From d4d6e4f57b1d738799d6f05caa67f90f0353d226 Mon Sep 17 00:00:00 2001
From: 4pr0n <admin@rarchives.com>
Date: Tue, 22 Dec 2015 07:47:58 -0800
Subject: [PATCH] v1.2.6: Fix 500px ripper to fetch higher-res, non-watermarked
 images.

Closes #260
---
 pom.xml                                       |  2 +-
 .../ripme/ripper/AbstractRipper.java          |  2 +-
 .../ripper/rippers/FivehundredpxRipper.java   | 53 ++++++++++++++-----
 .../com/rarchives/ripme/ui/UpdateUtils.java   |  2 +-
 4 files changed, 42 insertions(+), 17 deletions(-)

diff --git a/pom.xml b/pom.xml
index 5f3013c6..2a2f7eba 100644
--- a/pom.xml
+++ b/pom.xml
@@ -4,7 +4,7 @@
   <groupId>com.rarchives.ripme</groupId>
   <artifactId>ripme</artifactId>
   <packaging>jar</packaging>
-  <version>1.2.5</version>
+  <version>1.2.6</version>
   <name>ripme</name>
   <url>http://rip.rarchives.com</url>
   <properties>
diff --git a/src/main/java/com/rarchives/ripme/ripper/AbstractRipper.java b/src/main/java/com/rarchives/ripme/ripper/AbstractRipper.java
index 206161aa..398ee366 100644
--- a/src/main/java/com/rarchives/ripme/ripper/AbstractRipper.java
+++ b/src/main/java/com/rarchives/ripme/ripper/AbstractRipper.java
@@ -176,7 +176,7 @@ public abstract class AbstractRipper
      * @param url
      *      URL being retrieved
      */
-    public void retrievingSource(URL url) {
+    public void retrievingSource(String url) {
         RipStatusMessage msg = new RipStatusMessage(STATUS.LOADING_RESOURCE, url);
         observer.update(this,  msg);
     }
diff --git a/src/main/java/com/rarchives/ripme/ripper/rippers/FivehundredpxRipper.java b/src/main/java/com/rarchives/ripme/ripper/rippers/FivehundredpxRipper.java
index 702b919e..3da540cb 100644
--- a/src/main/java/com/rarchives/ripme/ripper/rippers/FivehundredpxRipper.java
+++ b/src/main/java/com/rarchives/ripme/ripper/rippers/FivehundredpxRipper.java
@@ -12,6 +12,8 @@ import java.util.regex.Pattern;
 
 import org.json.JSONArray;
 import org.json.JSONObject;
+import org.jsoup.nodes.Document;
+import org.jsoup.select.Elements;
 
 import com.rarchives.ripme.ripper.AbstractJSONRipper;
 import com.rarchives.ripme.ui.RipStatusMessage.STATUS;
@@ -184,21 +186,44 @@ public class FivehundredpxRipper extends AbstractJSONRipper {
         JSONArray photos = json.getJSONArray("photos");
         for (int i = 0; i < photos.length(); i++) {
             JSONObject photo = photos.getJSONObject(i);
-            String imageURL = photo.getString("image_url");
-            imageURL = imageURL.replaceAll("/4\\.", "/5.");
-            // See if there's larger images
-            for (String imageSize : new String[] { "2048" } ) {
-                String fsURL = imageURL.replaceAll("/5\\.", "/" + imageSize + ".");
-                sleep(10);
-                if (urlExists(fsURL)) {
-                    logger.info("Found larger image at " + fsURL);
-                    imageURL = fsURL;
-                    break;
-                }
+            String imageURL = null;
+            String rawUrl = "https://500px.com" + photo.getString("url");
+            Document doc;
+            Elements metas = new Elements();
+            try {
+            	logger.debug("Loading " + rawUrl);
+            	super.retrievingSource(rawUrl);
+            	doc = Http.url(rawUrl).get();
+				metas = doc.select("meta[property=\"twitter:image:src\"]");
             }
-            imageURLs.add(imageURL);
-            if (isThisATest()) {
-                break;
+            catch (IOException e) {
+            	logger.error("Error fetching full-size image from " + rawUrl, e);
+            }
+            if (metas.size() > 0) {
+            	imageURL = metas.first().attr("content");
+            }
+            else {
+				imageURL = photo.getString("image_url");
+				imageURL = imageURL.replaceAll("/4\\.", "/5.");
+				// See if there's larger images
+				for (String imageSize : new String[] { "2048" } ) {
+					String fsURL = imageURL.replaceAll("/5\\.", "/" + imageSize + ".");
+					sleep(10);
+					if (urlExists(fsURL)) {
+						logger.info("Found larger image at " + fsURL);
+						imageURL = fsURL;
+						break;
+					}
+				}
+            }
+            if (imageURL == null) {
+            	logger.error("Failed to find image for photo " + photo.toString());
+            }
+            else {
+				imageURLs.add(imageURL);
+				if (isThisATest()) {
+					break;
+				}
             }
         }
         return imageURLs;
diff --git a/src/main/java/com/rarchives/ripme/ui/UpdateUtils.java b/src/main/java/com/rarchives/ripme/ui/UpdateUtils.java
index bc6ce77c..ddc6dcec 100644
--- a/src/main/java/com/rarchives/ripme/ui/UpdateUtils.java
+++ b/src/main/java/com/rarchives/ripme/ui/UpdateUtils.java
@@ -21,7 +21,7 @@ import com.rarchives.ripme.utils.Utils;
 public class UpdateUtils {
 
     private static final Logger logger = Logger.getLogger(UpdateUtils.class);
-    private static final String DEFAULT_VERSION = "1.2.5";
+    private static final String DEFAULT_VERSION = "1.2.6";
     private static final String updateJsonURL = "http://rarchives.com/ripme.json";
     private static final String updateJarURL = "http://rarchives.com/ripme.jar";
     private static final String mainFileName = "ripme.jar";