kreuzberg 4.1.2 → 4.2.1

data/vendor/kreuzberg/tests/test_fastembed.rs

@@ -26,7 +26,7 @@ async fn test_fastembed_embedding_generation() {
     let result = model.embed(texts.clone(), None);
     assert!(result.is_ok(), "Failed to generate embeddings: {:?}", result.err());
 
-    let embeddings = result.unwrap();
+    let embeddings = result.expect("Operation failed");
     assert_eq!(embeddings.len(), 3, "Expected 3 embeddings");
 
     for (i, embedding) in embeddings.iter().enumerate() {
@@ -64,7 +64,7 @@ async fn test_fastembed_batch_processing() {
 
     assert!(result.is_ok(), "Batch embedding failed: {:?}", result.err());
 
-    let embeddings = result.unwrap();
+    let embeddings = result.expect("Operation failed");
     assert_eq!(embeddings.len(), 50, "Expected 50 embeddings");
 
     println!(
@@ -96,7 +96,7 @@ async fn test_fastembed_different_models() {
                 let result = m.embed(test_text.clone(), None);
                 assert!(result.is_ok(), "Failed to generate embedding for {}", description);
 
-                let embeddings = result.unwrap();
+                let embeddings = result.expect("Operation failed");
                 assert_eq!(embeddings.len(), 1);
                 assert_eq!(
                     embeddings[0].len(),
@@ -197,7 +197,7 @@ async fn test_generate_embeddings_for_chunks_basic() {
     for (i, chunk) in chunks.iter().enumerate() {
         assert!(chunk.embedding.is_some(), "Chunk {} missing embedding", i);
 
-        let embedding = chunk.embedding.as_ref().unwrap();
+        let embedding = chunk.embedding.as_ref().expect("Operation failed");
         assert_eq!(embedding.len(), 384, "Chunk {} has wrong embedding dimensions", i);
 
         let sum: f32 = embedding.iter().sum();
@@ -269,8 +269,8 @@ async fn test_generate_embeddings_for_chunks_normalization() {
 
     generate_embeddings_for_chunks(&mut chunks_norm, &config_norm).expect("Failed to generate normalized embeddings");
 
-    let embedding_no_norm = chunks_no_norm[0].embedding.as_ref().unwrap();
-    let embedding_norm = chunks_norm[0].embedding.as_ref().unwrap();
+    let embedding_no_norm = chunks_no_norm[0].embedding.as_ref().expect("Operation failed");
+    let embedding_norm = chunks_norm[0].embedding.as_ref().expect("Operation failed");
 
     let magnitude_no_norm: f32 = embedding_no_norm.iter().map(|x| x * x).sum::<f32>().sqrt();
     let magnitude_norm: f32 = embedding_norm.iter().map(|x| x * x).sum::<f32>().sqrt();
@@ -560,7 +560,7 @@ async fn test_generate_embeddings_for_chunks_batch_size() {
             i
         );
         assert_eq!(
-            chunk.embedding.as_ref().unwrap().len(),
+            chunk.embedding.as_ref().expect("Operation failed").len(),
             384,
             "Chunk {} has wrong dimensions",
             i
@@ -612,7 +612,7 @@ async fn test_generate_embeddings_chunking_integration() {
     for (i, chunk) in chunking_result.chunks.iter().enumerate() {
         assert!(chunk.embedding.is_some(), "Chunk {} missing embedding", i);
 
-        let embedding = chunk.embedding.as_ref().unwrap();
+        let embedding = chunk.embedding.as_ref().expect("Operation failed");
         assert_eq!(embedding.len(), 384, "Chunk {} has wrong embedding dimensions", i);
 
         let magnitude: f32 = embedding.iter().map(|x| x * x).sum::<f32>().sqrt();

data/vendor/kreuzberg/tests/typst_extractor_tests.rs

@@ -56,7 +56,7 @@ async fn test_simple_typst_document_extraction() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     assert_eq!(extraction.mime_type, "text/x-typst", "MIME type should be preserved");
 
@@ -145,7 +145,7 @@ async fn test_minimal_typst_document_extraction() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     assert!(
         !extraction.content.is_empty(),
@@ -189,7 +189,7 @@ async fn test_heading_hierarchy_extraction() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     assert!(!extraction.content.is_empty(), "Document should extract content");
 
