Minor patch

2026-10-17 05:07:28 +08:00 · 2026-07-09 09:57:08 +02:00
parent c1363dd44c
commit 478888b19e
3 changed files with 5 additions and 2 deletions
@@ -168,6 +168,8 @@ def crawl(target, post=None, cookie=None):
            finally:
                if found:
                    if items:
+                        # keep sitemap-derived URLs on-scope, exactly like the HTML-crawl path above
+                        items = OrderedSet(_ for _ in items if (re.search(conf.scope, _, re.I) if conf.scope else checkSameHost(_, target)))
                        for item in items:
                            if re.search(r"(.*?)\?(.+)", item):
                                threadData.shared.value.add(item)