diff options
Diffstat (limited to 'please')
| -rwxr-xr-x | please | 28 |
1 files changed, 27 insertions, 1 deletions
@@ -157,7 +157,9 @@ def run_statuscount(args): datetime.strftime(datetime.now(), "%Y-%m-%d-%H%M.%S")) cmd = """hadoop jar \ scalding/target/scala-2.11/sandcrawler-assembly-0.2.0-SNAPSHOT.jar \ - com.twitter.scalding.Tool sandcrawler.HBaseStatusCountJob \ + com.twitter.scalding.Tool \ + -Dmapred.task.timeout=3600000 \ + sandcrawler.HBaseStatusCountJob \ --hdfs \ --app.conf.path scalding/ia_cluster.conf \ --hbase-table wbgrp-journal-extract-0-{env} \ @@ -249,6 +251,27 @@ def run_dumpfilemeta(args): env=args.env) subprocess.call(cmd, shell=True) +def run_dumpgrobidstatuscode(args): + if args.rebuild: + rebuild_scalding() + print("Starting dumpgrobidstatuscode job...") + output = "{}/output-{}/{}-dumpgrobidstatuscode".format( + HDFS_DIR, + args.env, + datetime.strftime(datetime.now(), "%Y-%m-%d-%H%M.%S")) + cmd = """hadoop jar \ + scalding/target/scala-2.11/sandcrawler-assembly-0.2.0-SNAPSHOT.jar \ + com.twitter.scalding.Tool sandcrawler.DumpGrobidStatusCodeJob \ + --hdfs \ + --app.conf.path scalding/ia_cluster.conf \ + --hbase-table wbgrp-journal-extract-0-{env} \ + --zookeeper-hosts {zookeeper_hosts} \ + --output {output}""".format( + output=output, + zookeeper_hosts=ZOOKEEPER_HOSTS, + env=args.env) + subprocess.call(cmd, shell=True) + def run_dumpgrobidmetainsertable(args): if args.rebuild: rebuild_scalding() @@ -433,6 +456,9 @@ def main(): sub_dumpfilemeta = subparsers.add_parser('dump-file-meta') sub_dumpfilemeta.set_defaults(func=run_dumpfilemeta) + sub_dumpgrobidstatuscode = subparsers.add_parser('dump-grobid-status-code') + sub_dumpgrobidstatuscode.set_defaults(func=run_dumpgrobidstatuscode) + sub_dumpgrobidmetainsertable = subparsers.add_parser('dump-grobid-meta-insertable') sub_dumpgrobidmetainsertable.set_defaults(func=run_dumpgrobidmetainsertable) |