kreuzberg 4.2.2 → 4.2.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1765714785cbe89567dcb13ed0c1e1b79c2da7a2143a0d0b4653c5578a3ada84
-  data.tar.gz: 64d6db5e4d88992920f37fe9a3e28ab08b5bdd0b28385da570e8207e67d90f34
+  metadata.gz: 2c6fc44b151014f7e56c82bd191f55244a4294a259b24b95fc494dba6f8eaba6
+  data.tar.gz: 6e40a732814ff3e2a164e718cdb1c7a6ae838b2b2210a66b232f8675c7f79a80
 SHA512:
-  metadata.gz: cbb71395a285ddb1a74101fc935ebb8266b81c6172043a128721ab37fad583c7202e559f9e8cb2534bf110721bf20e2d0cbe6838554c772831c56bc09583bf75
-  data.tar.gz: b752cf56da8810211e5efd5e5d69f136eb7d0a3d5e27e985b81dff18bde442f0033b15962823ad7e3c5a27e080d02b6a6df1726bffe4aa21eaf89f56a5c6b56f
+  metadata.gz: f9c3a45f31c3ad9e3857872d8705b397b40c4317844ef421f4da4c2918e57411f5a626df4f6706d7db4916f33b8644c736e7b41508b398fd0197f1a87170fa3c
+  data.tar.gz: 8b05a75be261dbe583c4873d9d21079efff97d6c9c0340bbd8a73a43c9d15955431f4de20cd8b4a8b7956872f52e4467c253f5da03177a1e7d3b6a10d202b59d

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    kreuzberg (4.2.2)
+    kreuzberg (4.2.3)
 
