From d87e80f96b231091443882c224d5725802fa4df6 Mon Sep 17 00:00:00 2001
From: Chili Palmer <hugoms@me.com>
Date: Thu, 23 Jul 2026 09:55:19 +0200
Subject: [PATCH] Match bDS2 zero-byte route validation

---
 crates/bds-core/src/engine/validate_site.rs   | 8 +++++++-
 crates/bds-core/tests/m4_generation_engine.rs | 6 +++++-
 specs/editor_misc.allium                      | 3 ++-
 specs/generation.allium                       | 3 ++-
 4 files changed, 16 insertions(+), 4 deletions(-)

diff --git a/crates/bds-core/src/engine/validate_site.rs b/crates/bds-core/src/engine/validate_site.rs
index 986bfa4..7f323cc 100644
--- a/crates/bds-core/src/engine/validate_site.rs
+++ b/crates/bds-core/src/engine/validate_site.rs
@@ -45,19 +45,24 @@ pub fn validate_site(
         .collect::<HashSet<_>>();
 
     let mut actual = HashSet::new();
+    let mut zero_byte = HashSet::new();
     if output_dir.exists() {
         for entry in WalkDir::new(&output_dir).into_iter().filter_map(Result::ok) {
             if !entry.file_type().is_file() || entry.file_name() != "index.html" {
                 continue;
             }
+            let path = relative_path(&output_dir, entry.path());
             if entry.metadata().is_ok_and(|metadata| metadata.len() > 0) {
-                actual.insert(relative_path(&output_dir, entry.path()));
+                actual.insert(path);
+            } else {
+                zero_byte.insert(path);
             }
         }
     }
 
     let mut missing_pages = expected.difference(&actual).cloned().collect::<Vec<_>>();
     let mut extra_pages = actual.difference(&expected).cloned().collect::<Vec<_>>();
+    extra_pages.extend(zero_byte.difference(&expected).cloned());
     let generated_at = queries::generated_file_hash::list_generated_file_hashes(conn, project_id)?
         .into_iter()
         .map(|file| (file.relative_path, file.updated_at))
@@ -74,6 +79,7 @@ pub fn validate_site(
 
     missing_pages.sort();
     extra_pages.sort();
+    extra_pages.dedup();
     stale_pages.sort();
     stale_pages.dedup();
 
diff --git a/crates/bds-core/tests/m4_generation_engine.rs b/crates/bds-core/tests/m4_generation_engine.rs
index 546a60b..81b7584 100644
--- a/crates/bds-core/tests/m4_generation_engine.rs
+++ b/crates/bds-core/tests/m4_generation_engine.rs
@@ -782,12 +782,16 @@ fn site_validation_detects_missing_zero_byte_and_extra_index_routes() {
     std::fs::write(dir.path().join("index.html"), "").unwrap();
     let extra = dir.path().join("ghost/index.html");
     std::fs::create_dir_all(extra.parent().unwrap()).unwrap();
-    std::fs::write(extra, "ghost").unwrap();
+    std::fs::write(&extra, "ghost").unwrap();
+    let zero_byte_extra = dir.path().join("empty/index.html");
+    std::fs::create_dir_all(zero_byte_extra.parent().unwrap()).unwrap();
+    std::fs::write(&zero_byte_extra, "").unwrap();
 
     let report = validate_site(db.conn(), dir.path(), "p1").unwrap();
 
     assert!(report.missing_pages.contains(&"index.html".to_string()));
     assert!(report.extra_pages.contains(&"ghost/index.html".to_string()));
+    assert!(report.extra_pages.contains(&"empty/index.html".to_string()));
 }
 
 #[test]
diff --git a/specs/editor_misc.allium b/specs/editor_misc.allium
index 458560c..d9ffda9 100644
--- a/specs/editor_misc.allium
+++ b/specs/editor_misc.allium
@@ -450,7 +450,8 @@ surface SiteValidationSurface {
 rule SiteValidationScan {
     when: SiteValidationScanRequested()
     -- Parses <loc> entries from sitemap.xml into expected URL set
-    -- Scans HTML output dir for index.html files (zero-byte = missing)
+    -- Scans HTML output dir for index.html files (an expected zero-byte route
+    -- is missing; an unexpected zero-byte route is extra)
     -- Compares source .md mtime against generated HTML mtime
     ensures: SiteValidationReport
 }
diff --git a/specs/generation.allium b/specs/generation.allium
index 9a978f3..9c397db 100644
--- a/specs/generation.allium
+++ b/specs/generation.allium
@@ -267,7 +267,8 @@ rule ValidateSite {
     --   expected: sitemap-eligible routes derived from published snapshots
     --   actual: non-empty **/index.html files on disk
     --   missing: expected routes with no non-empty index.html
-    --   extra: non-empty index.html routes absent from the expected set
+    --   extra: index.html routes absent from the expected set, including
+    --          zero-byte files
     --   stale: existing post routes whose source mtime is more than one second
     --          newer than both the output mtime and tracked generation time
     -- Standalone HTML, XML, JSON, assets, Pagefind files, and content hashes are