https://github.com/nsk4762jp/OxygenNotIncluded-Japanese のpoファイルのタグに不整合が無いかチェックするスクリプト

check_po_tags.sh

#!/bin/bash
set -euo pipefail

if [ "$#" -ne 2 ]; then
    echo "Usage: $0 <poファイル> <出力JSONファイル>"
    exit 1
fi

input_file="$1"
output_file="$2"

# エラー情報とカウンターを直接awkスクリプト内で処理
awk -v output_file="$output_file" '
BEGIN {
    total = 0
    errors = 0
    printf "{\n  \"summary\": {\n    \"total\": 0,\n    \"errors\": 0\n  },\n  \"error_details\": [\n" > output_file
    current_msgctxt = ""
    current_msgid = ""
    current_msgstr = ""
    in_msgid = 0
    in_msgstr = 0
    first_error = 1
}

function check_tags(msgid, msgstr, msgctxt,    msgid_tags, msgstr_tags, msgid_count, msgstr_count) {
    # タグを抽出して配列に格納
    msgid_count = 0
    msgstr_count = 0
    split("", msgid_tags)
    split("", msgstr_tags)

    while (match(msgid, /<[^>]*>/)) {
        msgid_tags[++msgid_count] = substr(msgid, RSTART, RLENGTH)
        msgid = substr(msgid, RSTART + RLENGTH)
    }

    while (match(msgstr, /<[^>]*>/)) {
        msgstr_tags[++msgstr_count] = substr(msgstr, RSTART, RLENGTH)
        msgstr = substr(msgstr, RSTART + RLENGTH)
    }

    # タグ数のチェック
    if (msgid_count != msgstr_count) {
        if (!first_error) printf "," > output_file
        printf "{\n" > output_file
        printf "  \"type\": \"tag_count_mismatch\",\n" > output_file
        printf "  \"context\": %s,\n", quote(msgctxt) > output_file
        printf "  \"msgid\": %s,\n", quote(current_msgid) > output_file
        printf "  \"msgstr\": %s,\n", quote(current_msgstr) > output_file
        printf "  \"msgid_tag_count\": %d,\n", msgid_count > output_file
        printf "  \"msgstr_tag_count\": %d\n", msgstr_count > output_file
        printf "}" > output_file
        first_error = 0
        errors++
        return 0
    }

    # タグの整合性チェック
    if (msgid_count > 0) {
        for (i = 1; i <= msgid_count; i++) {
            found = 0
            for (j = 1; j <= msgstr_count; j++) {
                if (msgid_tags[i] == msgstr_tags[j]) {
                    found = 1
                    break
                }
            }
            if (!found) {
                if (!first_error) printf "," > output_file
                printf "{\n" > output_file
                printf "  \"type\": \"tag_elements_mismatch\",\n" > output_file
                printf "  \"context\": %s,\n", quote(msgctxt) > output_file
                printf "  \"msgid\": %s,\n", quote(current_msgid) > output_file
                printf "  \"msgstr\": %s,\n", quote(current_msgstr) > output_file
                printf "  \"msgid_elements\": %s,\n", quote(join(msgid_tags, msgid_count)) > output_file
                printf "  \"msgstr_elements\": %s\n", quote(join(msgstr_tags, msgstr_count)) > output_file
                printf "}" > output_file
                first_error = 0
                errors++
                return 0
            }
        }
    }
    return 1
}

function quote(str) {
    gsub(/\\/, "\\\\", str)
    gsub(/"/, "\\\"", str)
    gsub(/\n/, "\\n", str)
    return "\"" str "\""
}

function join(arr, count,    result, i) {
    result = ""
    for (i = 1; i <= count; i++) {
        if (i > 1) result = result "\\n"
        result = result arr[i]
    }
    return result
}

/^msgctxt/ {
    if (current_msgid != "" && current_msgstr != "") {
        total++
        check_tags(current_msgid, current_msgstr, current_msgctxt)
    }
    current_msgctxt = substr($0, 10, length($0) - 10)
    gsub(/"/, "", current_msgctxt)
    current_msgid = ""
    current_msgstr = ""
    in_msgid = 0
    in_msgstr = 0
    next
}

/^msgid/ {
    if ($0 == "msgid \"\"") {
        in_msgid = 1
        current_msgid = ""
    } else {
        current_msgid = substr($0, 8, length($0) - 8)
        gsub(/"/, "", current_msgid)
        in_msgid = 0
    }
    in_msgstr = 0
    next
}

/^msgstr/ {
    if ($0 == "msgstr \"\"") {
        in_msgstr = 1
        current_msgstr = ""
    } else {
        current_msgstr = substr($0, 9, length($0) - 9)
        gsub(/"/, "", current_msgstr)
        in_msgstr = 0
    }
    in_msgid = 0
    next
}

/^"/ {
    line = substr($0, 2, length($0) - 2)
    if (in_msgid) {
        current_msgid = current_msgid line
    } else if (in_msgstr) {
        current_msgstr = current_msgstr line
    }
    next
}

END {
    if (current_msgid != "" && current_msgstr != "") {
        total++
        check_tags(current_msgid, current_msgstr, current_msgctxt)
    }
    printf "\n  ]\n}" > output_file

    # 一時ファイルを使用してsummaryを更新
    system(sprintf("mv %s %s.tmp", output_file, output_file))
    system(sprintf("sed \"s/\\\"total\\\": 0/\\\"total\\\": %d/\" %s.tmp > %s", total, output_file, output_file))
    system(sprintf("sed -i \"\" \"s/\\\"errors\\\": 0/\\\"errors\\\": %d/\" %s", errors, output_file))
    system(sprintf("rm %s.tmp", output_file))

    # 結果を表示
    printf "チェックが完了しました。結果は %s に出力されました。\n", output_file > "/dev/stderr"
    printf "合計: %d エントリー\n", total > "/dev/stderr"
    printf "エラー: %d エントリー\n", errors > "/dev/stderr"
    exit (errors > 0 ? 1 : 0)
}' "$input_file"