/** checks the resources and pauses crawls if necessary */
public void resourceObserverJob() {
MemoryControl.setDHTMbyte(getMinFreeMemory());
normalizedDiskFree = getNormalizedDiskFree();
normalizedMemoryFree = getNormalizedMemoryFree();
if (normalizedDiskFree.compareTo(Space.HIGH) < 0
|| normalizedMemoryFree.compareTo(Space.HIGH) < 0) {
if (normalizedDiskFree.compareTo(Space.HIGH) < 0) { // pause crawls
if (!sb.crawlJobIsPaused(SwitchboardConstants.CRAWLJOB_LOCAL_CRAWL)) {
log.logInfo("pausing local crawls");
sb.pauseCrawlJob(SwitchboardConstants.CRAWLJOB_LOCAL_CRAWL);
}
if (!sb.crawlJobIsPaused(SwitchboardConstants.CRAWLJOB_REMOTE_TRIGGERED_CRAWL)) {
log.logInfo("pausing remote triggered crawls");
sb.pauseCrawlJob(SwitchboardConstants.CRAWLJOB_REMOTE_TRIGGERED_CRAWL);
}
}
if ((normalizedDiskFree == Space.LOW || normalizedMemoryFree.compareTo(Space.HIGH) < 0)
&& sb.getConfigBool(SwitchboardConstants.INDEX_RECEIVE_ALLOW, false)) {
log.logInfo("disabling index receive");
sb.setConfig(SwitchboardConstants.INDEX_RECEIVE_ALLOW, false);
sb.peers.mySeed().setFlagAcceptRemoteIndex(false);
sb.setConfig(SwitchboardConstants.INDEX_RECEIVE_AUTODISABLED, true);
}
} else {
if (sb.getConfigBool(
SwitchboardConstants.INDEX_RECEIVE_AUTODISABLED, false)) { // we were wrong!
log.logInfo("enabling index receive");
sb.setConfig(SwitchboardConstants.INDEX_RECEIVE_ALLOW, true);
sb.peers.mySeed().setFlagAcceptRemoteIndex(true);
sb.setConfig(SwitchboardConstants.INDEX_RECEIVE_AUTODISABLED, false);
}
log.logInfo("resources ok");
}
}
private static int importFromBookmarks(
BookmarksDB db,
final DigestURI baseURL,
final InputStreamReader input,
final String tag,
final boolean importPublic) {
int importCount = 0;
Map<MultiProtocolURI, Properties> links = new HashMap<MultiProtocolURI, Properties>();
String title;
MultiProtocolURI url;
Bookmark bm;
final Set<String> tags = ListManager.string2set(tag); // this allow multiple default tags
try {
// load the links
final ContentScraper scraper = new ContentScraper(baseURL);
// OutputStream os = new htmlFilterOutputStream(null, scraper, null, false);
final Writer writer = new TransformerWriter(null, null, scraper, null, false);
FileUtils.copy(input, writer);
writer.close();
links = scraper.getAnchors();
} catch (final IOException e) {
Log.logWarning(
"BOOKMARKS", "error during load of links: " + e.getClass() + " " + e.getMessage());
}
for (final Entry<MultiProtocolURI, Properties> link : links.entrySet()) {
url = link.getKey();
title = link.getValue().getProperty("name", "");
Log.logInfo("BOOKMARKS", "links.get(url)");
if ("".equals(title)) { // cannot be displayed
title = url.toString();
}
bm = db.new Bookmark(url.toString());
bm.setProperty(Bookmark.BOOKMARK_TITLE, title);
bm.setTags(tags);
bm.setPublic(importPublic);
db.saveBookmark(bm);
importCount++;
}
return importCount;
}
public ResourceObserver(final Switchboard sb) {
this.sb = sb;
this.path = sb.getDataPath(SwitchboardConstants.INDEX_PRIMARY_PATH, "");
log.logInfo("path for disc space measurement: " + this.path);
}
public CrawlSwitchboard(final String networkName, final Log log, final File queuesRoot) {
log.logInfo("Initializing Word Index for the network '" + networkName + "'.");
if (networkName == null || networkName.length() == 0) {
log.logSevere("no network name given - shutting down");
System.exit(0);
}
this.log = log;
this.profilesActiveCrawlsCache =
Collections.synchronizedMap(new TreeMap<byte[], CrawlProfile>(Base64Order.enhancedCoder));
// make crawl profiles database and default profiles
this.queuesRoot = queuesRoot;
this.queuesRoot.mkdirs();
this.log.logConfig("Initializing Crawl Profiles");
final File profilesActiveFile = new File(queuesRoot, DBFILE_ACTIVE_CRAWL_PROFILES);
this.profilesActiveCrawls = loadFromDB(profilesActiveFile);
for (final byte[] handle : this.profilesActiveCrawls.keySet()) {
CrawlProfile p;
try {
p = new CrawlProfile(this.profilesActiveCrawls.get(handle));
} catch (final IOException e) {
p = null;
} catch (final RowSpaceExceededException e) {
p = null;
}
if (p == null) continue;
if (!RegexHelper.isValidRegex(p.get(CrawlProfile.FILTER_URL_MUSTMATCH))) {
removeActive(handle);
Log.logWarning(
"CrawlProfiles",
"removed Profile "
+ p.handle()
+ ": "
+ p.name()
+ " from active crawls since "
+ CrawlProfile.FILTER_URL_MUSTMATCH
+ " is no valid regular expression: "
+ p.get(CrawlProfile.FILTER_URL_MUSTMATCH));
} else if (!RegexHelper.isValidRegex(p.get(CrawlProfile.FILTER_URL_MUSTNOTMATCH))) {
removeActive(handle);
Log.logWarning(
"CrawlProfiles",
"removed Profile "
+ p.handle()
+ ": "
+ p.name()
+ " from active crawls since "
+ CrawlProfile.FILTER_URL_MUSTNOTMATCH
+ " is no valid regular expression: "
+ p.get(CrawlProfile.FILTER_URL_MUSTNOTMATCH));
} else {
Log.logInfo("CrawlProfiles", "loaded Profile " + p.handle() + ": " + p.name());
}
}
initActiveCrawlProfiles();
log.logInfo(
"Loaded active crawl profiles from file "
+ profilesActiveFile.getName()
+ ", "
+ this.profilesActiveCrawls.size()
+ " entries");
final File profilesPassiveFile = new File(queuesRoot, DBFILE_PASSIVE_CRAWL_PROFILES);
this.profilesPassiveCrawls = loadFromDB(profilesPassiveFile);
for (final byte[] handle : this.profilesPassiveCrawls.keySet()) {
CrawlProfile p;
try {
p = new CrawlProfile(this.profilesPassiveCrawls.get(handle));
Log.logInfo("CrawlProfiles", "loaded Profile " + p.handle() + ": " + p.name());
} catch (final IOException e) {
continue;
} catch (final RowSpaceExceededException e) {
continue;
}
}
log.logInfo(
"Loaded passive crawl profiles from file "
+ profilesPassiveFile.getName()
+ ", "
+ this.profilesPassiveCrawls.size()
+ " entries"
+ ", "
+ profilesPassiveFile.length() / 1024);
}
