words_lists_stats.sh

#!/usr/bin/env bash

CURRENT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" >/dev/null 2>&1 && pwd)"
PROJECT_DIR="$(dirname "${CURRENT_DIR}")"

ASSETS_FOLDER="${PROJECT_DIR}/assets/files"

for LANG in fr
do
    echo "${LANG}"
    for LENGTH in 4 5 6 7 8
    do
        echo "${LANG} ${LENGTH}"
        for TYPE in dictionary easy normal
        do
            FILE="${ASSETS_FOLDER}/words-${LENGTH}-${LANG}-${TYPE}.txt"

            # Get errors
            PATTERN='^[A-Z]{'"${LENGTH}"'}$'
            ERRORS="$(cat "${FILE}" | grep -v -E "${PATTERN}")"

            # Force fix/clean file
            TMP="${FILE}.tmp"
            cat "${FILE}" | tr a-z A-Z | grep -E "${PATTERN}" | sort | uniq > "${TMP}"
            mv "${TMP}" "${FILE}"

            # Count
            COUNT="$(cat "${FILE}" | wc -l | awk '{print $1}')"
            echo "${LANG} ${LENGTH} ${TYPE} => ${COUNT}"
        done
    done
done