kreuzberg 4.1.1 → 4.2.0

data/vendor/kreuzberg/tests/jats_extractor_tests.rs

@@ -83,13 +83,13 @@ mod jats_extractor_tests {
             .await;
 
         assert!(result.is_ok());
-        let extraction = result.unwrap();
+        let extraction = result.expect("Operation failed");
 
         assert!(extraction.content.contains("Effects of Caffeine"));
         assert!(extraction.content.contains("Introduction"));
 
         assert!(extraction.metadata.subject.is_some());
-        let subject = extraction.metadata.subject.unwrap();
+        let subject = extraction.metadata.subject.expect("Operation failed");
         assert!(subject.contains("Effects of Caffeine"));
 
         assert!(subject.contains("10.1371"));
@@ -144,9 +144,9 @@ mod jats_extractor_tests {
             .await;
 
         assert!(result.is_ok());
-        let extraction = result.unwrap();
+        let extraction = result.expect("Operation failed");
 
-        let subject = extraction.metadata.subject.unwrap();
+        let subject = extraction.metadata.subject.expect("Operation failed");
         assert!(subject.contains("Alpha"));
         assert!(subject.contains("Beta"));
         assert!(subject.contains("Gamma"));
@@ -201,7 +201,7 @@ mod jats_extractor_tests {
             .await;
 
         assert!(result.is_ok());
-        let extraction = result.unwrap();
+        let extraction = result.expect("Operation failed");
 
         assert!(extraction.content.contains("Introduction"));
         assert!(extraction.content.contains("Methods"));
@@ -273,7 +273,7 @@ mod jats_extractor_tests {
             .await;
 
         assert!(result.is_ok());
-        let extraction = result.unwrap();
+        let extraction = result.expect("Operation failed");
 
         assert_eq!(extraction.tables.len(), 1);
         let table = &extraction.tables[0];
@@ -327,7 +327,7 @@ mod jats_extractor_tests {
             .await;
 
         assert!(result.is_ok());
-        let extraction = result.unwrap();
+        let extraction = result.expect("Operation failed");
 
         assert_eq!(extraction.tables.len(), 2);
         assert_eq!(extraction.tables[0].cells[0].len(), 2);
@@ -390,7 +390,7 @@ mod jats_extractor_tests {
             .await;
 
         assert!(result.is_ok());
-        let extraction = result.unwrap();
+        let extraction = result.expect("Operation failed");
 
         assert!(extraction.content.contains("Previous research"));
         assert!(extraction.content.contains("Other studies"));
@@ -429,9 +429,9 @@ mod jats_extractor_tests {
             .await;
 
         assert!(result.is_ok());
-        let extraction = result.unwrap();
+        let extraction = result.expect("Operation failed");
 
-        let subject = extraction.metadata.subject.unwrap();
+        let subject = extraction.metadata.subject.expect("Operation failed");
         assert!(subject.contains("background") || subject.contains("Background") || subject.contains("Abstract"));
     }
 
@@ -457,7 +457,7 @@ mod jats_extractor_tests {
             .await;
 
         assert!(result.is_ok());
-        let extraction = result.unwrap();
+        let extraction = result.expect("Operation failed");
 
         assert!(extraction.metadata.subject.is_some());
     }
@@ -486,7 +486,7 @@ mod jats_extractor_tests {
             .await;
 
         assert!(result.is_ok());
-        let extraction = result.unwrap();
+        let extraction = result.expect("Operation failed");
 
         assert!(extraction.metadata.created_at.is_some());
     }
@@ -511,7 +511,7 @@ mod jats_extractor_tests {
             .await;
 
         assert!(result.is_ok());
-        let extraction = result.unwrap();
+        let extraction = result.expect("Operation failed");
         assert!(extraction.content.is_empty() || extraction.content.trim().is_empty());
     }
 
@@ -578,7 +578,7 @@ mod jats_extractor_tests {
             .await;
 
         assert!(result.is_ok());
-        let extraction = result.unwrap();
+        let extraction = result.expect("Operation failed");
 
         assert!(extraction.content.contains("First paragraph"));
         assert!(extraction.content.contains("Second paragraph"));
@@ -611,9 +611,9 @@ mod jats_extractor_tests {
             .await;
 
         assert!(result.is_ok());
-        let extraction = result.unwrap();
+        let extraction = result.expect("Operation failed");
 
