Fix XvideosRipper: update URL pattern to match current URL scheme. (#2054)

* Fix XvideosRipper: update URL pattern to match current URL scheme. Also, extract patterns to be compiled once in static members to avoid repeating the match patterns and make the code read a little more clearly. Improve album name for the video case.

src/main/java/com/rarchives/ripme/ripper/rippers/XvideosRipper.java

@@ -9,18 +9,19 @@ import java.util.List;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 
-
 import com.rarchives.ripme.ripper.AbstractSingleFileRipper;
+
 import org.jsoup.nodes.Document;
 import org.jsoup.nodes.Element;
 import org.jsoup.select.Elements;
 
-import com.rarchives.ripme.utils.Http;
-
 public class XvideosRipper extends AbstractSingleFileRipper {
 
     private static final String HOST = "xvideos";
 
+    private static final Pattern videoPattern = Pattern.compile("^https?://[wm.]*xvideos\\.com/video\\.([^/]*)(.*)$");
+    private static final Pattern albumPattern = Pattern.compile("^https?://[wm.]*xvideos\\.com/(profiles|amateurs)/([a-zA-Z0-9_-]+)/photos/(\\d+)/([a-zA-Z0-9_-]+)$");
+
     public XvideosRipper(URL url) throws IOException {
         super(url);
     }
@@ -37,12 +38,12 @@ public class XvideosRipper extends AbstractSingleFileRipper {
 
     @Override
     public boolean canRip(URL url) {
-        Pattern p = Pattern.compile("^https?://[wm.]*xvideos\\.com/video[0-9]+.*$");
+        Pattern p = videoPattern;
         Matcher m = p.matcher(url.toExternalForm());
         if (m.matches()) {
             return true;
         }
-        p = Pattern.compile("^https?://[wm.]*xvideos\\.com/profiles/[a-zA-Z0-9_-]+/photos/\\d+/[a-zA-Z0-9_-]+$");
+        p = albumPattern;
         m = p.matcher(url.toExternalForm());
         if (m.matches()) {
             return true;
@@ -52,15 +53,15 @@ public class XvideosRipper extends AbstractSingleFileRipper {
 
     @Override
     public String getGID(URL url) throws MalformedURLException {
-        Pattern p = Pattern.compile("^https?://[wm.]*xvideos\\.com/video([0-9]+).*$");
+        Pattern p = videoPattern;
         Matcher m = p.matcher(url.toExternalForm());
         if (m.matches()) {
             return m.group(1);
         }
-        p = Pattern.compile("^https?://[wm.]*xvideos\\.com/profiles/[a-zA-Z0-9_-]+/photos/(\\d+)/[a-zA-Z0-9_-]+$");
+        p = albumPattern;
         m = p.matcher(url.toExternalForm());
         if (m.matches()) {
-            return m.group(1);
+            return m.group(3);
         }
 
         throw new MalformedURLException(
@@ -72,7 +73,7 @@ public class XvideosRipper extends AbstractSingleFileRipper {
     @Override
     public List<String> getURLsFromPage(Document doc) {
         List<String> results = new ArrayList<>();
-        Pattern p = Pattern.compile("^https?://[wm.]*xvideos\\.com/video([0-9]+).*$");
+        Pattern p = videoPattern;
         Matcher m = p.matcher(url.toExternalForm());
         if (m.matches()) {
             Elements scripts = doc.select("script");
@@ -106,12 +107,21 @@ public class XvideosRipper extends AbstractSingleFileRipper {
 
     @Override
     public String getAlbumTitle(URL url) throws MalformedURLException, URISyntaxException {
-        Pattern p = Pattern.compile("^https?://[wm.]*xvideos\\.com/profiles/([a-zA-Z0-9_-]+)/photos/(\\d+)/([a-zA-Z0-9_-]+)$");
+        Pattern p;
-        Matcher m = p.matcher(url.toExternalForm());
+        Matcher m;
+
+        p = videoPattern;
+        m = p.matcher(url.toExternalForm());
         if (m.matches()) {
-            return getHost() + "_" + m.group(1) + "_" + m.group(3) + "_" + m.group(2);
+            return getHost() + "_" + m.group(1) + "_" + m.group(2);
-        } else {
-            return super.getAlbumTitle(url);
         }
+
+        p = albumPattern;
+        m = p.matcher(url.toExternalForm());
+        if (m.matches()) {
+            return getHost() + "_" + m.group(1) + "_" + m.group(2) + "_" + m.group(4) + "_" + m.group(3);
+        }
+
+        return super.getAlbumTitle(url);
     }
-}
+}

src/test/java/com/rarchives/ripme/tst/ripper/rippers/XvideosRipperTest.java

@@ -9,9 +9,29 @@ import org.junit.jupiter.api.Test;
 
 public class XvideosRipperTest extends RippersTest {
     @Test
-    public void testXhamsterAlbum1() throws IOException, URISyntaxException {
+    public void testXvideosVideo1() throws IOException, URISyntaxException {
-        XvideosRipper ripper = new XvideosRipper(new URI("https://www.xvideos.com/video23515878/dee_s_pool_toys").toURL());
+        // This format is obsolete
+        // XvideosRipper ripper = new XvideosRipper(new URI("https://www.xvideos.com/video23515878/dee_s_pool_toys").toURL());
+        // The website now redirects that video to this page
+        XvideosRipper ripper = new XvideosRipper(new URI("https://www.xvideos.com/video.hppdiepcbfe/dee_s_pool_toys").toURL());
         testRipper(ripper);
     }
 
+    @Test
+    public void testXvideosVideo2() throws IOException, URISyntaxException {
+        XvideosRipper ripper = new XvideosRipper(new URI("https://www.xvideos.com/video.ufkmptkc4ae/big_tit_step_sis_made_me_cum_inside_her").toURL());
+        testRipper(ripper);
+    }
+
+    @Test
+    public void testXvideosAmateursAlbum() throws IOException, URISyntaxException {
+        XvideosRipper ripper = new XvideosRipper(new URI("https://www.xvideos.com/amateurs/nikibeee/photos/2476083/lanikki").toURL());
+        testRipper(ripper);
+    }
+
+    @Test
+    public void testXvideosProfilesAlbum() throws IOException, URISyntaxException {
+        XvideosRipper ripper = new XvideosRipper(new URI("https://www.xvideos.com/profiles/dmthate/photos/8259625/sexy").toURL());
+        testRipper(ripper);
+    }
 }