#!/bin/sh

#
# (s)tatic (b)log (g)enerator.
#
# Heavily inspired (ripped off) by Roman Zolotarev's ssg5 (https://rgz.ee/bin/ssg5)
# and rssg (https://rgz.ee/bin/rssg), as well as Luke Smith's lb and sup
# (https://github.com/LukeSmithxyz/lb).
#
# Aimed to correct 'flaws' of both sets of scripts, and made to work
# as how I wanted them both to work.
#
# This script uses lowdown (https://github.com/kristapsdz/lowdown) for
# the .md to .html translation.
#


print_help(){
	# $1: dir/file or parameter name, $2: dir/file name

	if [ "$1" = "dir" ]
	then echo "$(basename "$0"): $2: no such directory."
	elif [ "$1" = "file" ]
	then echo "$(basename "$0"): $2: no such file."
	else
		if test -n "$1"
		then echo "$(basename "$0"): missing -$1 parameter."
		fi
	fi
	echo ""
	echo "Usage: $(basename "$0") [-s] [-d] [-t] [-b] [other options]"
	echo "	-s 'src' directory."
	echo "	-d 'dst' directory."
	echo "	-t 'title' blog title."
	echo "	-b 'base url' blog base url."
	echo "	-f 'date format' date format as stated in 'find' command."
	echo "	-e 'extract' extract the header and footer from this .html file (replaces _header.html and _footer.html in src)."
	exit 1
}


readlink_f(){
	# $1: src/dst

	file="$1"
	cd "$(dirname "$file")"
	file=$(basename "$file")

	# Tests if src or dst is a symbolic link,
	# in which case resolves the symbolic link
	# and gets the actual directories.
	while test -L "$file"
	do
		file=$(readlink "$file")
		cd "$(dirname "$file")"
		file=$(basename "$file")
	done

	# Avoid all symbolic links and return the actual path.
	dir=$(pwd -P)
	echo "$dir/$file"
}


extract_hf_from_html(){
	# $1: html file name, $2: title

	# Get the header part, and replace title.
	sed "/<\/header/q" "$1" > "$src/_header.html"
	sed -i "s/<title>.*/<title>$2<\/title>/g" "$src/_header.html"

	# Not proud of this... get the footer part.
	tac  "$1" | sed "/<footer/q" | tac > "$src/_footer.html"
}


list_dirs(){
	# $1: dir path.

	cd "$1" && eval "find . -type d ! -name '.' ! -path '*/_*' $IGNORE"
}


list_files(){
	# $1: dir path.

	cd "$1" && eval "find . -type f ! -name '.' ! -path '*/_*' $IGNORE -exec stat -c '%Y 0 %n' {} \\;"
}


list_pages(){
	# $1: dir path.

	e="\\( -name '*.html' -o -name '*.md' \\)"
	cd "$1" && eval "find . -type f ! -path '*/.*' ! -path '*/_*' $IGNORE $e" |
	sed 's#^./##;s#.md$#.html#;s#/index.html$#/#'
}


list_files_add_mod(){
	# $1: dir path, $2: .files.
	# 'n' stands for new and 'o' for old.

	cd "$1" && fn=$(list_files "$1")
	fo=$(cat "$2")

	# Loop over new file list.
	while IFS= read -r fn_line
	do
		fn_n=$(echo "$fn_line" | cut -d' ' -f3)
		fn_ts=$(echo "$fn_line" | cut -d' ' -f1)

		fo_line=$(echo "$fo" | grep "$fn_n")
		if test -n "$fo_line"
		then
			# If file name exists in old list...
			fo_ts=$(echo "$fo_line" | cut -d' ' -f1)

			# Either add the new timestamp on the
			# second column (mod) or keep the old line.
			if [ $fn_ts -gt $fo_ts ]
			then echo $fo_ts $fn_ts $fn_n
			else echo $fo_line
			fi
		else
			# If not, just print the new file data.
			echo $fn_ts" 0 "$fn_n
		fi
	done <<EOF
	$fn
EOF
}


md_to_html(){
	# $1: src, $2: dst, $3: title, $4: .files

	while read -r line
	do
		# Only for .md files, duh.
		if echo $line | grep -q '\.md$'
		then
			# File name, birth ts and mod ts.
			f=$(echo "$line" | cut -d' ' -f3 | cut -d'/' -f2-)
			b=$(echo "$line" | cut -d' ' -f1)
			m=$(echo "$line" | cut -d' ' -f2)

