createEcNumbers.sh.in

#!/bin/bash
# Parses the ec numbers and names out of 2 source files

set -o pipefail -e
exec 3>&1

classes="http://ftp.ebi.ac.uk/pub/databases/enzyme/enzclass.txt"
numbers="http://ftp.ebi.ac.uk/pub/databases/enzyme/enzyme.dat"

# TODO remove dot at end of description
lines="$(wget -q "$classes" -O -       \
       | grep "^[1-9]"                 \
       | <<<CMD_SED>>> "s/\. *\([0-9\-]\)/.\1/g" \
       | <<<CMD_SED>>> "s/  */\t/"               \
       | nl -s "	"                  \
       | <<<CMD_SED>>> 's/^  *//'                \
       | <<<CMD_SED>>> 's/\.$//'                 \
       | tee >(cat >&3)                \
       | wc -l                         \
       )"

wget -q "$numbers" -O - | <<<CMD_AWK>>> -v lines="$lines" 'BEGIN {
    FS = "   "
    lines++
} /^ID/{
    if (id != "") {
    printf("%d\t%s\t%s\n", lines++, id, name)
}
    name = ""
    id = $2
} /^DE/{
    gsub(/.$/, "", $2)
    name = name $2
} END {
    printf("%d\t%s\t%s\n", lines++, id, name)
}'