kreuzberg 4.2.6 → 4.2.7

data/vendor/kreuzberg/tests/config_behavioral.rs

@@ -147,10 +147,9 @@ async fn test_chunking_max_chars_limits_chunk_size() {
 
     let config = ExtractionConfig {
         chunking: Some(ChunkingConfig {
-            max_chars: 100,
-            max_overlap: 20,
-            embedding: None,
-            preset: None,
+            max_characters: 100,
+            overlap: 20,
+            ..Default::default()
         }),
         ..Default::default()
     };
@@ -184,10 +183,9 @@ async fn test_chunking_overlap_creates_overlap() {
 
     let config = ExtractionConfig {
         chunking: Some(ChunkingConfig {
-            max_chars: 50,
-            max_overlap: 15,
-            embedding: None,
-            preset: None,
+            max_characters: 50,
+            overlap: 15,
+            ..Default::default()
         }),
         ..Default::default()
     };
@@ -351,10 +349,9 @@ async fn test_chunking_overlap_maximum() {
 
     let config = ExtractionConfig {
         chunking: Some(ChunkingConfig {
-            max_chars: 60,
-            max_overlap: 10,
-            embedding: None,
-            preset: None,
+            max_characters: 60,
+            overlap: 10,
+            ..Default::default()
         }),
         ..Default::default()
     };
@@ -385,10 +382,9 @@ async fn test_large_document_with_combined_config() {
     let config = ExtractionConfig {
         output_format: OutputFormat::Plain,
         chunking: Some(ChunkingConfig {
-            max_chars: 200,
-            max_overlap: 30,
-            embedding: None,
-            preset: None,
+            max_characters: 200,
+            overlap: 30,
+            ..Default::default()
         }),
         use_cache: true,
         enable_quality_processing: true,

data/vendor/kreuzberg/tests/config_features.rs

@@ -19,10 +19,12 @@ mod helpers;
 async fn test_chunking_enabled() {
     let config = ExtractionConfig {
         chunking: Some(ChunkingConfig {
-            max_chars: 50,
-            max_overlap: 10,
+            max_characters: 50,
+            overlap: 10,
             embedding: None,
             preset: None,
+            trim: true,
+            chunker_type: kreuzberg::chunking::ChunkerType::Text,
         }),
         ..Default::default()
     };
@@ -62,10 +64,12 @@ async fn test_chunking_enabled() {
 async fn test_chunking_with_overlap() {
     let config = ExtractionConfig {
         chunking: Some(ChunkingConfig {
-            max_chars: 100,
-            max_overlap: 20,
+            max_characters: 100,
+            overlap: 20,
             embedding: None,
             preset: None,
+            trim: true,
+            chunker_type: kreuzberg::chunking::ChunkerType::Text,
         }),
         ..Default::default()
     };
@@ -102,10 +106,12 @@ async fn test_chunking_with_overlap() {
 async fn test_chunking_custom_sizes() {
     let config = ExtractionConfig {
         chunking: Some(ChunkingConfig {
-            max_chars: 200,
-            max_overlap: 50,
+            max_characters: 200,
+            overlap: 50,
             embedding: None,
             preset: None,
+            trim: true,
+            chunker_type: kreuzberg::chunking::ChunkerType::Text,
         }),
         ..Default::default()
     };
@@ -512,10 +518,12 @@ async fn test_chunking_with_embeddings() {
 
     let config = ExtractionConfig {
         chunking: Some(ChunkingConfig {
-            max_chars: 100,
-            max_overlap: 20,
+            max_characters: 100,
+            overlap: 20,
             embedding: Some(EmbeddingConfig::default()),
             preset: None,
+            trim: true,
+            chunker_type: kreuzberg::chunking::ChunkerType::Text,
         }),
         ..Default::default()
     };
@@ -582,15 +590,15 @@ async fn test_chunking_with_fast_embeddings() {
 
     let config = ExtractionConfig {
         chunking: Some(ChunkingConfig {
-            max_chars: 100,
-            max_overlap: 20,
+            max_characters: 100,
+            overlap: 20,
             embedding: Some(EmbeddingConfig {
                 model: EmbeddingModelType::Preset {
                     name: "fast".to_string(),
                 },
                 ..Default::default()
             }),
-            preset: None,
+            ..Default::default()
         }),
         ..Default::default()
     };

data/vendor/kreuzberg/tests/config_loading_tests.rs

@@ -36,8 +36,8 @@ max_overlap = 100
     assert!(config.chunking.is_some(), "Should have chunking config");
 
     let chunking = config.chunking.expect("Operation failed");
-    assert_eq!(chunking.max_chars, 1000);
-    assert_eq!(chunking.max_overlap, 100);
+    assert_eq!(chunking.max_characters, 1000);
+    assert_eq!(chunking.overlap, 100);
 }
 