 GEM
   remote: https://rubygems.org/
@@ -61,7 +61,7 @@ GEM
     parser (3.3.10.1)
       ast (~> 2.4.1)
       racc
-    prism (1.8.0)
+    prism (1.9.0)
     pry (0.16.0)
       coderay (~> 1.1)
       method_source (~> 1.0)
@@ -98,7 +98,7 @@ GEM
       diff-lcs (>= 1.2.0, < 2.0)
       rspec-support (~> 3.13.0)
     rspec-support (3.13.6)
-    rubocop (1.82.1)
+    rubocop (1.84.0)
       json (~> 2.3)
       language_server-protocol (~> 3.17.0.2)
       lint_roller (~> 1.1.0)
@@ -106,7 +106,7 @@ GEM
       parser (>= 3.3.0.2)
       rainbow (>= 2.2.2, < 4.0)
       regexp_parser (>= 2.9.3, < 3.0)
-      rubocop-ast (>= 1.48.0, < 2.0)
+      rubocop-ast (>= 1.49.0, < 2.0)
       ruby-progressbar (~> 1.7)
       unicode-display_width (>= 2.4.0, < 4.0)
     rubocop-ast (1.49.0)
@@ -121,7 +121,7 @@ GEM
       rubocop (~> 1.81)
     ruby-progressbar (1.13.0)
     securerandom (0.4.1)
-    sorbet-runtime (0.6.12897)
+    sorbet-runtime (0.6.12903)
     steep (1.10.0)
       activesupport (>= 5.1)
       concurrent-ruby (>= 1.1.10)
@@ -207,7 +207,7 @@ CHECKSUMS
   i18n (1.14.8) sha256=285778639134865c5e0f6269e0b818256017e8cde89993fdfcbfb64d088824a5
   io-console (0.8.2) sha256=d6e3ae7a7cc7574f4b8893b4fca2162e57a825b223a177b7afa236c5ef9814cc
   json (2.18.0) sha256=b10506aee4183f5cf49e0efc48073d7b75843ce3782c68dbeb763351c08fd505
-  kreuzberg (4.2.2)
+  kreuzberg (4.2.3)
   language_server-protocol (3.17.0.5) sha256=fd1e39a51a28bf3eec959379985a72e296e9f9acfce46f6a79d31ca8760803cc
   lint_roller (1.1.0) sha256=2c0c845b632a7d172cb849cc90c1bce937a28c5c8ccccb50dfd46a485003cc87
   listen (3.10.0) sha256=c6e182db62143aeccc2e1960033bebe7445309c7272061979bb098d03760c9d2
@@ -217,7 +217,7 @@ CHECKSUMS
   mutex_m (0.3.0) sha256=cfcb04ac16b69c4813777022fdceda24e9f798e48092a2b817eb4c0a782b0751
   parallel (1.27.0) sha256=4ac151e1806b755fb4e2dc2332cbf0e54f2e24ba821ff2d3dcf86bf6dc4ae130
   parser (3.3.10.1) sha256=06f6a725d2cd91e5e7f2b7c32ba143631e1f7c8ae2fb918fc4cebec187e6a688
-  prism (1.8.0) sha256=84453a16ef5530ea62c5f03ec16b52a459575ad4e7b9c2b360fd8ce2c39c1254
+  prism (1.9.0) sha256=7b530c6a9f92c24300014919c9dcbc055bf4cdf51ec30aed099b06cd6674ef85
   pry (0.16.0) sha256=d76c69065698ed1f85e717bd33d7942c38a50868f6b0673c636192b3d1b6054e
   pry-byebug (3.12.0) sha256=594e094ae8a8390a7ad4c7b36ae36e13304ed02664c67417d108dc5f7213d1b7
   racc (1.8.1) sha256=4a7f6929691dbec8b5209a0b373bc2614882b55fc5d2e447a21aaa691303d62f
@@ -236,13 +236,13 @@ CHECKSUMS
   rspec-expectations (3.13.5) sha256=33a4d3a1d95060aea4c94e9f237030a8f9eae5615e9bd85718fe3a09e4b58836
   rspec-mocks (3.13.7) sha256=0979034e64b1d7a838aaaddf12bf065ea4dc40ef3d4c39f01f93ae2c66c62b1c
   rspec-support (3.13.6) sha256=2e8de3702427eab064c9352fe74488cc12a1bfae887ad8b91cba480ec9f8afb2
-  rubocop (1.82.1) sha256=09f1a6a654a960eda767aebea33e47603080f8e9c9a3f019bf9b94c9cab5e273
+  rubocop (1.84.0) sha256=88dec310153bb685a879f5a7cdb601f6287b8f0ee675d9dc63a17c7204c4190a
   rubocop-ast (1.49.0) sha256=49c3676d3123a0923d333e20c6c2dbaaae2d2287b475273fddee0c61da9f71fd
   rubocop-performance (1.26.1) sha256=cd19b936ff196df85829d264b522fd4f98b6c89ad271fa52744a8c11b8f71834
   rubocop-rspec (3.9.0) sha256=8fa70a3619408237d789aeecfb9beef40576acc855173e60939d63332fdb55e2
   ruby-progressbar (1.13.0) sha256=80fc9c47a9b640d6834e0dc7b3c94c9df37f08cb072b7761e4a71e22cff29b33
   securerandom (0.4.1) sha256=cc5193d414a4341b6e225f0cb4446aceca8e50d5e1888743fac16987638ea0b1
-  sorbet-runtime (0.6.12897) sha256=0348ab8803c4c3646977fee298083ded9b7e74d5b34b50c567c63eb7e36eb286
+  sorbet-runtime (0.6.12903) sha256=c23968c0dcf5a5db57f32c003fe3db7fb588c168cdd57d92ea4dceaba063118a
   steep (1.10.0) sha256=1b295b55f9aaff1b8d3ee42453ee55bc2a1078fda0268f288edb2dc014f4d7d1
   strscan (3.1.7) sha256=5f76462b94a3ea50b44973225b7d75b2cb96d4e1bee9ef1319b99ca117b72c8c
   terminal-table (4.0.0) sha256=f504793203f8251b2ea7c7068333053f0beeea26093ec9962e62ea79f94301d2

data/README.md

@@ -22,7 +22,7 @@
     <img src="https://img.shields.io/maven-central/v/dev.kreuzberg/kreuzberg?label=Java&color=007ec6" alt="Java">
   </a>
   <a href="https://github.com/kreuzberg-dev/kreuzberg/releases">
-    <img src="https://img.shields.io/github/v/tag/kreuzberg-dev/kreuzberg?label=Go&color=007ec6&filter=v4.2.2" alt="Go">
+    <img src="https://img.shields.io/github/v/tag/kreuzberg-dev/kreuzberg?label=Go&color=007ec6&filter=v4.2.3" alt="Go">
   </a>
   <a href="https://www.nuget.org/packages/Kreuzberg/">
     <img src="https://img.shields.io/nuget/v/Kreuzberg?label=C%23&color=007ec6" alt="C#">

data/lib/kreuzberg/api_proxy.rb

@@ -6,9 +6,9 @@ module Kreuzberg
   # @example Start the server
   # @example With block
   module APIProxy
-    Error = Class.new(Kreuzberg::Errors::Error)
-    MissingBinaryError = Class.new(Error)
-    ServerError = Class.new(Error)
+    class Error < Kreuzberg::Errors::Error; end
+    class MissingBinaryError < Error; end
+    class ServerError < Error; end
 
