#!/bin/bash

# Exit on error.
set -e

# TODO

function log { echo " $(date $DATE)" "$@" ; }

TMP=$$
#DATE="+%Y-%m-%dT%H:%M:%S%:z"
DATE="+%H:%M:%S"
CP="$(make_classpath $TDBROOT)"
USAGE="Usage: $(basename $0) --loc location datafile ..."
PKG=com.hp.hpl.jena.tdb.store.bulkloader2

if [ "$#" -lt 2 ] ; then echo "$USAGE" 1>&2 ; exit 1 ; fi

## Process --loc. Yuk.
ARG1="$1"
shift
if [ "$ARG1" = "-loc" -o  "$ARG1" = "--loc" ]
then
    LOC="$1"
    shift
else 
    LOC="${ARG1/-*loc=/}"
    if [ "$ARG1" = "$LOC" ] ; then echo $USAGE 1>&2 ; exit 1 ; fi
fi

if [ ! -e "$LOC" ] ; then mkdir "$LOC" ; fi
if [ ! -d "$LOC" ] ; then echo "Not a directory: $LOC" ; exit 1 ; fi

FILES="$@"
## Stdin?

# ---- Start
log "-- TDB Bulk Loader Start"
TIME1="$(date +%s)"

# ---- Data loading phase
log "Data phase"
# Produce nodes file and triples/quads text file.

DATA_TRIPLES="$LOC/data-triples.$TMP"
DATA_QUADS="$LOC/data-quads.$TMP"

java -cp "$CP" -server "$PKG".CmdNodeTableBuilder \
    "--loc=$LOC" "--triples=$DATA_TRIPLES" "--quads=$DATA_QUADS" $FILES

# ---- Index intermediates
## All files are writtern S P O but in different sort orders.
log "Index phase"
# Sort to SPO etc
## I1="SPO POS OSP"
## I2="GSPO GPOS GOSP SPOG POSG OSPG"
## I3="node2id prefixIdx prefix2id"

function process_rows
{
    local KEYS="$1"
    local DATA="$2"
    local IDX=$3
    local WORK="$LOC/$IDX-txt"

    if [ ! -s "$DATA" ]
    then
	return
	fi

    log "Index $IDX"
    sort $KEYS < "$DATA" > $WORK
    log "Build $IDX"
    rm -f "$LOC/$IDX.dat"
    rm -f "$LOC/$IDX.idn"
    java -cp "$CP" -server "$PKG".CmdIndexBuild "$LOC" "$IDX" "$WORK"
    # Remove intermediary file.
    rm "$WORK" 
}

#Better - sort, do index, delete intermediate file, repeat

process_rows "-k1 -k2 -k3" "$DATA_TRIPLES" SPO

process_rows "-k2 -k3 -k1" "$DATA_TRIPLES" POS

process_rows "-k3 -k1 -k2" "$DATA_TRIPLES" OSP

process_rows "-k1 -k2 -k3 -k4" "$DATA_QUADS" GSPO

process_rows "-k1 -k3 -k4 -k2" "$DATA_QUADS" GPOS

process_rows "-k1 -k4 -k2 -k3" "$DATA_QUADS" GOSP

process_rows "-k2 -k3 -k4 -k1" "$DATA_QUADS" SPOG

process_rows "-k3 -k4 -k2 -k1" "$DATA_QUADS" POSG

process_rows "-k4 -k2 -k3 -k1" "$DATA_QUADS" OSPG

log "Index phase end"
TIME2="$(date +%s)"
## # ---- Index loading phase
## 
## for IDX in $I1
## do
##     log "Build index $IDX"
##     ## The B+Tree builder creates a tree from no file files.
##     rm -f "$LOC/$IDX.dat"
##     rm -f "$LOC/$IDX.idn"
##     java -cp "$CP" -server "$PKG".CmdIndexBuild "$LOC" "$IDX" "$LOC/$IDX-txt"
##     done
## 

# ---- Clean up.

rm -f "$DATA_TRIPLES" "$DATA_QUADS" 

# ---- End
log "-- TDB Bulk Loader Finish"
ELAPSED=$(($TIME2-$TIME1))
log "-- $ELAPSED seconds"