Filter by mime

crates/pile-dataset/src/dataset.rs

@@ -2,6 +2,7 @@ use chrono::{DateTime, Utc};
 use pile_config::{ConfigToml, Label, Source, objectpath::ObjectPath};
 use pile_toolbox::cancelabletask::{CancelFlag, CancelableTaskError};
 use pile_value::{
+	extract::traits::ExtractState,
 	source::{DataSource, DirDataSource, S3DataSource, misc::path_ts_earliest},
 	value::{Item, PileValue},
 };
@@ -187,6 +188,7 @@ impl Datasets {
 	/// Returns `None` if the item or field is not found.
 	pub async fn get_field(
 		&self,
+		state: &ExtractState,
 		source: &Label,
 		key: &str,
 		path: &ObjectPath,
@@ -196,11 +198,11 @@ impl Datasets {
 		};
 
 		let item = PileValue::Item(item);
-		let Some(value) = item.query(path).await? else {
+		let Some(value) = item.query(state, path).await? else {
 			return Ok(None);
 		};
 
-		Ok(Some(value.to_json().await?))
+		Ok(Some(value.to_json(state).await?))
 	}
 
 	//
@@ -210,6 +212,7 @@ impl Datasets {
 	/// Refresh this dataset's fts index.
 	pub async fn fts_refresh(
 		&self,
+		state: &ExtractState,
 		_threads: usize,
 		flag: Option<CancelFlag>,
 	) -> Result<(), CancelableTaskError<DatasetError>> {
@@ -265,9 +268,10 @@ impl Datasets {
 
 				let item = item_result.map_err(DatasetError::from)?;
 				let db = Arc::clone(&db_index);
+				let state = state.clone();
 				join_set.spawn(async move {
 					let key = item.key();
-					let result = db.entry_to_document(&item).await;
+					let result = db.entry_to_document(&state, &item).await;
 					(key, result)
 				});
 

crates/pile-dataset/src/index/index_fts.rs

@@ -1,5 +1,8 @@
 use pile_config::{ConfigToml, DatasetFts, Label};
-use pile_value::value::{Item, PileValue};
+use pile_value::{
+	extract::traits::ExtractState,
+	value::{Item, PileValue},
+};
 use std::{path::PathBuf, sync::LazyLock};
 use tantivy::{
 	DocAddress, Index, ReloadPolicy, TantivyDocument, TantivyError,
@@ -63,6 +66,7 @@ impl DbFtsIndex {
 	/// Turn an entry into a tantivy document
 	pub async fn entry_to_document(
 		&self,
+		state: &ExtractState,
 		item: &Item,
 	) -> Result<Option<TantivyDocument>, TantivyError> {
 		let mut doc = TantivyDocument::default();
@@ -75,7 +79,7 @@ impl DbFtsIndex {
 
 		let mut empty = true;
 		for name in self.fts_cfg().fields.keys() {
-			let x = self.get_field(&item, name).await?;
+			let x = self.get_field(state, &item, name).await?;
 
 			let val = match x {
 				Some(x) => x,
@@ -99,6 +103,7 @@ impl DbFtsIndex {
 
 	pub async fn get_field(
 		&self,
+		state: &ExtractState,
 		extractor: &PileValue,
 		field_name: &Label,
 	) -> Result<Option<String>, std::io::Error> {
@@ -112,7 +117,7 @@ impl DbFtsIndex {
 
 		// Try paths in order, using the first value we find
 		'outer: for path in field.path.as_slice() {
-			let val = match extractor.query(path).await? {
+			let val = match extractor.query(state, path).await? {
 				Some(x) => x,
 				None => return Ok(None),
 			};

crates/pile-dataset/src/serve/field.rs

@@ -5,7 +5,7 @@ use axum::{
 	response::{IntoResponse, Response},
 };
 use pile_config::{Label, objectpath::ObjectPath};
-use pile_value::value::PileValue;
+use pile_value::{extract::traits::ExtractState, value::PileValue};
 use serde::Deserialize;
 use std::{sync::Arc, time::Instant};
 use tracing::debug;
@@ -62,8 +62,10 @@ pub async fn get_field(
 		return StatusCode::NOT_FOUND.into_response();
 	};
 
+	let state = ExtractState { ignore_mime: false };
+
 	let item = PileValue::Item(item);
-	let value = match item.query(&path).await {
+	let value = match item.query(&state, &path).await {
 		Ok(Some(v)) => v,
 		Ok(None) => return StatusCode::NOT_FOUND.into_response(),
 		Err(e) => return (StatusCode::INTERNAL_SERVER_ERROR, format!("{e:?}")).into_response(),
@@ -90,7 +92,7 @@ pub async fn get_field(
 			bytes.as_ref().clone(),
 		)
 			.into_response(),
-		_ => match value.to_json().await {
+		_ => match value.to_json(&state).await {
 			Ok(json) => (StatusCode::OK, Json(json)).into_response(),
 			Err(e) => (StatusCode::INTERNAL_SERVER_ERROR, format!("{e:?}")).into_response(),
 		},