-        let subject = extraction.metadata.subject.unwrap();
+        let subject = extraction.metadata.subject.expect("Operation failed");
         assert!(subject.contains("keyword") || subject.contains("Keyword"));
     }
 
@@ -630,7 +630,7 @@ mod jats_extractor_tests {
                 .await;
 
             assert!(result.is_ok());
-            let extraction = result.unwrap();
+            let extraction = result.expect("Operation failed");
 
             assert!(!extraction.content.is_empty());
             assert!(extraction.metadata.subject.is_some());

data/vendor/kreuzberg/tests/jupyter_extractor_tests.rs

@@ -61,7 +61,7 @@ async fn test_jupyter_simple_notebook_extraction() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     assert_eq!(
         extraction.mime_type, "application/x-ipynb+json",
@@ -156,7 +156,7 @@ async fn test_jupyter_mime_notebook_extraction() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     assert_eq!(
         extraction.mime_type, "application/x-ipynb+json",
@@ -261,7 +261,7 @@ async fn test_jupyter_mime_out_notebook_extraction() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     assert_eq!(
         extraction.mime_type, "application/x-ipynb+json",
@@ -351,7 +351,7 @@ async fn test_jupyter_rank_notebook_extraction() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     assert_eq!(
         extraction.mime_type, "application/x-ipynb+json",
@@ -440,7 +440,7 @@ async fn test_jupyter_metadata_aggregation() {
             continue;
         }
 
-        let extraction = result.unwrap();
+        let extraction = result.expect("Operation failed");
 
         assert!(
             !extraction.content.is_empty(),
@@ -491,7 +491,7 @@ async fn test_jupyter_cell_content_aggregation() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     let code_indicators = ["class", "def", "import", "from", "python"];
     let code_count = code_indicators
@@ -563,7 +563,7 @@ async fn test_jupyter_mime_output_handling() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     assert!(
         extraction.content.contains("image")
@@ -620,7 +620,7 @@ async fn test_jupyter_notebook_structure_preservation() {
         return;
     }
 
-    let extraction = result.unwrap();
+    let extraction = result.expect("Operation failed");
 
     let cell_id_patterns = ["uid1", "uid2", "uid3", "uid4", "uid6"];
     let id_count = cell_id_patterns
@@ -672,7 +672,7 @@ async fn test_jupyter_pandoc_baseline_alignment() {
             continue;
         }
 
-        let extraction = result.unwrap();
+        let extraction = result.expect("Operation failed");
 
         assert!(
             extraction.content.contains("cell")

data/vendor/kreuzberg/tests/keywords_integration.rs

@@ -38,7 +38,7 @@ El procesamiento del lenguaje natural es un campo de la inteligencia artificial
 #[test]
 fn test_yake_basic_extraction() {
     let config = KeywordConfig::yake();
-    let keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
 
     assert!(!keywords.is_empty(), "Should extract keywords from document");
     assert!(
@@ -80,7 +80,7 @@ fn test_yake_basic_extraction() {
 #[test]
 fn test_rake_basic_extraction() {
     let config = KeywordConfig::rake();
-    let keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
 
     assert!(!keywords.is_empty(), "Should extract keywords from document");
     assert!(
@@ -122,8 +122,8 @@ fn test_yake_vs_rake_comparison() {
     let yake_config = KeywordConfig::yake().with_max_keywords(5);
     let rake_config = KeywordConfig::rake().with_max_keywords(5);
 
-    let yake_keywords = extract_keywords(ML_DOCUMENT, &yake_config).unwrap();
-    let rake_keywords = extract_keywords(ML_DOCUMENT, &rake_config).unwrap();
+    let yake_keywords = extract_keywords(ML_DOCUMENT, &yake_config).expect("Operation failed");
+    let rake_keywords = extract_keywords(ML_DOCUMENT, &rake_config).expect("Operation failed");
 
     assert!(!yake_keywords.is_empty(), "YAKE should extract keywords");
     assert!(!rake_keywords.is_empty(), "RAKE should extract keywords");
@@ -161,7 +161,7 @@ fn test_yake_vs_rake_comparison() {
 #[test]
 fn test_yake_with_max_keywords() {
     let config = KeywordConfig::yake().with_max_keywords(3);
-    let keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
 
     assert!(keywords.len() <= 3, "Should respect max_keywords=3 limit");
 
