adding unit tests for more sites and also fixing the domain that wasn…

…'t being set correctly on the article object
amir343 · Jan 25, 2011 · 3617b84 · 3617b84
1 parent fde6572
commit 3617b84
Show file tree

Hide file tree

Showing 3 changed files with 35 additions and 5 deletions.
diff --git a/src/main/java/com/jimplush/goose/Article.java b/src/main/java/com/jimplush/goose/Article.java
@@ -8,11 +8,12 @@
 import org.apache.log4j.Logger;
 import org.jsoup.nodes.Element;
 
+import java.net.MalformedURLException;
+import java.net.URL;
 import java.util.ArrayList;
 
 /**
  * This class represents the extraction of an Article from a website
- *
  */
 public class Article {
 
@@ -81,6 +82,7 @@ public class Article {
 
   /**
    * returns the title of the webpage
+   *
    * @return
    */
   public String getTitle() {
@@ -91,7 +93,7 @@ public void setTitle(String title) {
     this.title = title;
   }
 
-    public String getMetaDescription() {
+  public String getMetaDescription() {
     return metaDescription;
   }
 
@@ -119,7 +121,16 @@ public String getDomain() {
     return domain;
   }
 
-  public void setDomain(String domain) {
+  public void setDomain(String urlToParse) {
+    String domain = "";
+
+    URL url = null;
+    try {
+      url = new URL(urlToParse);
+      domain = url.getHost();
+    } catch (MalformedURLException e) {
+      logger.error(e.toString(), e);
+    }
     this.domain = domain;
   }
 
@@ -140,7 +151,6 @@ public void setMovies(ArrayList<Element> movies) {
   }
 
 
-
   public ArrayList<String> getImageCandidates() {
     return imageCandidates;
   }

diff --git a/src/test/java/com/jimplush/goose/ArticleTest.java b/src/test/java/com/jimplush/goose/ArticleTest.java
@@ -15,9 +15,19 @@ public void testArticle()
   {
     Article article = new Article();
     article.setTitle("This is a title");
+    assertEquals("This is a title", article.getTitle());
+  }
 
 
-    assertEquals("This is a title", article.getTitle());
+  public void testSettingDomainOnArticle() {
+
+    Article article = new Article();
+    article.setDomain("http://grapevinyl.com/v/84/magnetic-morning/getting-nowhere");
+    assertEquals("grapevinyl.com", article.getDomain());
+
+    article.setDomain("http://www.economist.com/v/84/magnetic-morning/getting-nowhere");
+    assertEquals("www.economist.com", article.getDomain());
   }
 
+
 }
diff --git a/src/test/java/com/jimplush/goose/GoldSitesTest.java b/src/test/java/com/jimplush/goose/GoldSitesTest.java
@@ -466,5 +466,15 @@ public void testTheVacationGals() {
     assertTrue(article.getCleanedArticleText().startsWith("Editors&rsquo; Note: We are huge proponents"));
     assertTrue(article.getTopImage().getImageSrc().equals("http://thevacationgals.com/wp-content/uploads/2010/11/Gemmel-Family-Reunion-at-a-Vacation-Rental-Home1-300x225.jpg"));
   }
+
+  // test the extraction of videos from a page
+  public void testGettingVideosFromGraphVinyl() {
+    String url = "http://grapevinyl.com/v/84/magnetic-morning/getting-nowhere";
+    ContentExtractor contentExtractor = new ContentExtractor();
+    Article article = contentExtractor.extractContent(url);
+    assertTrue(article.getMovies().get(0).attr("src").equals("http://www.youtube.com/v/dsVWVtGWoa4&hl=en_US&fs=1&color1=d6d6d6&color2=ffffff&autoplay=1&iv_load_policy=3&rel=0&showinfo=0&hd=1"));
+
+  }
+
 }