void setConfig(Configuration config) {
log.debug("config: " + config);
proxyHost = config.get(PARAM_PROXY_HOST);
proxyPort = config.getInt(PARAM_PROXY_PORT, DEFAULT_PROXY_PORT);
if (StringUtil.isNullString(proxyHost) || proxyPort <= 0) {
String http_proxy = System.getenv("http_proxy");
if (!StringUtil.isNullString(http_proxy)) {
try {
HostPortParser hpp = new HostPortParser(http_proxy);
proxyHost = hpp.getHost();
proxyPort = hpp.getPort();
} catch (HostPortParser.InvalidSpec e) {
log.warning("Can't parse http_proxy environment var, ignoring: " + http_proxy + ": " + e);
}
}
}
if (StringUtil.isNullString(proxyHost) || proxyPort <= 0) {
proxyHost = null;
} else {
log.info("Proxying through " + proxyHost + ":" + proxyPort);
}
userAgent = config.get(PARAM_USER_AGENT);
if (StringUtil.isNullString(userAgent)) {
userAgent = null;
} else {
log.debug("Setting User-Agent to " + userAgent);
}
}
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;
}