@@ -176,7 +176,7 @@ fn test_yake_with_max_keywords() {
 #[test]
 fn test_rake_with_max_keywords() {
     let config = KeywordConfig::rake().with_max_keywords(3);
-    let keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
 
     assert!(keywords.len() <= 3, "Should respect max_keywords=3 limit");
 
@@ -191,7 +191,7 @@ fn test_rake_with_max_keywords() {
 #[test]
 fn test_yake_with_min_score() {
     let config = KeywordConfig::yake().with_min_score(0.5);
-    let keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
 
     for keyword in &keywords {
         assert!(
@@ -207,7 +207,7 @@ fn test_yake_with_min_score() {
 #[test]
 fn test_rake_with_min_score() {
     let config = KeywordConfig::rake().with_min_score(0.2);
-    let keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
 
     for keyword in &keywords {
         assert!(
@@ -223,7 +223,7 @@ fn test_rake_with_min_score() {
 #[test]
 fn test_yake_with_ngram_range() {
     let config = KeywordConfig::yake().with_ngram_range(1, 1);
-    let keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
 
     for keyword in &keywords {
         let word_count = keyword.text.split_whitespace().count();
@@ -231,7 +231,7 @@ fn test_yake_with_ngram_range() {
     }
 
     let config = KeywordConfig::yake().with_ngram_range(2, 3);
-    let keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
 
     for keyword in &keywords {
         let word_count = keyword.text.split_whitespace().count();
@@ -248,7 +248,7 @@ fn test_yake_with_ngram_range() {
 #[test]
 fn test_rake_with_ngram_range() {
     let config = KeywordConfig::rake().with_ngram_range(1, 1);
-    let keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
 
     for keyword in &keywords {
         let word_count = keyword.text.split_whitespace().count();
@@ -256,7 +256,7 @@ fn test_rake_with_ngram_range() {
     }
 
     let config = KeywordConfig::rake().with_ngram_range(2, 2);
-    let keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
 
     for keyword in &keywords {
         let word_count = keyword.text.split_whitespace().count();
@@ -268,7 +268,7 @@ fn test_rake_with_ngram_range() {
 #[test]
 fn test_rake_with_spanish() {
     let config = KeywordConfig::rake().with_language("es");
-    let keywords = extract_keywords(SPANISH_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(SPANISH_DOCUMENT, &config).expect("Operation failed");
 
     assert!(!keywords.is_empty(), "Should extract Spanish keywords");
 
@@ -294,7 +294,7 @@ fn test_rake_with_spanish() {
 #[test]
 fn test_yake_with_spanish() {
     let config = KeywordConfig::yake().with_language("es");
-    let keywords = extract_keywords(SPANISH_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(SPANISH_DOCUMENT, &config).expect("Operation failed");
 
     assert!(!keywords.is_empty(), "Should extract Spanish keywords");
 
@@ -308,7 +308,7 @@ fn test_yake_with_spanish() {
 #[test]
 fn test_rake_empty_document() {
     let config = KeywordConfig::rake();
-    let keywords = extract_keywords("", &config).unwrap();
+    let keywords = extract_keywords("", &config).expect("Operation failed");
 
     assert!(keywords.is_empty(), "Empty document should yield no keywords");
 }
@@ -317,7 +317,7 @@ fn test_rake_empty_document() {
 #[test]
 fn test_yake_empty_document() {
     let config = KeywordConfig::yake();
-    let keywords = extract_keywords("", &config).unwrap();
+    let keywords = extract_keywords("", &config).expect("Operation failed");
 
     assert!(keywords.is_empty(), "Empty document should yield no keywords");
 }
@@ -327,7 +327,7 @@ fn test_yake_empty_document() {
 fn test_rake_short_document() {
     let short_text = "Machine learning algorithms.";
     let config = KeywordConfig::rake();
-    let keywords = extract_keywords(short_text, &config).unwrap();
+    let keywords = extract_keywords(short_text, &config).expect("Operation failed");
 
