Filter by mime

crates/pile-value/src/extract/item/pdf/pdf_pages.rs

@@ -7,7 +7,7 @@ use std::{
 use tracing::trace;
 
 use crate::{
-	extract::traits::ListExtractor,
+	extract::traits::{ExtractState, ListExtractor},
 	value::{Item, PileValue, SyncReadBridge},
 };
 
@@ -34,12 +34,20 @@ impl PdfPagesExtractor {
 
 #[async_trait::async_trait]
 impl ListExtractor for PdfPagesExtractor {
-	async fn get(&self, idx: usize) -> Result<Option<PileValue>, std::io::Error> {
+	async fn get(
+		&self,
+		state: &ExtractState,
+		idx: usize,
+	) -> Result<Option<PileValue>, std::io::Error> {
 		trace!(
 			key = self.item.key().as_str(),
 			"Getting index {idx} from PdfPagesExtractor",
 		);
 
+		if !state.ignore_mime && self.item.mime().essence_str() != "application/pdf" {
+			return Ok(None);
+		}
+
 		let bytes = self.get_bytes().await?;
 		let png = tokio::task::spawn_blocking(move || {
 			let pdfium = Pdfium::default();
@@ -81,7 +89,11 @@ impl ListExtractor for PdfPagesExtractor {
 		Ok(Some(value))
 	}
 
-	async fn len(&self) -> Result<usize, std::io::Error> {
+	async fn len(&self, state: &ExtractState) -> Result<usize, std::io::Error> {
+		if !state.ignore_mime && self.item.mime().essence_str() != "application/pdf" {
+			return Ok(0);
+		}
+
 		let bytes = self.get_bytes().await?;
 		let count = tokio::task::spawn_blocking(move || {
 			let pdfium = Pdfium::default();
@@ -103,10 +115,10 @@ impl ListExtractor for PdfPagesExtractor {
 
 	// Override, extracting all pages is very slow,
 	// and we can't display binary in json anyway
-	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> {
 		Ok(serde_json::Value::String(format!(
 			"<PdfPages ({} pages)>",
-			self.len().await?
+			self.len(state).await?
 		)))
 	}
 }