Filter by mime

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

@@ -1,3 +1,13 @@
+#[derive(Debug, Clone)]
+pub struct ExtractState {
+	/// If true, extract all fields from all items.
+	/// Do not pre-filter using mime type.
+	///
+	/// This may detect additional fields, but
+	/// makes extraction take much longer
+	pub ignore_mime: bool,
+}
+
 /// An attachment that extracts metadata from an [Item].
 ///
 /// Metadata is exposed as an immutable map of {label: value},
@@ -15,6 +25,7 @@ pub trait ObjectExtractor: Send + Sync {
 	/// this fn should return `Ok(Some(None))`.
 	async fn field(
 		&self,
+		state: &ExtractState,
 		name: &pile_config::Label,
 		args: Option<&str>,
 	) -> Result<Option<crate::value::PileValue>, std::io::Error>;
@@ -25,15 +36,15 @@ pub trait ObjectExtractor: Send + Sync {
 	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> {
+	async fn to_json(&self, state: &ExtractState) -> 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, None).await? {
+			let v = match self.field(state, k, None).await? {
 				Some(x) => x,
 				None => continue,
 			};
-			map.insert(k.to_string(), Box::pin(v.to_json()).await?);
+			map.insert(k.to_string(), Box::pin(v.to_json(state)).await?);
 		}
 
 		Ok(serde_json::Value::Object(map))
@@ -49,25 +60,25 @@ pub trait ListExtractor: Send + Sync {
 	/// 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 get(
+		&self,
+		state: &ExtractState,
+		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)
-	}
+	async fn len(&self, state: &ExtractState) -> Result<usize, std::io::Error>;
 
 	/// Convert this list to a JSON value.
-	async fn to_json(&self) -> Result<serde_json::Value, std::io::Error> {
-		let len = self.len().await?;
+	async fn to_json(&self, state: &ExtractState) -> Result<serde_json::Value, std::io::Error> {
+		let len = self.len(state).await?;
 		let mut list = Vec::with_capacity(len);
 		for i in 0..len {
 			#[expect(clippy::expect_used)]
 			let v = self
-				.get(i)
+				.get(state, i)
 				.await?
 				.expect("value must be present according to length");
-			list.push(Box::pin(v.to_json()).await?);
+			list.push(Box::pin(v.to_json(state)).await?);
 		}
 
 		Ok(serde_json::Value::Array(list))