     println!(
         "Keywords from short text: {:?}",
@@ -340,7 +340,7 @@ fn test_rake_short_document() {
 fn test_yake_short_document() {
     let short_text = "Machine learning algorithms.";
     let config = KeywordConfig::yake();
-    let keywords = extract_keywords(short_text, &config).unwrap();
+    let keywords = extract_keywords(short_text, &config).expect("Operation failed");
 
     println!(
         "YAKE keywords from short text: {:?}",
@@ -353,13 +353,13 @@ fn test_yake_short_document() {
 fn test_rake_different_domains() {
     let config = KeywordConfig::rake().with_max_keywords(5);
 
-    let ml_keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let ml_keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
     println!("\nML domain keywords:");
     for kw in &ml_keywords {
         println!("  {} (score: {:.3})", kw.text, kw.score);
     }
 
-    let climate_keywords = extract_keywords(CLIMATE_DOCUMENT, &config).unwrap();
+    let climate_keywords = extract_keywords(CLIMATE_DOCUMENT, &config).expect("Operation failed");
     println!("\nClimate domain keywords:");
     for kw in &climate_keywords {
         println!("  {} (score: {:.3})", kw.text, kw.score);
@@ -395,13 +395,13 @@ fn test_rake_different_domains() {
 fn test_yake_different_domains() {
     let config = KeywordConfig::yake().with_max_keywords(5);
 
-    let ml_keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let ml_keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
     println!("\nYAKE ML domain keywords:");
     for kw in &ml_keywords {
         println!("  {} (score: {:.3})", kw.text, kw.score);
     }
 
-    let climate_keywords = extract_keywords(CLIMATE_DOCUMENT, &config).unwrap();
+    let climate_keywords = extract_keywords(CLIMATE_DOCUMENT, &config).expect("Operation failed");
     println!("\nYAKE Climate domain keywords:");
     for kw in &climate_keywords {
         println!("  {} (score: {:.3})", kw.text, kw.score);
@@ -415,7 +415,7 @@ fn test_yake_different_domains() {
 #[test]
 fn test_rake_score_distribution() {
     let config = KeywordConfig::rake();
-    let keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
 
     if keywords.is_empty() {
         return;
@@ -439,7 +439,7 @@ fn test_rake_score_distribution() {
 #[test]
 fn test_yake_score_distribution() {
     let config = KeywordConfig::yake();
-    let keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
 
     if keywords.is_empty() {
         return;
@@ -463,7 +463,7 @@ fn test_yake_score_distribution() {
 #[test]
 fn test_keyword_struct_properties() {
     let config = KeywordConfig::default();
-    let keywords = extract_keywords(ML_DOCUMENT, &config).unwrap();
+    let keywords = extract_keywords(ML_DOCUMENT, &config).expect("Operation failed");
 
     if keywords.is_empty() {
         return;

data/vendor/kreuzberg/tests/keywords_quality.rs

@@ -189,7 +189,7 @@ Global warming is the long-term heating of Earth's climate system. Climate scien
 #[test]
 fn test_yake_quality_ml_document_default_config() {
     let config = KeywordConfig::yake();
-    let keywords = extract_keywords(ML_DOC_SAMPLE, &config).unwrap();
+    let keywords = extract_keywords(ML_DOC_SAMPLE, &config).expect("Operation failed");
 
     assert!(!keywords.is_empty(), "Should extract keywords with default config");
 
@@ -239,7 +239,7 @@ fn test_yake_quality_ml_document_default_config() {
 #[test]
 fn test_rake_quality_ml_document_default_config() {
     let config = KeywordConfig::rake();
-    let keywords = extract_keywords(ML_DOC_SAMPLE, &config).unwrap();
+    let keywords = extract_keywords(ML_DOC_SAMPLE, &config).expect("Operation failed");
 
     assert!(!keywords.is_empty(), "Should extract keywords with default config");
 
