osubdl/osubdl

#!/bin/sh

# --------------------------------------
# generic

# NIL --> STRING
# Read from stdin until eof hit; return all input--
# good for writing functions that take piped info.
function reade {
	local stack=""

	while read input; do
		stack="$(printf '%s\n%s' "$stack" "$input")"
	done

	echo "$stack"
}

# STRING STRING --> STRING
# Replace the file-extension of a filename with another.
function replace_file_ext {
	local filename="$1"
	local ext="$2"
	
	local new_filename="$(echo "$filename" | sed 's%\....$%\.'"$ext"'%')"

	if test "$new_filename" = "$filename"; then
		echo "${filename}.${ext}"
	else
		echo "${new_filename}"
	fi	
}	



#---------------------------------------

# STRING --> STRING
# Hash a video file.
function vhash {
	local file="$1"
	oshash "$file"
}



#---------------------------------------

# STRING STRING --> STRING
# Get search results/subtitle-page of a file of certain hash and language.
# Make sure the second argument, $lang, is set to the 3-char language code,
# *not* the 2-char. I.E., "esp" over "es"
function search {
	local hash="$1"
	local lang="$2"

	local url="https://www.opensubtitles.org/en/search"
	url="${url}/sublanguageid-${lang}"
	url="${url}/moviehash-${hash}"

	gendl "$url"
}

# |STRING --> [BOOLEAN]
# Return 0 if page is a search-results; 1 if it's a subtitle page.
# Sometimes, when searching for a file, it'll redirect you to the only result;
# other times, it lists results. So we need this. :p
function search_resultsp {
	local html="$(reade)"

	local result="$(echo "$html" | grep "Subtitles -" | grep -v "img wid")"

	if test -z "$result"; then
		return 1
	else
		return 0
	fi
}

# --------------------------------------

# |STRING --> STRING
# Return the dl.opensubtitles URL, from the given subtitle page HTML.
function subtitle_dl_url {
	local html="$(reade)"

	echo "$html" \
	| grep "subtitleserve" \
	| grep "directUrl" \
	| head -1 \
	| sed 's/.*=.//' \
	| sed 's/.;//' \
	| sed 's%.*/%%' \
	| sed 's%^%https://dl.opensubtitles.org/en/download/sub/%'
}

# STRING --> STRING
# Return the regular subtitle-page URL from the dl.opensubtitles URL.
function subtitle_url {
	local dl_url="$1"
	local temp="$(mktemp)"

	curl --verbose "$dl_url" \
	2> $temp

	grep "location:" $temp \
	| sed 's%.*location: %%'
}

# --------------------------------------

# STRING STRING --> NIL
# Download the subtitle zip of a video, from subpage URL and dl URL.
function subtitle_get {
	local url="$1"
	local dl_url="$2"
	local dest="$3"

	local temp="/tmp/$(echo "$RANDOM")"

	mkdir "$temp"
	local useragent='Mozilla/5.0 (Windows NT 6.1; rv:52.0) Gecko/20100101'
	useragent="$useragent Firefox/52.0"

	# curl line shamelessly borrowed from jwilk/opensubtitles-dl <3
	curl --show-error --user-agent "$useragent" --referer "$url" "$dl_url" \
	> "${temp}sub.zip"
	
	unzip -qq -d "$temp" "${temp}/sub.zip"
	mv ${temp}/*.srt "$dest"
	rm -rf "${temp}"
}

# --------------------------------------

# STRING STRING PATH --> NIL
# Download a file's subtitles in given language to given path.
function get_file_subs {
	local file="$1"
	local language="$2"
	local dest="$3"

	local hash="$(vhash "$file")"
	local html="$(search "$hash" "$language")"

	if echo "$html" | search_resultsp; then
#		echo "$html"
		echo "No results found."
		return 3
	fi

	local dl_url="$(echo "$html" | subtitle_dl_url | tr -d '\r')"
	local url="$(subtitle_url "$dl_url" | tr -d '\r')"

	subtitle_get "$url" "$dl_url" "$dest"
}



# --------------------------------------
# invocation

function usage {
	echo "usage: osubdl file [language] [path]"
	echo "	'file' should be a video file."
	echo "	'language' should be a three-char language code."
	echo "	'path' should be the destination of the srt file."

	exit 2
}

# --------------------------------------

FILE="$1"
LANGUAGE="$2"
DESTINATION="$3"
if test -z "$FILE"; then
	usage
elif test -z "$LANG"; then
	LANGUAGE="eng"
elif test -z "$DESTINATION"; then
	DESTINATION="$(replace_file_ext "$FILE" "srt")"
fi
	
# --------------------------------------

echo "$FILE --> $DESTINATION"

get_file_subs "$FILE" "$LANGUAGE" "$DESTINATION"