From 0783a991102f9dce85fe85389a7f2125e290f8b0 Mon Sep 17 00:00:00 2001
From: 4pr0n <admin@rarchives.com>
Date: Mon, 26 May 2014 02:13:56 -0700
Subject: [PATCH] 1.0.47 - E-hentai: URL validation, logs more info

For troubleshooting #41
---
 pom.xml                                       |  2 +-
 .../ripme/ripper/rippers/EHentaiRipper.java   | 41 ++++++++++++-------
 .../com/rarchives/ripme/ui/UpdateUtils.java   |  2 +-
 3 files changed, 28 insertions(+), 17 deletions(-)

diff --git a/pom.xml b/pom.xml
index d3461f4b..71b35946 100644
--- a/pom.xml
+++ b/pom.xml
@@ -4,7 +4,7 @@
   <groupId>com.rarchives.ripme</groupId>
   <artifactId>ripme</artifactId>
   <packaging>jar</packaging>
-  <version>1.0.46</version>
+  <version>1.0.47</version>
   <name>ripme</name>
   <url>http://rip.rarchives.com</url>
   <properties>
diff --git a/src/main/java/com/rarchives/ripme/ripper/rippers/EHentaiRipper.java b/src/main/java/com/rarchives/ripme/ripper/rippers/EHentaiRipper.java
index 67b9bdf9..96e067b7 100644
--- a/src/main/java/com/rarchives/ripme/ripper/rippers/EHentaiRipper.java
+++ b/src/main/java/com/rarchives/ripme/ripper/rippers/EHentaiRipper.java
@@ -22,10 +22,6 @@ public class EHentaiRipper extends AlbumRipper {
 
     private Document albumDoc = null;
 
-    private URL prevUrl = null;
-    private URL cursorUrl = null;
-    private Document cursorDoc = null;
-
     public EHentaiRipper(URL url) throws IOException {
         super(url);
     }
@@ -80,20 +76,32 @@ public class EHentaiRipper extends AlbumRipper {
         if (albumDoc == null) {
             albumDoc = Jsoup.connect(this.url.toExternalForm()).get();
         }
-        if (cursorDoc == null) {
-            Elements select = albumDoc.select("#gdt > .gdtm");
-            Element first = select.first();
-            String href = first.select("a").attr("href");
-            cursorUrl = new URL(href);
+        Elements select = albumDoc.select("#gdt > .gdtm");
+        Element first = select.first();
+        String href = first.select("a").attr("href");
+        if (href.equals("")) {
+            throw new IOException("Could not find 'href' inside elements under #gdt > .gdtm > a");
         }
+        URL cursorUrl = new URL(href), prevUrl = null;
 
         while (!cursorUrl.equals(prevUrl)) {
-            cursorDoc = Jsoup.connect(this.cursorUrl.toExternalForm()).get();
+            Document cursorDoc = Jsoup.connect(cursorUrl.toExternalForm())
+                                      .userAgent(USER_AGENT)
+                                      .get();
 
             Elements a = cursorDoc.select(".sni > a");
-            Elements img = a.select("img");
+            Elements images = a.select("img");
+            if (images.size() == 0) {
+                logger.error("No images found at " + cursorUrl);
+                break;
+            }
 
-            String imgsrc = img.attr("src");
+            String imgsrc = images.get(0).attr("src");
+            if (imgsrc.equals("")) {
+                logger.warn("Image URL is empty via " + images.get(0));
+                continue;
+            }
+            logger.info("Found URL " + imgsrc + " via " + images.get(0));
             Pattern p = Pattern.compile("^http://.*/ehg/image.php.*&n=([^&]+).*$");
             Matcher m = p.matcher(imgsrc);
             if (m.matches()) {
@@ -114,10 +122,13 @@ public class EHentaiRipper extends AlbumRipper {
                 addURLToDownload(new URL(imgsrc), prefix);
             }
 
-            String href = a.attr("href");
-
             prevUrl = cursorUrl;
-            cursorUrl = new URL(href);
+            String nextUrl = a.attr("href");
+            if (nextUrl.equals("")) {
+                logger.warn("Next page URL is empty, via " + a);
+                break;
+            }
+            cursorUrl = new URL(nextUrl);
 
             index++;
         }
diff --git a/src/main/java/com/rarchives/ripme/ui/UpdateUtils.java b/src/main/java/com/rarchives/ripme/ui/UpdateUtils.java
index 3a42b458..bb988d20 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 = "1.0.46";
+    private static final String DEFAULT_VERSION = "1.0.47";
     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";