&self) -> Result<Vec<u8>, std::io::Error> {
+		let reader = SyncReadBridge::new_current(self.item.read().await?);
+		tokio::task::spawn_blocking(move || {
+			let mut b = Vec::new();
+			std::io::Read::read_to_end(&mut BufReader::new(reader), &mut b)?;
+			Ok::<_, std::io::Error>(b)
+		})
+		.await
+		.map_err(std::io::Error::other)?
+	}
+}
+
+#[async_trait::async_trait]
+impl ListExtractor for PdfPagesExtractor {
+	async fn get(&self, idx: usize) -> Result<Option<PileValue>, std::io::Error> {
+		let bytes = self.get_bytes().await?;
+		let png = tokio::task::spawn_blocking(move || {
+			let pdfium = Pdfium::default();
+			let doc = pdfium
+				.load_pdf_from_byte_slice(&bytes, None)
+				.map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidData, e.to_string()))?;
+			if idx >= doc.pages().len() as usize {
+				return Ok::<_, std::io::Error>(None);
+			}
+			let render_config = PdfRenderConfig::new().set_target_width(1024);
+			let page = doc
+				.pages()
+				.get(idx as u16)
+				.map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidData, e.to_string()))?;
+			let image = page
+				.render_with_config(&render_config)
+				.map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidData, e.to_string()))?
+				.as_image();
+			let mut png_bytes = Vec::new();
+			image
+				.write_to(&mut Cursor::new(&mut png_bytes), ImageFormat::Png)
+				.map_err(|e| std::io::Error::other(e.to_string()))?;
+			Ok(Some(png_bytes))
+		})
+		.await
+		.map_err(std::io::Error::other)?;
+
+		let value = match png {
+			Ok(None) => return Ok(None),
+			Ok(Some(bytes)) => PileValue::Blob {
+				mime: mime::IMAGE_PNG,
+				bytes: Arc::new(bytes),
+			},
+			Err(error) => {
+				trace!(message = "Could not render pdf page", ?error, idx, key = ?self.item.key());
+				PileValue::Null
+			}
+		};
+		Ok(Some(value))
+	}
+
+	async fn len(&self) -> Result<usize, std::io::Error> {
+		let bytes = self.get_bytes().await?;
+		let count = tokio::task::spawn_blocking(move || {
+			let pdfium = Pdfium::default();
+			let doc = pdfium
+				.load_pdf_from_byte_slice(&bytes, None)
+				.map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidData, e.to_string()))?;
+			Ok::<_, std::io::Error>(doc.pages().len() as usize)
+		})
+		.await
+		.map_err(std::io::Error::other)?;
+		match count {
+			Ok(n) => Ok(n),
+			Err(error) => {
+				trace!(message = "Could not read pdf page count", ?error, key = ?self.item.key());
+				Ok(0)
+			}
+		}
+	}
+
+	// Override, extracting all pages is very slow,
+	// and we can't display binary in json anyway
+	async fn to_json(&self) -> Result<serde_json::Value, std::io::Error> {
+		Ok(serde_json::Value::String(format!(
+			"<PdfPages ({} pages)>",
+			self.len().await?
+		)))
+	}
+}

