/**
* Changes existing entry in a blacklist.
*
* @param blacklistToUse The blacklist which contains the entry.
* @param supportedBlacklistTypes Types of blacklists which the entry is to changed in.
* @param oldEntry Entry to be changed.
* @param newEntry Changed entry.
* @return The length of the new entry.
*/
private static int alterEntries(
final String blacklistToUse,
final String[] supportedBlacklistTypes,
final String[] oldEntry,
final String[] newEntry) {
removeEntries(blacklistToUse, supportedBlacklistTypes, oldEntry);
PrintWriter pw = null;
try {
pw = new PrintWriter(new FileWriter(new File(ListManager.listsPath, blacklistToUse), true));
String host, path;
for (final String n : newEntry) {
int pos = n.indexOf('/');
if (pos < 0) {
host = n;
path = ".*";
} else {
host = n.substring(0, pos);
path = n.substring(pos + 1);
}
pw.println(host + "/" + path);
for (final String s : supportedBlacklistTypes) {
if (ListManager.listSetContains(s + ".BlackLists", blacklistToUse)) {
Switchboard.urlBlacklist.add(s, host, path);
}
}
SearchEventCache.cleanupEvents(true);
}
pw.close();
} catch (final IOException e) {
Log.logSevere("BLACKLIST-CLEANER", "error on writing altered entries to blacklist", e);
}
return newEntry.length;
}
/**
* Removes existing entries from a blacklist.
*
* @param blacklistToUse The blacklist which contains the
* @param supportedBlacklistTypes Types of blacklists which the entry is to changed in.
* @param entries Array of entries to be deleted.
* @return Length of the list of entries to be removed.
*/
private static int removeEntries(
final String blacklistToUse, final String[] supportedBlacklistTypes, final String[] entries) {
// load blacklist data from file
final List<String> list =
FileUtils.getListArray(new File(ListManager.listsPath, blacklistToUse));
boolean listChanged = false;
// delete the old entry from file
for (final String entry : entries) {
String s = entry;
if (list != null) {
// get rid of escape characters which make it impossible to
// properly use contains()
if (s.contains("\\\\")) {
s = s.replaceAll(Pattern.quote("\\\\"), Matcher.quoteReplacement("\\"));
}
if (list.contains(s)) {
listChanged = list.remove(s);
}
}
// remove the entry from the running blacklist engine
for (final String supportedBlacklistType : supportedBlacklistTypes) {
if (ListManager.listSetContains(supportedBlacklistType + ".BlackLists", blacklistToUse)) {
final String host = (s.indexOf('/') == -1) ? s : s.substring(0, s.indexOf('/'));
final String path = (s.indexOf('/') == -1) ? ".*" : s.substring(s.indexOf('/') + 1);
try {
Switchboard.urlBlacklist.remove(supportedBlacklistType, host, path);
} catch (final RuntimeException e) {
Log.logSevere("BLACKLIST-CLEANER", e.getMessage() + ": " + host + "/" + path);
}
}
}
SearchEventCache.cleanupEvents(true);
}
if (listChanged) {
FileUtils.writeList(
new File(ListManager.listsPath, blacklistToUse), list.toArray(new String[list.size()]));
}
return entries.length;
}
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);
}
