// crawlertaskdetailview
public List<CrawlerTask> getBatchTasks(int batchID) throws Throwable {
PreparedStatement statement =
conn.prepareStatement(
"SELECT * FROM `yamaloo`.`crawlertaskdetailview` WHERE `BatchID` = ?");
statement.setInt(1, batchID);
List<CrawlerTask> list = new ArrayList<CrawlerTask>();
ResultSet rs = statement.executeQuery();
while (rs.next()) {
CrawlerTask task = CrawlerTask.Parse(rs);
list.add(task);
}
statement.close();
return list;
}
// TODO: Priority field is not used yet, f**k mysql
// Order by RAND() to shuttle Urls from same site
// Otherwise, you may have a package full of Urls from the same site, which
// will slow the performance due to QPS limit
public List<CrawlerTask> getCrawlerTaskPackage(int crawlerID, int count) throws Throwable {
PreparedStatement statement =
conn.prepareStatement(
"UPDATE `yamaloo`.`crawlertask`"
+ " SET `Status` = ?,"
+ " `CrawlerID` = ?"
+ " WHERE `Status` = ?"
+ " ORDER BY Depth, RAND()"
+ " LIMIT ?");
statement.setString(1, Statuses.Crawling.toString());
statement.setInt(2, crawlerID);
statement.setString(3, Statuses.None.toString());
statement.setInt(4, count);
statement.execute();
statement.close();
// FIFO
statement =
conn.prepareStatement(
"SELECT B.SiteID, T.*"
+ " FROM (SELECT * FROM `yamaloo`.`crawlertask` WHERE `Status` = ? AND `CrawlerID` = ?) AS T"
+ " INNER JOIN `yamaloo`.`batch` AS B"
+ " ON T.BatchID = B.BatchID"
+ " ORDER BY T.Depth, CrawlerTaskID; ");
statement.setString(1, Statuses.Crawling.toString());
statement.setInt(2, crawlerID);
List<CrawlerTask> list = new ArrayList<CrawlerTask>();
ResultSet rs = statement.executeQuery();
while (rs.next()) {
CrawlerTask task = CrawlerTask.Parse(rs);
list.add(task);
}
statement.close();
return list;
}