crates/pile-value/src/extract/item/pdf/pdf_text.rs

@@ -0,0 +1,112 @@
+use pdf::content::{Op, TextDrawAdjusted};
+use pdf::file::FileOptions;
+use pile_config::Label;
+use std::{
+	collections::HashMap,
+	io::BufReader,
+	sync::{Arc, OnceLock},
+};
+use tracing::trace;
+
+use crate::{
+	extract::traits::ObjectExtractor,
+	value::{Item, PileValue, SyncReadBridge},
+};
+
+pub struct PdfTextExtractor {
+	item: Item,
+	output: OnceLock<HashMap<Label, PileValue>>,
+}
+
+impl PdfTextExtractor {
+	pub fn new(item: &Item) -> Self {
+		Self {
+			item: item.clone(),
+			output: OnceLock::new(),
+		}
+	}
+
+	async fn get_inner(&self) -> Result<&HashMap<Label, PileValue>, std::io::Error> {
+		if let Some(x) = self.output.get() {
+			return Ok(x);
+		}
+
+		let reader = SyncReadBridge::new_current(self.item.read().await?);
+		let raw_text = tokio::task::spawn_blocking(move || {
+			let mut bytes = Vec::new();
+			std::io::Read::read_to_end(&mut BufReader::new(reader), &mut bytes)?;
+
+			let file = match FileOptions::cached().load(bytes) {
+				Ok(x) => x,
+				Err(pdf::PdfError::Io { source }) => return Err(source),
+				Err(error) => {
+					return Err(std::io::Error::new(
+						std::io::ErrorKind::InvalidData,
+						error.to_string(),
+					));
+				}
+			};
+
+			let mut text_parts: Vec<String> = Vec::new();
+
+			for page in file.pages() {
+				let page = page.map_err(|e| {
+					std::io::Error::new(std::io::ErrorKind::InvalidData, e.to_string())
+				})?;
+
+				if let Some(content) = &page.contents {
+					let ops = content.operations(&file.resolver()).map_err(|e| {
+						std::io::Error::new(std::io::ErrorKind::InvalidData, e.to_string())
+					})?;
+
+					for op in ops {
+						match op {
+							Op::TextDraw { text } => {
+								text_parts.push(text.to_string_lossy());
+							}
+							Op::TextDrawAdjusted { array } => {
+								for item in array {
+									if let TextDrawAdjusted::Text(text) = item {
+										text_parts.push(text.to_string_lossy());
+									}
+								}
+							}
+							_ => {}
+						}
+					}
+				}
+			}
+
+			Ok::<_, std::io::Error>(text_parts.join(" "))
+		})
+		.await
+		.map_err(std::io::Error::other)?;
+
+		let raw_text = match raw_text {
+			Ok(x) => x,
+			Err(error) => {
+				trace!(message = "Could not process pdf", ?error, key = ?self.item.key());
+				return Ok(self.output.get_or_init(HashMap::new));
+			}
+		};
+
+		#[expect(clippy::unwrap_used)]
+		let output = HashMap::from([(
+			Label::new("text").unwrap(),
+			PileValue::String(Arc::new(raw_text.into())),
+		)]);
+
+		return Ok(self.output.get_or_init(|| output));
+	}
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for PdfTextExtractor {
+	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> {
+		Ok(self.get_inner().await?.keys().cloned().collect())
+	}
+}

crates/pile-value/src/extract/item/sidecar.rs

@@ -0,0 +1,45 @@
+use pile_config::Label;
+use std::sync::OnceLock;
+
+use super::TomlExtractor;
+use crate::{
+	extract::traits::ObjectExtractor,
+	value::{Item, PileValue},
+};
+
+pub struct SidecarExtractor {
+	item: Item,
+	output: OnceLock<Option<TomlExtractor>>,
+}
+
+impl SidecarExtractor {
+	pub fn new(item: &Item) -> Self {
+		Self {
+			item: item.clone(),
+			output: OnceLock::new(),
+		}
+	}
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for SidecarExtractor {
+	async fn field(&self, name: &Label) -> Result<Option<PileValue>, std::io::Error> {
+		match self
+			.output
+			.get_or_init(|| self.item.sidecar().map(TomlExtractor::new))
+		{
+			Some(x) => Ok(x.field(name).await?),
+			None => Ok(Some(PileValue::Null)),
+		}
+	}
+
+	async fn fields(&self) -> Result<Vec<Label>, std::io::Error> {
+		match self
+			.output
+			.get_or_init(|| self.item.sidecar().map(TomlExtractor::new))
+		{
+			Some(x) => Ok(x.fields().await?),
+			None => Ok(Vec::new()),
+		}
+	}
+}

