From: Mischa POSLAWSKY <perl@shiar.org>
Date: Sat, 19 Oct 2019 16:01:05 +0000 (+0200)
Subject: sitemap: skip subdirectories containing .gitignore
X-Git-Tag: v4.1~16
X-Git-Url: http://git.shiar.net/minimedit.git/commitdiff_plain/bc6019f39a712d18437161942174649589635cb9

sitemap: skip subdirectories containing .gitignore

Assume html files stored in (partially) untracked directories are not meant
for public distribution.  Fixes unintended inclusion of profile/ pages and
archived documents in data/, and avoids a costly search in thumb/.
---

diff --git a/article.inc.php b/article.inc.php
index 6fcb640..d42e498 100644
--- a/article.inc.php
+++ b/article.inc.php
@@ -157,9 +157,12 @@ class PageSearch
 					# ignore symlinks, original contents only
 					return FALSE;
 				}
+				if ($current->isDir()) {
+					# traverse subdirectories unless untracked in any amount
+					return !file_exists("$current/.gitignore");
+				}
 				# match **/*.html
-				return $current->isDir()
-					|| preg_match('/(?<!\.inc)\.html$/', $current->getFilename());
+				return preg_match('/(?<!\.inc)\.html$/', $current->getFilename());
 			}
 		);
 	}