     # API server instance
     class Server

data/lib/kreuzberg/cli_proxy.rb

@@ -5,8 +5,8 @@ require 'open3'
 module Kreuzberg
   # @example
   module CLIProxy
-    Error = Class.new(Kreuzberg::Errors::Error)
-    MissingBinaryError = Class.new(Error)
+    class Error < Kreuzberg::Errors::Error; end
+    class MissingBinaryError < Error; end
 
     # CLI execution error with stderr and exit status
     class CLIExecutionError < Error

data/lib/kreuzberg/mcp_proxy.rb

@@ -6,9 +6,9 @@ require 'json'
 module Kreuzberg
   # @example Start MCP server
   module MCPProxy
-    Error = Class.new(Kreuzberg::Errors::Error)
-    MissingBinaryError = Class.new(Error)
-    ServerError = Class.new(Error)
+    class Error < Kreuzberg::Errors::Error; end
+    class MissingBinaryError < Error; end
+    class ServerError < Error; end
 
     # MCP server instance
     class Server

data/lib/kreuzberg/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Kreuzberg
-  VERSION = '4.2.2'
+  VERSION = '4.2.3'
 end

data/vendor/Cargo.toml

@@ -3,7 +3,7 @@ members = ["kreuzberg", "kreuzberg-tesseract", "kreuzberg-ffi"]
 resolver = "2"
 
 [workspace.package]
-version = "4.2.2"
+version = "4.2.3"
 edition = "2024"
 rust-version = "1.91"
 authors = ["Na'aman Hirschfeld <nhirschfeld@gmail.com>"]

data/vendor/kreuzberg/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "kreuzberg"
-version = "4.2.2"
+version = "4.2.3"
 edition = "2024"
 rust-version = "1.91"
 authors = ["Na'aman Hirschfeld <nhirschfeld@gmail.com>"]
@@ -198,7 +198,7 @@ rake = { version = "0.3.6", optional = true }
 axum = { version = "0.8", features = ["macros", "json", "multipart"], optional = true }
 tower = { version = "0.5", optional = true }
 tower-http = { version = "0.6", features = ["cors", "trace", "limit"], optional = true }
-utoipa = { version = "5.3", features = ["axum_extras"], optional = true }
+utoipa = { version = "5.4", features = ["axum_extras"], optional = true }
 rmcp = { version = "0.14.0", features = [
     "server",
     "macros",

data/vendor/kreuzberg/README.md

@@ -17,7 +17,7 @@ High-performance document intelligence library for Rust. Extract text, metadata,
 
 This is the core Rust library that powers the Python, TypeScript, and Ruby bindings.
 
-> **🚀 Version 4.2.2 Release**
+> **🚀 Version 4.2.3 Release**
 > This is a pre-release version. We invite you to test the library and [report any issues](https://github.com/kreuzberg-dev/kreuzberg/issues) you encounter.
 >
 > **Note**: The Rust crate is not currently published to crates.io for this RC. Use git dependencies or language bindings (Python, TypeScript, Ruby) instead.

data/vendor/kreuzberg/src/api/error.rs

@@ -2,6 +2,7 @@
 
 use axum::{
     Json,
+    body::to_bytes,
     extract::{FromRequest, Request, rejection::JsonRejection},
     http::StatusCode,
     response::{IntoResponse, Response},
@@ -16,6 +17,9 @@ use super::types::ErrorResponse;
 ///
 /// This wraps axum's `Json` extractor but uses `ApiError` as the rejection type,
 /// ensuring that all JSON parsing errors are returned as JSON with proper content type.
+///
+/// Additionally, this extractor validates that the root JSON value is an object (not an array),
+/// which prevents serde from incorrectly deserializing JSON arrays into struct fields.
 #[derive(Debug, Clone, Copy, Default)]
 pub struct JsonApi<T>(pub T);
 
