// Called by RegistryPlugin iff any config below RegistryPlugin.PREFIX
// has changed
protected void setConfig(
Configuration config, Configuration prevConfig, Configuration.Differences changedKeys) {
m_maxRefetchDepth =
config.getInt(
NewContentCrawler.PARAM_MAX_CRAWL_DEPTH, NewContentCrawler.DEFAULT_MAX_CRAWL_DEPTH);
fetchRateLimiter = recomputeFetchRateLimiter(fetchRateLimiter);
enablePolls = config.getBoolean(PARAM_ENABLE_REGISTRY_POLLS, DEFAULT_ENABLE_REGISTRY_POLLS);
}
private boolean startCrawl(ArchivalUnit au, boolean force, boolean deep)
throws CrawlManagerImpl.NotEligibleException {
CrawlManagerImpl cmi = (CrawlManagerImpl) crawlMgr;
if (force) {
RateLimiter limit = cmi.getNewContentRateLimiter(au);
if (!limit.isEventOk()) {
limit.unevent();
}
}
cmi.checkEligibleToQueueNewContentCrawl(au);
String delayMsg = "";
String deepMsg = "";
try {
cmi.checkEligibleForNewContentCrawl(au);
} catch (CrawlManagerImpl.NotEligibleException e) {
delayMsg = ", Start delayed due to: " + e.getMessage();
}
Configuration config = ConfigManager.getCurrentConfig();
int pri = config.getInt(PARAM_CRAWL_PRIORITY, DEFAULT_CRAWL_PRIORITY);
CrawlReq req;
try {
req = new CrawlReq(au);
req.setPriority(pri);
if (deep) {
int d = Integer.parseInt(formDepth);
if (d < 0) {
errMsg = "Illegal refetch depth: " + d;
return false;
}
req.setRefetchDepth(d);
deepMsg = "Deep (" + req.getRefetchDepth() + ") ";
}
} catch (NumberFormatException e) {
errMsg = "Illegal refetch depth: " + formDepth;
return false;
} catch (RuntimeException e) {
log.error("Couldn't create CrawlReq: " + au, e);
errMsg = "Couldn't create CrawlReq: " + e.toString();
return false;
}
cmi.startNewContentCrawl(req, null);
statusMsg = deepMsg + "Crawl requested for " + au.getName() + delayMsg;
return true;
}
public void loadAuConfigDescrs(Configuration config) throws ConfigurationException {
super.loadAuConfigDescrs(config);
this.m_registryUrl = config.get(ConfigParamDescr.BASE_URL.getKey());
// Now we can construct a valid CC permission checker.
m_permissionCheckers =
// ListUtil.list(new CreativeCommonsPermissionChecker(m_registryUrl));
ListUtil.list(new CreativeCommonsPermissionChecker());
paramMap.putLong(
KEY_AU_NEW_CONTENT_CRAWL_INTERVAL,
CurrentConfig.getTimeIntervalParam(
PARAM_REGISTRY_CRAWL_INTERVAL, DEFAULT_REGISTRY_CRAWL_INTERVAL));
if (log.isDebug2()) {
log.debug2(
"Setting Registry AU recrawl interval to "
+ StringUtil.timeIntervalToString(
paramMap.getLong(KEY_AU_NEW_CONTENT_CRAWL_INTERVAL)));
}
}