crates/pile-value/src/extract/item/toml.rs

@@ -0,0 +1,78 @@
+use pile_config::Label;
+use std::{
+	collections::HashMap,
+	sync::{Arc, OnceLock},
+};
+
+use crate::{
+	extract::traits::ObjectExtractor,
+	value::{AsyncReader, Item, PileValue},
+};
+
+fn toml_to_pile(value: toml::Value) -> PileValue {
+	match value {
+		toml::Value::String(s) => PileValue::String(Arc::new(s.into())),
+		toml::Value::Integer(i) => PileValue::String(Arc::new(i.to_string().into())),
+		toml::Value::Float(f) => PileValue::String(Arc::new(f.to_string().into())),
+		toml::Value::Boolean(b) => PileValue::String(Arc::new(b.to_string().into())),
+		toml::Value::Datetime(d) => PileValue::String(Arc::new(d.to_string().into())),
+		toml::Value::Array(a) => {
+			PileValue::Array(Arc::new(a.into_iter().map(toml_to_pile).collect()))
+		}
+		toml::Value::Table(_) => PileValue::Null,
+	}
+}
+
+pub struct TomlExtractor {
+	item: Item,
+	output: OnceLock<HashMap<Label, PileValue>>,
+}
+
+impl TomlExtractor {
+	pub fn new(item: &Item) -> Self {
+		Self {
+			item: item.clone(),
+			output: OnceLock::new(),
+		}
+	}
+
+	async fn get_inner(&self) -> Result<&HashMap<Label, PileValue>, std::io::Error> {
+		if let Some(x) = self.output.get() {
+			return Ok(x);
+		}
+
+		let mut reader = match self.item.read().await {
+			Ok(r) => r,
+			Err(e) if e.kind() == std::io::ErrorKind::NotFound => {
+				return Ok(self.output.get_or_init(HashMap::new));
+			}
+			Err(e) => return Err(e),
+		};
+		let bytes = reader.read_to_end().await?;
+		let toml: toml::Value = match toml::from_slice(&bytes) {
+			Ok(x) => x,
+			Err(_) => return Ok(self.output.get_or_init(HashMap::new)),
+		};
+
+		let output: HashMap<Label, PileValue> = match toml {
+			toml::Value::Table(t) => t
+				.into_iter()
+				.filter_map(|(k, v)| Label::new(&k).map(|label| (label, toml_to_pile(v))))
+				.collect(),
+			_ => HashMap::new(),
+		};
+
+		return Ok(self.output.get_or_init(|| output));
+	}
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for TomlExtractor {
+	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> {
+		Ok(self.get_inner().await?.keys().cloned().collect())
+	}
+}

crates/pile-value/src/extract/misc/list.rs

@@ -0,0 +1,24 @@
+use std::sync::Arc;
+
+use crate::{extract::traits::ListExtractor, value::PileValue};
+
+pub struct ArrayExtractor {
+	inner: Arc<Vec<PileValue>>,
+}
+
+impl ArrayExtractor {
+	pub fn new(inner: Arc<Vec<PileValue>>) -> Self {
+		Self { inner }
+	}
+}
+
+#[async_trait::async_trait]
+impl ListExtractor for ArrayExtractor {
+	async fn get(&self, idx: usize) -> Result<Option<PileValue>, std::io::Error> {
+		Ok(self.inner.get(idx).cloned())
+	}
+
+	async fn len(&self) -> Result<usize, std::io::Error> {
+		Ok(self.inner.len())
+	}
+}

crates/pile-value/src/extract/misc/map.rs

@@ -0,0 +1,20 @@
+use pile_config::Label;
+use std::collections::HashMap;
+
+use crate::{extract::traits::ObjectExtractor, value::PileValue};
+
+#[derive(Default)]
+pub struct MapExtractor {
+	pub inner: HashMap<Label, PileValue>,
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for MapExtractor {
+	async fn field(&self, name: &Label) -> Result<Option<PileValue>, std::io::Error> {
+		Ok(self.inner.get(name).cloned())
+	}
+
+	async fn fields(&self) -> Result<Vec<Label>, std::io::Error> {
+		Ok(self.inner.keys().cloned().collect())
+	}
+}

crates/pile-value/src/extract/misc/mod.rs

@@ -0,0 +1,8 @@
+mod list;
+pub use list::*;
+
+mod vec;
+pub use vec::*;
+
+mod map;
+pub use map::*;

crates/pile-value/src/extract/misc/vec.rs

@@ -0,0 +1,17 @@
+use crate::{extract::traits::ListExtractor, value::PileValue};
+
+#[derive(Default)]
+pub struct VecExtractor {
+	pub inner: Vec<PileValue>,
+}
+
+#[async_trait::async_trait]
+impl ListExtractor for VecExtractor {
+	async fn get(&self, idx: usize) -> Result<Option<PileValue>, std::io::Error> {
+		Ok(self.inner.get(idx).cloned())
+	}
+
+	async fn len(&self) -> Result<usize, std::io::Error> {
+		Ok(self.inner.len())
+	}
+}

