private void initializeOnStartOrResume(String type, String crawlerId) {
if (!CrawlerEvent.CRAWLER_STARTED.equals(type) && !CrawlerEvent.CRAWLER_RESUMED.equals(type)) {
return;
}
// Create new file on crawler start/resume
outputFile =
new File(
outputDir,
fileNamePrefix
+ FileUtil.toSafeFileName(crawlerId)
+ "-"
+ System.currentTimeMillis()
+ ".tsv");
try {
FileUtil.createDirsForFile(outputFile);
} catch (IOException e) {
throw new CollectorException("Cannot create output directory for file: " + outputFile, e);
}
writeLine("Referrer", "URL", "Status", "Reason", false);
// Parse status codes
if (StringUtils.isBlank(statusCodes)) {
parsedCodes.clear();
return;
}
String[] ranges = statusCodes.split("\\s*,\\s*");
for (String range : ranges) {
String[] endPoints = range.split("\\s*-\\s*");
if (endPoints.length == 1) {
parsedCodes.add(toInt(endPoints[0]));
} else if (endPoints.length == 2) {
int start = toInt(endPoints[0]);
int end = toInt(endPoints[1]);
if (start >= end) {
throw new IllegalArgumentException(
"Invalid statusCode range: "
+ range
+ ". Start value must be higher than end value.");
}
while (start <= end) {
parsedCodes.add(start);
start++;
}
} else {
throw new IllegalArgumentException("Invalid statusCode range: " + range);
}
}
}
@Override
public ICrawlDataStore createCrawlDataStore(ICrawlerConfig config, boolean resume) {
String storeDir =
config.getWorkDir().getPath()
+ "/crawlstore/mvstore/"
+ FileUtil.toSafeFileName(config.getId())
+ "/";
return new MVStoreCrawlDataStore(storeDir, resume);
}
