MAKEMANUALHTML

#!/usr/bin/env bash
#
# MAKEMANUAL: MANUAL/*.txt から MANUALHTML を作成する
#
# : written by 柳 日向 (h-yanagi@usp-lab.com; 2024-02-11)
#

####################################################################
#実行用環境変数の設定
LANG=ja_JP.UTF-8

####################################################################
#ベースとなる一時ファイルの指定
tmp=$(mktemp)
exec 2> >(cat > /dev/stderr; rm "${tmp}"*)

####################################################################
#作業ディレクトリとコンテンツ区分の指定
commands=COMMANDS
dir=$(realpath .)

if hash gawk 2> /dev/null; then
	awk() { gawk "$@"; }
elif [ "$(uname)" = Darwin ]; then
	echo "GNU awk is required." >&2
	exit 1
fi

if ! hash tidy; then
	echo '`tidy` command is required to tidy MANUALHTML up.' >&2
	tidy() { cat "$1"; }
fi

ERROR_CHECK() {
	if awk '{sum = 0; for (i = 1; i <= NF; i++) if ($(i) > 0) sum = 1} END {exit !sum}' <<< "${PIPESATUS[@]}"; then
		echo  "An error occured." >&2
		exit 1
	fi
}

html2man() {
	TARGET=${1}

	if [ -z "${uec:-}" ]; then
		mkdir -p MANUALHTML
		OUT=MANUALHTML/${TARGET}.html
	else
		mkdir -p MANUALUEC
		SECTION=$(sed -n 3p "MANUAL/${TARGET}.txt" |
		          sed -n -e 's|^.*(\([15]\))$|\1|p');
		OUT=MANUALUEC/MAN${SECTION}_${TARGET}
	fi

	OUTDIR=${2}
	
	####################################################################
	#テンプレートから基本コンテンツを生成
	{
		if [ -z "${uec:-}" ]; then
			awk < ${dir}/MANUALHTML/COMMON/HTML/TEMPLATE.HTML '/^<!--CONTENTS_CONTENTS-->$/ {exit} 1'
		else
			echo  -n
		fi                                                                |
		sed 's/<!--CONTENTS_TITLE-->/'"${TARGET} - "'/'

		version=`
			egrep 2> /dev/null "^_version = " "${dir}/COMMANDS/${TARGET}" |
			cut -d" " -f 3-9                                              |
			sed -e 's/^.\(.*\)./\1/'                                      `

		commands_pattern=$(
			find "${dir}/COMMANDS" -type f -printf "%f\n" |
			awk '
				NR == 1 {printf $0; next}
				{printf "|" $0}'
		)

		awk -vtmp="${tmp}" '
			NR <= 2 {next}
			/^last modified:/ {
				sub(/^last modified: /, "");
					# MANUAL/* の最終更新日時を $tmp-last-modified に書き込む。
					print > (tmp "-lastmodified")
					exit
				} 1' "MANUAL/${TARGET}.txt"  |
		awk  -vuec="${uec:-}"                                         \
		     -vcommands_pattern="(${commands_pattern}|sorter)[(]1[)]" \
			 -vcommand="${TARGET}" '
			function hyperlink_category5(content) {
				if (uec) {
					format = "?POMPA=MAN5_%s"
				} else {
					format = "%s.html"
				}

				content = gensub(/フィールド形式\(5\)/, "<a href=\"" sprintf(format, "field-format") "\" class=\"command\">\\0</a>", "g", content)
				content = gensub(/マスタファイル\(5\)/, "<a href=\"" sprintf(format, "master-file") "\" class=\"command\">\\0</a>", "g", content)
				content = gensub(/ネームファイル\(5\)/, "<a href=\"" sprintf(format, "name-file") "\" class=\"command\">\\0</a>", "g", content)
				content = gensub(/ネーム形式\(5\)/, "<a href=\"" sprintf(format, "name-format") "\" class=\"command\">\\0</a>", "g", content)
				content = gensub(/タグ形式\(5\)/, "<a href=\"" sprintf(format, "tag-format") "\" class=\"command\">\\0</a>", "g", content)
				content = gensub(/num=&lt;n&gt;/, "<code>\\0</code>", "g", content)
				return gensub(/トランザクションファイル\(5\)/, "<a href=\"" sprintf(format, "transaction-file") "\" class=\"command\">\\0</a>", "g", content)
			}

			function prettyprint(content) {
				# `&lt;\/?タグ[^&lt;]*&gt;` は formhame のため。

				if (match(content, /&lt;タグ \/&gt;|&lt;タグ&gt;|&lt;\/タグ&gt;|input type="checkbox"/)) {
					return gensub(/&lt;タグ \/&gt;|&lt;タグ&gt;|&quot;.&quot;|&lt;\/タグ&gt;|(\w+ )?\w+=&quot;\w+&quot;/, "<code>\\0</code>", "g", content);
				}

