lib.sh

#!/usr/bin/env bash

set -eEuo pipefail
shopt -s inherit_errexit

# A simple trap handler for exiting on errors in subshells
err_trap_handler() {
	echo "ERROR on line $(caller)!" >&2
	exit 13
}
trap err_trap_handler ERR

#shellcheck disable=SC2034
VERSION="0.5.1"
#shellcheck disable=SC2034
GREEN='\033[0;32m'
#shellcheck disable=SC2034
RED='\033[0;31m'
#shellcheck disable=SC2034
BOLD='\033[1m'
#shellcheck disable=SC2034
NC='\033[0m'

: "${VERBOSE:=false}"
: "${DRY_RUN:=false}"
: "${SYMLINK_ONLY:=false}"
: "${KEEP_METADATA:=false}"

: "${TESTED_ARCHIVE_EXTENSIONS:=^(7z|bz2|chm|arj|cab|gz|tgz|gzip|zip|rar|xz|tar|epub|docx|odt|ods|cbr|cbz|maff|iso)\$}"

# Horizontal whitespace and dash-like ASCII and Unicode characters that are
# used for better matching of ISBNs in (badly) OCR-ed books. Gathered from:
# - https://en.wikipedia.org/wiki/Whitespace_character
# - https://en.wikipedia.org/wiki/Dash#Similar_Unicode_characters
# - https://en.wikipedia.org/wiki/Dash#Common_dashes
: "${WSD:="[\\x{0009}\\x{0020}\\x{00A0}\\x{1680}\\x{2000}\
\\x{2001}\\x{2002}\\x{2003}\\x{2004}\\x{2005}\\x{2006}\\x{2007}\\x{2008}\
\\x{2009}\\x{200A}\\x{202F}\\x{205F}\\x{3000}\\x{180E}\\x{200B}\\x{200C}\
\\x{200D}\\x{2060}\\x{FEFF}\\x{002D}\\x{005F}\\x{007E}\\x{00AD}\\x{00AF}\
\\x{02C9}\\x{02CD}\\x{02D7}\\x{02DC}\\x{2010}\\x{2011}\\x{2012}\\x{203E}\
\\x{2043}\\x{207B}\\x{208B}\\x{2212}\\x{223C}\\x{23AF}\\x{23E4}\\x{2500}\
\\x{2796}\\x{2E3A}\\x{2E3B}\\x{10191}\\x{2012}\\x{2013}\\x{2014}\\x{2015}\
\\x{2053}\\x{058A}\\x{05BE}\\x{1428}\\x{1B78}\\x{3161}\\x{30FC}\\x{FE63}\
\\x{FF0D}\\x{10110}\\x{1104B}\\x{11052}\\x{110BE}\\x{1D360}]?"}"

# This regular expression should match most ISBN10/13-like sequences in
# texts. To minimize false-positives, matches should be passed through
# is_isbn_valid() or another ISBN validator
: "${ISBN_REGEX:="(?<![0-9])(${WSD}9${WSD}7${WSD}[789]${WSD})?+((${WSD}[0-9]${WSD}){9}[0-9xX])(?![0-9])"}"
: "${ISBN_DIRECT_GREP_FILES:="^(text/(plain|xml|html)|application/xml)\$"}"
: "${ISBN_IGNORED_FILES:="^(image/(gif|svg.+)|application/(x-shockwave-flash|CDFV2|vnd.ms-opentype|x-font-ttf|x-dosexec|vnd.ms-excel|x-java-applet)|audio/.+|video/.+)\$"}"
: "${ISBN_RET_SEPARATOR:=,}"

# This is matched against normalized valid-looking ISBNs and any numbers that
# match it are discarded.
# The default value should match 0123456789 and any ISBN-10 that uses only one
# digit (e.g. 1111111111 or 3333333333)
: "${ISBN_BLACKLIST_REGEX="^(0123456789|([0-9xX])\\2{9})\$"}"

# These options specify if and how we should reorder ISBN_DIRECT_GREP files
# before passing them to find_isbns(). If true, the first
# ISBN_GREP_RF_SCAN_FIRST lines of the files are passed as is, then we pass
# the last ISBN_GREP_RF_REVERSE_LAST in reverse order and finally we pass the
# remainder in the middle. There is no issue if files have fewer lines, there
# will be no duplicate lines passed to grep.
: "${ISBN_GREP_REORDER_FILES:=true}"
: "${ISBN_GREP_RF_SCAN_FIRST:=400}"
: "${ISBN_GREP_RF_REVERSE_LAST:=50}"