 /// Test loading config from YAML file.
@@ -51,8 +51,8 @@ ocr:
   enabled: true
   backend: tesseract
 chunking:
-  max_chars: 1000
-  max_overlap: 100
+  max_characters: 1000
+  overlap: 100
 "#;
 
     fs::write(&config_path, yaml_content).expect("Operation failed");
@@ -65,8 +65,8 @@ chunking:
     assert!(config.chunking.is_some(), "Should have chunking config");
 
     let chunking = config.chunking.expect("Operation failed");
-    assert_eq!(chunking.max_chars, 1000);
-    assert_eq!(chunking.max_overlap, 100);
+    assert_eq!(chunking.max_characters, 1000);
+    assert_eq!(chunking.overlap, 100);
 }
 
 /// Test loading config from JSON file.
@@ -98,8 +98,8 @@ fn test_from_file_json_succeeds() {
     assert!(config.chunking.is_some(), "Should have chunking config");
 
     let chunking = config.chunking.expect("Operation failed");
-    assert_eq!(chunking.max_chars, 1000);
-    assert_eq!(chunking.max_overlap, 100);
+    assert_eq!(chunking.max_characters, 1000);
+    assert_eq!(chunking.overlap, 100);
 }
 
 /// Test loading config from .yml extension.
@@ -420,6 +420,6 @@ max_overlap = -100
     if let Ok(config) = result
         && let Some(chunking) = config.chunking
     {
-        assert!(chunking.max_chars > 0, "max_chars should be positive");
+        assert!(chunking.max_characters > 0, "max_characters should be positive");
     }
 }

data/vendor/kreuzberg/tests/contract_mcp.rs

@@ -96,8 +96,8 @@ fn test_mcp_chunking_config_nested_matches_rust_core() {
     assert!(config.chunking.is_some(), "Chunking config should be present");
 
     if let Some(chunking) = &config.chunking {
-        assert_eq!(chunking.max_chars, 500, "max_chars should be 500");
-        assert_eq!(chunking.max_overlap, 50, "max_overlap should be 50");
+        assert_eq!(chunking.max_characters, 500, "max_chars should be 500");
+        assert_eq!(chunking.overlap, 50, "max_overlap should be 50");
     }
 
     // Verify roundtrip

data/vendor/kreuzberg/tests/core_integration.rs

@@ -370,8 +370,8 @@ max_overlap = 300
     assert_eq!(ocr_config.language, "deu");
 
     let chunking_config = config.chunking.expect("Operation failed");
-    assert_eq!(chunking_config.max_chars, 2000);
-    assert_eq!(chunking_config.max_overlap, 300);
+    assert_eq!(chunking_config.max_characters, 2000);
+    assert_eq!(chunking_config.overlap, 300);
 }
 
 /// Test config discovery in parent directories.
@@ -481,10 +481,9 @@ async fn test_extraction_with_chunking_config() {
 
     let config = ExtractionConfig {
         chunking: Some(kreuzberg::ChunkingConfig {
-            max_chars: 100,
-            max_overlap: 20,
-            embedding: None,
-            preset: None,
+            max_characters: 100,
+            overlap: 20,
+            ..Default::default()
         }),
         ..Default::default()
     };

data/vendor/kreuzberg/tests/opml_extractor_tests.rs

@@ -576,7 +576,7 @@ async fn test_opml_extraction_statistics() {
                     println!("  Metadata fields: {}", result.metadata.additional.len());
 
                     if !result.metadata.additional.is_empty() {
-                        let keys: Vec<String> = result.metadata.additional.keys().cloned().collect();
+                        let keys: Vec<String> = result.metadata.additional.keys().map(|k| k.to_string()).collect();
                         println!("  Keys: {}", keys.join(", "));
                     }
 