@@ -269,7 +269,7 @@ async fn test_metadata_extraction() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     if let Some(title) = extraction.metadata.additional.get("title") {
         assert!(
@@ -330,7 +330,7 @@ async fn test_advanced_typst_document_extraction() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     assert!(
         extraction.metadata.additional.contains_key("title"),
@@ -411,7 +411,7 @@ async fn test_typst_reader_extraction() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     assert!(
         !extraction.content.is_empty(),
@@ -454,7 +454,7 @@ async fn test_undergradmath_extraction() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     assert!(
         !extraction.content.is_empty(),
@@ -534,7 +534,7 @@ async fn test_formatting_preservation() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     assert!(
         extraction.content.contains("*") || extraction.content.contains("bold"),
@@ -576,7 +576,7 @@ async fn test_large_document_extraction() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     assert!(
         !extraction.content.is_empty(),

data/vendor/kreuzberg/tests/xlsx_metadata_extraction_test.rs

@@ -7,9 +7,9 @@ use kreuzberg::extraction::excel::read_excel_file;
 fn test_xlsx_full_metadata_extraction() {
     let workspace_root = std::path::Path::new(env!("CARGO_MANIFEST_DIR"))
         .parent()
-        .unwrap()
+        .expect("Operation failed")
         .parent()
-        .unwrap();
+        .expect("Operation failed");
     let test_file = workspace_root.join("test_documents/office/excel.xlsx");
 
     if !test_file.exists() {
@@ -17,7 +17,8 @@ fn test_xlsx_full_metadata_extraction() {
         return;
     }
 
-    let result = read_excel_file(test_file.to_str().unwrap()).expect("Should extract XLSX successfully");
+    let file_path = test_file.to_str().expect("File path should be valid UTF-8");
+    let result = read_excel_file(file_path).expect("Should extract XLSX successfully");
 
     assert!(!result.sheets.is_empty(), "Should have at least one sheet");
 
@@ -34,9 +35,9 @@ fn test_xlsx_full_metadata_extraction() {
 fn test_xlsx_multi_sheet_metadata() {
     let workspace_root = std::path::Path::new(env!("CARGO_MANIFEST_DIR"))
         .parent()
-        .unwrap()
+        .expect("Operation failed")
         .parent()
-        .unwrap();
+        .expect("Operation failed");
     let test_file = workspace_root.join("test_documents/spreadsheets/excel_multi_sheet.xlsx");
 
     if !test_file.exists() {
@@ -44,7 +45,8 @@ fn test_xlsx_multi_sheet_metadata() {
         return;
     }
 
-    let result = read_excel_file(test_file.to_str().unwrap()).expect("Should extract multi-sheet XLSX successfully");
+    let file_path = test_file.to_str().expect("File path should be valid UTF-8");
+    let result = read_excel_file(file_path).expect("Should extract multi-sheet XLSX successfully");
 
     assert!(
         result.sheets.len() > 1,
@@ -65,9 +67,9 @@ fn test_xlsx_multi_sheet_metadata() {
 fn test_xlsx_minimal_metadata_extraction() {
     let workspace_root = std::path::Path::new(env!("CARGO_MANIFEST_DIR"))
         .parent()
-        .unwrap()
+        .expect("Operation failed")
         .parent()
-        .unwrap();
+        .expect("Operation failed");
     let test_file = workspace_root.join("test_documents/spreadsheets/test_01.xlsx");
 
     if !test_file.exists() {
@@ -75,7 +77,8 @@ fn test_xlsx_minimal_metadata_extraction() {
         return;
     }
 
-    let result = read_excel_file(test_file.to_str().unwrap()).expect("Should extract XLSX successfully");
+    let file_path = test_file.to_str().expect("File path should be valid UTF-8");
+    let result = read_excel_file(file_path).expect("Should extract XLSX successfully");
 
     assert!(!result.sheets.is_empty(), "Content should not be empty");
     assert!(

data/vendor/kreuzberg-tesseract/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "kreuzberg-tesseract"
-version = "4.1.2"
+version = "4.2.1"
 edition = "2024"
 rust-version = "1.91"
 authors = ["Na'aman Hirschfeld <nhirschfeld@gmail.com>"]

