ClickHouse/utils/check-style/check-doc-aspell

#!/usr/bin/env bash

# Perform spell checking on the docs
# Files casesensitive.txt and caseinsensitive.txt contains words to ignore (case insensitive and sensitive respectively)
# File todo.txt needs to be revised which words is actual misspellings

ROOT_PATH=$(git rev-parse --show-toplevel)

CHECK_LANG=${1:-en}

ASPELL_IGNORE_PATH="${ROOT_PATH}/utils/check-style/aspell-ignore/${CHECK_LANG}"

STATUS=0
for fname in ${ROOT_PATH}/docs/${CHECK_LANG}/**/*.md; do
    #                       vvv ---- remove anchors   ---- vvv
    errors=$(cat "$fname" | sed -E 's/(^#.*) \{#[a-z-]+\}$/\1/' \
        | aspell list --add-sgml-skip=code --encoding=utf-8 --mode=markdown -W 3 --lang=${CHECK_LANG} --home-dir=${ASPELL_IGNORE_PATH} \
        | grep -Ewv -f "${ASPELL_IGNORE_PATH}/todo.txt" \
        | grep -Ewvi -f "${ASPELL_IGNORE_PATH}/caseinsensitive.txt" \
        | grep -Ewv -f "${ASPELL_IGNORE_PATH}/casesensitive.txt" \
        | grep -Ewv "[A-Z]+" \
        | sort | uniq)
    if [ ! -z "$errors" ]; then
        STATUS=1
        echo "====== $fname ======"
        echo "$errors"
    fi
done

exit ${STATUS}