fix #2866 Prevent URL decoding of filenames for specific schemes.

codelibs · Jan 9, 2025 · 162081c · 162081c
1 parent 541319b
commit 162081c
Show file tree

Hide file tree

Showing 2 changed files with 27 additions and 1 deletion.
diff --git a/src/main/java/org/codelibs/fess/crawler/transformer/FessTransformer.java b/src/main/java/org/codelibs/fess/crawler/transformer/FessTransformer.java
@@ -187,7 +187,9 @@ default String getFileName(final String url, final String encoding) {
                 u = u.substring(0, idx);
             }
         }
-        u = decodeUrlAsName(u, u.startsWith("file:"));
+        if (!u.startsWith("smb:") && !u.startsWith("smb1:") && !u.startsWith("ftp:")) {
+            u = decodeUrlAsName(u, u.startsWith("file:"));
+        }
         idx = u.lastIndexOf('/');
         if (idx >= 0) {
             if (u.length() > idx + 1) {

diff --git a/src/test/java/org/codelibs/fess/crawler/transformer/FessFileTransformerTest.java b/src/test/java/org/codelibs/fess/crawler/transformer/FessFileTransformerTest.java
@@ -109,6 +109,30 @@ public void test_getFileName_ok() throws Exception {
         url = "smb://example.com/test?.txt";
         exp = "test?.txt";
         assertEquals(exp, transformer.getFileName(url, Constants.UTF_8));
+
+        url = "https://example.com/test%E3%81%82.txt";
+        exp = "testあ.txt";
+        assertEquals(exp, transformer.getFileName(url, Constants.UTF_8));
+
+        url = "file://example.com/test%E3%81%82.txt";
+        exp = "testあ.txt";
+        assertEquals(exp, transformer.getFileName(url, Constants.UTF_8));
+
+        url = "storage://example.com/test%E3%81%82.txt";
+        exp = "testあ.txt";
+        assertEquals(exp, transformer.getFileName(url, Constants.UTF_8));
+
+        url = "smb://example.com/test%E3%81%82.txt";
+        exp = "test%E3%81%82.txt";
+        assertEquals(exp, transformer.getFileName(url, Constants.UTF_8));
+
+        url = "smb1://example.com/test%E3%81%82.txt";
+        exp = "test%E3%81%82.txt";
+        assertEquals(exp, transformer.getFileName(url, Constants.UTF_8));
+
+        url = "ftp://example.com/test%E3%81%82.txt";
+        exp = "test%E3%81%82.txt";
+        assertEquals(exp, transformer.getFileName(url, Constants.UTF_8));
     }
 
     public void test_decodeUrl_null() throws Exception {