@@ -27,6 +31,31 @@ where
     type Rejection = ApiError;
 
     async fn from_request(req: Request, state: &S) -> Result<Self, Self::Rejection> {
+        // First, extract the body to check if it's a valid JSON object (not array)
+        let (parts, body) = req.into_parts();
+        let bytes = to_bytes(body, usize::MAX).await.map_err(|_| {
+            ApiError::new(
+                StatusCode::BAD_REQUEST,
+                KreuzbergError::Other("Failed to read request body".to_string()),
+            )
+        })?;
+
+        // Validate that the root JSON is an object, not an array
+        if !bytes.is_empty() {
+            let trimmed = std::str::from_utf8(&bytes).unwrap_or("").trim_start();
+            if trimmed.starts_with('[') {
+                return Err(ApiError::new(
+                    StatusCode::BAD_REQUEST,
+                    KreuzbergError::validation(
+                        "Expected JSON object, but received JSON array. \
+                         Please wrap your data in an object with appropriate fields.",
+                    ),
+                ));
+            }
+        }
+
+        // Reconstruct the request and use the standard Json extractor
+        let req = Request::from_parts(parts, axum::body::Body::from(bytes));
         match Json::<T>::from_request(req, state).await {
             Ok(Json(value)) => Ok(JsonApi(value)),
             Err(rejection) => Err(ApiError::from(rejection)),

data/vendor/kreuzberg/src/mcp/format.rs

@@ -72,28 +72,12 @@ pub(super) fn build_config(
     }
 }
 
-/// Format extraction result as human-readable text.
+/// Format extraction result as JSON string.
+///
+/// Serializes the full `ExtractionResult` to JSON, ensuring 1:1 parity
+/// with the API and CLI JSON output.
 pub(super) fn format_extraction_result(result: &KreuzbergResult) -> String {
-    let mut response = String::new();
-
-    response.push_str(&format!("Content ({} characters):\n", result.content.len()));
-    response.push_str(&result.content);
-    response.push_str("\n\n");
-
-    response.push_str("Metadata:\n");
-    response.push_str(&serde_json::to_string_pretty(&result.metadata).unwrap_or_default());
-    response.push_str("\n\n");
-
-    if !result.tables.is_empty() {
-        response.push_str(&format!("Tables ({}):\n", result.tables.len()));
-        for (i, table) in result.tables.iter().enumerate() {
-            response.push_str(&format!("\nTable {} (page {}):\n", i + 1, table.page_number));
-            response.push_str(&table.markdown);
-            response.push('\n');
-        }
-    }
-
-    response
+    serde_json::to_string_pretty(result).unwrap_or_default()
 }
 
 #[cfg(test)]
@@ -303,7 +287,7 @@ mod tests {
     }
 
     #[test]
-    fn test_format_extraction_result_with_content() {
+    fn test_format_extraction_result_is_valid_json() {
         let result = KreuzbergResult {
             content: "Sample extracted text".to_string(),
             mime_type: "text/plain".to_string(),
@@ -318,36 +302,27 @@ mod tests {
         };
 
         let formatted = format_extraction_result(&result);
+        let parsed: serde_json::Value = serde_json::from_str(&formatted).expect("Should be valid JSON");
 
-        assert!(formatted.contains("Content (21 characters)"));
-        assert!(formatted.contains("Sample extracted text"));
-        assert!(formatted.contains("Metadata:"));
+        assert_eq!(parsed["content"], "Sample extracted text");
+        assert_eq!(parsed["mime_type"], "text/plain");
+        assert!(parsed["metadata"].is_object());
     }
 
     #[test]
-    fn test_format_extraction_result_with_tables() {
+    fn test_format_extraction_result_includes_tables() {
         let result = KreuzbergResult {
             content: "Document with tables".to_string(),
             mime_type: "application/pdf".to_string(),
             metadata: crate::Metadata::default(),
-            tables: vec![
-                crate::Table {
-                    cells: vec![
-                        vec!["Col1".to_string(), "Col2".to_string()],
-                        vec!["A".to_string(), "B".to_string()],
-                    ],
-                    page_number: 1,
-                    markdown: "| Col1 | Col2 |\n|------|------|\n| A    | B    |".to_string(),
-                },
-                crate::Table {
-                    cells: vec![
-                        vec!["X".to_string(), "Y".to_string()],
-                        vec!["1".to_string(), "2".to_string()],
-                    ],
-                    page_number: 2,
-                    markdown: "| X | Y |\n|---|---|\n| 1 | 2 |".to_string(),
-                },
-            ],
+            tables: vec![crate::Table {
+                cells: vec![
+                    vec!["Col1".to_string(), "Col2".to_string()],
+                    vec!["A".to_string(), "B".to_string()],
+                ],
+                page_number: 1,
+                markdown: "| Col1 | Col2 |\n|------|------|\n| A    | B    |".to_string(),
+            }],
             detected_languages: None,
             chunks: None,
             images: None,
@@ -357,23 +332,33 @@ mod tests {
         };
 
         let formatted = format_extraction_result(&result);
+        let parsed: serde_json::Value = serde_json::from_str(&formatted).expect("Should be valid JSON");
 
