From 5cd67f38d623ddfbecbb5192aeadfbfb8cb401c9 Mon Sep 17 00:00:00 2001
From: Yujong Lee <yujong@berri.ai>
Date: Sat, 12 Sep 2026 10:16:42 -0700
Subject: [PATCH] fix(ocr): normalize DeepSeek document content

---
 .../src/ocr/codecs/deepseek/transformation.rs |  6 ++++-
 .../crates/core/tests/deepseek_ocr.rs         | 22 +++++++++++++++++++
 .../core/tests/vertex_ai_deepseek_ocr.rs      |  2 +-
 3 files changed, 28 insertions(+), 2 deletions(-)

diff --git a/litellm-rust/crates/core/src/ocr/codecs/deepseek/transformation.rs b/litellm-rust/crates/core/src/ocr/codecs/deepseek/transformation.rs
index 98cfc0db78d..c3a341a2921 100644
--- a/litellm-rust/crates/core/src/ocr/codecs/deepseek/transformation.rs
+++ b/litellm-rust/crates/core/src/ocr/codecs/deepseek/transformation.rs
@@ -14,11 +14,15 @@ pub(crate) fn transform_ocr_request(
     if document.source().is_empty() {
         return Err(OcrRequestError::MissingField("document URL"));
     }
+    let content = OcrDocument::ImageUrl {
+        image_url: document.source().to_string(),
+        extra_fields: serde_json::Map::new(),
+    };
     Ok(DeepSeekOcrRequest {
         model: provider_model.to_string(),
         messages: vec![DeepSeekOcrMessage {
             role: UserRole::User,
-            content: vec![document],
+            content: vec![content],
         }],
         params: params.clone(),
     })
diff --git a/litellm-rust/crates/core/tests/deepseek_ocr.rs b/litellm-rust/crates/core/tests/deepseek_ocr.rs
index 875fc9e3dc6..4ba39561dcd 100644
--- a/litellm-rust/crates/core/tests/deepseek_ocr.rs
+++ b/litellm-rust/crates/core/tests/deepseek_ocr.rs
@@ -34,6 +34,28 @@ fn request_mapping_matches_python(#[case] name: &str, #[case] value: Value) {
     assert!(result.get("ignored").is_none());
 }
 
+#[rstest]
+#[case(json!({"type":"image_url","image_url":"data:image/png;base64,AA=="}))]
+#[case(json!({"type":"document_url","document_url":"data:application/pdf;base64,AA=="}))]
+fn request_maps_both_document_types_to_image_content(#[case] document: Value) {
+    let source = document
+        .get("image_url")
+        .or_else(|| document.get("document_url"))
+        .unwrap()
+        .clone();
+    let request = transform_ocr_request(
+        "deepseek-ai/deepseek-ocr-maas",
+        serde_json::from_value(document).unwrap(),
+        &DeepSeekOcrParams::default(),
+    )
+    .unwrap();
+    let result = serde_json::to_value(request).unwrap();
+    assert_eq!(
+        result["messages"][0]["content"][0],
+        json!({"type":"image_url","image_url":source})
+    );
+}
+
 #[rstest]
 #[case(json!("# hello"), "# hello")]
 #[case(json!("{broken"), "{broken")]
diff --git a/litellm-rust/crates/core/tests/vertex_ai_deepseek_ocr.rs b/litellm-rust/crates/core/tests/vertex_ai_deepseek_ocr.rs
index 6d3061d8f5d..676799eb2fe 100644
--- a/litellm-rust/crates/core/tests/vertex_ai_deepseek_ocr.rs
+++ b/litellm-rust/crates/core/tests/vertex_ai_deepseek_ocr.rs
@@ -49,7 +49,7 @@ async fn facade_executes_vertex_deepseek_at_the_openai_endpoint() {
     assert!(body.get("extra_body").is_none());
     assert_eq!(
         body["messages"][0]["content"][0],
-        json!({"type":"document_url","document_url":"gs://bucket/document.pdf"})
+        json!({"type":"image_url","image_url":"gs://bucket/document.pdf"})
     );
 }