# Whether to use OCR on image files, pdfs and djvu files for ISBN searching
# and conversion to txt
: "${OCR_ENABLED:=false}"
: "${OCR_ONLY_FIRST_LAST_PAGES:=7,3}"
: "${OCR_COMMAND:=tesseract_wrapper}"

# Set to empty string if using Calibre versions before 2.84, previous versions did not support the
# --allowed-plugin option that we use to search for metadata in the order specified below, so they
# can search in all enabled sources in the GUI.
: "${ISBN_METADATA_FETCH_ORDER="Goodreads,Amazon.com,Google,ISBNDB,WorldCat xISBN,OZON.ru"}"
: "${ORGANIZE_WITHOUT_ISBN_SOURCES="Goodreads,Amazon.com,Google"}"

# Should be matched against a lowercase filename.ext, lines that start with #
# and newlines are removed. The default value should filter out most periodicals
: "${RE_YEAR:="(19[0-9]|20[0-$(date '+%Y' | cut -b 3)])[0-9]"}"
: "${WITHOUT_ISBN_IGNORE:=$(echo "
# Perdiodicals with filenames that contain something like 2010-11, 199010, 2015_7, 20110203:
(^|[^0-9])${RE_YEAR}[ _\\.-]*(0?[1-9]|10|11|12)([0-9][0-9])?(\$|[^0-9])
# Periodicals with month numbers before the year
|(^|[^0-9])([0-9][0-9])?(0?[1-9]|10|11|12)[ _\\.-]*${RE_YEAR}(\$|[^0-9])
# Periodicals with months or issues
|((^|[^a-z])(jan(uary)?|feb(ruary)?|mar(ch)?|apr(il)?|may|june?|july?|aug(ust)?|sep(tember)?|oct(ober)?|nov(ember)?|dec(ember)?|mag(azine)?|issue|#[ _\\.-]*[0-9]+)+(\$|[^a-z]))
# Periodicals with seasons and years
|((spr(ing)?|sum(mer)?|aut(umn)?|win(ter)?|fall)[ _\\.-]*${RE_YEAR})
|(${RE_YEAR}[ _\\.-]*(spr(ing)?|sum(mer)?|aut(umn)?|win(ter)?|fall))
" | grep -v '^#' | tr -d '\n')}"


: "${TOKEN_MIN_LENGTH:=3}"
: "${TOKENS_TO_IGNORE:="ebook|book|novel|series|ed(ition)?|vol(ume)?|${RE_YEAR}"}"
[ -z "${FILE_SORT_FLAGS:+x}" ] && FILE_SORT_FLAGS=()

#shellcheck disable=SC2016
[ -z "${OUTPUT_FILENAME_TEMPLATE:+x}" ] && OUTPUT_FILENAME_TEMPLATE='"${d[AUTHORS]// & /, } - ${d[SERIES]:+[${d[SERIES]//:/ -}] - }${d[TITLE]//:/ -}${d[PUBLISHED]:+ (${d[PUBLISHED]%%-*})}${d[ISBN]:+ [${d[ISBN]}]}.${d[EXT]}"'
: "${OUTPUT_METADATA_EXTENSION:=meta}"

: "${DEBUG_PREFIX_LENGTH:=40}"

# Handle parsing from arguments and setting all the common config vars
#shellcheck disable=SC2034
handle_script_arg() {
	local arg="$1"
	case "$arg" in
		-v|--verbose) VERBOSE=true ;;
		-d|--dry-run) DRY_RUN=true ;;
		-sl|--symlink-only) SYMLINK_ONLY=true ;;
		-km|--keep-metadata) KEEP_METADATA=true ;;