-        assert!(formatted.contains("Tables (2)"));
-        assert!(formatted.contains("Table 1 (page 1)"));
-        assert!(formatted.contains("Table 2 (page 2)"));
-        assert!(formatted.contains("| Col1 | Col2 |"));
-        assert!(formatted.contains("| X | Y |"));
+        assert_eq!(parsed["tables"].as_array().unwrap().len(), 1);
+        assert_eq!(parsed["tables"][0]["page_number"], 1);
     }
 
     #[test]
-    fn test_format_extraction_result_empty_content() {
+    fn test_format_extraction_result_includes_chunks_when_present() {
         let result = KreuzbergResult {
-            content: String::new(),
+            content: "Chunked text".to_string(),
             mime_type: "text/plain".to_string(),
             metadata: crate::Metadata::default(),
             tables: vec![],
             detected_languages: None,
-            chunks: None,
+            chunks: Some(vec![crate::Chunk {
+                content: "Chunk 1".to_string(),
+                embedding: None,
+                metadata: crate::ChunkMetadata {
+                    byte_start: 0,
+                    byte_end: 7,
+                    token_count: None,
+                    chunk_index: 0,
+                    total_chunks: 1,
+                    first_page: None,
+                    last_page: None,
+                },
+            }]),
             images: None,
             pages: None,
             elements: None,
@@ -381,13 +366,14 @@ mod tests {
         };
 
         let formatted = format_extraction_result(&result);
+        let parsed: serde_json::Value = serde_json::from_str(&formatted).expect("Should be valid JSON");
 
-        assert!(formatted.contains("Content (0 characters)"));
-        assert!(formatted.contains("Metadata:"));
+        assert_eq!(parsed["chunks"].as_array().unwrap().len(), 1);
+        assert_eq!(parsed["chunks"][0]["content"], "Chunk 1");
     }
 
     #[test]
-    fn test_format_extraction_result_no_tables() {
+    fn test_format_extraction_result_omits_none_fields() {
         let result = KreuzbergResult {
             content: "Simple text".to_string(),
             mime_type: "text/plain".to_string(),
@@ -402,8 +388,11 @@ mod tests {
         };
 
         let formatted = format_extraction_result(&result);
+        let parsed: serde_json::Value = serde_json::from_str(&formatted).expect("Should be valid JSON");
 
-        assert!(formatted.contains("Simple text"));
-        assert!(!formatted.contains("Tables"));
+        // None fields should be omitted via skip_serializing_if
+        assert!(parsed.get("chunks").is_none());
+        assert!(parsed.get("images").is_none());
+        assert!(parsed.get("detected_languages").is_none());
     }
 }

data/vendor/kreuzberg/src/mcp/server.rs

@@ -144,7 +144,7 @@ impl KreuzbergMcp {
         Parameters(params): Parameters<super::params::BatchExtractFilesParams>,
     ) -> Result<CallToolResult, rmcp::ErrorData> {
         use super::errors::map_kreuzberg_error_to_mcp;
-        use super::format::{build_config, format_extraction_result};
+        use super::format::build_config;
         use crate::{batch_extract_file, batch_extract_file_sync};
 
         let config =
@@ -158,13 +158,7 @@ impl KreuzbergMcp {
             batch_extract_file_sync(params.paths.clone(), &config).map_err(map_kreuzberg_error_to_mcp)?
         };
 
-        let mut response = String::new();
-        for (i, result) in results.iter().enumerate() {
-            response.push_str(&format!("=== Document {}: {} ===\n", i + 1, params.paths[i]));
-            response.push_str(&format_extraction_result(result));
-            response.push_str("\n\n");
-        }
-
+        let response = serde_json::to_string_pretty(&results).unwrap_or_default();
         Ok(CallToolResult::success(vec![Content::text(response)]))
     }
 

data/vendor/kreuzberg/src/mcp/tools/extraction.rs

@@ -99,13 +99,7 @@ pub(in crate::mcp) trait ExtractionTool {
             batch_extract_file_sync(params.paths.clone(), &config).map_err(map_kreuzberg_error_to_mcp)?
         };
 
