save_company

type -P miceweb &>/dev/null && ISMWINST=1 || ISMWINST=0
if [ $ISMWINST -eq 0 ]; then
    >&2 echo "Error: miceweb is not installed (seeing https://github.com/Robotizing/MiceWeb/)"
fi

get_pages() {
	prevfilesize=1000000000
	minfilesize="$prevfilesize"
	for p in {2..2000}
	do
		sleep 0.1
		wget -E $1/page$p
		if [ $? -ne 0 ]
		then
			break
		fi
		filesize="$(wc -c <"page$p.html")"
		# echo "$filesize $prevfilesize $minfilesize $p"
		if [ "$filesize" -eq "$prevfilesize" ]
		then
			let minfilesizeext=$minfilesize+8
			if [ "$filesize" -le "$minfilesizeext" ]
			then
				if [ "$p" -ge "10" ]
				then
					break
				fi
			fi
		fi
		prevfilesize="$filesize"
		if [ "$filesize" -lt "$minfilesize" ]
		then
			minfilesize="$filesize"
		fi
	done
}

COMPANY=$1
if [[ "$COMPANY" == "" ]]
then
	echo "The script needs username as a parameter."
	exit 1	# TODO: adjust error code
fi

SUBDIR="company_$COMPANY"
ADDR="https://habr.com/ru/companies/$COMPANY"

miceweb_save_urls() {
	if [ $ISMWINST -eq 1 ]; then
		miceweb save urls - <<< "$(echo $ADDR/profile/ ; echo $ADDR/articles/ ; echo $ADDR/news/ ; echo $ADDR/fans/all/rating/ ; echo $ADDR/workers/all/rating/)"
	fi
}

create_directory_and_get_pages() {
	local directory="$1"
	local URL="$2"
	sleep 2.5
	mkdir "$directory"
	cd "$directory"
	wget -E "$URL"
	get_pages "$URL"
	cd ..
}

if [ -e "$SUBDIR" ]; then
    echo "$SUBDIR already exist (completely downloaded, but when?)"
	miceweb_save_urls
elif [ -e "$SUBDIR.zip" ]; then
    echo "$SUBDIR already exist (completely downloaded and zipped)"
	miceweb_save_urls
    exit
elif [ -e "~$SUBDIR" ]; then
    echo "~$SUBDIR already exist (incompletely downloaded), you should remove it"
    exit 1
else    
	mkdir "~$SUBDIR"
	cd "~$SUBDIR"
	wget -E "$ADDR/profile/"
	create_directory_and_get_pages articles "$ADDR/articles"
	create_directory_and_get_pages news "$ADDR/news"
	create_directory_and_get_pages fans "$ADDR/fans/all/rating"
	create_directory_and_get_pages workers "$ADDR/workers/all/rating"
    if [ $ISMWINST -eq 1 ]; then
		#TODO: use --archives=all, save posts from next pages
		miceweb_save_urls
		miceweb save urls index.html
		miceweb save urls articles/articles.html --base="https://habr.com" --grep="articles/"
		miceweb save urls news/news.html --base="https://habr.com" --grep="news/"
    fi
    cd ..
    mv "~$SUBDIR" "$SUBDIR"
fi

ZIPFILE="$SUBDIR.zip"
if [ -d "$SUBDIR" ]; then
	if [ ! -e "$ZIPFILE" ]; then
		zip -1 -T -r $ZIPFILE $SUBDIR
		if [ $? -eq 0 ]; then
			if [ ! -e "~$SUBDIR" ]; then
				echo -n "Deleting ~$SUBDIR..."
				mv "$SUBDIR" "~$SUBDIR"
				rm -r "~$SUBDIR"
				if [ $? -eq 0 ]; then
				    echo " Done"
				else
				    echo ""
				    exit 1
				fi
			fi
		fi
	fi
fi