		--tested-archive-extensions=*) TESTED_ARCHIVE_EXTENSIONS="${arg#*=}" ;;
		-i=*|--isbn-regex=*) ISBN_REGEX="${arg#*=}" ;;
		--isbn-blacklist-regex=*) ISBN_BLACKLIST_REGEX="${arg#*=}" ;;
		--isbn-direct-grep-files=*) ISBN_DIRECT_GREP_FILES="${arg#*=}" ;;
		--isbn-ignored-files=*) ISBN_IGNORED_FILES="${arg#*=}" ;;
		--reorder-files-for-grep=*)
			i="${arg#*=}"
			if [[ "$arg" == "false" ]]; then
				ISBN_GREP_REORDER_FILES=false
			else
				ISBN_GREP_REORDER_FILES=true
				ISBN_GREP_RF_SCAN_FIRST="${arg%,*}"
				ISBN_GREP_RF_REVERSE_LAST="${arg##*,}"
			fi
		;;
		-ocr=*|--ocr-enabled=*) OCR_ENABLED="${arg#*=}" ;;
		-ocrop=*|--ocr-only-first-last-pages=*) OCR_ONLY_FIRST_LAST_PAGES="${arg#*=}" ;;
		-ocrc=*|--ocr-command=*) OCR_COMMAND="${arg#*=}" ;;

		--token-min-length=*) TOKEN_MIN_LENGTH="${arg#*=}" ;;
		--tokens-to-ignore=*) TOKENS_TO_IGNORE="${arg#*=}" ;;

		-mfo=*|--metadata-fetch-order=*) ISBN_METADATA_FETCH_ORDER="${arg#*=}" ;;
		-owis=*|--organize-without-isbn-sources=*) ORGANIZE_WITHOUT_ISBN_SOURCES="${arg#*=}" ;;
		-wii=*|--without-isbn-ignore=*) WITHOUT_ISBN_IGNORE="${arg#*=}" ;;
		-fsf=*|--file-sort-flags=*) IFS=" " read -r -a FILE_SORT_FLAGS <<< "${arg#*=}" ;;

		-oft=*|--output-filename-template=*) OUTPUT_FILENAME_TEMPLATE="${arg#*=}" ;;
		-ome=*|--output-metadata-extension=*) OUTPUT_METADATA_EXTENSION="${arg#*=}" ;;

		--debug-prefix-length=*) DEBUG_PREFIX_LENGTH="${arg#*=}" ;;

		--lib-hook=*)
			# shellcheck source=/dev/null
			source "${arg#*=}"
		;;

		-*) echo "Invalid option '$arg'"; exit 4; ;;
	esac
}


# If the VERBOSE flag is on, outputs the arguments to stderr
decho () {
	if [[ "${VERBOSE:-false}" == true ]]; then
		echo "$@" >&2
	fi
}


# If the VERBOSE flag is on, prefixes the stdin with the supplied prefix
# (shortened/padded or not) and outputs the result to stderr
#
# Arguments:
#	prefix:	the string with which we will prefix the lines
#	[should_fit_in]: number of characters to which we want to shorten or pad
#		the prefix so it fits; 0 is disabled
#	[...]: everything else is passed to the fmt command
debug_prefixer() {
	local prefix="$1"
	if (( $# > 1 )); then
		local should_fit_in="$2"
		if (( should_fit_in > 0 )); then
			if (( ${#prefix} > should_fit_in )); then
				prefix="${prefix:0:10}..${prefix:(-$((should_fit_in - 12)))}"
			else
				prefix="$(printf "%-${should_fit_in}s" "$prefix")"
			fi
		fi
		shift
	fi
	shift

	( if [[ "$#" != "0" ]]; then fmt "$@"; else cat; fi ) |
	while IFS= read -r line || [[ -n "$line" ]] ; do
		decho "${prefix}${line}"
	done
}


# Converts to lowercase (with unicode support)
to_lower() {
	sed -E 's/[[:upper:]]+/\L&/g'
}

# Prints only the first instance of any line
uniq_no_sort() {
	awk '!x[$0]++'
}

# Concatenate the passed arguments with $1 (can be multiple characters)
str_concat () {
	local od="$1"
	shift
	[[ "$#" == "0" ]] && return
	echo -n "$1"
	shift
	printf "%s" "${@/#/$od}";
}

# Read values from stdin (separated by $2 or newline by default) and
# concatenate them with $1. The delimiter can be multiple characters
stream_concat () {
	local od="$1" id="${2:-$'\n'}" val
	read -d "$id" -r val || return 0
	echo -n "$val"
	while read -d "$id" -r val || [[ -n "$val" ]]; do
		echo -n "${od}${val}"
	done
}

