Add json extractor

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

@@ -0,0 +1,87 @@
+use pile_config::Label;
+use std::{
+	collections::HashMap,
+	sync::{Arc, OnceLock},
+};
+
+use crate::{
+	extract::traits::{ExtractState, ObjectExtractor},
+	value::{AsyncReader, Item, PileValue},
+};
+
+fn json_to_pile(value: serde_json::Value) -> PileValue {
+	match value {
+		serde_json::Value::Null => PileValue::Null,
+		serde_json::Value::Bool(b) => PileValue::String(Arc::new(b.to_string().into())),
+		serde_json::Value::Number(n) => PileValue::String(Arc::new(n.to_string().into())),
+		serde_json::Value::String(s) => PileValue::String(Arc::new(s.into())),
+		serde_json::Value::Array(a) => {
+			PileValue::Array(Arc::new(a.into_iter().map(json_to_pile).collect()))
+		}
+		serde_json::Value::Object(_) => PileValue::Null,
+	}
+}
+
+pub struct JsonExtractor {
+	item: Item,
+	output: OnceLock<HashMap<Label, PileValue>>,
+}
+
+impl JsonExtractor {
+	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 = self.item.read().await?;
+		let bytes = reader.read_to_end().await?;
+		let json: serde_json::Value = match serde_json::from_slice(&bytes) {
+			Ok(x) => x,
+			Err(_) => return Ok(self.output.get_or_init(HashMap::new)),
+		};
+
+		let output: HashMap<Label, PileValue> = match json {
+			serde_json::Value::Object(map) => map
+				.into_iter()
+				.filter_map(|(k, v)| Label::new(&k).map(|label| (label, json_to_pile(v))))
+				.collect(),
+			_ => HashMap::new(),
+		};
+
+		return Ok(self.output.get_or_init(|| output));
+	}
+}
+
+#[async_trait::async_trait]
+impl ObjectExtractor for JsonExtractor {
+	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::APPLICATION
+				&& self.item.mime().type_() != mime::TEXT)
+		{
+			return Ok(None);
+		}
+
+		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

@@ -18,6 +18,9 @@ pub use exif::*;
 mod pdf;
 pub use pdf::*;
 
+mod json;
+pub use json::*;
+
 mod toml;
 use pile_config::Label;
 pub use toml::*;
@@ -66,6 +69,10 @@ impl ItemExtractor {
 					Label::new("pdf").unwrap(),
 					PileValue::ObjectExtractor(Arc::new(PdfExtractor::new(item))),
 				),
+				(
+					Label::new("json").unwrap(),
+					PileValue::ObjectExtractor(Arc::new(JsonExtractor::new(item))),
+				),
 				(
 					Label::new("toml").unwrap(),
 					PileValue::ObjectExtractor(Arc::new(TomlExtractor::new(item))),
@@ -101,6 +108,7 @@ impl ObjectExtractor for ItemExtractor {
 			Label::new("epub").unwrap(),
 			Label::new("exif").unwrap(),
 			Label::new("pdf").unwrap(),
+			Label::new("json").unwrap(),
 			Label::new("sidecar").unwrap(),
 		]);
 	}