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

@@ -0,0 +1,4 @@
+pub mod item;
+pub mod misc;
+pub mod string;
+pub mod traits;

crates/pile-value/src/extract/string.rs

@@ -0,0 +1,51 @@
+use pile_config::Label;
+use smartstring::{LazyCompact, SmartString};
+use std::sync::Arc;
+
+use crate::{extract::traits::ObjectExtractor, value::PileValue};
+
+pub struct StringExtractor {
+	item: Arc<SmartString<LazyCompact>>,
+}
+
+impl StringExtractor {
+	pub fn new(item: &Arc<SmartString<LazyCompact>>) -> Self {
+		Self { item: item.clone() }
+	}
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for StringExtractor {
+	async fn field(&self, name: &Label) -> Result<Option<PileValue>, std::io::Error> {
+		Ok(match name.as_str() {
+			"trim" => Some(PileValue::String(Arc::new(
+				self.item.as_str().trim().into(),
+			))),
+
+			"upper" => Some(PileValue::String(Arc::new(
+				self.item.as_str().to_lowercase().into(),
+			))),
+
+			"lower" => Some(PileValue::String(Arc::new(
+				self.item.as_str().to_uppercase().into(),
+			))),
+
+			"nonempty" => Some(match self.item.is_empty() {
+				true => PileValue::Null,
+				false => PileValue::String(self.item.clone()),
+			}),
+
+			_ => None,
+		})
+	}
+
+	#[expect(clippy::unwrap_used)]
+	async fn fields(&self) -> Result<Vec<Label>, std::io::Error> {
+		return Ok(vec![
+			Label::new("trim").unwrap(),
+			Label::new("upper").unwrap(),
+			Label::new("lower").unwrap(),
+			Label::new("nonempty").unwrap(),
+		]);
+	}
+}

crates/pile-value/src/extract/traits.rs

@@ -0,0 +1,68 @@
+/// An attachment that extracts metadata from an [Item].
+///
+/// Metadata is exposed as an immutable map of {label: value},
+/// much like a json object.
+#[async_trait::async_trait]
+pub trait ObjectExtractor: 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
+	async fn field(
+		&self,
+		name: &pile_config::Label,
+	) -> Result<Option<crate::value::PileValue>, std::io::Error>;
+
+	/// Return all fields in this extractor.
+	/// `Self::field` must return [Some] for all these keys
+	/// and [None] for all others.
+	async fn fields(&self) -> Result<Vec<pile_config::Label>, std::io::Error>;
+
+	/// Convert this to a JSON value.
+	async fn to_json(&self) -> Result<serde_json::Value, std::io::Error> {
+		let keys = self.fields().await?;
+		let mut map = serde_json::Map::new();
+		for k in &keys {
+			let v = match self.field(k).await? {
+				Some(x) => x,
+				None => continue,
+			};
+			map.insert(k.to_string(), Box::pin(v.to_json()).await?);
+		}
+
+		Ok(serde_json::Value::Object(map))
+	}
+}
+
+/// An attachment that extracts metadata from an [Item].
+///
+/// Metadata is exposed as an immutable list of values.
+#[async_trait::async_trait]
+pub trait ListExtractor: Send + Sync {
+	/// Get the item at index `idx`.
+	/// Indices start at zero, and must be consecutive.
+	/// - returns `None` if `idx` is out of range
+	/// - returns `Some(Null)` if `None` is at `idx`
+	async fn get(&self, idx: usize) -> Result<Option<crate::value::PileValue>, std::io::Error>;
+
+	async fn len(&self) -> Result<usize, std::io::Error>;
+
+	async fn is_empty(&self) -> Result<bool, std::io::Error> {
+		Ok(self.len().await? == 0)
+	}
+
+	/// Convert this list to a JSON value.
+	async fn to_json(&self) -> Result<serde_json::Value, std::io::Error> {
+		let len = self.len().await?;
+		let mut list = Vec::with_capacity(len);
+		for i in 0..len {
+			#[expect(clippy::expect_used)]
+			let v = self
+				.get(i)
+				.await?
+				.expect("value must be present according to length");
+			list.push(Box::pin(v.to_json()).await?);
+		}
+
+		Ok(serde_json::Value::Array(list))
+	}
+}