Add text extractor

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

@@ -28,6 +28,9 @@ pub use toml::*;
 mod group;
 pub use group::*;
 
+mod text;
+pub use text::*;
+
 use crate::{
 	extract::{
 		misc::MapExtractor,
@@ -77,6 +80,10 @@ impl ItemExtractor {
 					Label::new("toml").unwrap(),
 					PileValue::ObjectExtractor(Arc::new(TomlExtractor::new(item))),
 				),
+				(
+					Label::new("text").unwrap(),
+					PileValue::ObjectExtractor(Arc::new(TextExtractor::new(item))),
+				),
 				(
 					Label::new("groups").unwrap(),
 					PileValue::ObjectExtractor(Arc::new(GroupExtractor::new(item))),
@@ -110,6 +117,7 @@ impl ObjectExtractor for ItemExtractor {
 			Label::new("pdf").unwrap(),
 			Label::new("json").unwrap(),
 			Label::new("toml").unwrap(),
+			Label::new("text").unwrap(),
 			Label::new("groups").unwrap(),
 		]);
 	}

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

@@ -0,0 +1,67 @@
+use pile_config::Label;
+use std::sync::{Arc, OnceLock};
+
+use crate::{
+	extract::traits::{ExtractState, ObjectExtractor},
+	value::{AsyncReader, Item, PileValue},
+};
+
+pub struct TextExtractor {
+	item: Item,
+	output: OnceLock<PileValue>,
+}
+
+impl TextExtractor {
+	pub fn new(item: &Item) -> Self {
+		Self {
+			item: item.clone(),
+			output: OnceLock::new(),
+		}
+	}
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for TextExtractor {
+	async fn field(
+		&self,
+		state: &ExtractState,
+		name: &Label,
+		args: Option<&str>,
+	) -> Result<Option<PileValue>, std::io::Error> {
+		if args.is_some() {
+			return Ok(None);
+		}
+
+		if !state.ignore_mime
+			&& (self.item.mime().type_() != mime::TEXT
+				&& self.item.mime().type_() != mime::APPLICATION)
+		{
+			return Ok(None);
+		}
+
+		if name.as_str() != "text" {
+			return Ok(None);
+		}
+
+		{
+			if let Some(x) = self.output.get() {
+				return Ok(Some(x.clone()));
+			}
+
+			let mut reader = self.item.read().await?;
+			let bytes = reader.read_to_end().await?;
+			let string = String::from_utf8(bytes).ok();
+			let value = match string {
+				Some(x) => PileValue::String(Arc::new(x.into())),
+				None => PileValue::Null,
+			};
+
+			return Ok(Some(self.output.get_or_init(|| value).clone()));
+		}
+	}
+
+	#[expect(clippy::unwrap_used)]
+	async fn fields(&self) -> Result<Vec<Label>, std::io::Error> {
+		Ok(vec![Label::new("text").unwrap()])
+	}
+}