libby rewrite by stewie410 w/ modifications

libby-rewrite

#!/usr/bin/env bash
#
# A simple CLI tool to quickly download books from Library Genesis

cleanup() {
	rm --recursive --force "${stage}"
	unset stage
}

show_help() {
	cat << EOF
A simple CLI tool to quickly download books from Library Genesis

USAGE: ${0##*/} [OPTIONS] QUERY

OPTIONS:
	-h, --help		Show this help message
	-f, --fzf		Use fzf to select a book (default)
	-r, --rofi		Use rofi to select a book
	-d, --dmenu		Use dmenu to select a book
	-m, --mirror NUM	Use an alternative libgen mirror (default: 3)
	-V, --no-view		Don't view the result
	-k, --to-kindle		Convert file to .mobi and send to \$LIBBY_KINDLE_EMAIL via neo-/mutt (implies -V)
	-u, --to-usb		Copy file to root of \$LIBBY_USB via rsync (impies -V)
	-o, --outdir PATH	Output book to PATH instead of \$LIBBY_OUTPUT_DIR
EOF
}

trim() {
	sed 's/^\s*//;s/\s*$//' | awk '{$1=$1};1' | xargs
}

require() {
	command -v "${*}" &>/dev/null && return
	printf '%s\n' "Missing required utility: '${*}'" >&2
	return 1
}

getRawResults() {
	local uri
	uri="https://libgen.gs/index.php?req="
	uri+="$(jq --slurp --raw-input --raw-output '@uri' <<< "${*}")"
	uri+='&columns%5B%5D=t&columns%5B%5D=a&columns%5B%5D=s&columns%5B%5D=y&columns%5B%5D=i&objects%5B%5D=f&topics%5B%5D=l&topics%5B%5D=f&res=25'

	curl --user-agent "${curlAgent}" --silent --fail --location "${uri}" | pup 'table:nth-last-of-type(1)'
}

getBookField() {
	local ln filter
	filter="tr:nth-of-type(${1}) td:nth-of-type(${2})"

	case "${2}" in
		1 | 2 )
			(($1 == 1)) && ln="8"
			pup --file "${3}" "${filter} text{}" | \
				sed '/^\s*$/d' | \
				awk --assign "num=${ln:-1}" '
					NR == num {
						print gensub(/^\s*|\s*$/, "", "G", $0)
					}
				'
			;;
		9 )
			pup --file "${3}" "${filter} a:nth-of-type(${mirrornum}) attr{href}" | \
				trim
			;;
		* )
			pup --file "${3}" "${filter} text{}" | \
				trim
			;;
	esac
}

parseRawResults() {
	local i j idx
	local -a book
	idx="1"

	for ((i = 1; i > 0; i++)); do
		book[0]="$(getBookField "${i}" "1" "${*}")"
		[[ -n "${book[0]}" ]] || return

		for j in {2..5} {7..9}; do
			book+=("$(getBookField "${i}" "${j}" "${*}")")
		done

		if ! [[ "${book[6],,}" =~ (epub|pdf) ]]; then
			((idx++))
			unset book
			continue
		fi

		printf '%s;' "0" "${idx}"
		for ((j = 0; j < ${#book[@]}; j++)); do
			printf '%s;' "$((j + 1))" "${book[${j}]}"
		done
		printf '%s\n' ""

		unset line
		((idx++))
	done
}

getUserSelection() {
	local options userSelection
	options="${stage}/userSelectionOptions"
	# I made the separator in the style of ';2;' because I wanted a string that I was pretty sure could not appear in a title. A standalone semicolon may (?) appear in a title somewhere, so it might be best to make this two semicolons if possible. But thank you for the awk suggestion, that's definitely better.
	awk --field-separator ";" '{
		line = $2 ": " $4 " - " $6 " (" $8 ", " $10 ") [" $12 "]"
		print gensub(/^\s*|\s*$/, "", "G", line)
	}' "${*}" | \
		recode --quiet "html..ascii" > "${options}"
		# this used to say "record" instead of "recode", probably a mistake

	case "${mode,,}" in
		rofi )
			userSelection="$(rofi \
					-i \
					-hover-select \
					-me-select-entry "" \
					-me-accept-entry "MousePrimary" \
					-dmenu \
					-p "Select book: " \
					-theme-str '* {font: "monospace 12"; width:80%;}' \
					-l "26" < "${options}")" || \
				return
			;;
		dmenu )
			userSelection="$(dmenu -l 26 -fn "monospace" -p "Select book: " < "${options}")" || \
				return
			;;
		fzf )
			userSelection="$(fzf < "${options}")" || \
				return
			;;
	esac

	echo "\n\n${userSelection}\n\n" > wow.txt
	# this used to be a ';', but the string for prompting the user usually won't contain any semicolons. So I replaced this with a colon.
	sed --quiet "$(cut --fields='1' --delimiter=':' <<< "${userSelection}")p" "${*}"
}

sendKindleEmail() {
	local mobi emlCmd
	mobi="${stage}/$(basename "${*%.*}").mobi"

	require "ebook-converter" || return
	if ! ebook-convert "${*}" "${mobi}"; then
		printf '%s\n' "Failed to convert book: '${*}' -> '${mobi}'" >&2
		return 1
	fi

	for i in "mutt" "neomutt" "mailx" "mail"; do
		if command -v "${i}" &>/dev/null; then
			emlCmd="${i}"
			break
		fi
	done

	if [[ -z "${emlCmd}" ]]; then
		printf '%s\n' "Missing application to send file as email attachment: '${mobi}'"
		return 1
	fi

