private void checkSubstance(ArchivalUnit au) {
SubstanceChecker subChecker = new SubstanceChecker(au);
if (!subChecker.isEnabled()) {
errMsg = "No substance patterns defined for plugin.";
return;
}
AuState auState = AuUtil.getAuState(au);
SubstanceChecker.State oldState = auState.getSubstanceState();
SubstanceChecker.State newState = subChecker.findSubstance();
String chtxt = (newState == oldState ? "(unchanged)" : "(was " + oldState.toString() + ")");
switch (newState) {
case Unknown:
log.error("Shouldn't happen: SubstanceChecker returned Unknown");
errMsg = "Error in SubstanceChecker; see log.";
break;
case Yes:
statusMsg = "AU has substance " + chtxt + ": " + au.getName();
auState.setSubstanceState(SubstanceChecker.State.Yes);
break;
case No:
statusMsg = "AU has no substance " + chtxt + ": " + au.getName();
auState.setSubstanceState(SubstanceChecker.State.No);
break;
}
}
private boolean startReindexingMetadata(ArchivalUnit au, boolean force) {
if (metadataMgr == null) {
errMsg = "Metadata processing is not enabled.";
return false;
}
if (!force) {
if (!AuUtil.hasCrawled(au)) {
errMsg = "Au has never crawled. Click again to reindex metadata";
showForceReindexMetadata = true;
return false;
}
AuState auState = AuUtil.getAuState(au);
switch (auState.getSubstanceState()) {
case No:
errMsg = "Au has no substance. Click again to reindex metadata";
showForceReindexMetadata = true;
return false;
case Unknown:
errMsg = "Unknown substance for Au. Click again to reindex metadata.";
showForceReindexMetadata = true;
return false;
case Yes:
// fall through
}
}
// Fully reindex metadata with the highest priority.
Connection conn = null;
PreparedStatement insertPendingAuBatchStatement = null;
try {
conn = dbMgr.getConnection();
insertPendingAuBatchStatement = metadataMgr.getPrioritizedInsertPendingAuBatchStatement(conn);
if (metadataMgr.enableAndAddAuToReindex(
au, conn, insertPendingAuBatchStatement, false, true)) {
statusMsg = "Reindexing metadata for " + au.getName();
return true;
}
} catch (DbException dbe) {
log.error("Cannot reindex metadata for " + au.getName(), dbe);
} finally {
DbManager.safeCloseStatement(insertPendingAuBatchStatement);
DbManager.safeRollbackAndClose(conn);
}
if (force) {
errMsg = "Still cannot reindex metadata for " + au.getName();
} else {
errMsg = "Cannot reindex metadata for " + au.getName();
}
return false;
}
private boolean disableMetadataIndexing(ArchivalUnit au, boolean force) {
if (metadataMgr == null) {
errMsg = "Metadata processing is not enabled.";
return false;
}
try {
metadataMgr.disableAuIndexing(au);
statusMsg = "Disabled metadata indexing for " + au.getName();
return true;
} catch (Exception e) {
errMsg = "Cannot reindex metadata for " + au.getName() + ": " + e.getMessage();
return false;
}
}
private void crawlPluginRegistries() {
StringBuilder sb = new StringBuilder();
for (ArchivalUnit au : pluginMgr.getAllRegistryAus()) {
sb.append(au.getName());
sb.append(": ");
try {
startCrawl(au, true, false);
sb.append("Queued.");
} catch (CrawlManagerImpl.NotEligibleException e) {
sb.append("Failed: ");
sb.append(e.getMessage());
}
sb.append("\n");
}
statusMsg = sb.toString();
}
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;
}
private void callV3ContentPoll(ArchivalUnit au) throws PollManager.NotEligibleException {
log.debug("Enqueuing a V3 Content Poll on " + au.getName());
PollSpec spec = new PollSpec(au.getAuCachedUrlSet(), Poll.V3_POLL);
pollManager.enqueueHighPriorityPoll(au, spec);
statusMsg = "Enqueued V3 poll for " + au.getName();
}