Edge Case fixes, bug fixes, and UI Cleanup.

Sitemap Builder/sitemap_builder.py

@@ -175,6 +175,15 @@ def get_log_path(output_path: Path) -> Path:
     return output_path.with_suffix(output_path.suffix + DEFAULT_LOG_SUFFIX)
 
 
+def cleanup_run_files(output_path: Path) -> list[Path]:
+    removed_paths: list[Path] = []
+    for path in (Path(output_path), get_state_path(Path(output_path)), get_log_path(Path(output_path))):
+        if path.exists():
+            path.unlink()
+            removed_paths.append(path)
+    return removed_paths
+
+
 def log_message(log_path: Path, message: str) -> None:
     log_path.parent.mkdir(parents=True, exist_ok=True)
     timestamp = time.strftime("%Y-%m-%d %H:%M:%S")
@@ -599,6 +608,7 @@ def crawl_site(
                 break
 
             current = resolve_alias(state.queue.popleft(), state.alias_to_canonical)
+            state.queued.discard(current)
             if current in state.visited:
                 continue
 
@@ -647,6 +657,7 @@ def crawl_site(
 
                 while state.queue and len(pending) < workers and len(state.visited) < max_pages:
                     current = resolve_alias(state.queue.popleft(), state.alias_to_canonical)
+                    state.queued.discard(current)
                     if current in state.visited:
                         continue
 
@@ -790,6 +801,11 @@ def run_crawl(
                 "The saved crawl state uses a different document setting. "
                 "Keep the same choice or start a fresh crawl."
             )
+        if state.include_subdomains != include_subdomains:
+            raise ValueError(
+                "The saved crawl state uses a different subdomain setting. "
+                "Keep the same choice or start a fresh crawl."
+            )
     else:
         state = initialize_state(normalized_start, include_subdomains, include_documents)