#!/bin/sh
# Copyright (c) 2014-2026 Sentieon Inc. All rights reserved
# bamslice -i input.bam --shard chr1 --shard chr2 -o output --slice_end 0/1/2/3 --slice_margin 10000
# slice_end mode:
# 0: no slice for end, produce output_$i.bam for each shard
# 1: slice for end, produce output_$i.bam, output_$i_L.bam, output_$i_R.bam for each shard,
# 2: slice for end only, produce output_$i_L.bam, ouitput_base_$i_R.bam
# 3: slice for end only, produce output_L.bam, ouitput_base_R.bam, similar to 2, but different naming
# 

usage()
{
    echo "Usage: $1 [options]"
    echo "Required options:"  
    echo "    -i, --input"
    echo "    -o, --output"
    echo "    -s, --shard"
    echo "optional options:"  
    echo "    -r, --reference"
    echo "    -e, --slice_end"
    echo "    -m, --slice_margin"
    echo "    -t, --threads"
    echo "    -c, --bam_compression"
    echo "    -d, --dry_run"
    echo "slice_end mode:"
    echo "    0: no slice for end, produce output_\$i.bam for each shard"
    echo "    1: slice for end, produce output_\$i.bam, output_\$i_L.bam, output_\$i_R.bam"
    echo "       for each shard"
    echo "    2: slice for end only, produce output_\$i_L.bam, output_\$i_R.bam"
    echo "    3: slice for end only, produce output_L.bam, output_R.bam, similar to 2,"
    echo "       but different naming"
}

#return in $slicing_shard
create_slicing_shard()
{
    shard=$1; j=$(($2+0))
    if [ $j -eq 0 ]; then
        echo $shard
        return
    fi    
    # if j != 0, creat slicing shard for left margin or right margin
    if [ $j -eq 1 ]; then
        s=${shard%%,*}
    else
        s=${shard##*,}
    fi
   
    # parse shard as contig:start-end    
    contig=${s%%:*}
    range=${s#*:}
    if [ "$range" = "$contig" ]; then
        range=
    fi
    if [ -n "$range" ]; then
        start=${range%%-*}
        end=${range#*-}
    fi
    if [ -z "$start" ]; then
        start=1
    fi
    if [ -z "$end" ]; then 
        end=2147483647
    fi
    start=$((start-1))
    end=$((end+0))
    if [ $j -eq 1 ]; then
        if [ "$contig" != "NO_COOR" ]; then
            end=$((start+slice_margin))
            shard="$contig:$((start+1))"-"$end"
        else
            shard="NO_COOR_DUMMY_LEFT"
        fi
    else
        if [ "$contig" != "NO_COOR" ]; then
            start=$((end-slice_margin))
            if [ $start -lt 0 ]; then
                start=0
            fi
            shard="$contig:$((start+1))"-"$end"
        else
            shard="NO_COOR_DUMMY_RIGHT"
        fi
    fi
    echo $shard
}

reference=
slice_end=0
slice_margin=10000
inputs=
output=
shards=
dry_run="false"
threads=
comp_level=

# parsing options, not gnu convention, need be like '-i input' or '--input input'
while [ $# -gt 0 ]; do
    opt="$1"
    shift;
    case "$opt" in 
    -r|--reference) reference="$1"; shift;; 
    -i|--input) inputs="$inputs $1"; shift;;
    -o|--output) output="$1"; shift;;
    -s|--shard) shards="$shards $1"; shift;;
    -e|--slice_end) slice_end="$1"; shift;;
    -m|--slice_margin) slice_margin="$1"; shift;;
    -t|--threads) threads="$1"; shift;;
    -c|--bam_compression) comp_level="$1"; shift;;
    -d|--dry_run) dry_run="$1"; shift;;
    -h|--help) usage $0; exit;;
    *) echo "Unknown option $opt"
       usage $0
       exit 2;;
    esac
done

if [ -z "$inputs" -o -z "$output" -o -z "$shards" ]; then
    usage $0
    exit 2
fi

slice_end=$((slice_end+0))
slice_margin=$((slice_margin+0))

concur_count=$BAMSLICE_CONCUR_COUNT
if [ -z "$concur_count" ]; then
    concur_count=6
fi
concur_count=$((concur_count+0))
nproc=0
k=-1
for shard in $shards; do
    k=$((k+1))  
    for j in $(seq 0 2); do
        j=$((j+0))
        if [ $slice_end -eq 0 -a $j -ne 0 ]; then
            continue
        fi
        if [ $slice_end -ge 2 -a $j -eq 0 ]; then
            continue
        fi
        cmd="$SENTIEON_INSTALL_DIR/bin/sentieon driver"
        if [ -n "$reference" ]; then
            cmd="$cmd -r $reference"
        fi
        if [ -n "$threads" ]; then
            cmd="$cmd -t $threads"
        fi
        for i in $inputs; do
            cmd="$cmd -i $i"
        done
        slicing_shard=$(create_slicing_shard $shard $j)
        cmd="$cmd --shard $slicing_shard"
        cmd="$cmd --algo ReadWriter"
        if [ -n "$comp_level" ]; then
            cmd="$cmd --bam_compression $comp_level"
        fi
        if [ $slice_end -eq 3 ]; then
            out_nameroot=${output%.*}
            if [ $j -eq 1 ]; then
                out="$out_nameroot""_L.bam"
            elif [ $j -eq 2 ]; then
                out="$out_nameroot""_R.bam"
            fi        
        else
            if [ $j -eq 0 ]; then
                out="$output""_$k.bam"
            elif [ $j -eq 1 ]; then      
                out="$output""_$k""_L.bam"
            elif [ $j -eq 2 ]; then
                out="$output""_$k""_R.bam"
            fi
        fi
        cmd="$cmd $out"
        if [ "$dry_run" = "true" ]; then
            echo $cmd
        else
            sh -c "$cmd" &
            nproc=$((nproc+1))
            if [ $nproc -eq $concur_count ]; then
                wait
                nproc=0
            fi
        fi
    done
done
wait
