From 3b51636ecbc8580b5ff8a5e3e94ac92d2baad6a9 Mon Sep 17 00:00:00 2001
From: Michael Peter Christen <mc@yacy.net>
Date: Mon, 12 Jan 2015 00:35:47 +0100
Subject: [PATCH] fix for mediawiki import

---
 htroot/IndexImportMediawiki_p.html        |  2 +-
 htroot/IndexImportMediawiki_p.java        | 23 +++++++++++++++--------
 source/net/yacy/search/Switchboard.java   |  2 +-
 source/net/yacy/search/index/Segment.java |  2 +-
 4 files changed, 18 insertions(+), 11 deletions(-)

diff --git a/htroot/IndexImportMediawiki_p.html b/htroot/IndexImportMediawiki_p.html
index f106e278d..3bf333b76 100644
--- a/htroot/IndexImportMediawiki_p.html
+++ b/htroot/IndexImportMediawiki_p.html
@@ -20,7 +20,7 @@
           <a href="http://dumps.wikimedia.org/dewiki/latest/dewiki-latest-pages-articles.xml.bz2">
           http://dumps.wikimedia.org/dewiki/latest/dewiki-latest-pages-articles.xml.bz2</a>.
           <br />
-          Dumps must be in XML format and may be compressed in gz or bz2. Place the file in the YaCy folder or in one of its sub-folders.
+          Dumps must be stored in the local file system in XML format and may be compressed in gz or bz2.
           <br />
           <input name="file" type="text" value="" size="80" />
           <input name="submit" type="submit" value="Import MediaWiki Dump" />
diff --git a/htroot/IndexImportMediawiki_p.java b/htroot/IndexImportMediawiki_p.java
index f36fd4b57..37a1c6b87 100644
--- a/htroot/IndexImportMediawiki_p.java
+++ b/htroot/IndexImportMediawiki_p.java
@@ -53,15 +53,22 @@ public class IndexImportMediawiki_p {
                 prop.put("import_status", 0);
             } else {
                 if (post.containsKey("file")) {
-                    final File sourcefile = new File(post.get("file"));
-                    if (sourcefile.exists()) {
-                        MediawikiImporter.job = new MediawikiImporter(sourcefile, sb.surrogatesInPath);
-                        MediawikiImporter.job.start();
-                        prop.put("import_dump", MediawikiImporter.job.source());
-                        prop.put("import_thread", "started");
-                    } else {
+                    String file = post.get("file");
+                    if (file.startsWith("file://")) file = file.substring(7);
+                    if (file.startsWith("http")) {
                         prop.put("import_dump", "");
-                        prop.put("import_thread", "Error: file not found ["+sourcefile+"]");
+                        prop.put("import_thread", "Error: file argument must be a path to a document in the local file system");
+                    } else {
+                        final File sourcefile = new File(file);
+                        if (sourcefile.exists()) {
+                            MediawikiImporter.job = new MediawikiImporter(sourcefile, sb.surrogatesInPath);
+                            MediawikiImporter.job.start();
+                            prop.put("import_dump", MediawikiImporter.job.source());
+                            prop.put("import_thread", "started");
+                        } else {
+                            prop.put("import_dump", "");
+                            prop.put("import_thread", "Error: file not found ["+sourcefile+"]");
+                        }
                     }
                     prop.put("import", 1);
                     prop.put("import_count", 0);
diff --git a/source/net/yacy/search/Switchboard.java b/source/net/yacy/search/Switchboard.java
index 4b9f29f33..d735dba3a 100644
--- a/source/net/yacy/search/Switchboard.java
+++ b/source/net/yacy/search/Switchboard.java
@@ -2840,7 +2840,7 @@ public final class Switchboard extends serverSwitch {
         }
 
         // remove stopwords
-        this.log.info("Excluded " + condenser.excludeWords(stopwords) + " words in URL " + url);
+        this.log.info("Excluded " + condenser.excludeWords(stopwords) + " words in URL " + url.toNormalform(true));
 
         // STORE WORD INDEX
         SolrInputDocument newEntry =
diff --git a/source/net/yacy/search/index/Segment.java b/source/net/yacy/search/index/Segment.java
index 898d1886b..295ca338d 100644
--- a/source/net/yacy/search/index/Segment.java
+++ b/source/net/yacy/search/index/Segment.java
@@ -718,7 +718,7 @@ public class Segment {
         final long indexingEndTime = System.currentTimeMillis();
 
         if (this.log.isInfo()) {
-            this.log.info("*Indexed " + condenser.words().size() + " words in URL " + url +
+            this.log.info("*Indexed " + condenser.words().size() + " words in URL " + url.toNormalform(true) +
                     " [" + id + "]" +
                     "\n\tDescription:  " + dc_title +
                     "\n\tMimeType: "  + document.dc_format() + " | Charset: " + document.getCharset() + " | " +