data/vendor/kreuzberg/tests/orgmode_extractor_tests.rs

@@ -778,7 +778,7 @@ async fn test_orgmode_extraction_statistics() {
                     println!("  Metadata fields: {}", result.metadata.additional.len());
 
                     if !result.metadata.additional.is_empty() {
-                        let keys: Vec<String> = result.metadata.additional.keys().cloned().collect();
+                        let keys: Vec<String> = result.metadata.additional.keys().map(|k| k.to_string()).collect();
                         println!("  Keys: {}", keys.join(", "));
                     }
 

data/vendor/kreuzberg/tests/pipeline_integration.rs

@@ -13,6 +13,7 @@ use kreuzberg::plugins::{Plugin, PostProcessor, ProcessingStage};
 use kreuzberg::types::{ExtractionResult, Metadata};
 use kreuzberg::{KreuzbergError, Result};
 use serial_test::serial;
+use std::borrow::Cow;
 use std::sync::Arc;
 
 struct OrderTrackingProcessor {
@@ -74,7 +75,7 @@ impl PostProcessor for MetadataAddingProcessor {
         result
             .metadata
             .additional
-            .insert(self.key.clone(), serde_json::json!(self.value));
+            .insert(Cow::Owned(self.key.clone()), serde_json::json!(self.value));
         Ok(())
     }
 
@@ -134,7 +135,7 @@ async fn test_pipeline_empty_no_processors() {
 
     let result = ExtractionResult {
         content: "original content".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -181,7 +182,7 @@ async fn test_pipeline_single_processor_per_stage() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -228,7 +229,7 @@ async fn test_pipeline_multiple_processors_per_stage() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -266,7 +267,7 @@ async fn test_pipeline_all_stages_enabled() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -302,7 +303,7 @@ async fn test_pipeline_postprocessing_disabled() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -353,7 +354,7 @@ async fn test_pipeline_early_stage_runs_first() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -395,7 +396,7 @@ async fn test_pipeline_middle_stage_runs_second() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -433,7 +434,7 @@ async fn test_pipeline_late_stage_runs_last() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -471,7 +472,7 @@ async fn test_pipeline_within_stage_priority_order() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -522,7 +523,7 @@ async fn test_pipeline_cross_stage_data_flow() {
         #[async_trait]
         impl PostProcessor for MiddleProcessor {
             async fn process(&self, result: &mut ExtractionResult, _: &ExtractionConfig) -> Result<()> {
-                if let Some(stage) = result.metadata.additional.get("stage") {
+                if let Some(stage) = result.metadata.additional.get(&Cow::Borrowed("stage")) {
                     result.content.push_str(&format!(
                         "[saw:{}]",
                         stage.as_str().expect("Failed to extract string from value")
@@ -541,7 +542,7 @@ async fn test_pipeline_cross_stage_data_flow() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -601,7 +602,7 @@ async fn test_pipeline_early_stage_error_recorded() {
 
     let result = ExtractionResult {
         content: "content".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -645,7 +646,7 @@ async fn test_pipeline_middle_stage_error_propagation() {
 
     let result = ExtractionResult {
         content: "content".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -719,7 +720,7 @@ async fn test_pipeline_late_stage_error_doesnt_affect_earlier_stages() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -809,7 +810,7 @@ async fn test_pipeline_processor_error_doesnt_stop_other_processors() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -889,7 +890,7 @@ async fn test_pipeline_multiple_processor_errors() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -933,7 +934,7 @@ async fn test_pipeline_error_context_preservation() {
 
     let result = ExtractionResult {
         content: "content".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -991,8 +992,11 @@ async fn test_pipeline_metadata_added_in_early_visible_in_middle() {
         #[async_trait]
         impl PostProcessor for MiddleReadingProcessor {
             async fn process(&self, result: &mut ExtractionResult, _: &ExtractionConfig) -> Result<()> {
-                if let Some(val) = result.metadata.additional.get("early_key") {
-                    result.metadata.additional.insert("middle_saw".to_string(), val.clone());
+                if let Some(val) = result.metadata.additional.get(&Cow::Borrowed("early_key")) {
+                    result
+                        .metadata
+                        .additional
+                        .insert(Cow::Borrowed("middle_saw"), val.clone());
                 }
                 Ok(())
             }
@@ -1008,7 +1012,7 @@ async fn test_pipeline_metadata_added_in_early_visible_in_middle() {
 