	"${emlCmd}" -s "wow" -a "${mobi}" -- "${kindleEmail}" <<< ""
}

sendToUSB() {
	require "rsync" || return

	if ! mount | grep --quiet "${usb}"; then
		printf '%s\n' "Usb is not mounted: '${usb}'" >&2
		return 1
	fi

	mkdir --parents "${usb}"
	rsync \
		--times \
		--executability \
		--recursive \
		--archive \
		--verbose \
		--human-readable \
		--perms \
		--progress \
		"${*}" "${usb}/${*##*/}" || return
	sync
}

getSelectedField() {
	local num

	case "${1,,}" in
		title )			num="4";;
		author )		num="6";;
		publisher )		num="8";;
		year )			num="10";;
		size )			num="12";;
		filetype )		num="16";; # offset by 2, originally 14
		mirror )		num="18";; # offset by 2, originally 16
		* )
			printf '%s\n' "Unknown selected field: '${1,,}'" >&2
			return 1
			;;
	esac

	case "${1,,}" in
		4 )
			awk --field-separator ";" '{
				title = gensub(/[0-9]{6,}\s?,?;?/, "", "G", $4)
				title = gensub(/^\s|\s[1Xb]$/, "", "G", title)
				title = gensub(/\s+/, "-", "G", title)
				title = gensub(/[^a-zA-Z0-9._-]/, "", "G", title)
				print gensub(/-+/, "-", "G", title)
			}' <<< "${2,,}" | recode --quiet "html..ascii"
			;;
		* )
			cut --fields="${num}" --delimiter=";" <<< "${2}"
			;;
	esac
}

download() {
	local mirror tmp
	tmp="${stage}/$(getSelectedField "title" "${*}").$(getSelectedField "filetype" "${*}")"
	mirror="$(getSelectedField "mirror" "${*}")"

	if [[ -z "${mirror}" ]]; then
		printf '%s\n' "No mirror found!" >&2
		return 1
	fi

	printf '%s\n' "Fetching mirror URI..."
	# --agent is not an option in curl, I think you meant --user-agent
	mirror="$(curl --silent --fail --user-agent "${curlAgent}" --location "${mirror}" | \
		pup "#download a attr{href}" | \
		sed 1q)"

	if [[ -z "${mirror}" ]]; then
		printf '%s\n' "Failed to retrieve mirror!" >&2
		return 1
	fi

	printf '\n\n%s\n\n' "Downloading book from mirror: '${mirror}' -> '${tmp}'"
	if ! curl --progress-bar --fail --location "${mirror}" > "${tmp}"; then
		printf '%s\n' "Failed to download file!" >&2
		return 1
	fi

	mkdir --parents "${outdir}"
	mv --force "${tmp}" "${outdir}/${tmp##*/}"
	printf '%s\n' "${outdir}/${tmp##*/}"
}

process() {
	local raw books selected outfile err
	raw="${stage}/raw.html"
	books="${stage}/books.list"

	printf '%s\n' "Fetching data from query..."
	getRawResults "${*}" > "${raw}"
	parseRawResults "${raw}" > "${books}"
	if ! [[ -s "${books}" ]]; then
		printf '%s\n' "No results found for query: '${*}'" >&2
		return 1
	fi

	printf '%s\n' "Prompting user for selection..."
	selected="$(getUserSelection "${books}")" || return

	# nothing in the download function gets printed/shown to the user because the whole thing is called in a command subtitution
	outfile="$(download "${selected}")" || return

	if [[ -n "${toKindle}" ]] && ! sendKindleEmail "${outfile}"; then
		printf '%s\n' "Failed to send kindle email" >&2
		err+="1"
	fi

	if [[ -n "${toUSB}" ]] && ! sendToUSB "${outfile}"; then
		printf '%s\n' "Failed to send to USB" >&2
		err+="2"
	fi
	# Many of the function calls along the way include some print statements, and this interferes with the output of the download function.
	# The $outfile variable includes the entire program output, not just the path to the outfile.
	# As a hacky workaround, I just get the last line of output.
	outfile="$(tail -1 <<< ${outfile})"
	[[ -n "${view}" ]] && "${view}" "${outfile}"
	return "${err:-0}"
}

main() {
	local opts outdir usb kindleEmail mode mirrornum view toKindle toUSB curlAgent
	outdir="${LIBBY_OUTPUT_DIR:-${HOME}/books}"
	usb="${LIBBY_USB_DIR}"
	kindleEmail="${LIBBY_KINDLE_EMAIL}"
	mode="fzf"
	mirrornum="3"
	view="${LIBBY_VIEWER:-xdg-open}"
	curlAgent='Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) QtWebEngine/5.15.5 Chrome/87.0.4280.144 Safari/537.36'

	opts="$(getopt --options hfrdm:Vkuo: --longoptions help,fzf,rofi,dmenu,mirror:,no-view,to-kindle,to-usb,outdir: --name "${0##*/}" -- "${@}")"
	eval set -- "${opts}"
	while true; do
		case "${1}" in
			-h | --help )		show_help; return 0;;
			-f | --fzf )		mode="fzf";;
			-r | --rofi )		mode="rofi";;
			-d | --dmenu )		mode="dmenu";;
			-m | --mirror )		mirrornum="${2}"; shift;;
			-V | --no-view )	unset view;;
			-k | --to-kindle )	toKindle="1"; unset view;;
			-u | --to-usb )		toUSB="1"; unset view;;
			-o | --outdir )		outdir="${2}"; shift;;
			-- )				shift; break;;
			* )					break;;
		esac
		shift
	done

	if [[ -z "${*}" ]]; then
		printf '%s\n' "No query specified" >&2
		show_help
		return 1
	fi

	require "pup" || return
	require "curl" || return
	require "${mode,,}" || return

	process "${*}"
}

declare stage
stage="$(mktemp --directory)"
trap cleanup EXIT

main "${@}"