@@ -289,7 +289,7 @@ fn test_rake_quality_ml_document_default_config() {
 #[test]
 fn test_yake_quality_climate_document_default_config() {
     let config = KeywordConfig::yake();
-    let keywords = extract_keywords(CLIMATE_DOC_SAMPLE, &config).unwrap();
+    let keywords = extract_keywords(CLIMATE_DOC_SAMPLE, &config).expect("Operation failed");
 
     assert!(!keywords.is_empty(), "Should extract keywords with default config");
 
@@ -331,7 +331,7 @@ fn test_yake_quality_climate_document_default_config() {
 #[test]
 fn test_rake_quality_climate_document_default_config() {
     let config = KeywordConfig::rake();
-    let keywords = extract_keywords(CLIMATE_DOC_SAMPLE, &config).unwrap();
+    let keywords = extract_keywords(CLIMATE_DOC_SAMPLE, &config).expect("Operation failed");
 
     assert!(!keywords.is_empty(), "Should extract keywords with default config");
 
@@ -375,8 +375,8 @@ fn test_yake_vs_rake_quality_comparison() {
     let yake_config = KeywordConfig::yake();
     let rake_config = KeywordConfig::rake();
 
-    let yake_keywords = extract_keywords(ML_DOC_SAMPLE, &yake_config).unwrap();
-    let rake_keywords = extract_keywords(ML_DOC_SAMPLE, &rake_config).unwrap();
+    let yake_keywords = extract_keywords(ML_DOC_SAMPLE, &yake_config).expect("Operation failed");
+    let rake_keywords = extract_keywords(ML_DOC_SAMPLE, &rake_config).expect("Operation failed");
 
     let yake_extracted: Vec<&str> = yake_keywords.iter().map(|k| k.text.as_str()).collect();
     let rake_extracted: Vec<&str> = rake_keywords.iter().map(|k| k.text.as_str()).collect();
@@ -414,7 +414,7 @@ fn test_yake_quality_with_optimized_config() {
         .with_ngram_range(1, 3)
         .with_min_score(0.0);
 
-    let keywords = extract_keywords(ML_DOC_SAMPLE, &config).unwrap();
+    let keywords = extract_keywords(ML_DOC_SAMPLE, &config).expect("Operation failed");
 
     let extracted: Vec<&str> = keywords.iter().map(|k| k.text.as_str()).collect();
     let ground_truth = get_ml_ground_truth();
@@ -441,7 +441,7 @@ fn test_rake_quality_with_optimized_config() {
         .with_ngram_range(1, 3)
         .with_min_score(0.0);
 
-    let keywords = extract_keywords(ML_DOC_SAMPLE, &config).unwrap();
+    let keywords = extract_keywords(ML_DOC_SAMPLE, &config).expect("Operation failed");
 
     let extracted: Vec<&str> = keywords.iter().map(|k| k.text.as_str()).collect();
     let ground_truth = get_ml_ground_truth();
@@ -464,7 +464,7 @@ fn test_rake_quality_with_optimized_config() {
 #[test]
 fn test_extracted_keywords_are_domain_relevant() {
     let config = KeywordConfig::default();
-    let keywords = extract_keywords(ML_DOC_SAMPLE, &config).unwrap();
+    let keywords = extract_keywords(ML_DOC_SAMPLE, &config).expect("Operation failed");
 
     let ml_terms = [
         "machine",

data/vendor/kreuzberg/tests/latex_extractor_tests.rs

@@ -28,9 +28,9 @@ fn test_file_path(filename: &str) -> PathBuf {
     let manifest_dir = env!("CARGO_MANIFEST_DIR");
     PathBuf::from(manifest_dir)
         .parent()
-        .unwrap()
+        .expect("Operation failed")
         .parent()
-        .unwrap()
+        .expect("Operation failed")
         .join("test_documents")
         .join("latex")
         .join(filename)

data/vendor/kreuzberg/tests/markdown_extractor_tests.rs

@@ -486,5 +486,5 @@ async fn test_special_characters_in_metadata() {
 
     let title = result.metadata.additional.get("title").and_then(|v| v.as_str());
     assert!(title.is_some());
-    assert!(title.unwrap().contains("&") || title.unwrap().contains("Part"));
+    assert!(title.expect("Operation failed").contains("&") || title.expect("Operation failed").contains("Part"));
 }