			# Here, parse-no-autolink is used
			# so no link starting with 'www'
			# is converted to an actual a tag with an
			# actual link, ew.
			html=$(
				lowdown \
				--parse-no-autolink \
				--html-no-skiphtml \
				--html-no-escapehtml \
				--html-no-owasp \
				--html-no-head-ids < "$1/$f")

			if echo "$f" | grep -q "index"
			then html=$(add_article_list "$html" "$1" "$4")
			else html=$(add_html_ts "$html" "$b" "$m")
			fi

			add_hf_html "$html" > "$2/${f%\.md}.html"
		fi
	done <<EOF
	$4
EOF
}


add_article_list() {
	# $1: html text, $2: src, $3: .files

	# Reverse sorted file list.
	fs_reversed="$(echo "$3" | sort -r -k 1)"

	# echo the current html text and add the ul tag for the list.
	echo "$1"
	echo ""
	echo '<ul class="articles">'

	c_month_year=
	while IFS= read -r line
	do
		# Some necessary varaibles
		fn=$(echo $line | cut -d' ' -f3 | cut -d'/' -f2-)
		d=$(echo $line | cut -d' ' -f1)
		month_year=$(date --date @$d +"%B %Y")
		date=$(date --date @$d +"%b %d")

		# Only work on non index files.
		if ! echo "$fn" | grep -q 'index'
		then
			if [ "$c_month_year" = "" ]
			then
				# For the first appearing year.
				c_month_year=$month_year
				echo ""
				echo "<h3>$month_year</h3>"
				echo ""
			else
				# Each time the year changes, print it.
				if [ "$c_month_year" != "$month_year" ]
				then
					echo ""
					echo "<h3>$month_year</h3>"
					echo ""
				fi
			fi

			page_title=$(head -n 1 "$2/$fn" | cut -c 3-)
			echo "<li>$date - <a href=https://blog.luevano.xyz/${fn%\.md}>$page_title</a></li>"
		fi
	done <<EOF
	$fs_reversed
EOF
	echo "</ul>"
}


add_html_ts(){
	# $1: html text, $2: birth ts, $3: mod ts.

	echo "$1"

	if [ $m -gt $b ]
	then
		printf "\n<div class="timestamp">\n<hr>\n<p>Created: $(date --date @$b +"$date_f"); modified: $(date --date @$m +"$date_f")</p>\n</div>"
	else
		printf "\n<div class="timestamp">\n<hr>\n<p>Created: $(date --date @$b +"$date_f")</p>\n</div>"
	fi
}


add_hf_html(){
	# $1: html text.

	# lol.
	echo "$HEADER"
	echo "$1"
	echo "$FOOTER"
}


html_add_hf(){
	# $1: src, $2: dst, $3: .files.

	while IFS= read -r line
	do
		if echo "$line" | grep -q '\.html$'
		then
			f=$(echo "$line" | cut -d' ' -f3 | cut -d'/' -f2-)

			html=$(cat "$src/$f")

			add_hf_html "$html" > "$2/$f"
		fi
	done <<EOF
	$3
EOF
}


print_status() {
	# $1: singular, $2: plural, $3: list of something (urls, files, etc.)

	# Prints a simple status line.
	test -z "$3" && printf 'no %s' "$2" && return

	echo "$3" | awk -v singular="$1" -v plural="$2" '
	END {
		if (NR==1) printf NR " " singular
		if (NR>1) printf NR " " plural
	}'
}


create_sitemap() {
	# $1: .files, $2: urls, $3: base_url.

	# $1 = urls, $2 = base_url, $3 = date, $4 = .files_ts

	date=$(date +%Y-%m-%d)
	# Since fs saves lists as .md files, change .md to .html
	# to be able to parse it.
	fs="$(echo "$1" | sed 's#\.md#\.html#')"

	echo '<?xml version="1.0" encoding="UTF-8"?>'
	echo '<urlset'
	echo 'xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"'
	echo 'xsi:schemaLocation="http://www.sitemaps.org/schemas/sitemap/0.9'
	echo 'http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd"'
	echo 'xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'
	# Read each line on the urls list.
	while IFS= read -r line
	do
		# Get the corresponding .files_ts line.
		fd=$(echo "$fs" | grep "$line")

		if test -n "$fd"
		then
			# Get timestamps present in .files_ts.
			fdb=$(echo "$fd" | cut -d' ' -f1)
			fdm=$(echo "$fd" | cut -d' ' -f2)

