Adjusted hash computation and toNormalform for file:// protocol to deliver

same hash same file on Windows filesystem path with forward- and backslash in path.
Background see http://mantis.tokeek.de/view.php?id=671
+Test case

source/net/yacy/cora/document/id/DigestURL.java

@@ -244,11 +244,12 @@ public class DigestURL extends MultiProtocolURL implements Serializable {
         // find rootpath
         int rootpathStart = 0;
         int rootpathEnd = this.path.length() - 1;
-        if (!this.path.isEmpty() && this.path.charAt(0) == '/')
+        if (!this.path.isEmpty() && (this.path.charAt(0) == '/' || this.path.charAt(0) == '\\'))
             rootpathStart = 1;
         if (this.path.endsWith("/"))
             rootpathEnd = this.path.length() - 2;
         p = this.path.indexOf('/', rootpathStart);
+        if (this.isFile() && p < 0) p = this.path.indexOf('\\', rootpathStart); // double-check for windows path (if it's a file url)
         String rootpath = "";
         if (p > 0 && p < rootpathEnd) {
             rootpath = this.path.substring(rootpathStart, p);
@@ -264,7 +265,7 @@ public class DigestURL extends MultiProtocolURL implements Serializable {
         final StringBuilder hashs = new StringBuilder(12);
         assert hashs.length() == 0;
         // form the 'local' part of the hash
-        final String normalform = toNormalform(true, true);
+        final String normalform = toNormalform(true, true); // normalizes also Windows backslash in path to '/' for file url
         final String b64l = Base64Order.enhancedCoder.encode(Digest.encodeMD5Raw(normalform));
         if (b64l.length() < 5) return null;
         hashs.append(b64l.substring(0, 5)); // 5 chars

source/net/yacy/cora/document/id/MultiProtocolURL.java

@@ -832,7 +832,7 @@ public class MultiProtocolURL implements Serializable, Comparable<MultiProtocolU
     }
 
     /**
-     * Get extension out of a filename
+     * Get extension out of a filename in lowercase
      * cuts off query part
      * @param fileName
      * @return extension or ""
@@ -1064,8 +1064,14 @@ public class MultiProtocolURL implements Serializable, Comparable<MultiProtocolU
         return toNormalform(excludeAnchor, false);
     }
 
+    /**
+     * Generates a normal form of the URL.
+     * For file: url it normalizes also path delimiter to be '/' (replace possible Windows '\'
+     * @param excludeAnchor
+     * @param removeSessionID
+     * @return
+     */
     public String toNormalform(final boolean excludeAnchor, final boolean removeSessionID) {
-        // generates a normal form of the URL
         boolean defaultPort = false;
         if (this.protocol.equals("mailto")) {
             return this.protocol + ":" + this.userInfo + "@" + this.host;
@@ -1096,6 +1102,9 @@ public class MultiProtocolURL implements Serializable, Comparable<MultiProtocolU
             u.append(":");
             u.append(this.port);
         }
+        if (isFile() && urlPath.indexOf('\\') >= 0) { // normalize windows backslash (important for hash computation)
+            urlPath = urlPath.replace('\\', '/');
+        }
         u.append(urlPath);
         String result = u.toString();
         

test/java/net/yacy/cora/document/id/DigestURLTest.java

@@ -2,6 +2,7 @@ package net.yacy.cora.document.id;
 
 import java.net.MalformedURLException;
 import junit.framework.TestCase;
+import net.yacy.cora.document.encoding.ASCII;
 import org.junit.Test;
 
 public class DigestURLTest extends TestCase {
@@ -30,4 +31,23 @@ public class DigestURLTest extends TestCase {
         }
     }
 
+    /**
+     * Test hash() of DigestURL and File protocol to deliver same hash for
+     * allowed Windows or Java notation of same file
+     */
+    @Test
+    public void testHash_ForFile() throws MalformedURLException {
+        String winUrlStr = "file:///C:\\tmp\\test.html"; // allowed Windows notation
+        String javaUrlStr = "file:///C:/tmp/test.html"; // allowed Java notation for Windows file system
+
+        DigestURL winUrl = new DigestURL(winUrlStr);
+        DigestURL javaUrl = new DigestURL(javaUrlStr);
+
+        String winHashResult = ASCII.String(winUrl.hash());
+        String javaHashResult = ASCII.String(javaUrl.hash());
+        
+        assertEquals("hash for same file url", javaHashResult, winHashResult);
+
+    }
+
 }