Handling of Attachments is improved. Now we treat responses only have…

…n and Content-Type header of application/octetstream in the same way as if the Content-Disposition header was set to attachment. This is was real browsers also do (#789)
HtmlUnit · May 18, 2024 · 31e8574 · 31e8574
1 parent 5f85dcd
commit 31e8574
Show file tree

Hide file tree

Showing 3 changed files with 71 additions and 1 deletion.
diff --git a/src/changes/changes.xml b/src/changes/changes.xml
@@ -18,6 +18,10 @@
                 for the XMLHttpRequest ActiveX object that is no longer there.
             </action>
 
+            <action type="update" dev="rbri" issue="#789">
+                Handling of Attachments is improved. Now we treat responses only haven and Content-Type header of application/octetstream
+                in the same way as if the Content-Disposition header was set to attachment. This is was real browsers also do.
+            </action>
             <action type="update" dev="rbri">
                 Upgrade commons-logging to 1.3.2
             </action>

diff --git a/src/main/java/org/htmlunit/attachment/AttachmentHandler.java b/src/main/java/org/htmlunit/attachment/AttachmentHandler.java
@@ -20,6 +20,7 @@
 import org.htmlunit.HttpHeader;
 import org.htmlunit.Page;
 import org.htmlunit.WebResponse;
+import org.htmlunit.util.MimeType;
 
 /**
  * <p>A handler for attachments, which represent pages received from the server which contain
@@ -82,7 +83,12 @@ default boolean handleAttachment(final WebResponse response, final String attach
     default boolean isAttachment(final WebResponse response) {
         final String disp = response.getResponseHeaderValue(HttpHeader.CONTENT_DISPOSITION);
         if (disp == null) {
-            return false;
+            // if there is no content disposition header and content type application/octet-stream
+            // is handled like an attachment by the browsers
+            // https://developer.mozilla.org/en-US/docs/Web/HTTP/Basics_of_HTTP/MIME_types#applicationoctet-stream
+            // They treat it as if the Content-Disposition header was set to attachment, and propose a "Save As" dialog.
+            final String contentType = response.getResponseHeaderValue(HttpHeader.CONTENT_TYPE);
+            return MimeType.APPLICATION_OCTET_STREAM.equals(contentType.toLowerCase(Locale.ROOT));
         }
         return disp.toLowerCase(Locale.ROOT).startsWith("attachment");
     }

diff --git a/src/test/java/org/htmlunit/attachment/AttachmentTest.java b/src/test/java/org/htmlunit/attachment/AttachmentTest.java
@@ -428,4 +428,64 @@ public void handleAttachment(final Page page, final String attachmentFilename) {
 
         assertEquals("htmlunit.zip", attachmentFilenames.get(0));
     }
+
+    /**
+     * Tests attachment callbacks and the contents of attachments.
+     * @throws Exception if an error occurs
+     */
+    @Test
+    public void handleResponseOnlyApplicationOctetstream() throws Exception {
+        final String content1 = "<html><body>\n"
+            + "<form method='POST' name='form' action='" + URL_SECOND + "'>\n"
+            + "<input type='submit' value='ok'>\n"
+            + "</form>\n"
+            + "<a href='#' onclick='document.form.submit()'>click me</a>\n"
+            + "</body></html>";
+        final String content2 = "download file contents";
+
+        final WebClient client = getWebClient();
+        final List<WebResponse> attachments = new ArrayList<>();
+        final List<String> attachmentFilenames = new ArrayList<>();
+
+        client.setAttachmentHandler(new AttachmentHandler() {
+            @Override
+            public boolean handleAttachment(final WebResponse response, final String attachmentFilename) {
+                attachments.add(response);
+                attachmentFilenames.add(attachmentFilename);
+                return true;
+            }
+
+            @Override
+            public void handleAttachment(final Page page, final String attachmentFilename) {
+                throw new IllegalAccessError("handleAttachment(Page, String) called");
+            }
+        });
+
+        final List<NameValuePair> headers = new ArrayList<>();
+        headers.add(new NameValuePair("Content-Type", MimeType.APPLICATION_OCTET_STREAM));
+
+        final MockWebConnection conn = new MockWebConnection();
+        conn.setResponse(URL_FIRST, content1);
+        conn.setResponse(URL_SECOND, content2, 200, "OK", MimeType.TEXT_HTML, headers);
+        client.setWebConnection(conn);
+        assertTrue(attachments.isEmpty());
+        assertTrue(attachmentFilenames.isEmpty());
+
+        final HtmlPage result = client.getPage(URL_FIRST);
+        final HtmlAnchor anchor = result.getAnchors().get(0);
+        final Page clickResult = anchor.click();
+        assertEquals(result, clickResult);
+        assertEquals(1, attachments.size());
+        assertEquals(1, attachmentFilenames.size());
+        assertEquals(1, client.getWebWindows().size());
+
+        final WebResponse attachmentResponse = attachments.get(0);
+        final InputStream attachmentStream = attachmentResponse.getContentAsStream();
+        HttpWebConnectionTest.assertEquals(new ByteArrayInputStream(content2.getBytes()), attachmentStream);
+        assertEquals(MimeType.APPLICATION_OCTET_STREAM, attachmentResponse.getContentType());
+        assertEquals(200, attachmentResponse.getStatusCode());
+        assertEquals(URL_SECOND, attachmentResponse.getWebRequest().getUrl());
+
+        assertNull(attachmentFilenames.get(0));
+    }
 }