more strict protocol handling when opening an url

Author: rbri

src/main/java/org/htmlunit/WebClient.java

@@ -514,7 +514,8 @@ <P extends Page> P getPage(final WebWindow webWindow, final WebRequest webReques
     /**
      * Convenient method to build a URL and load it into the current WebWindow as it would be done
      * by {@link #getPage(WebWindow, WebRequest)}.
-     * @param url the URL of the new content
+     * @param url the URL of the new content; in contrast to real browsers plain file url's are not supported.
+     *        You have to use the 'file', 'data', 'blob', 'http' or 'https' protocol.
      * @param <P> the page type
      * @return the new page
      * @throws FailingHttpStatusCodeException if the server returns a failing status code AND the property
@@ -530,7 +531,8 @@ public <P extends Page> P getPage(final String url) throws IOException, FailingH
     /**
      * Convenient method to load a URL into the current top WebWindow as it would be done
      * by {@link #getPage(WebWindow, WebRequest)}.
-     * @param url the URL of the new content
+     * @param url the URL of the new content; in contrast to real browsers plain file url's are not supported.
+     *        You have to use the 'file', 'data', 'blob', 'http' or 'https' protocol.
      * @param <P> the page type
      * @return the new page
      * @throws FailingHttpStatusCodeException if the server returns a failing status code AND the property
@@ -541,7 +543,6 @@ public <P extends Page> P getPage(final URL url) throws IOException, FailingHttp
         final WebRequest request = new WebRequest(url, getBrowserVersion().getHtmlAcceptHeader(),
                                                           getBrowserVersion().getAcceptEncodingHeader());
         request.setCharset(UTF_8);
-
         return getPage(getCurrentWindow().getTopWindow(), request);
     }
 
@@ -1546,7 +1547,8 @@ private WebResponse makeWebResponseForJavaScriptUrl(final WebWindow webWindow, f
      * @return the WebResponse
      */
     public WebResponse loadWebResponse(final WebRequest webRequest) throws IOException {
-        switch (webRequest.getUrl().getProtocol()) {
+        final String protocol = webRequest.getUrl().getProtocol();
+        switch (protocol) {
             case UrlUtils.ABOUT:
                 return makeWebResponseForAboutUrl(webRequest);
 
@@ -1559,8 +1561,12 @@ public WebResponse loadWebResponse(final WebRequest webRequest) throws IOExcepti
             case "blob":
                 return makeWebResponseForBlobUrl(webRequest);
 
-            default:
+            case "http":
+            case "https":
                 return loadWebResponseFromWebConnection(webRequest, ALLOWED_REDIRECTIONS_SAME_URL);
+
+            default:
+                throw new IOException("Unsupported protocol '" + protocol + "'");
         }
     }
 

src/test/java/org/htmlunit/WebClient2Test.java

@@ -15,7 +15,11 @@
 package org.htmlunit;
 
 import static org.htmlunit.httpclient.HtmlUnitBrowserCompatCookieSpec.EMPTY_COOKIE_NAME;
+import static org.junit.Assert.fail;
 
+import java.io.File;
+import java.io.IOException;
+import java.net.URL;
 import java.time.ZonedDateTime;
 import java.time.format.DateTimeFormatter;
 import java.time.temporal.ChronoUnit;
@@ -378,4 +382,61 @@ public void toLocaleLowerCase() throws Exception {
         page = loadPage(html);
         assertEquals("\u0069", page.getTitleText());
     }
+
+    /**
+     * This is supported by reals browsers but not with HtmlUnit.
+     * @throws Exception if the test fails
+     */
+    @Test
+    public void localFile() throws Exception {
+        final URL url = getClass().getClassLoader().getResource("simple.html");
+        String file = url.getFile();
+        if (file.startsWith("/")) {
+            file = file.substring(1);
+        }
+
+        assertTrue(new File(file).exists());
+
+        try (WebClient webClient = new WebClient(getBrowserVersion())) {
+            final HtmlPage page = webClient.getPage(file);
+            fail("IOException expected");
+        }
+        catch (final IOException e) {
+            assertEquals("Unsupported protocol 'c'", e.getMessage());
+        }
+    }
+
+    /**
+     * @throws Exception if the test fails
+     */
+    @Test
+    @Alerts("titel - simple.html")
+    public void localFileFile() throws Exception {
+        final URL url = getClass().getClassLoader().getResource("simple.html");
+        String file = url.getFile();
+        if (file.startsWith("/")) {
+            file = file.substring(1);
+        }
+
+        assertTrue(new File(file).exists());
+
+        try (WebClient webClient = new WebClient(getBrowserVersion())) {
+            final HtmlPage page = webClient.getPage("file://" + file);
+            assertEquals(getExpectedAlerts()[0], page.getTitleText());
+        }
+    }
+
+    /**
+     * @throws Exception if the test fails
+     */
+    @Test
+    public void unknownProtocol() throws Exception {
+        try (WebClient webClient = new WebClient(getBrowserVersion())) {
+            final HtmlPage page = webClient.getPage("unknown://simple.html");
+            fail("IOException expected");
+        }
+        catch (final IOException e) {
+            assertEquals("Unsupported protocol 'unknown'", e.getMessage());
+        }
+    }
 }

src/test/resources/simple.html

@@ -0,0 +1,9 @@
+<!DOCTYPE html>
+<html>
+  <head>
+    <title>titel - simple.html</title>
+  </head>
+  <body>
+    <p>Page Content</p>
+  </body>
+</html>