harmonize used lastmodified date for rwi and fulltext in storeDocument

source/net/yacy/document/Document.java

@@ -95,7 +95,7 @@ public class Document {
     private final double lon, lat;
     private final Parser parserObject; // the source object that was used to create the Document
     private final Map<String, Set<String>> generic_facets; // a map from vocabulary names to the set of tags for that vocabulary which apply for this document
-    private final Date lastModified;
+    private final Date lastModified; // creation or last modification date of the source document
     private int crawldepth;
 
     public Document(final DigestURL location, final String mimeType, final String charset,
@@ -513,6 +513,9 @@ dc_rights
         return this.emaillinks;
     }
 
+    /**
+     * @return last modification date of the source document
+     */
     public Date getLastModified() {
         return this.lastModified;
     }

source/net/yacy/search/index/Segment.java

@@ -582,9 +582,11 @@ public class Segment {
         final String urlNormalform = url.toNormalform(true);
         final String language = votedLanguage(url, urlNormalform, document, condenser); // identification of the language
 
-        // STORE URL TO LOADED-URL-DB
-        Date modDate = responseHeader == null ? new Date() : responseHeader.lastModified();
+        // get last modified date of the document to be used for the rwi index
+        // (the lastmodified document propery should be the same in rwi and fulltext (calculated in yacy2solr))
+        Date modDate = responseHeader == null ? document.getLastModified() : responseHeader.lastModified();
         if (modDate == null) modDate = new Date();
+        if (document.getLastModified().before(modDate)) modDate = document.getLastModified();
         if (modDate.getTime() > loadDate.getTime()) modDate = loadDate;
         char docType = Response.docType(document.dc_format());