sb.crawlQueues.noticeURL.clear(NoticedURL.StackType.CORE);
try { sb.cleanProfiles(); } catch (final InterruptedException e) {/* ignore this */}
} else if (option > INVALID) {
try {
// compiling the regular expression
final Pattern compiledPattern = Pattern.compile(pattern);
if (option == PROFILE) {
// search and delete the crawl profile (_much_ faster, independant of queue size)
// XXX: what to do about the annoying LOST PROFILE messages in the log?
CrawlProfile entry;
for (final byte[] handle: sb.crawler.getActive()) {
entry = sb.crawler.getActive(handle);
final String name = entry.name();
if (name.equals(CrawlSwitchboard.CRAWL_PROFILE_PROXY) ||
name.equals(CrawlSwitchboard.CRAWL_PROFILE_REMOTE) ||
name.equals(CrawlSwitchboard.CRAWL_PROFILE_SNIPPET_LOCAL_TEXT) ||
name.equals(CrawlSwitchboard.CRAWL_PROFILE_SNIPPET_GLOBAL_TEXT) ||
name.equals(CrawlSwitchboard.CRAWL_PROFILE_SNIPPET_LOCAL_MEDIA) ||
name.equals(CrawlSwitchboard.CRAWL_PROFILE_SNIPPET_GLOBAL_MEDIA) ||
name.equals(CrawlSwitchboard.CRAWL_PROFILE_SURROGATE))
continue;
if (compiledPattern.matcher(name).find()) sb.crawler.removeActive(entry.handle().getBytes());
}
} else {
// iterating through the list of URLs
final Iterator<Request> iter = sb.crawlQueues.noticeURL.iterator(NoticedURL.StackType.CORE);
Request entry;
final List<byte[]> removehashes = new ArrayList<byte[]>();
while (iter.hasNext()) {
if ((entry = iter.next()) == null) continue;
String value = null;
location: switch (option) {
case URL: value = (entry.url() == null) ? null : entry.url().toString(); break location;
case ANCHOR: value = entry.name(); break location;
case DEPTH: value = Integer.toString(entry.depth()); break location;
case INITIATOR:
value = (entry.initiator() == null || entry.initiator().length == 0) ? "proxy" : ASCII.String(entry.initiator());
break location;
case MODIFIED: value = daydate(entry.appdate()); break location;
default: value = null; break location;
}
if (value != null && compiledPattern.matcher(value).matches()) removehashes.add(entry.url().hash());
}
Log.logInfo("IndexCreateWWWLocalQueue", "created a remove list with " + removehashes.size() + " entries for pattern '" + pattern + "'");
for (final byte[] b: removehashes) {
sb.crawlQueues.noticeURL.removeByURLHash(b);
}