mirror of
https://github.com/yacy/yacy_search_server.git
synced 2025-07-18 08:36:07 -04:00
do not write frame links to webgraph
This commit is contained in:
source/net/yacy/document
@ -62,7 +62,6 @@ import net.yacy.cora.util.ByteBuffer;
|
||||
import net.yacy.cora.util.ConcurrentLog;
|
||||
import net.yacy.crawler.retrieval.Request;
|
||||
import net.yacy.document.parser.html.ContentScraper;
|
||||
import net.yacy.document.parser.html.EmbedEntry;
|
||||
import net.yacy.document.parser.html.ImageEntry;
|
||||
import net.yacy.kelondro.util.FileUtils;
|
||||
|
||||
@ -914,6 +913,7 @@ dc_rights
|
||||
|
||||
public final static String CANONICAL_MARKER = "canonical";
|
||||
public final static String IFRAME_MARKER = "iframe";
|
||||
public final static String FRAME_MARKER = "frame";
|
||||
public final static String EMBED_MARKER = "embed";
|
||||
|
||||
public static Map<AnchorURL, String> getHyperlinks(final Document[] documents, boolean includeNofollow) {
|
||||
@ -935,6 +935,7 @@ dc_rights
|
||||
if (canonical != null) {
|
||||
result.put(canonical, CANONICAL_MARKER);
|
||||
}
|
||||
for (AnchorURL u: html.getFrames()) result.put(u, FRAME_MARKER);
|
||||
for (AnchorURL u: html.getIFrames()) result.put(u, IFRAME_MARKER);
|
||||
for (AnchorURL u: html.getEmbeds().keySet()) result.put(u, EMBED_MARKER);
|
||||
}
|
||||
|
@ -388,7 +388,7 @@ public class ContentScraper extends AbstractScraper implements Scraper {
|
||||
final AnchorURL src = absolutePath(tag.opts.getProperty("src", EMPTY_STRING));
|
||||
tag.opts.put("src", src.toNormalform(true));
|
||||
src.setAll(tag.opts);
|
||||
this.anchors.add(src);
|
||||
//this.anchors.add(src); // don't add the frame to the anchors because the webgraph should not contain such links (by definition)
|
||||
this.frames.add(src);
|
||||
this.evaluationScores.match(Element.framepath, src.toNormalform(true));
|
||||
} else if (tag.name.equalsIgnoreCase("body")) {
|
||||
|
Reference in New Issue
Block a user