			# echo "b: $fdb, m: $fdm"
			# If modification timestamp is greater than creation timestamp...
			if [ $fdm -gt $fdb ]
			then
				# Use modification timestamp.
				echo $line |
				sed -E 's#^(.*)$#<url><loc>'"$3"'/\1</loc><lastmod>'"$(date --date @$fdm +%Y-%m-%d)"'</lastmod><priority>1.0</priority></url>#'
			else
				# Use creation timestamp.
				echo $line |
				sed -E 's#^(.*)$#<url><loc>'"$3"'/\1</loc><lastmod>'"$(date --date @$fdb +%Y-%m-%d)"'</lastmod><priority>1.0</priority></url>#'
			fi
		else
			# Use 'current' timestamp.
			echo $line |
			sed -E 's#^(.*)$#<url><loc>'"$3"'/\1</loc><lastmod>'"$date"'</lastmod><priority>1.0</priority></url>#'
		fi
	done <<EOF
	$2
EOF
	echo '</urlset>'
}

### Parameter catching.
while getopts ":s:d:t:b:f:e:" opt
do
	case "$opt" in
		s) src="$OPTARG" ;;
		d) dst="$OPTARG" ;;
		t) title="$OPTARG" ;;
		b) base_url="$OPTARG" ;;
		f) date_f="$OPTARG" ;;
		e) extract_hf="$OPTARG" ;;
		\?) echo "$(basename "$0"): invalid option -$OPTARG."
			print_help ;;
		:) echo "$(basename "$0"): invalid option -$OPTARG: requires an argument."
			print_help ;;
	esac
done

### Handle exeptions, prepare variables and add default values.
test -n "$src" || print_help "s"
test -n "$dst" || print_help "d"
test -n "$title" || print_help "t"
test -n "$base_url" || print_help "b"
test -d "$src" || print_help "dir" "$src"
test -d "$dst" || print_help "dir" "$dst"
test -z "$extract_hf" || test -f "$extract_hf" || print_help "file" "$extract_hf"

# Only tests if there is a string in date_f,
# doesn't really check if it has the correct formatting.
if test -z "$date_f"
then date_f="%a, %b %d, %+4Y @ %R %Z"
fi

# Convert src and dst to full paths, avoiding symlinks.
src=$(readlink_f "$src")
dst=$(readlink_f "$dst")

### Actual program logic.
# General ignores for 'find'.
IGNORE=$(
	if test -f "$src/.sbgignore"
	then
		# Currently not working...
		while read -r x
		do
			test -n "$x" || continue
			printf ' ! -path "*/%s*"' "$x"
		done < "$src/.sbgignore"
	else
		# If no .sbgignore, just ignore all dotfiles.
		printf ' ! -path "*/.*"'
	fi
)


# Get the _header.html and _footer.html.
if test -n "$extract_hf"
then extract_hf_from_html "$extract_hf" "$title"
fi

h_file="$src/_header.html"
f_file="$src/_footer.html"
test -f "$h_file" && HEADER=$(cat "$h_file")

test -f "$f_file" && FOOTER=$(cat "$f_file")

# Get list of directories inside src
# and create them inside dst.
list_dirs "$src" | (cd "$src" && cpio -pdu "$dst")

# Get file list.
echo "[file list]"
fs=$(
	if test -f "$src/.files"
	then list_files_add_mod "$src" "$src/.files"
	else list_files "$src"
	fi
)
echo "$fs" | tee "$src/.files"

# If the file list contains .md files,
# parse them with lowdown.
if echo "$fs" | grep -q '\.md$'
then
	if test -x "$(which lowdown 2> /dev/null)"
	then
		md_to_html "$src" "$dst" "$title" "$fs"
	else
		echo "$(basename "$0"): couldn't find lowdown."
		exit 1
	fi
fi

# If the file list contains .html files,
# just add the header and the footer.
html_add_hf "$src" "$dst" "$fs"

printf '[sbg] ' >&2
print_status 'file, ' 'files, ' "$fs" >&2

# Create sitemap.
urls=$(list_pages "$src")
test -n "$urls" && test -n "$fs" &&
create_sitemap "$fs" "$urls" "$base_url" > "$dst/sitemap.xml"

print_status 'url' 'urls' "$urls" >&2
echo >&2