# Validates ISBN-10 and ISBN-13 numbers
is_isbn_valid() {
	local isbn i number sum=0
	isbn="$(echo "$1" | tr -d ' -' | tr '[:lower:]' '[:upper:]')"

	if [ "${#isbn}" == "10" ]; then
		for i in {0..9}; do
			number="${isbn:$i:1}"
			if [[ "$i" == "9" && "$number" == "X" ]]; then
				number=10
			fi
			sum=$(( sum + (number * ( 10 - i )) ))
		done
		if (( sum % 11 == 0 )); then
			return 0
		fi
	elif [ "${#isbn}" == "13" ]; then
		if [[ "${isbn:0:3}" = "978" || "${isbn:0:3}" = "979" ]]; then
			for i in {0..12..2}; do
				sum=$(( sum + ${isbn:$i:1} ))
			done
			for i in {1..11..2}; do
				sum=$(( sum + (${isbn:$i:1} * 3) ))
			done
			if (( sum % 10 == 0 )); then
				return 0
			fi
		fi
	fi
	return 1
}


# Reads and echoes only n lines from STDIN, without consuming the rest
cat_n() {
	local line="" lines=0
	while ((lines++ < $1 )); do
		if read -r line; then
			echo "$line"
		elif [[ $line != "" ]]; then
			echo -n "$line"
		fi
	done
}


# If ISBN_GREP_REORDER_FILES is enabled, reorders the specified file according
# to the values of ISBN_GREP_RF_SCAN_FIRST and ISBN_GREP_RF_REVERSE_LAST
cat_file_for_isbn_grep() {
	if [[ "$ISBN_GREP_REORDER_FILES" == true ]]; then
		decho "Reordering input file (if possible, read first $ISBN_GREP_RF_SCAN_FIRST lines normally, then read last $ISBN_GREP_RF_REVERSE_LAST lines in reverse and then read the rest"

		{ cat_n "$ISBN_GREP_RF_SCAN_FIRST"; tac | { cat_n "$ISBN_GREP_RF_REVERSE_LAST"; tac; } } < "$1"
	else
		cat "$1"
	fi
}


# Searches STDIN for ISBN-like sequences and removes duplicates (preserving
# the order) and finally validates them using is_isbn_valid() and returns
# them separated by $ISBN_RET_SEPARATOR
find_isbns() {
	local isbn
	{ grep -oP "$ISBN_REGEX" || true; } | tr -c -d '0-9xX\n' | uniq_no_sort | {
		while IFS='' read -r isbn || [[ -n "$isbn" ]]; do
			if is_isbn_valid "$isbn"; then
				echo "$isbn"
			fi
		done
	} | {
		if [ "$ISBN_BLACKLIST_REGEX" != "" ]; then
			grep -vP "$ISBN_BLACKLIST_REGEX" || true
		else
			cat
		fi
	} | stream_concat "$ISBN_RET_SEPARATOR"
}


# Returns non-zero status if the supplied command does not exist
command_exists() {
	command -v "$1" >/dev/null 2>&1
}


# Return "$1/$2" if no file exists at this path. Otherwrise, sequentially
# insert " ($n)" before the extension of $2 and return the first path for
# which no file is present.
unique_filename() {
	local new_path="$1/$2" counter=0
	# while [[ -e "$new_path" ]]; do
	# 	counter="$((counter+1))"
	# 	decho "File '$new_path' already exists in destination '$1', trying with counter $counter!"
	# 	new_path="${1}/${2%.*} ($counter).${2##*.}"
	# done
	echo "$new_path"
}


# Returns a single value by key by parsing the calibre-style text metadata
# hashmap that is passed to stdin
grep_meta_val() {
	{ grep --max-count=1 "^$1" || true; } | awk -F' : ' '{ print $2 }'
}


# Splits the stdin stream into alpha or numeric tokens with length at least
# $3 (or $TOKEN_MIN_LENGTH), converts them to lowercase, optionally
# deduplicates them (if $2 is true or not specified) and finally concatenates
# them with $1 (or ' ' if not specified)
tokenize() {
	local separator="${1:- }" dedup="${2:-true}"
	local lenr="${3:-$TOKEN_MIN_LENGTH}" tokens_to_ignore="${4:-$TOKENS_TO_IGNORE}"
	{ grep -oE "[[:alpha:]]{${lenr},}|[[:digit:]]{${lenr},}" || true; } | to_lower | {
		if [[ "$dedup" == true ]]; then
			uniq_no_sort
		else
			cat
		fi
	} | {
		if [[ "$tokens_to_ignore" != "" ]]; then
			grep -ivE "^($tokens_to_ignore)\$" || true
		else
			cat
		fi
	} | stream_concat "$separator"
}


