documents pushed over the api/push_p.html interface will have their

unique flag set by default

source/net/yacy/crawler/CrawlSwitchboard.java

@@ -67,7 +67,6 @@ public final class CrawlSwitchboard {
     public static final String CRAWL_PROFILE_SNIPPET_LOCAL_MEDIA = "snippetLocalMedia";
     public static final String CRAWL_PROFILE_SNIPPET_GLOBAL_MEDIA = "snippetGlobalMedia";
     public static final String CRAWL_PROFILE_SURROGATE = "surrogates";
-    public static final String CRAWL_PROFILE_PUSH_STUB = "push_";
 
     public static Set<String> DEFAULT_PROFILES = new HashSet<String>();
     static {
@@ -517,7 +516,7 @@ public final class CrawlSwitchboard {
         CrawlProfile genericPushProfile = this.defaultPushProfiles.get(collection);
         if (genericPushProfile != null) return genericPushProfile;
         genericPushProfile = new CrawlProfile(
-                CRAWL_PROFILE_PUSH_STUB + collection,
+                CrawlProfile.CRAWL_PROFILE_PUSH_STUB + collection,
                 CrawlProfile.MATCH_ALL_STRING,   //crawlerUrlMustMatch
                 CrawlProfile.MATCH_NEVER_STRING, //crawlerUrlMustNotMatch
                 CrawlProfile.MATCH_ALL_STRING,   //crawlerIpMustMatch

source/net/yacy/crawler/data/CrawlProfile.java

@@ -58,6 +58,8 @@ public class CrawlProfile extends ConcurrentHashMap<String, String> implements M
     public static final Pattern MATCH_ALL_PATTERN   = Pattern.compile(MATCH_ALL_STRING);
     public static final Pattern MATCH_NEVER_PATTERN = Pattern.compile(MATCH_NEVER_STRING);
 
+    public static final String CRAWL_PROFILE_PUSH_STUB = "push_";
+    
     // this is a simple record structure that hold all properties of a single crawl start
     private static final String HANDLE          = "handle";
     public static final String AGENT_NAME       = "agentName";
@@ -649,6 +651,10 @@ public class CrawlProfile extends ConcurrentHashMap<String, String> implements M
         return new StringBuilder(host.length() + 20).append(protocol).append("://(www.)?").append(Pattern.quote(host)).append(url.getPath()).append(".*").toString();
     }
     
+    public boolean isPushCrawlProfile() {
+        return this.name().startsWith(CrawlProfile.CRAWL_PROFILE_PUSH_STUB);
+    }
+
     public void putProfileEntry(
     		final String CRAWL_PROFILE_PREFIX,
             final serverObjects prop,

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

@@ -568,7 +568,7 @@ public class Segment {
         
         // CREATE SOLR DOCUMENT
         final CollectionConfiguration collectionConfig = this.fulltext.getDefaultConfiguration();
-        final CollectionConfiguration.SolrVector vector = collectionConfig.yacy2solr(this, collections, responseHeader, document, condenser, referrerURL, language, this.fulltext().useWebgraph() ? this.fulltext.getWebgraphConfiguration() : null, sourceName);
+        final CollectionConfiguration.SolrVector vector = collectionConfig.yacy2solr(this, collections, responseHeader, document, condenser, referrerURL, language, crawlProfile.isPushCrawlProfile(), this.fulltext().useWebgraph() ? this.fulltext.getWebgraphConfiguration() : null, sourceName);
         
         // ENRICH DOCUMENT WITH RANKING INFORMATION
         this.fulltext.getDefaultConfiguration().postprocessing_references(this.getReferenceReportCache(), vector, url, null);

source/net/yacy/search/schema/CollectionConfiguration.java

@@ -409,7 +409,7 @@ public class CollectionConfiguration extends SchemaConfiguration implements Seri
     public SolrVector yacy2solr(
             final Segment segment,
             final Map<String, Pattern> collections, final ResponseHeader responseHeader,
-            final Document document, final Condenser condenser, final DigestURL referrerURL, final String language,
+            final Document document, final Condenser condenser, final DigestURL referrerURL, final String language, final boolean setUnique,
             final WebgraphConfiguration webgraph, final String sourceName) {
         // we use the SolrCell design as index schema
         SolrVector doc = new SolrVector();
@@ -521,8 +521,6 @@ public class CollectionConfiguration extends SchemaConfiguration implements Seri
                     add(doc, CollectionSchema.date_in_content_max_dt, date_in_content_max_dt);
                 }
             }
-            
-            
         }
         if (allAttr || contains(CollectionSchema.keywords)) {
             String keywords = document.dc_subject(' ');
@@ -537,8 +535,8 @@ public class CollectionConfiguration extends SchemaConfiguration implements Seri
         }
 
         // unique-fields; these values must be corrected during postprocessing. (the following logic is !^ (not-xor) but I prefer to write it that way as it is)
-        add(doc, CollectionSchema.http_unique_b, UNIQUE_HEURISTIC_PREFER_HTTPS ? digestURL.isHTTPS() : digestURL.isHTTP()); // this must be corrected afterwards during storage!
-        add(doc, CollectionSchema.www_unique_b, host != null && (UNIQUE_HEURISTIC_PREFER_WWWPREFIX ? host.startsWith("www.") : !host.startsWith("www."))); // this must be corrected afterwards during storage!
+        add(doc, CollectionSchema.http_unique_b, setUnique || UNIQUE_HEURISTIC_PREFER_HTTPS ? digestURL.isHTTPS() : digestURL.isHTTP()); // this must be corrected afterwards during storage!
+        add(doc, CollectionSchema.www_unique_b, setUnique || host != null && (UNIQUE_HEURISTIC_PREFER_WWWPREFIX ? host.startsWith("www.") : !host.startsWith("www."))); // this must be corrected afterwards during storage!
         
         add(doc, CollectionSchema.exact_signature_l, condenser.exactSignature());
         add(doc, CollectionSchema.exact_signature_unique_b, true); // this must be corrected afterwards during storage!