fix: resolve url without path but searchpart

e.g. http://yacy.net?q=test was resolved as host "yacy.net?q=test" now host="yacy.net" path="/"
fixes http://mantis.tokeek.de/view.php?id=47

added test case for getHost

2 files changed, 32 insertions, 3 deletions

diff --git a/source/net/yacy/cora/document/id/MultiProtocolURL.java b/source/net/yacy/cora/document/id/MultiProtocolURL.java
index c9af0763e..1fe482bc2 100644
--- a/source/net/yacy/cora/document/id/MultiProtocolURL.java
+++ b/source/net/yacy/cora/document/id/MultiProtocolURL.java
@@ -164,7 +164,8 @@ public class MultiProtocolURL implements Serializable, Comparable<MultiProtocolU
         if (url.length() < p + 4) throw new MalformedURLException("URL not parseable: '" + url + "'");
         if (!this.protocol.equals("file") && url.substring(p + 1, p + 3).equals("//")) {
             // identify host, userInfo and file for http and ftp protocol
-            final int q = url.indexOf('/', p + 3);
+            int q = url.indexOf('/', p + 3);
+            if (q < 0) q = url.indexOf("?", p + 3); // check for www.test.com?searchpart
             int r;
             if (q < 0) {
                 if ((r = url.indexOf('@', p + 3)) < 0) {
@@ -183,11 +184,11 @@ public class MultiProtocolURL implements Serializable, Comparable<MultiProtocolU
                     this.userInfo = this.host.substring(0, r);
                     this.host = this.host.substring(r + 1).intern();
                 }
-                this.path = url.substring(q);
+                this.path = url.substring(q); // may result in "?searchpart" (resolveBackpath prepends a "/" )
             }
             if (this.host.length() < 4 && !this.protocol.equals("file")) throw new MalformedURLException("host too short: '" + this.host + "', url = " + url);
             if (this.host.indexOf('&') >= 0) throw new MalformedURLException("invalid '&' in host");
-            this.path = resolveBackpath(this.path);
+            this.path = resolveBackpath(this.path); // adds "/" if missing
             identPort(url, (isHTTP() ? 80 : (isHTTPS() ? 443 : (isFTP() ? 21 : (isSMB() ? 445 : -1)))));
             identAnchor();
             identSearchpart();
diff --git a/test/net/yacy/cora/document/id/MultiProtocolURLTest.java b/test/net/yacy/cora/document/id/MultiProtocolURLTest.java
index 046083b8d..aae13fa1a 100644
--- a/test/net/yacy/cora/document/id/MultiProtocolURLTest.java
+++ b/test/net/yacy/cora/document/id/MultiProtocolURLTest.java
@@ -113,6 +113,34 @@ public class MultiProtocolURLTest {
 
         }
     }
+
+    @Test
+    public void testGetHost() throws MalformedURLException {
+        String[][] testStrings = new String[][]{
+            // teststring , expectedresult
+            new String[]{"http://www.yacy.net", "www.yacy.net"},
+            new String[]{"http://www.yacy.net:8090", "www.yacy.net"},
+            new String[]{"http://www.yacy.net/test?query=test", "www.yacy.net"},
+            new String[]{"http://www.yacy.net/?query=test", "www.yacy.net"},
+            new String[]{"http://www.yacy.net?query=test", "www.yacy.net"},
+            new String[]{"http://www.yacy.net:?query=test", "www.yacy.net"},
+            new String[]{"//www.yacy.net:?query=test", "www.yacy.net"},
+        };
+
+        for (int i = 0; i < testStrings.length; i++) {
+            // desired conversion result
+            System.out.print("testGetHost: " + testStrings[i][0]);
+            String shouldBe = testStrings[i][1];
+
+            // conversion result
+            String resolvedHost = new MultiProtocolURL(testStrings[i][0]).getHost();
+
+            // test if equal
+            assertEquals(shouldBe, resolvedHost);
+            System.out.println(" -> " + resolvedHost);
+
+        }
+    }
 }