Extend duplicate detection to non-adjacent paths

1 files changed, 28 insertions, 6 deletions

diff --git a/src/main.rs b/src/main.rs
index bf687c6..7455e23 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -1,9 +1,10 @@
 use std::{
+    collections::HashSet,
     env,
     ffi::OsStr,
     io::{self, Read, Write},
     os::unix::ffi::OsStrExt,
-    path::{Path, PathBuf},
+    path::Path,
 };
 
 use repo2markdown::{
@@ -74,7 +75,7 @@ pub fn run<R: Read, W: Write>(
     let project_name = project_name.unwrap_or_else(|| derive_project_name(root));
     renderer.render_header(project_name)?;
 
-    let mut last_rendered_path = PathBuf::from("");
+    let mut seen_paths = HashSet::new();
     for segment in buf.split(|b| *b == 0) {
         if segment.is_empty() {
             continue;
@@ -82,15 +83,14 @@ pub fn run<R: Read, W: Write>(
 
         let path = Path::new(OsStr::from_bytes(segment));
         let normalized_path = normalizer.normalize(path)?;
-        if normalized_path.relative == last_rendered_path {
+        if !seen_paths.insert(normalized_path.relative.clone()) {
             logger.warn(format!(
                 "Duplicate file detected: {:?}",
                 normalized_path.relative
             ));
-        } else {
-            renderer.render_path(&normalized_path)?;
-            last_rendered_path = normalized_path.relative;
+            continue;
         }
+        renderer.render_path(&normalized_path)?;
     }
     Ok(())
 }
@@ -270,6 +270,28 @@ mod tests {
     }
 
     #[test]
+    fn duplicate_files_are_skipped_with_preserved_display_order_even_if_not_adjacent() {
+        let temp_dir = tempdir().unwrap();
+        let origin = temp_dir.path();
+        let root = temp_dir.path();
+
+        fs::write(origin.join("a.rs"), "A").unwrap();
+        fs::write(origin.join("b.rs"), "B").unwrap();
+
+        let input = Cursor::new(b"a.rs\0b.rs\0a.rs\0");
+        let mut output = Vec::new();
+
+        run_with_default_logger(input, &mut output, root, origin, None).unwrap();
+
+        let output = String::from_utf8(output).unwrap();
+        assert_eq!(output.matches("## File: a.rs").count(), 1);
+        assert_eq!(output.matches("## File: b.rs").count(), 1);
+        let a_pos = output.find("a.rs").unwrap();
+        let b_pos = output.find("b.rs").unwrap();
+        assert!(a_pos < b_pos);
+    }
+
+    #[test]
     fn project_name_is_derived_from_root_by_default_even_if_directory_does_not_exist() {
         let temp_dir = tempdir().unwrap();
         let origin_base = temp_dir.path();