Skip to content

A fast, helpful, and open-source document parser

test: add unit tests for find_pdf_in_dir helper

kiran-kumar957bca6c6741e7
diff --git a/crates/liteparse/src/conversion.rs b/crates/liteparse/src/conversion.rs
index a37da74..0b3d787 100644
--- a/crates/liteparse/src/conversion.rs
+++ b/crates/liteparse/src/conversion.rs
@@ -381,12 +381,17 @@ pub async fn convert_office_document(
args.push(file_path);
execute_command(&libre_office_cmd, args, 120_000).await?;
+ find_pdf_in_dir(output_dir).await
+}
- // LibreOffice may sanitise the input filename (e.g. strip parentheses, leading
- // digits, spaces) so the expected `<file_stem>.pdf` may not match what was
- // actually produced. Since `output_dir` is always a fresh temp directory that
- // contains exactly one file after a successful conversion, scanning it for any
- // `.pdf` entry is the most robust approach.
+/// Scan `output_dir` for the first `.pdf` file and return its path.
+///
+/// LibreOffice sanitises filenames during conversion (e.g. strips parentheses,
+/// leading digit sequences, spaces) so the output PDF stem often differs from
+/// the input file stem. Since `output_dir` is always a fresh temp directory
+/// that holds exactly one file after a successful conversion, scanning for any
+/// `.pdf` entry is more robust than constructing a fixed `<stem>.pdf` path.
+async fn find_pdf_in_dir(output_dir: &str) -> Result<String, LiteParseError> {
let mut read_dir = tokio::fs::read_dir(output_dir)
.await
.map_err(|e| LiteParseError::Conversion(format!("cannot read output dir: {e}")))?;
@@ -616,4 +621,47 @@ mod tests {
assert!(r.is_err());
assert!(r.unwrap_err().to_string().contains("unsupported"));
}
+
+ // ── find_pdf_in_dir ──────────────────────────────────────────────────────
+
+ /// Regression test for the LibreOffice filename-sanitisation bug.
+ ///
+ /// LibreOffice strips characters like parentheses and leading digit
+ /// sequences from filenames, so the output PDF stem often differs from the
+ /// input file stem. `find_pdf_in_dir` must locate the PDF regardless of
+ /// what LibreOffice chose to call it.
+ ///
+ /// Scenario: input was `52304751_AnuragLahare_E2 (1).docx`; LibreOffice
+ /// wrote `AnuragLahare_E2_1_.pdf` instead of the expected
+ /// `52304751_AnuragLahare_E2 (1).pdf`.
+ #[tokio::test]
+ async fn test_find_pdf_in_dir_returns_sanitised_name() {
+ let tmp = tempfile::tempdir().unwrap();
+ // Simulate LibreOffice writing a sanitised filename.
+ let sanitised = tmp.path().join("AnuragLahare_E2_1_.pdf");
+ tokio::fs::write(&sanitised, b"%PDF-1.4").await.unwrap();
+
+ let result = find_pdf_in_dir(tmp.path().to_str().unwrap()).await;
+ assert!(result.is_ok(), "should find PDF even with sanitised name");
+ assert!(
+ result.unwrap().ends_with("AnuragLahare_E2_1_.pdf"),
+ "returned path should point to the sanitised file"
+ );
+ }
+
+ /// When LibreOffice somehow produces no PDF (unexpected failure), the
+ /// helpful error message must still be returned.
+ #[tokio::test]
+ async fn test_find_pdf_in_dir_empty_dir_returns_error() {
+ let tmp = tempfile::tempdir().unwrap();
+ let result = find_pdf_in_dir(tmp.path().to_str().unwrap()).await;
+ assert!(result.is_err());
+ assert!(
+ result
+ .unwrap_err()
+ .to_string()
+ .contains("output PDF not found"),
+ "error message should mention missing PDF"
+ );
+ }
}