-        let mut response = String::new();
-        for (i, result) in results.iter().enumerate() {
-            response.push_str(&format!("=== Document {}: {} ===\n", i + 1, params.paths[i]));
-            response.push_str(&format_extraction_result(result));
-            response.push_str("\n\n");
-        }
-
+        let response = serde_json::to_string_pretty(&results).unwrap_or_default();
         Ok(CallToolResult::success(vec![Content::text(response)]))
     }
 }

data/vendor/kreuzberg/tests/api_chunk.rs

@@ -321,3 +321,28 @@ async fn test_chunk_custom_config() {
     assert_eq!(chunk_response.config.overlap, 5);
     assert!(!chunk_response.config.trim);
 }
+
+#[tokio::test]
+async fn test_chunk_rejects_json_array() {
+    let app = create_router(ExtractionConfig::default());
+
+    // Send a JSON array instead of object
+    let response = app
+        .oneshot(
+            Request::builder()
+                .uri("/chunk")
+                .method("POST")
+                .header("content-type", "application/json")
+                .body(Body::from(r#"[["text"], {"text": "content"}]"#))
+                .expect("Operation failed"),
+        )
+        .await
+        .expect("Operation failed");
+
+    // Should reject with 400 or 422, NOT 200
+    assert!(
+        response.status() == StatusCode::BAD_REQUEST || response.status() == StatusCode::UNPROCESSABLE_ENTITY,
+        "Expected 400 or 422, got {}",
+        response.status()
+    );
+}

data/vendor/kreuzberg/tests/api_embed.rs

@@ -255,6 +255,66 @@ async fn test_embed_malformed_json() {
     assert_eq!(response.status(), StatusCode::BAD_REQUEST);
 }
 
+/// Test embed endpoint rejects JSON array at root level.
+#[tokio::test]
+async fn test_embed_rejects_json_array() {
+    let app = create_router(ExtractionConfig::default());
+
+    // Send a JSON array instead of object
+    let response = app
+        .oneshot(
+            Request::builder()
+                .method("POST")
+                .uri("/embed")
+                .header("content-type", "application/json")
+                .body(Body::from(r#"[["text1"], {"texts": ["text2"]}]"#))
+                .expect("Operation failed"),
+        )
+        .await
+        .expect("Operation failed");
+
+    // Should reject with 400 or 422, NOT 200
+    assert!(
+        response.status() == StatusCode::BAD_REQUEST || response.status() == StatusCode::UNPROCESSABLE_ENTITY,
+        "Expected 400 or 422, got {}",
+        response.status()
+    );
+}
+
+/// Test embed endpoint rejects simple JSON array with strings.
+#[tokio::test]
+async fn test_embed_rejects_simple_json_array() {
+    let app = create_router(ExtractionConfig::default());
+
+    // Send a simple string array instead of object with texts field
+    let response = app
+        .oneshot(
+            Request::builder()
+                .method("POST")
+                .uri("/embed")
+                .header("content-type", "application/json")
+                .body(Body::from(r#"["text1", "text2", "text3"]"#))
+                .expect("Operation failed"),
+        )
+        .await
+        .expect("Operation failed");
+
+    assert_eq!(response.status(), StatusCode::BAD_REQUEST);
+
+    // Check that error response contains helpful message
+    let body = axum::body::to_bytes(response.into_body(), usize::MAX)
+        .await
+        .expect("Failed to read response body");
+    let error_response: serde_json::Value = serde_json::from_slice(&body).expect("Failed to parse error response");
+
+    assert!(
+        error_response["message"]
+            .as_str()
+            .map(|msg| msg.contains("array") || msg.contains("object"))
+            .unwrap_or(false)
+    );
+}
+
 /// Test embed endpoint preserves embedding vector values across calls.
 #[tokio::test]
 async fn test_embed_deterministic() {

data/vendor/kreuzberg-tesseract/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "kreuzberg-tesseract"
-version = "4.2.2"
+version = "4.2.3"
 edition = "2024"
 rust-version = "1.91"
 authors = ["Na'aman Hirschfeld <nhirschfeld@gmail.com>"]

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: kreuzberg
 version: !ruby/object:Gem::Version
-  version: 4.2.2
+  version: 4.2.3
 platform: ruby
 authors:
 - Na'aman Hirschfeld