From 9d9ff332fbbaf5194047bbd056d624a5e999b8c7 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Mon, 20 Aug 2018 18:51:41 -0700 Subject: please support for status-code-count --- please | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) diff --git a/please b/please index 8a46978..c83f867 100755 --- a/please +++ b/please @@ -95,6 +95,27 @@ def run_rowcount(args): env=args.env) subprocess.call(cmd, shell=True) +def run_statuscodecount(args): + if args.rebuild: + rebuild_scalding() + print("Starting statuscodecount job...") + output = "{}/output-{}/{}-statuscodecount".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.HBaseStatusCodeCountJob \ + --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_statuscount(args): if args.rebuild: rebuild_scalding() @@ -219,6 +240,9 @@ def main(): sub_statuscount = subparsers.add_parser('status-count') sub_statuscount.set_defaults(func=run_statuscount) + sub_statuscodecount = subparsers.add_parser('status-code-count') + sub_statuscodecount.set_defaults(func=run_statuscodecount) + sub_matchcrossref = subparsers.add_parser('match-crossref') sub_matchcrossref.set_defaults(func=run_matchcrossref) sub_matchcrossref.add_argument('crossref_input', -- cgit v1.2.3