     let result = ExtractionResult {
         content: "content".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -1082,7 +1086,7 @@ async fn test_pipeline_content_modified_in_middle_visible_in_late() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -1135,7 +1139,7 @@ async fn test_pipeline_multiple_processors_modifying_same_metadata() {
                     result
                         .metadata
                         .additional
-                        .insert("shared_key".to_string(), serde_json::json!(self.value));
+                        .insert(Cow::Borrowed("shared_key"), serde_json::json!(self.value));
                     Ok(())
                 }
                 fn processing_stage(&self) -> ProcessingStage {
@@ -1153,7 +1157,7 @@ async fn test_pipeline_multiple_processors_modifying_same_metadata() {
 
     let result = ExtractionResult {
         content: "content".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -1213,13 +1217,13 @@ async fn test_pipeline_processors_reading_previous_output() {
                 let current_count = result
                     .metadata
                     .additional
-                    .get("count")
+                    .get(&Cow::Borrowed("count"))
                     .and_then(|v| v.as_i64())
                     .unwrap_or(0);
                 result
                     .metadata
                     .additional
-                    .insert("count".to_string(), serde_json::json!(current_count + 1));
+                    .insert(Cow::Borrowed("count"), serde_json::json!(current_count + 1));
                 Ok(())
             }
             fn processing_stage(&self) -> ProcessingStage {
@@ -1243,7 +1247,7 @@ async fn test_pipeline_processors_reading_previous_output() {
 
     let result = ExtractionResult {
         content: "content".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -1310,7 +1314,7 @@ async fn test_pipeline_large_content_modification() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -1348,7 +1352,7 @@ async fn test_pipeline_enabled_processors_whitelist() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -1397,7 +1401,7 @@ async fn test_pipeline_disabled_processors_blacklist() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -1446,7 +1450,7 @@ async fn test_pipeline_no_filtering_runs_all() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -1486,7 +1490,7 @@ async fn test_pipeline_empty_whitelist_runs_none() {
 
     let result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,

data/vendor/kreuzberg/tests/plugin_ocr_backend_test.rs

@@ -54,9 +54,10 @@ impl OcrBackend for MockOcrBackend {
             return Err(KreuzbergError::validation("Empty image data".to_string()));
         }
 
+        use std::borrow::Cow;
         Ok(ExtractionResult {
             content: format!("{} (lang: {})", self.return_text, config.language),
-            mime_type: "text/plain".to_string(),
+            mime_type: Cow::Borrowed("text/plain"),
             metadata: Metadata::default(),
             tables: vec![],
             detected_languages: None,
@@ -152,9 +153,10 @@ impl OcrBackend for ValidatingOcrBackend {
             )));
         }
 
+        use std::borrow::Cow;
         Ok(ExtractionResult {
             content: format!("Processed {} bytes", image_bytes.len()),
-            mime_type: "text/plain".to_string(),
+            mime_type: Cow::Borrowed("text/plain"),
             metadata: Metadata::default(),
             tables: vec![],
             detected_languages: None,
@@ -201,19 +203,23 @@ impl Plugin for MetadataOcrBackend {
 impl OcrBackend for MetadataOcrBackend {
     async fn process_image(&self, image_bytes: &[u8], config: &OcrConfig) -> Result<ExtractionResult> {
         let mut metadata = Metadata::default();
-        metadata
-            .additional
-            .insert("ocr_backend".to_string(), serde_json::json!(self.name()));
-        metadata
-            .additional
-            .insert("image_size".to_string(), serde_json::json!(image_bytes.len()));
-        metadata
-            .additional
-            .insert("ocr_language".to_string(), serde_json::json!(config.language));
-
+        metadata.additional.insert(
+            std::borrow::Cow::Borrowed("ocr_backend"),
+            serde_json::json!(self.name()),
+        );
+        metadata.additional.insert(
+            std::borrow::Cow::Borrowed("image_size"),
+            serde_json::json!(image_bytes.len()),
+        );
+        metadata.additional.insert(
+            std::borrow::Cow::Borrowed("ocr_language"),
+            serde_json::json!(config.language),
+        );
+
+        use std::borrow::Cow;
         Ok(ExtractionResult {
             content: "OCR processed text".to_string(),
-            mime_type: "text/plain".to_string(),
+            mime_type: Cow::Borrowed("text/plain"),
             metadata,
             tables: vec![],
             detected_languages: None,

