Nomad changes

e2e/rust/tests/format_specific_test.rs

@@ -0,0 +1,93 @@
+// 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: format_specific
+
+use kreuzberg::{extract_bytes_sync, extract_file_sync};
+
+#[test]
+fn test_format_docx_standalone() {
+    // Standalone DOCX extraction using extract_bytes_sync
+    let content = std::fs::read(concat!(
+        env!("CARGO_MANIFEST_DIR"),
+        "/../../test_documents/docx/fake.docx"
+    ))
+    .expect("test_documents/docx/fake.docx must exist");
+    let mime_type = r#"application/vnd.openxmlformats-officedocument.wordprocessingml.document"#;
+    let config = Default::default();
+    let result = extract_bytes_sync(&content, mime_type, &config).expect("should succeed");
+    assert!(
+        result.content.len() >= 20,
+        "expected length >= 20, got {}",
+        result.content.len()
+    );
+}
+
+#[test]
+fn test_format_hwpx_standalone() {
+    // Standalone HWPX extraction using extract_bytes_sync
+    let content = std::fs::read(concat!(
+        env!("CARGO_MANIFEST_DIR"),
+        "/../../test_documents/hwpx/simple.hwpx"
+    ))
+    .expect("test_documents/hwpx/simple.hwpx must exist");
+    let mime_type = r#"application/haansofthwpx"#;
+    let config = Default::default();
+    let result = extract_bytes_sync(&content, mime_type, &config).expect("should succeed");
+    assert!(
+        result.content.len() >= 20,
+        "expected length >= 20, got {}",
+        result.content.len()
+    );
+    assert!(
+        format!("{:?}", result.content).contains(r#"Hello from HWPX"#),
+        "expected to contain: {}",
+        r#"Hello from HWPX"#
+    );
+}
+
+#[test]
+fn test_format_pdf_text() {
+    // Standalone PDF text extraction using extract_bytes_sync
+    let content = std::fs::read(concat!(
+        env!("CARGO_MANIFEST_DIR"),
+        "/../../test_documents/pdf/fake_memo.pdf"
+    ))
+    .expect("test_documents/pdf/fake_memo.pdf must exist");
+    let mime_type = r#"application/pdf"#;
+    let config = Default::default();
+    let result = extract_bytes_sync(&content, mime_type, &config).expect("should succeed");
+    assert!(
+        result.content.len() >= 50,
+        "expected length >= 50, got {}",
+        result.content.len()
+    );
+    assert!(
+        result.content.contains(r#"Mallori"#) || result.content.contains(r#"May"#),
+        "expected to contain at least one of the specified values"
+    );
+}
+
+#[test]
+fn test_format_pptx() {
+    // PPTX presentation extraction using extract_file_sync
+    let path: &str = concat!(env!("CARGO_MANIFEST_DIR"), "/../../test_documents/", "pptx/simple.pptx");
+    let mime_type = Some(r#"application/vnd.openxmlformats-officedocument.presentationml.presentation"#);
+    let config = Default::default();
+    let _ = extract_file_sync(path, mime_type.as_deref(), &config).expect("should succeed");
+}
+
+#[test]
+fn test_format_xlsx() {
+    // XLSX spreadsheet extraction using extract_file_sync
+    let path: &str = concat!(
+        env!("CARGO_MANIFEST_DIR"),
+        "/../../test_documents/",
+        "xlsx/stanley_cups.xlsx"
+    );
+    let mime_type = Some(r#"application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"#);
+    let config = Default::default();
+    let _ = extract_file_sync(path, mime_type.as_deref(), &config).expect("should succeed");
+}