From 48262bab487f5a1e6ded33242ee5d37b30751baa Mon Sep 17 00:00:00 2001
From: rm-dr <96270320+rm-dr@users.noreply.github.com>
Date: Sat, 21 Mar 2026 09:32:22 -0700
Subject: [PATCH] Exclude large strings

---
 .../pile-value/src/extract/item/epub/mod.rs   | 29 +++++++++++++++++++
 crates/pile-value/src/extract/item/pdf/mod.rs | 29 +++++++++++++++++++
 2 files changed, 58 insertions(+)

diff --git a/crates/pile-value/src/extract/item/epub/mod.rs b/crates/pile-value/src/extract/item/epub/mod.rs
index ccf96a7..a67e4f4 100644
--- a/crates/pile-value/src/extract/item/epub/mod.rs
+++ b/crates/pile-value/src/extract/item/epub/mod.rs
@@ -54,4 +54,33 @@ impl ObjectExtractor for EpubExtractor {
 			Label::new("meta").unwrap(),
 		])
 	}
+
+	async fn to_json(&self, state: &ExtractState) -> Result<serde_json::Value, std::io::Error> {
+		let keys = self.fields().await?;
+		let mut map = serde_json::Map::new();
+		for k in &keys {
+			let v = match self.field(state, k, None).await? {
+				Some(x) => x,
+				None => continue,
+			};
+
+			if k.as_str() == "text" {
+				map.insert(
+					k.to_string(),
+					serde_json::Value::String(format!(
+						"<String ({} bytes)",
+						match v {
+							PileValue::String(x) => x.len(),
+							_ => 0,
+						}
+					)),
+				);
+				continue;
+			}
+
+			map.insert(k.to_string(), Box::pin(v.to_json(state)).await?);
+		}
+
+		Ok(serde_json::Value::Object(map))
+	}
 }
diff --git a/crates/pile-value/src/extract/item/pdf/mod.rs b/crates/pile-value/src/extract/item/pdf/mod.rs
index f603535..cdb23a4 100644
--- a/crates/pile-value/src/extract/item/pdf/mod.rs
+++ b/crates/pile-value/src/extract/item/pdf/mod.rs
@@ -68,4 +68,33 @@ impl ObjectExtractor for PdfExtractor {
 			Label::new("pages").unwrap(),
 		])
 	}
+
+	async fn to_json(&self, state: &ExtractState) -> Result<serde_json::Value, std::io::Error> {
+		let keys = self.fields().await?;
+		let mut map = serde_json::Map::new();
+		for k in &keys {
+			let v = match self.field(state, k, None).await? {
+				Some(x) => x,
+				None => continue,
+			};
+
+			if k.as_str() == "text" {
+				map.insert(
+					k.to_string(),
+					serde_json::Value::String(format!(
+						"<String ({} bytes)",
+						match v {
+							PileValue::String(x) => x.len(),
+							_ => 0,
+						}
+					)),
+				);
+				continue;
+			}
+
+			map.insert(k.to_string(), Box::pin(v.to_json(state)).await?);
+		}
+
+		Ok(serde_json::Value::Object(map))
+	}
 }