Indicate whether a redirect happened when retrieving info about the l…

…ink, replace url with final url loaded. Retrieve canonical URL if specified.

README.md

@@ -7,4 +7,8 @@ Currently this library retrieves the basic information for a link like title, de
 # Tag documentation
 
 * Twitter Cards: https://dev.twitter.com/cards/markup - Photo specific: https://dev.twitter.com/cards/types/photo
-* Facebook OpenGraph: http://ogp.me/
+* Facebook OpenGraph: http://ogp.me/
+
+# Build instructions
+
+Just run ```gradle jar```

build.gradle

@@ -1,5 +1,5 @@
 group 'com.larvalabs'
-version '0.1'
+version '0.11'
 
 apply plugin: 'java'
 

src/main/java/com/larvalabs/linkunfurl/LinkInfo.java

@@ -6,6 +6,8 @@
 public class LinkInfo {
 
     private String url;
+    private boolean redirected = false;
+    private String canonicalUrl;
     private String type;
     private String site;
     private String title;
@@ -19,6 +21,11 @@ public class LinkInfo {
     private Integer videoWidth;
     private Integer videoHeight;
 
+    /**
+     * This is the URL that the content was ultimately loaded from. If isRedirected() is true then some number
+     * of redirects were necessary to get here.
+     * @return
+     */
     public String getUrl() {
         return url;
     }
@@ -27,6 +34,26 @@ public void setUrl(String url) {
         this.url = url;
     }
 
+    public boolean isRedirected() {
+        return redirected;
+    }
+
+    public void setRedirected(boolean redirected) {
+        this.redirected = redirected;
+    }
+
+    /**
+     * This is the canonical URL reported by the site, usually by way of a <link rel='canonical'/> tag.
+     * @return
+     */
+    public String getCanonicalUrl() {
+        return canonicalUrl;
+    }
+
+    public void setCanonicalUrl(String canonicalUrl) {
+        this.canonicalUrl = canonicalUrl;
+    }
+
     public String getType() {
         return type;
     }

src/main/java/com/larvalabs/linkunfurl/LinkUnfurl.java

@@ -2,6 +2,7 @@
 
 import org.jsoup.Connection;
 import org.jsoup.Jsoup;
+import org.jsoup.helper.HttpConnection;
 import org.jsoup.nodes.Document;
 import org.jsoup.select.Elements;
 
@@ -64,27 +65,40 @@ public static LinkInfo unfurl(String url, int timeoutMillis) throws IOException
                 .followRedirects(true)
                 .execute();
 
+        String finalUrl = response.url().toString();
+        boolean didRedirect = !finalUrl.equals(url);
+
         String contentType = response.contentType();
         if (contentType != null && contentType.toLowerCase().contains(TYPE_IMAGE)) {
             // Don't attempt to parse, this is an image file - set as hero image
             LinkInfo info = new LinkInfo();
             info.setType(TYPE_IMAGE);
-            info.setImageUrl(url);
+            info.setImageUrl(finalUrl);
+            info.setRedirected(didRedirect);
             setImageLength(info);
             return info;
         } else if (contentType != null && contentType.toLowerCase().contains(TYPE_VIDEO)) {
             // Don't attempt to parse, this is a direct video file link
             LinkInfo info = new LinkInfo();
             info.setType(TYPE_VIDEO);
-            info.setVideoUrl(url);
+            info.setVideoUrl(finalUrl);
+            info.setRedirected(didRedirect);
             setImageLength(info);
             return info;
         }
 
         Document document = response.parse();
 
         LinkInfo info = new LinkInfo();
-        info.setUrl(url);
+        info.setUrl(finalUrl);
+        info.setRedirected(didRedirect);
+
+        {
+            Elements elements = document.select("link[rel='canonical']");
+            if (elements.size() > 0) {
+                info.setCanonicalUrl(elements.get(0).attr("href"));
+            }
+        }
 
         {
             String twitterTitle = getMetaElementIfExists(document, METANAME_TWITTER_TITLE);

src/test/java/com/larvalabs/linkunfurl/UnfurlTest.java

@@ -104,6 +104,18 @@ public void testInstagramVideo() throws Exception {
         assertEquals("video", info.getType());
     }
 
+    @Test
+    public void testShortenedLinkRedirect() throws Exception {
+        LinkInfo info = LinkUnfurl.unfurl("https://t.co/vnEZZzmsGF", 30000);
+        assertEquals("http://us.battle.net/sc2/en/blog/20325209/blizzcon-in-game-goodies-dva-announcer-and-portrait-10-17-2016", info.getUrl());
+    }
+
+    @Test
+    public void testCanonicalUrl() throws Exception {
+        LinkInfo info = LinkUnfurl.unfurl("http://www.ibtimes.co.uk/zetta-spanish-phone-brand-under-fire-allegedly-passing-off-rebranded-xiaomi-phones-their-own-1586988", 30000);
+        assertEquals("http://www.ibtimes.co.uk/zetta-spanish-phone-brand-under-fire-allegedly-passing-off-rebranded-xiaomi-phones-their-own-1586988", info.getCanonicalUrl());
+    }
+
     @Test
     public void testDirectVideoUrl() throws Exception {
         String url = "http://www.html5videoplayer.net/videos/toystory.mp4";