				# スクリプト以外の複数行にわたるコード断片は multiline クラスを付加する。

				if (match(content, /\n\n/)) {
					return gensub(/\n*([-&;A-Za-z\\-][ -&*-~\n]*[!-&*-~])\n*/, "<code class=\"multiline\">\\1</code>", "g", content)
				}

				# コード断片に準じる用語。

				content = gensub(/（/, "(",  "g", content)
				content = gensub(/）/, ")",  "g", content)
				content = gensub(/\s*(keycut\.c|'"${commands_pattern}"'|(lock|mime|counter)-file|&lt;(file|filename|(json|lock)-file|counter-file)&gt;|%0D%0A|NFS|NF(-[0-9x]+)?|-&lt;.&gt;|\\_|+(age|ng|yarr|[a-z])|\x27.{1,2}\x27|\\n|+&lt;n&gt;h|--|--?[a-z]+( &lt;[-a-z_]+&gt;)*|(&lt;\w+&gt;)+( +&lt;\w+&gt;)*|([0-9][\/@][0-9]\x20?)+|&quot;(.|NF)&quot;|(&quot;)?[a-z]+=([\x2F0-9a-z;=@.]|&quot;|&lt;|&gt;|&amp;)*)|&lt;\w{5,}&gt;|awk[(]1\)|\x27..?\x27)\s*/, "<code>\\1</code>", "g", content)
				content = hyperlink_category5(content)
				content = gensub(/touch [(]1[)]/, "<a href=\"https://linuxjm.osdn.jp/html/gnumaniak/man1/touch.1.html\">\\0</a>", "g", content)
				content = gensub(/([^0-9])([0-9]{6,})([^0-9])/, "\\1<code>\\2</code>\\3", "g", content)
				content = gensub(/(\w)オプション/, "<code>\\1</code>オプション", "g", content)
				content = gensub(/(^|[^-A-Za-z0-9])([-])([^-A-Za-z0-9&])/, "\\1<code>\\2</code>\\3", "g", content)

				content = gensub(/<\x2Fcode><code>/, " ", "g", content)
				content = gensub(/<\x2Fcode> /, "</code>", "g", content)
				return content
			}

			FNR == 1 { sub(/\xef\xbb\xbf/, "") }

			/^Version :/ {
			  content = content "Version : '"${version}"'\n"
			  next
			}

			function print_content() {
				if (content == "") {
					return
				}

				gsub(/&/, "\\&amp;", content);
				gsub(/</, "\\&lt;", content);
				gsub(/>/, "\\&gt;", content);
				gsub(/"/, "\\&quot;", content);
				gsub(/[ \t]+$/, "", content);

				if (mode == "list" || mode == "prettyprint") {
					gsub(/\s+$/, "", content)
					printf "<pre class=\"%s\">%s</pre>\n", mode, content
				} else if (mode == "unordered_list" || mode == "ordered_list") {
					tag = mode == "unordered_list" ? "ul" : "ol"
					print "	  <" tag ">"
					split(content, items, /\s*\n\s*/)
					label = ""

					for (position = 0; position < length(items); position++) {
						if (items[position] ~ /^\s+$/) {
							continue
						}

						if (match(items[position], /^\s*([0-9]+)\.\s*(.*)\s*$/, matches)) {
							if (label !~ /^\s*$/) {
								printf "<li>%s</li>\n", prettyprint(label)
							}
							label = matches[2]
						} else {
							label = label items[position] "\n"
						}
					}

