traite.ksh

#!/usr/bin/env ksh93

# Traité de la tabula: HTML (and maybe PDF?) generator for Markdown
# documentation in a "tabula-style".

# Include the global configuration file.
. /usr/etc/traite.conf

# Include some libraries that we will be using. 
. "${ADDERE}/posix-alt.shi"

# Cuz We're real
documents="$(realpath "$DOCUMENTS")"
www_output_directory="$(realpath "$WWW_DIRECTORY")"

# Create a temporary directory for temporary files
temporary_dir="$(mktemp -d /tmp/traite.$$.XXXXXX)"

# Cheap hack for getting a list for all the directories in the $documents
# directory, containing or not a valid tabula.
# This probably isn't bad-name safe, be aware.
docdir=( $(cd "$documents"; echo *) )

# Get the current time, for the log file.
current_time="$(date '+%Y-%m-%d %H.%M.%S')"

# This will be used when we will generate a custom HTML footer for the document
# in question.

# Except for this one, which will only be used in print_help()
program_name="$0"
traite_version='0.1-a'
ksh_version="${.sh.version}"
pandoc_version="$(pandoc -v | sed 1q)"
kernel_name="$(uname -s)"
kernel_release="$(uname -r)"
machine_host="$(uname -n)"

function main {
	# Non-UNIX(?) command-line options. Not using getopt() for now, maybe in
	# the future.
	action="$1"
	case "$action" in
		build) generate_html_documents ;;
		nuke) nuke_html_documents $2;;
		*|help) print_help ;;
	esac
}

function generate_html_documents {
	output="${OUTPUT:-index.html}"
	owd="$PWD"

	for ((i=0; i < $(n ${docdir[@]}); i++)); do
		# First of all: check if tabula.conf exist in all the
		# directories; if not, it's not a tabula.
		test -f "$documents/${docdir[$i]}/tabula.conf" \
			|| { printL 'ERROR: %s is not a tabula\n' \
		       	"${docdir[$i]}"; continue; } \
			&&  cd "$documents/${docdir[$i]}" \
		&& printL 'ACTION: Entered directory %s\n' "$documents/${docdir[$i]}"
		printL 'ACTION: Reading tabula.conf\n'
		. ./tabula.conf
		printL 'INFO:\nTitle: %s\nNumber of files: %s\nFiles: %s\n' \
			"$title" $(n ${files[@]}) "${files[@]}"

		# Determine absolute paths for files listed in tabula.conf
		for ((j=0; j < $(n ${files[@]}); j++)); do
			realfiles+=" $(realpath "${files[$j]}")"
		done
		printL 'ACTION: realpath'\''d files.\n'
		printL 'INFO:\nOriginal files: %s\nrealpath'\''d files: %s\n' \
			"${files[@]}" "${realfiles[@]}"
	
		deploy_directory="$www_output_directory/${docdir[$i]}"
		printL 'ACTION: Creating directory %s for the compiled files\n' \
		       	"$deploy_directory"
		test -d "$deploy_directory" -a -w "$deploy_directory" \
			|| { mkdir -pv "$deploy_directory" 2>&1 | tee "$LOG"; }
	
		{ echo "$USE_EXTERNAL_CSS" | grep -i '^y' 1>&2; } \
			&& pandoc_opts+="-c "$EXTERNAL_CSS""
		{ echo "$USE_FOOTER" | grep -i '^y' 1>&2; } \
			&& generate_html_footer
		
		printL 'ACTION: Compiling documentation\n'
		md2html "${realfiles[@]}" "$deploy_directory/$output" "$title" \
			"$lang" \
			&& { echo "$USE_MAIN_PAGE" | grep -i '^y' 1>&2; } \
				&& hydrate_tabulas
	
		# Go back to that same old working place...
		cd "$owd"
	done
}

function nuke_html_documents {
	# This function shall clean a specific directory that already contains
	# a built tabula --- in other words, that is already in HTML.
	printL 'NUKE: Nuking %s, which is expected to be present at %s\n' \
		"$2" "$www_output_directory/$2"
	find "$www_output_directory/$2" -type f -name '*.htm?' \
		-exec rm -iv {} \;
}

function print_help {
	printf 'usage: %s build\n       %s nuke [tabula name...]\n' \
		$program_name $program_name 1>&2
	return 1
}