# Checks the supplied file for different kinds of corruption:
#  - If it's zero-sized or contains only \0
#  - If it's has a pdf extension but different mime type
#  - If it's a pdf and pdfinfo returns an error
#  - If it has an archive extension but `7z t` returns an error
check_file_for_corruption() {
	local file_path="$1"
	decho "Testing '$file_path' for corruption..."

	if [[ "$(tr -d '\0' < "$file_path" | head -c 1)" == "" ]]; then
		echo "The file is empty or contains only zeros!"
		return
	fi

	local ext="${1##*.}" mimetype
	mimetype="$(file --brief --mime-type "$file_path")"

	if [[ "$mimetype" == "application/octet-stream" && "$ext" =~ ^(pdf|djv|djvu)$ ]]; then
		echo "The file has a .$ext extension but '$mimetype' MIME type!"
	elif [[ "$mimetype" == "application/pdf" ]]; then
		decho "Checking pdf file for integrity..."
		if ! command_exists pdfinfo; then
			decho "pdfinfo does not exist, could not check if pdf is OK"
		else
			local pdfinfo_output
			if ! pdfinfo_output="$(pdfinfo "$file_path" 2> >(tail | debug_prefixer "[pdfinfo-err|tail] " 0 --width=80 -s))"; then
				decho "pdfinfo returned an error!"
				echo "$pdfinfo_output" | debug_prefixer "[pdfinfo] " 0 --width=80 -t
				echo "Has pdf MIME type or extension, but pdfinfo returned an error!"
				return
			else
				decho "pdfinfo returned successfully"
				echo "$pdfinfo_output" | debug_prefixer "[pdfinfo] " 0 --width=80 -t
				if echo "$pdfinfo_output" | grep --quiet -E '^Page size:\s*0 x 0 pts$'; then
					decho "pdf is corrupt anyway, page size property is empty!"
					echo "pdf can be parsed, but page size is 0 x 0 pts!"
				fi
			fi
		fi
	fi

	if [[ "$ext" =~ $TESTED_ARCHIVE_EXTENSIONS ]]; then
		decho "The file has a '.$ext' extension, testing with 7z..."
		local log

		if ! log="$(7z t "$file_path" 2>&1)"; then
			decho "Test failed!"
			echo "$log" |  debug_prefixer "[7z-test-log] " 0 --width=80 -s
			echo "Looks like an archive, but testing it with 7z failed!"
		fi
	fi
}


# Tries to convert the supplied ebook file into .txt. It uses calibre's
# ebook-convert tool. For optimization, if present, it will use pdftotext
# for pdfs, catdoc for word files and djvutxt for djvu files.
# Arguments: input path, output path (shloud have .txt extension), mimetype
convert_to_txt() {
	local if="$1" of="$2" mimetype="$3"
	if [[ "$mimetype" == "application/pdf" ]] && command_exists pdftotext; then
		decho "The file looks like a pdf, using pdftotext to extract the text"
		pdftotext "$if" "$of"
	elif [[ "$mimetype" == "application/msword" ]] && command_exists catdoc; then
		decho "The file looks like a doc, using catdoc to extract the text"
		catdoc "$if" > "$of"
	elif [[ "$mimetype" == "image/vnd.djvu"* ]] && command_exists djvutxt; then
		decho "The file looks like a djvu, using djvutxt to extract the text"
		djvutxt "$if" "$of"
	elif [[ "$mimetype" != "image/vnd.djvu"* && "$mimetype" == "image/"* ]]; then
		decho "The file looks like a normal image ($mimetype), skipping ebook-convert usage!"
		return 1
	else
		decho "Trying to use calibre's ebook-convert to convert the '$mimetype' file to .txt"
		ebook-convert "$if" "$of"
	fi
}

tesseract_wrapper () {
	tesseract "$1" stdout --psm 12 > "$2" || exit 1
}

ocr_file() {
	local if="$1" of="$2" mimetype="$3"
	local ocr_first_pages="${OCR_ONLY_FIRST_LAST_PAGES%,*}" ocr_last_pages="${OCR_ONLY_FIRST_LAST_PAGES##*,}"
	local num_pages page_convert_cmd

	convert_pdf_page() {
		gs -dSAFER -q -r300 -dFirstPage="$3" -dLastPage="$3" -dNOPAUSE -dINTERPOLATE -sDEVICE=png16m -sOutputFile="$2" "$1" -c quit
	}
	convert_djvu_page() {
		ddjvu -page="$3" -format=tif "$1" "$2"
	}