					if (label !~ /^\s*$/) {
						printf " <li>%s</li>\n", prettyprint(label)
					}

					print "</" tag ">"
				} else {
					content = gensub(/[ \t]+/, " ", "g", content)
					content = gensub(/\n+/, "", "g", content)
					content = gensub(/[(（] +([^)）]+) +[)）]/, "(\\1)", "g", content)
					content = gensub(/ {2,}/, " ", "g", content)

					if (content == "") {
						next
					}

					if (name_has_occured) {
						if (uec) {
							content = gensub(commands_pattern, "<a href=\"?POMPA=MAN1_\\1\" class=\"command\">\\0</a>", "g", content)
						} else {
							content = gensub(commands_pattern, "<a href=\"\\1.html\" class=\"command\">\\0</a>", "g", content)
						}

						if (match(content, /([-a-z0-9_]+)\(1\)|[ぁ-ん亜-熙]+\(5\)/)) {
							content = hyperlink_category5(content)
							content = gensub(/-?&lt;数値&gt;/, "<code>\\0</code>","g", content)
							content = gensub(/gunzip *[(]1[)]/, "<a href=\"https://linuxjm.osdn.jp/html/GNU_gzip/man1/zcat.1.html\">\\0</a>", "g", content)
							content = gensub(/paste *[(]1[)]/, "<a href=\"https://linuxjm.osdn.jp/html/gnumaniak/man1/paste.1.html\">\\0</a>", "g", content)
							content = gensub(/(^|\n+)([-&;A-Za-z][ !#-&*-;=?-~(]*[!#-&*-;=?-~])\n*/, "<code>\\2</code>", "g", content)
							content = gensub(/<\x2Fcode> /, "</code>", "g", content)
						} else {
							content = gensub(/\n+/, "", "g", content)
							content = prettyprint(content);
						}
					} else {
						content = gensub(/\s*\n+\s*/, "", "g", content)
						content = gensub(/\n*([-&;0-9A-Za-z][!-&*-~]*) :/, "<code>\\1</code> : ", "g", content)
					}
	
					name_has_occured = 1
					printf "<p>%s</p>\n", gensub(/\n+/, "", "g", content)
				}
	
				content = ""
				mode = ""
			}

			NR == 1 {
				print_content()
				level = uec ? 1 : 2
				printf "<h%d>%s</h%d>\n", level, $0, level
				next
			}
			$0 ~ /^Usage +:/ || $0 ~ /^  /{mode = "list"}

			command == "check_cmp_name" && $0 ~ /^    / {
				if (mode != "check_cmp_name") {
					print "<pre>"
				}

				mode = "check_cmp_name"
				print $0
				next
			}
			
			command == "formhame" && $0 ~ /^【備考】$/ {
				inside_note = 1
			}

			command == "formhame" && inside_note && $0 ~ /^name_(radio|checkbox|pulldown) [a-z0-9]+$/ {
				if (mode != "formhame") {
					print "<pre style=\"margin-left: 3.5em;\">"
				}

				mode = "formhame"
				print $0
				next
			}

			command == "getfirst" && $0 ~ /^【注意】/ {
				inside_note = 1
			}

			command == "getfirst" && inside_note && $0 ~ /^0000007/{
				if (mode != "getfirst") {
					print "<pre style=\"margin-left: 3.5em;\">"
				}

				mode = "getfirst"
				print $0
				next
			}

			command == "msort" && $0 ~ /^  msort/ { 
				if (mode != "msort") {
					print "<pre style=\"margin-left: 3.5em;\">"
				}

				mode  = "msort"
				print $0
				next
			}

			command == "ulock" && match($0, /^  (.--) (.+)$/, matches) {
				print "<p><em style=\"margin-right: 0.5em; letter-spacing: 0.5em; font-family: monospace;\">" matches[1] "</em>" matches[2]"</p>"
				next
			}

			/^([a-z]+) コマンド ＝ (.+)$/ {
				print "<p><code>" $1 "</code> コマンド: " $4 "</p>"
				next
			}

