Nomad changes

e2e/rust/tests/error_test.rs

@@ -0,0 +1,82 @@
+// This file is auto-generated by alef — DO NOT EDIT.
+// alef:hash:4e15143f4af1ae8bafbdb1506ef057da924484c66a19483966333558ad437e75
+// To regenerate: alef generate
+// To verify freshness: alef verify --exit-code
+// Issues & docs: https://github.com/kreuzberg-dev/alef
+//! E2e tests for category: error
+
+use kreuzberg::extract_bytes_sync;
+
+#[test]
+fn test_error_empty_bytes() {
+    // Graceful handling of empty bytes (should not error)
+    let content = std::fs::read(concat!(
+        env!("CARGO_MANIFEST_DIR"),
+        "/../../test_documents/text/empty.txt"
+    ))
+    .expect("test_documents/text/empty.txt must exist");
+    let mime_type = r#"text/plain"#;
+    let config_json: serde_json::Value = serde_json::from_str(r#"{}"#).unwrap();
+    let config = serde_json::from_value(config_json).unwrap();
+    let _ = extract_bytes_sync(&content, mime_type, &config).expect("should succeed");
+}
+
+#[test]
+fn test_error_empty_mime() {
+    // Error when extracting with empty MIME type
+    let content = std::fs::read(concat!(
+        env!("CARGO_MANIFEST_DIR"),
+        "/../../test_documents/text/plain.txt"
+    ))
+    .expect("test_documents/text/plain.txt must exist");
+    let mime_type = r#""#;
+    let config_json: serde_json::Value = serde_json::from_str(r#"{}"#).unwrap();
+    let config = serde_json::from_value(config_json).unwrap();
+    let result = extract_bytes_sync(&content, mime_type, &config);
+    assert!(result.is_err(), "expected call to fail");
+}
+
+#[test]
+fn test_error_extract_bytes_conflicting_ocr() {
+    // extract_bytes force+disable OCR
+    let content = std::fs::read(concat!(
+        env!("CARGO_MANIFEST_DIR"),
+        "/../../test_documents/text/fake_text.txt"
+    ))
+    .expect("test_documents/text/fake_text.txt must exist");
+    let mime_type = r#"text/plain"#;
+    let config_json: serde_json::Value = serde_json::from_str(r#"{"disable_ocr":true,"force_ocr":true}"#).unwrap();
+    let config = serde_json::from_value(config_json).unwrap();
+    let result = extract_bytes_sync(&content, mime_type, &config);
+    assert!(result.is_err(), "expected call to fail");
+}
+
+#[test]
+fn test_error_invalid_mime_format() {
+    // Error when extracting with invalid MIME type format
+    let content = std::fs::read(concat!(
+        env!("CARGO_MANIFEST_DIR"),
+        "/../../test_documents/text/plain.txt"
+    ))
+    .expect("test_documents/text/plain.txt must exist");
+    let mime_type = r#"not-a-mime"#;
+    let config_json: serde_json::Value = serde_json::from_str(r#"{}"#).unwrap();
+    let config = serde_json::from_value(config_json).unwrap();
+    let result = extract_bytes_sync(&content, mime_type, &config);
+    assert!(result.is_err(), "expected call to fail");
+}
+
+#[test]
+fn test_error_unsupported_mime() {
+    // Error when extracting with unsupported MIME type
+    let content = std::fs::read(concat!(
+        env!("CARGO_MANIFEST_DIR"),
+        "/../../test_documents/text/plain.txt"
+    ))
+    .expect("test_documents/text/plain.txt must exist");
+    let mime_type = r#"application/x-nonexistent"#;
+    let config_json: serde_json::Value = serde_json::from_str(r#"{}"#).unwrap();
+    let config = serde_json::from_value(config_json).unwrap();
+    let result = extract_bytes_sync(&content, mime_type, &config);
+    assert!(result.is_err(), "expected call to fail");
+}