def run(self):
hdfs = luigi.contrib.hdfs.hadoopcli_clients.create_hadoopcli_client()
dates = self.version_date()
if len(dates) == 0:
raise Exception("These's no user data in[%s]" % self.input()[0].fn)
exit_code = -1
mr_path = os.path.dirname(self.bin)
copyfile("%s/mapreduce.properties.user" % mr_path, "%s/mapreduce.properties" % mr_path)
if not os.path.exists(self.local_version):
hdfs.remove(self.user_root)
hdfs.mkdir(self.user_root)
exit_code = mr_cmd(self.bin, "pr.user")
if exit_code != 0 or not check_mr_success(self.merge):
raise Exception("GetExternalUser failed")
hdfs.rename(self.merge, self.output()["user"].path)
else:
local_dates = [line.strip() for line in open(self.local_version)]
latest_dates = set(dates) - set(local_dates)
if len(latest_dates) == 0:
raise Exception("These's no new arrival user data")
hdfs.remove(self.user_root)
hdfs.mkdir(self.user_root)
hdfs.mkdir(self.increamental_archive)
for d in latest_dates:
hdfs.copy("%s/%s" % (self.input()[0].path, d), "%s/%s" % (self.increamental_archive, d))
exit_code = mr_cmd(self.bin, "pr.latest.user")
if exit_code != 0 or not check_mr_success(self.output()["user"].path):
raise Exception("GetExternalUser failed")
# make version tag
with self.output()["version"].open("w") as version_fd:
dates.sort()
for d in dates:
print >> version_fd, d
def run(self):
hdfs = luigi.contrib.hdfs.hadoopcli_clients.create_hadoopcli_client()
hdfs.remove(self.paper_root)
hdfs.mkdir(self.paper_root)
mr_path = os.path.dirname(self.bin)
copyfile("%s/mapreduce.properties.doc" % mr_path, "%s/mapreduce.properties" % mr_path)
exit_code = mr_cmd(self.bin, "pr.paper")
if exit_code != 0 or not check_mr_success(self.output().path):
raise Exception("GetExternalPaper failed")
def to_hbase(data_path, jarbin):
mr_conf_fn = '%s/DataCreate.xml' % os.path.dirname(jarbin)
with open(mr_conf_fn, 'w') as conf_fd:
conf_str = '''<?xml version="1.0" encoding="UTF-8"?>
<jobs>
<hbase name="recommendation 2 hbase" hbasename="user_recommendation_info" regions="10" input="%s" method="create_hfile"/>
</jobs>'''
conf_str = conf_str % (data_path)
print >> conf_fd, conf_str
exit_code = mr_cmd(jarbin, "DataCreateBehavior")
if exit_code != 0:
raise Exception('failed to write recommendation to hbase')