			/^-([a-z]{4}) +[-A-Za-z0-9]+ *から *[-A-Za-z0-9]+$/ {
				if (mode != "uconv") {
					mode = "uconv"
					print "<ul>"
				}

				print "<li><code>" $1 "</code>: <code>" $2 "</code> から <code>" $4 "</code></li>"
				next
			}

			substr($0, 0, 1) == "$"        ||
			$0 ~ /^   [^ ]/                ||
			substr($0, 0, 4) == "[usp"     ||
			substr($0, 0, 7) =="before1"   ||
			substr($0, 0, 8) == "yyyymmdd" {
				if (mode == "" || mode == "unordered_list" || mode == "ordered_list") {
					print_content()
				}

				mode = "prettyprint"
			}

			/^【.+】.*$/ {
				print_content()
				gsub(/&/, "\\&amp;");
				gsub(/</, "\\&lt;");
				gsub(/</, "\\&gt;");
				level = uec ? 2 : 3
				printf "<h%d>%s</h%d>\n", level, $0, level
				mode = ""
				next
			}

			match($0, /^<([亜-熙ぁ-んァ-ヶ].+)>$/, title) {
				print_content()
				level = uec ? 3 : 4
				printf "<h%d>%s</h%d>\n", level, title[1], level
				mode = ""
				next
			}
	
			/^[ \t]*1:.+$/ {
				print_content()

				print "<dl class=\"layout\">"

				for (position = 1; position <= NF; position++) {
					split($(position), label, ":")
					printf "<dt>%s</dt>\n",  label[1]
					printf "<dd>%s</dd>\n",  label[2]
				}

				print "</dl>"

				next
			}

			/^- / {
				mode = "unordered_list";
				$1 = "";
				print "<li>" hyperlink_category5($0) "</li>";
				next;
			}

			/^[ \t]+1\..+$/ {
				mode = "ordered_list"
			}

			NF == 0 && (                 \
				mode == ""               ||
				mode == "prettyprint"    ||
				mode == "unordered_list" ||
				mode == "ordered_list"   ||
				mode == "formhame"       ||
				mode == "getfirst"       ||
				mode == "msort"          ||
				mode == "uconv"          ) {

				print_content()

				if (mode == "ordered_list") {
					print "</ol>"
				}

				if (mode == "unordered_list" || mode =="uconv") {
					print "</ul>"
				}

				if (mode == "formhame" ||
				    mode == "getfirst" ||
				    mode == "msort") {
					print "</pre>"
				}

				mode = ""
				next
			}

			mode != "" {
				content = content $0 "\n"
				next
			}

			mode == "" || mode == "unordered_list" || mode == "ordered_list" {
				content = content $0 "\n"
			}

			END {print_content()}'

		if [ -z "${uec:-}" ]; then
			sed -n '/CONTENTS_CONTENTS/,$p' ${dir}/MANUALHTML/COMMON/HTML/TEMPLATE.HTML
		else
			echo -n
		fi                                                        |
		tail -n+2                                                 |
		sed 's/<!--YEAR-->/'"$(date +%Y)"'/'                      |
		sed 's/<!--LASTMODIFIED-->/'"$(
			egrep '^last modified:' "${dir}/MANUAL/${TARGET}.txt" |
			cut -d " " -f3-                                       )"'/'

		ERROR_CHECK
	}                                                                               |

	# マニュアル・メニューを作成する。
	awk -vcommands_string="$(find ${dir}/COMMANDS -not -path "${dir}/COMMANDS/.*" -type f -printf "%f " | sort -u)" '
		 /^<!-- CONTENTS_MENU -->/ {
			template = ""
	
			for (position = 4; position <= NF; position++) {
				template = template $(position) (position < NF ? " " : "")
			}

			split(commands_string, commands, " ")

			printf("              ")

			for (position = 1; position <= length(commands); position++) {
				printf(template "\n              ", commands[position], commands[position])
			}

			for (position = 0; position <= NR % 3 - 1; position++) {
				print "<li></li>\n"
			}

			next
		 }

	1' > "${OUTDIR}/${OUT}"
}

ERROR_CHECK

####################################################################
#コンテンツ出力

mkdir -p "${dir}"

export uec

if [ "$#" -eq 1 ]; then
	# 単一コマンドのマニュアルを作成する

	html2man "$1" "${dir}"
else
	# アルファベット順にコマンド一覧を列挙する

	for command in $(tr ' ' '\n' <<< 'field-format master-file name-file name-format tag-format'; find ${commands} -type f -printf "%f\n" | sort); do
		echo building a MANUAL"$([ ! -s "${uec:-}" ] && echo -n UEC || echo -n HTML)" of ${command} >&2
		html2man "${command}" "${dir}"
	done
fi