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: kreuzberg
 version: !ruby/object:Gem::Version
-  version: 4.1.2
+  version: 4.2.1
 platform: ruby
 authors:
 - Na'aman Hirschfeld
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-01-25 00:00:00.000000000 Z
+date: 2026-01-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -209,6 +209,7 @@ files:
 - ext/kreuzberg_rb/native/include/msvc_compat/strings.h
 - ext/kreuzberg_rb/native/include/strings.h
 - ext/kreuzberg_rb/native/include/unistd.h
+- ext/kreuzberg_rb/native/libpdfium.so
 - ext/kreuzberg_rb/native/src/batch.rs
 - ext/kreuzberg_rb/native/src/config/mod.rs
 - ext/kreuzberg_rb/native/src/config/types.rs
@@ -232,6 +233,7 @@ files:
 - lib/kreuzberg/cli.rb
 - lib/kreuzberg/cli_proxy.rb
 - lib/kreuzberg/config.rb
+- lib/kreuzberg/djot_content.rb
 - lib/kreuzberg/error_context.rb
 - lib/kreuzberg/errors.rb
 - lib/kreuzberg/extraction_api.rb
@@ -271,6 +273,7 @@ files:
 - spec/fixtures/config.toml
 - spec/fixtures/config.yaml
 - spec/fixtures/invalid_config.toml
+- spec/serialization_spec.rb
 - spec/smoke/package_spec.rb
 - spec/spec_helper.rb
 - spec/unit/config/chunking_config_spec.rb
@@ -283,6 +286,7 @@ files:
 - spec/unit/config/keyword_config_spec.rb
 - spec/unit/config/language_detection_config_spec.rb
 - spec/unit/config/ocr_config_spec.rb
+- spec/unit/config/output_format_spec.rb
 - spec/unit/config/page_config_spec.rb
 - spec/unit/config/pdf_config_spec.rb
 - spec/unit/config/postprocessor_config_spec.rb
@@ -588,6 +592,7 @@ files:
 - vendor/kreuzberg/src/plugins/registry/ocr.rs
 - vendor/kreuzberg/src/plugins/registry/processor.rs
 - vendor/kreuzberg/src/plugins/registry/validator.rs
+- vendor/kreuzberg/src/plugins/startup_validation.rs
 - vendor/kreuzberg/src/plugins/traits.rs
 - vendor/kreuzberg/src/plugins/validator/mod.rs
 - vendor/kreuzberg/src/plugins/validator/registry.rs
@@ -705,6 +710,7 @@ files:
 - vendor/kreuzberg/stopwords/zh_stopwords.json
 - vendor/kreuzberg/stopwords/zu_stopwords.json
 - vendor/kreuzberg/tests/api_chunk.rs
+- vendor/kreuzberg/tests/api_consistency.rs
 - vendor/kreuzberg/tests/api_embed.rs
 - vendor/kreuzberg/tests/api_extract_multipart.rs
 - vendor/kreuzberg/tests/api_large_pdf_extraction.rs
@@ -716,9 +722,11 @@ files:
 - vendor/kreuzberg/tests/batch_processing.rs
 - vendor/kreuzberg/tests/bibtex_parity_test.rs
 - vendor/kreuzberg/tests/concurrency_stress.rs
+- vendor/kreuzberg/tests/config_behavioral.rs
 - vendor/kreuzberg/tests/config_features.rs
 - vendor/kreuzberg/tests/config_integration_test.rs
 - vendor/kreuzberg/tests/config_loading_tests.rs
+- vendor/kreuzberg/tests/contract_mcp.rs
 - vendor/kreuzberg/tests/core_integration.rs
 - vendor/kreuzberg/tests/csv_integration.rs
 - vendor/kreuzberg/tests/data/hierarchy_ground_truth.json
@@ -740,6 +748,7 @@ files:
 - vendor/kreuzberg/tests/keywords_quality.rs
 - vendor/kreuzberg/tests/latex_extractor_tests.rs
 - vendor/kreuzberg/tests/markdown_extractor_tests.rs
+- vendor/kreuzberg/tests/mcp_integration.rs
 - vendor/kreuzberg/tests/mime_detection.rs
 - vendor/kreuzberg/tests/ocr_configuration.rs
 - vendor/kreuzberg/tests/ocr_errors.rs
@@ -766,6 +775,7 @@ files:
 - vendor/kreuzberg/tests/rst_extractor_tests.rs
 - vendor/kreuzberg/tests/rtf_extractor_tests.rs
 - vendor/kreuzberg/tests/security_validation.rs
+- vendor/kreuzberg/tests/serialization_integration.rs
 - vendor/kreuzberg/tests/stopwords_integration_test.rs
 - vendor/kreuzberg/tests/test_fastembed.rs
 - vendor/kreuzberg/tests/typst_behavioral_tests.rs