data/vendor/kreuzberg/tests/plugin_postprocessor_test.rs

@@ -11,6 +11,7 @@ use kreuzberg::plugins::{Plugin, PostProcessor, ProcessingStage};
 use kreuzberg::types::ExtractionResult;
 use kreuzberg::{KreuzbergError, Result, extract_file_sync};
 use serial_test::serial;
+use std::borrow::Cow;
 use std::sync::Arc;
 use std::sync::atomic::{AtomicBool, AtomicUsize, Ordering};
 
@@ -82,9 +83,9 @@ impl PostProcessor for MetadataAddingProcessor {
         result
             .metadata
             .additional
-            .insert("processed_by".to_string(), serde_json::json!(self.name()));
+            .insert(Cow::Borrowed("processed_by"), serde_json::json!(self.name()));
         result.metadata.additional.insert(
-            "word_count".to_string(),
+            Cow::Borrowed("word_count"),
             serde_json::json!(result.content.split_whitespace().count()),
         );
         Ok(())

data/vendor/kreuzberg/tests/plugin_system.rs

@@ -11,6 +11,7 @@ use kreuzberg::plugins::registry::{
 use kreuzberg::plugins::{DocumentExtractor, Plugin, PostProcessor, ProcessingStage, Validator};
 use kreuzberg::types::{ExtractionResult, Metadata};
 use kreuzberg::{KreuzbergError, Result};
+use std::borrow::Cow;
 use std::sync::Arc;
 
 struct FailingExtractor {
@@ -52,7 +53,7 @@ impl DocumentExtractor for FailingExtractor {
         } else {
             Ok(ExtractionResult {
                 content: "success".to_string(),
-                mime_type: "text/plain".to_string(),
+                mime_type: Cow::Borrowed("text/plain"),
                 metadata: Metadata::default(),
                 tables: vec![],
                 detected_languages: None,
@@ -299,7 +300,7 @@ fn test_extractor_priority_ordering_complex() {
         async fn extract_bytes(&self, _: &[u8], _: &str, _: &ExtractionConfig) -> Result<ExtractionResult> {
             Ok(ExtractionResult {
                 content: "test".to_string(),
-                mime_type: "text/plain".to_string(),
+                mime_type: Cow::Borrowed("text/plain"),
                 metadata: Metadata::default(),
                 tables: vec![],
                 detected_languages: None,
@@ -461,7 +462,7 @@ async fn test_processor_execution_order_within_stage() {
 
     let mut result = ExtractionResult {
         content: "start".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -498,7 +499,7 @@ async fn test_processor_error_propagation() {
 
     let mut result = ExtractionResult {
         content: "test".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -672,7 +673,7 @@ async fn test_validator_content_validation() {
 
     let short_result = ExtractionResult {
         content: "short".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,
@@ -688,7 +689,7 @@ async fn test_validator_content_validation() {
 
     let long_result = ExtractionResult {
         content: "this is long enough content".to_string(),
-        mime_type: "text/plain".to_string(),
+        mime_type: Cow::Borrowed("text/plain"),
         metadata: Metadata::default(),
         tables: vec![],
         detected_languages: None,

data/vendor/kreuzberg/tests/plugin_validator_test.rs

@@ -157,7 +157,7 @@ impl Plugin for MetadataValidator {
 #[async_trait]
 impl Validator for MetadataValidator {
     async fn validate(&self, result: &ExtractionResult, _config: &ExtractionConfig) -> Result<()> {
-        if !result.metadata.additional.contains_key(&self.required_key) {
+        if !result.metadata.additional.contains_key(self.required_key.as_str()) {
             Err(KreuzbergError::validation(format!(
                 "Required metadata key '{}' missing",
                 self.required_key