	case "$mimetype" in
		application/pdf)
			num_pages=$(pdfinfo "$if" | sed -n -E 's/^Pages:\s+([0-9]+)/\1/p')
			page_convert_cmd=convert_pdf_page
		;;
		image/vnd.djvu*)
			num_pages=$( djvused -e "n" "$if")
			page_convert_cmd=convert_djvu_page
		;;
		image/*) "$OCR_COMMAND" "$if" "$of" ;;
		*) decho "Unsupported mimetype '$mimetype'!"; return 4 ;;
	esac

	decho "Running OCR on file '$if' $num_pages pages and with mimetype '$mimetype'..."

	local page=1 tmp_file tmp_file_txt
	while (( page <= num_pages )); do
		if [[ "$OCR_ONLY_FIRST_LAST_PAGES" == false ]] ||
			(( page <= ${ocr_first_pages:-0} )) ||
			(( page > num_pages - ${ocr_last_pages:-0} ));
		then
			tmp_file=$(mktemp)
			tmp_file_txt=$(mktemp --suffix='.txt')
			decho "Running OCR of page $page, using tmp files '$tmp_file' and '$tmp_file_txt' ..."

			"$page_convert_cmd" "$if" "$tmp_file" "$page"
			"$OCR_COMMAND" "$tmp_file" "$tmp_file_txt"
			cat "$tmp_file_txt"

			decho "Cleaning up tmp files '$tmp_file' and '$tmp_file_txt'"
			rm "$tmp_file" "$tmp_file_txt"
		fi
		page=$(( page + 1))
	done > "$of"
}


# Arguments: the path to the archive file
get_all_isbns_from_archive() {
	local file_path="$1" isbns tmpdir
	tmpdir="$(mktemp -d)"

	decho "Trying to decompress '$file_path' into tmp folder '$tmpdir' and recursively scan the contents"
	if ! 7z x -o"$tmpdir" "$file_path" 2>&1 | debug_prefixer "[7zx] " 0 --width=80 -s; then
		decho "Error extracting the file (probably not an archive)! Removing tmp dir..."
		rm -rf "$tmpdir"
		return 1
	fi

	decho "Archive extracted successfully in '$tmpdir', scanning contents recursively..."
	while IFS= read -r -d '' file_to_check; do
		#decho "Searching '$file_to_check' for ISBNs..."
		isbns="$(search_file_for_isbns "$file_to_check" 2> >(debug_prefixer "[${file_to_check#$tmpdir}] " "${DEBUG_PREFIX_LENGTH:-40}") )"
		if [[ "$isbns" != "" ]]; then
			decho "Found ISBNs $isbns!"
			echo "$isbns" | tr "$ISBN_RET_SEPARATOR" '\n'
		fi
		decho "Removing '$file_to_check'..."
		rm "$file_to_check"
	done < <(find "$tmpdir" -type f  -print0 | sort -z ${FILE_SORT_FLAGS[@]:+"${FILE_SORT_FLAGS[@]}"})

	decho "Removing temporary folder '$tmpdir' (should be empty)..."
	find "$tmpdir" -type d -empty -delete
}


# Tries to find ISBN numbers in the given ebook file by using progressively
# more "expensive" tactics. If at some point ISBN numbers are found, they
# are echoed to stdout and the function returns.
# These are the steps:
#   - Check the supplied file name for ISBNs (the path is ignored)
#   - If the MIME type of the file matches ISBN_DIRECT_GREP_FILES, search
#     the file contents directly for ISBNs
#   - If the MIME type matches ISBN_IGNORED_FILES, the function returns
#     early with no results
#   - Check the file metadata from calibre's `ebook-meta` for ISBNs
#   - Try to extract the file as an archive with `7z`; if successful,
#     recursively call search_file_for_isbns for all the extracted files
#   - If the file is not an archive, try to convert it to a .txt file
#     via convert_to_txt()
#   - If OCR is enabled and convert_to_txt() fails or its result is empty,
#     try OCR-ing the file. If the result is non-empty but does not contain
#     ISBNs and OCR_ENABLED is set to "always", run OCR as well.
search_file_for_isbns() {
	local file_path="$1" isbns
	decho "Searching file '$file_path' for ISBN numbers..."

