seqlen

#!/usr/bin/env bash
set -e
helpstr="SEQLEN - get length of specified sequence from FASTA file, or of all
sequences combined
Usage: headgrep FILE [SEQ_NUM]

Options:
 -h: Print this help message and exit."

unset file
while [[ $OPTIND -le $# ]]; do
    while getopts "h" opt; do
        case "$opt" in
            h)
                echo "$helpstr"
                exit
                ;;
            *)
                echo "Invalid argument:" $opt >&2
                echo "$helpstr"
                exit 1
        esac
    done
    if [[ -z "$file" ]]; then
        file=${@:$OPTIND:1}
        OPTIND=$(expr $OPTIND + 1)
    elif [[ -z "$seqnum" ]]; then
        seqnum=${@:$OPTIND:1}
        OPTIND=$(expr $OPTIND + 1)
    fi
done # TODO: Fix so doesn't cycle if add extra options
shift $((OPTIND-1))

# Argument errors and warnings:
if [[ -z $file ]]; then
    >&2 echo "Error: Input file required."
    echo "$helpstr"
    exit 1
fi
#if [[ -z $seqnum ]]; then
#    >&2 echo "Error: Sequence number required."
#    echo "$helpstr"
#    exit 1
#fi

# Script:
headchar=$(awk '/./{print;exit}' $file | cut -c1)
if [[ "$headchar" != ">" ]]; then
    >&2 echo "Error: input file must be in FASTA format."
    exit 1
fi
if [[ -n "$seqnum" ]]; then
    echo $(index -xs $seqnum $file 1 len | wc -c)
else
    # Join sequence lines together and get total length
    echo $(grep -v "$headchar" "$file" | paste -sd "" - | wc -c)
fi
exit