BrowsertrixGatherer: Cleanup pywb QA collection directory

nla · Nov 13, 2023 · 3ff91ba · 3ff91ba
1 parent 72522b6
commit 3ff91ba
Showing 1 changed file with 6 additions and 0 deletions.
diff --git a/gatherer/src/pandas/gatherer/heritrix/BrowsertrixGatherer.java b/gatherer/src/pandas/gatherer/heritrix/BrowsertrixGatherer.java
@@ -228,6 +228,12 @@ public void shutdown() {
     @Override
     public void delete(Instance instance) throws IOException {
         workingArea.deleteInstance(instance.getTitle().getPi(), instance.getDateString());
+        Path pywbDir = pywbService.directoryFor(instance);
+        if (Files.isSymbolicLink(pywbDir)) {
+            Files.deleteIfExists(pywbDir);
+        } else if (Files.isDirectory(pywbDir)) {
+            workingArea.deleteRecursivelyIfExists(pywbDir);
+        }
     }
 
     @Override