#!/bin/bash

#input_dir=pd-ef-json-files
input_dir="hdfs://10.10.0.52:9000/user/htrc/pd-ef-json-files"
output_dir=pd-solr-json-files

#master_opt="--master local[4]"
master_opt="--master spark://10.10.0.52:7077"

self_contained_jar=target/htrc-ef-ingest-0.9-jar-with-dependencies.jar
base_cmd="spark-submit --class org.hathitrust.PrepareForIngest $master_opt $self_contained_jar"

if [ $# -ge 1 ] ; then
    file_listing=$1
    shift
    $base_cmd --json-filelist="$file_listing" $input_dir $output_dir $*
else
    echo "****"
    echo "* Processing all files in: $input_dir"
    echo "****"
    $base_cmd $input_dir/*.json.bz2 $output_dir $*
fi

#    spark-submit --class org.hathitrust.PrepareForIngest --master local[4] target/htrc-ef-ingest-0.9-jar-with-dependencies.jar --json-filelist=pd-file-listing-step10000.txt pd-ef-json-files pd-solr-json-files $* 

# spark-submit --class org.hathitrust.PrepareForIngest --master local[4] target\htrc-ef-ingest-0.9-jar-with-dependencies.jar --json-filelist=pd-file-listing-step1000.txt json-files solr-files $*