Serve item bytes & fields

crates/pile-dataset/src/dataset.rs

@@ -1,6 +1,7 @@
 use chrono::{DateTime, Utc};
-use pile_config::{ConfigToml, Label, Source};
+use pile_config::{ConfigToml, Label, Source, objectpath::ObjectPath};
 use pile_toolbox::cancelabletask::{CancelFlag, CancelableTaskError};
+use serde_json::Value;
 use std::{collections::HashMap, io::ErrorKind, path::PathBuf, sync::Arc, time::Instant};
 use tantivy::{Executor, Index, IndexWriter, TantivyError, collector::TopDocs};
 use thiserror::Error;
@@ -9,7 +10,8 @@ use tokio_stream::{StreamExt, wrappers::ReceiverStream};
 use tracing::{debug, info, trace, warn};
 
 use crate::{
-	DataSource, Item,
+	DataSource, Item, PileValue,
+	extract::MetaExtractor,
 	index::{DbFtsIndex, FtsLookupResult},
 	path_ts_earliest,
 	source::{DirDataSource, S3DataSource},
@@ -170,6 +172,25 @@ impl Datasets {
 		self.sources.get(source)?.get(key).await
 	}
 
+	/// Extract a field from an item by object path.
+	/// Returns `None` if the item or field is not found.
+	pub async fn get_field(
+		&self,
+		source: &Label,
+		key: &str,
+		path: &ObjectPath,
+	) -> Result<Option<Value>, std::io::Error> {
+		let Some(item) = self.get(source, key).await else {
+			return Ok(None);
+		};
+		let extractor = MetaExtractor::new(&item);
+		let root = PileValue::Extractor(Arc::new(extractor));
+		let Some(value) = root.query(path).await? else {
+			return Ok(None);
+		};
+		Ok(Some(value.to_json().await?))
+	}
+
 	//
 	// MARK: fts
 	//