# Just a boilerplate for calling Pandoc, of course.
function md2html {
	# It must have a better way of doing this.
	files=( $(echo $1) )
	output="$2"
	title="$3"
	lang="$4"
	pandoc --highlight-style="$THEME" \
		--metadata title="$title" \
			--metadata lang="$lang"\
		${files[@]} \
		--output "$output" \
		--table-of-contents \
		-s --self-contained \
		--file-scope \
		--pdf-engine="${pdfengine:-xelatex}" \
		--verbose \
		${pandoc_opts[@]} \
		2>&1 | tee "$LOG"
}

function hydrate_tabulas {
	# Create a temporary files
	tabulas_list="$(mktemp $temporary_dir/aqua.XXXXXX.html)"
	temp_main_page="$(mktemp $temporary_dir/temp.index.XXXXXX.html)"

	for ((i=0; i<$(n ${docdir[@]}); i++)); do
			# Test if there's a tabula.conf that correspond to the
			# generated tabula 
			{ test -f "$documents/${docdir[$i]}/tabula.conf"; \
		       	test -d "$www_output_directory/${docdir[$i]}"; }\
			|| { printL \
			'ERROR: %s was not generated from a tabula...
			Strange enough, it'\''s present on the directory meant for HTML files built from tabulas\n' \
		       	"${docdir[$i]}"; continue; } \
			&&  cd "$documents/${docdir[$i]}" \
		&& printL 'ACTION: Entered directory %s\n' "$documents/${docdir[$i]}"
		printL 'ACTION: Reading tabula.conf\n'
		. ./tabula.conf
cat >> "$tabulas_list" << EOF
		<dl>
			<dt>📔 <a href="./${docdir[$i]}">$title</a></dt>
			<dd>Description: $desc</dd>
			<dd>Authors: ${author[@]} </dd>
			<dd>Languages: $lang</dd>
		</dl>
EOF
	done

	printL 'ACTION: Adding tabulas to the main page.\n' 
	nawk -vaqua="$tabulas_list" \
		'/<!--TABULA LIST END-->/{flag=0} !flag;
		sub(/<!--TABULA LIST START-->/, ""){
			# Insert the HTML files containing all the tabulas
			system("cat '\''"aqua"'\''");
    			flag=1
  		}
	' < "$(realpath "$MAIN_PAGE")" > "$temp_main_page" \
	&& mv "$temp_main_page" "$www_output_directory/${MAIN_PAGE##*/}"
	printL 'INFO: Now you can publish everything.\nThe compiled tabulas with the new main page are available at %s.\n' \
		"$www_output_directory"
}

function generate_html_footer {
	# I really hope all this works (in fact, it does).
	footer="$(realpath "$FOOTER")"
	tmpfooter="$(mktemp $temporary_dir/footer.XXXXXX.html)"
 
	printL \
	'ACTION: Generating a footer for the final HTML at %s, using %s as a base\n' \
	"$tmpfooter" "$footer"

	sed -e "s@%%author%%@$author@" \
	-e "s@%%project%%@$project@" \
	-e "s@%%copyright_year%%@$copyright_year@" \
	-e "s@%%date%%@$date@" \
	-e "s@%%pandoc_version%%@$pandoc_version@" \
	-e "s@%%ksh_version%%@$ksh_version@" \
	-e "s@%%machine_host%%@$machine_host@" \
       	-e "s@%%kernel_name%%@$kernel_name@" \
	-e "s@%%kernel_release%%@$kernel_release@" < "$footer" > "$tmpfooter"

	pandoc_opts+=" -A "$tmpfooter""
}

# printL - print to log
# I did not put this into another file (library) because I don't think it's
# worth for now.

function printL { printf "$@" | tee "$LOG"; }

main $@