#!/bin/bash

HOST=<%= host %>
DEBUG=<%= debug %>
TFILE=<%= tfile %>
TOPTS="<%= topts %>"
TOPTS_ODIR=<%= output_dir %>
TMPDIR=<%= tmpdir %>
QUERY="<%= query %>"
HADOOP_HOME=<%= hadoop_home %>
HIVE_HOME=<%= hive_home %>
OVERWRITE=<%= overwrite %>
HIVE_TABLE_NAME=<%= hive_table_name %>
HDFS_TMPDIR=<%= hdfs_tmpdir %>

HDFS_PATH=hdfs://$HDFS_TMPDIR/$HIVE_TABLE_NAME/$HOST

function debug { 
    if [ "$DEBUG" == '1' ]; then
	echo $HOST $(date +'%Y-%m-%d %H:%M:%S'): $1
    fi
}

function error {
    echo $HOST $(date +'%Y-%m-%d %H:%M:%S'): ERROR - $1
    exit 1
}

function check_result {
    if [ $? != 0 ]; then
	error "$1"
    fi
}

function hdfs_run {
    debug "$2"
    $HADOOP_HOME/bin/hadoop dfs $1
    check_result "$2"
}

function remote_exec {
    debug "$2"
    ssh $HOST "$1"
    check_result "$2"
}

$HADOOP_HOME/bin/hadoop dfs -test -e $HDFS_PATH || hdfs_run "-mkdir $HDFS_PATH" "Creating $HDFS_PATH"

if [ "$OVERWRITE" == '0' ]; then
    ssh $HOST "test ! -x $TOPTS_ODIR"
    check_result "The directory $HOST:$TOPTS_ODIR exists, indicating a past failure.  Try a full reimport with the -o option."
    FILE_COUNT=$($HADOOP_HOME/bin/hadoop dfs -ls $HDFS_PATH | wc -l)
    if [ "$FILE_COUNT" != "0" ]; then
	error "$HDFS_PATH is not empty, indicating a past failure.  Try a full reimport with the -o option."
    fi
else
    remote_exec "rm -rf $TOPTS_ODIR" "removing directory $TOPTS_ODIR"
    hdfs_run "-rmr $HDFS_PATH" "Clearing directory $HDFS_PATH"
fi

remote_exec "mkdir -p $TMPDIR" "making directory $HOST:$TMPDIR if it does not exist"

debug "copying $TFILE to $HOST:$TMPDIR/transpart"
scp -q $TFILE $HOST:$TMPDIR/transpart
check_result "could not copy $TFILE to $HOST:$TMPDIR/transpart"

remote_exec "$QUERY | ruby $TMPDIR/transpart $TOPTS" "executing query: $QUERY"

remote_exec "$HADOOP_HOME/bin/hadoop dfs -moveFromLocal $TOPTS_ODIR $HDFS_PATH" "uploading files in $TOPTS_ODIR to HDFS"

remote_exec "rm -rf $TOPTS_ODIR" "removing directory $TOPTS_ODIR"

debug "finished import"