Examples of net.yacy.kelondro.data.meta.URIMetadataRow

net.yacy.kelondro.data.meta.URIMetadataRow

                        if (this.format == 1) pw.println("<a href=\"http://" + host + "\">" + host + "</a><br>");
                        this.count++;
                    }
                } else {
                    final Iterator<URIMetadataRow> i = entries(); // iterates indexURLEntry objects
                    URIMetadataRow entry;
                    URIMetadataRow.Components metadata;
                    String url;
                    while (i.hasNext()) {
                        entry = i.next();
                        if (this.set != null && !this.set.has(entry.hash())) continue;
                        metadata = entry.metadata();
                        url = metadata.url().toNormalform(true, false);
                        if (!url.matches(this.filter)) continue;
                        if (this.format == 0) {
                            pw.println(url);
                        }
                        if (this.format == 1) {
                            pw.println("<a href=\"" + url + "\">" + CharacterCoding.unicode2xml(metadata.dc_title(), true) + "</a><br>");
                        }
                        if (this.format == 2) {
                            pw.println("<item>");
                            pw.println("<title>" + CharacterCoding.unicode2xml(metadata.dc_title(), true) + "</title>");
                            pw.println("<link>" + MultiProtocolURI.escape(url) + "</link>");
                            if (metadata.dc_creator().length() > 0) pw.println("<author>" + CharacterCoding.unicode2xml(metadata.dc_creator(), true) + "</author>");
                            if (metadata.dc_subject().length() > 0) pw.println("<description>" + CharacterCoding.unicode2xml(metadata.dc_subject(), true) + "</description>");
                            pw.println("<pubDate>" + entry.moddate().toString() + "</pubDate>");
                            pw.println("<yacy:size>" + entry.size() + "</yacy:size>");
                            pw.println("<guid isPermaLink=\"false\">" + ASCII.String(entry.hash()) + "</guid>");
                            pw.println("</item>");
                        }
                        this.count++;
                    }
                }

View Full Code Here

        final ReferenceOrder order = new ReferenceOrder(query.ranking, UTF8.getBytes(query.targetlang));
        final RWIProcess rankedCache = new RWIProcess(query, order, SearchEvent.max_results_preparation);
        rankedCache.start();


        // search is running; retrieve results
        URIMetadataRow row;
        final ArrayList<DigestURI> files = new ArrayList<DigestURI>();
        Components metadata;
        while ((row = rankedCache.takeURL(false, 1000)) != null) {
            metadata = row.metadata();
            if (metadata == null) continue;
            files.add(metadata.url());
            count--;
            if (count == 0) break;
        }

View Full Code Here

            this.links = new ArrayList<URIMetadataRow>(this.urlcount);
            for (int n = 0; n < this.urlcount; n++) {
                // get one single search result
                final String resultLine = resultMap.get("resource" + n);
                if (resultLine == null) continue;
                final URIMetadataRow urlEntry = URIMetadataRow.importEntry(resultLine);
                if (urlEntry == null) continue;
                this.links.add(urlEntry);
            }
        }

View Full Code Here

            }
        }


        // create a new loaded URL db entry
        if (modDate.getTime() > loadDate.getTime()) modDate = loadDate;
        final URIMetadataRow newEntry = new URIMetadataRow(
                url,                                       // URL
                dc_title,                                  // document description
                document.dc_creator(),                     // author
                document.dc_subject(' '),                  // tags
                document.dc_publisher(),                   // publisher (may be important to get location data)

View Full Code Here

     */
    public int removeAllUrlReferences(final byte[] urlhash, final LoaderDispatcher loader, final CacheStrategy cacheStrategy) {


        if (urlhash == null) return 0;
        // determine the url string
        final URIMetadataRow entry = urlMetadata().load(urlhash);
        if (entry == null) return 0;
        final URIMetadataRow.Components metadata = entry.metadata();
        if (metadata == null || metadata.url() == null) return 0;


        try {
            // parse the resource
            final Document document = Document.mergeDocuments(metadata.url(), null, loader.loadDocuments(loader.request(metadata.url(), true, false), cacheStrategy, 10000, Integer.MAX_VALUE));

View Full Code Here

0 1 2 3 4 5 6 7

TOP

Related Classes of net.yacy.kelondro.data.meta.URIMetadataRow

Bookmarks

crawlReceipt

CrawlResults

de.anomic.crawler.CrawlStacker

de.anomic.crawler.ResultURLs

de.anomic.crawler.retrieval.Request

de.anomic.crawler.SitemapImporter

de.anomic.data.ymark.YMarkMetadata

de.anomic.search.DocumentIndex

de.anomic.search.DocumentIndex$Worker

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.