	isbns="$(basename "$file_path" | find_isbns)"
	if [[ "$isbns" != "" ]]; then
		decho "Extracted ISBNs '$isbns' from the file name!"
		echo -n "$isbns"
		return
	fi

	local mimetype
	mimetype="$(file --brief --mime-type "$file_path")"
	decho "Ebook MIME type: $mimetype"
	if [[ "$mimetype" =~ $ISBN_DIRECT_GREP_FILES ]]; then
		# TODO: maybe decode entities in HTML/XML files since ISBNs can
		# be intersected by &nbsp; or other escaped characters;
		# recode/perl/php can be used for this
		decho "Ebook is in text format, trying to find ISBN directly"
		isbns="$(cat_file_for_isbn_grep "$file_path" | find_isbns)"
		if [[ "$isbns" != "" ]]; then
			decho "Extracted ISBNs '$isbns' from the text file contents!"
			echo -n "$isbns"
		else
			decho "Did not find any ISBNs"
		fi
		return
	elif [[ "$mimetype" =~ $ISBN_IGNORED_FILES ]]; then
		decho "The file type in the blacklist, ignoring..."
		return
	fi

	local ebookmeta
	ebookmeta="$(ebook-meta "$file_path")"
	decho "Ebook metadata:"
	echo "$ebookmeta" | debug_prefixer "	" 0 --width=80 -t
	isbns="$(echo "$ebookmeta" | find_isbns)"
	if [[ "$isbns" != "" ]]; then
		decho "Extracted ISBNs '$isbns' from calibre ebook metadata!"
		echo -n "$isbns"
		return
	fi


	if isbns="$(get_all_isbns_from_archive "$file_path" | uniq_no_sort | stream_concat "$ISBN_RET_SEPARATOR")"; then
		decho "Extracted ISBNs '$isbns' from the archive file"
		echo -n "$isbns"
		return
	fi

	local tmptxtfile try_ocr=false
	tmptxtfile="$(mktemp --suffix='.txt')"
	decho "Converting ebook to text format in file '$tmptxtfile'..."

	if convert_to_txt "$file_path" "$tmptxtfile" "$mimetype" 2>&1 | debug_prefixer "[ebook2txt] " 0 --width=80 -s; then
		decho "Conversion to text was successful, checking the result..."
		if ! grep -qiE "[[:alnum:]]+" "$tmptxtfile"; then
			decho "The converted txt with size $(stat -c '%s' "$tmptxtfile") bytes does not seem to contain text:"
			#xxd -a $tmptxtfile | head | debug_prefixer "[cat tmp-txt] "
			try_ocr=true
		else
			isbns="$(cat_file_for_isbn_grep "$tmptxtfile" | find_isbns)"
			if [[ "$isbns" != "" ]]; then
				decho "Text output contains ISBNs '$isbns'!"
			elif [[ "$OCR_ENABLED" == "always" ]]; then
				decho "We will try OCR because the successfully converted text did not have any ISBNs"
				try_ocr=true
			else
				decho "Did not find any ISBNs and will NOT try OCR"
			fi
		fi
	else
		decho "There was an error converting the book to txt format"
		try_ocr=true
	fi

	if [[ "$isbns" == "" && "$OCR_ENABLED" != false && "$try_ocr" == true ]]; then
		decho "Trying to run OCR on the file..."
		if ocr_file "$file_path" "$tmptxtfile" "$mimetype" 2>&1 | debug_prefixer "[ocr] " 0 --width=80 -t; then
			decho "OCR was successful, checking the result..."
			isbns="$(cat_file_for_isbn_grep "$tmptxtfile" | find_isbns)"
			if [[ "$isbns" != "" ]]; then
				decho "OCR output contains ISBNs '$isbns'!"
			else
				decho "Did not find any ISBNs in the OCR output"
			fi
		else
			decho "There was an error while running OCR!"
		fi
	fi

	decho "Removing '$tmptxtfile'..."
	rm "$tmptxtfile"

	if [[ "$isbns" != "" ]]; then
		decho "Returning the found ISBNs '$isbns'!"
		echo -n "$isbns"
	else
		decho "Could not find any ISBNs in '$file_path' :("
	fi
}


move_or_link_file() {
	local current_path="$1" new_path="$2" new_folder="${2%/*}"

	$DRY_RUN && decho "(DRY RUN! All operations except metadata deletion are skipped!)"

	if [[ ! -d "$new_folder" ]]; then
		decho "Creating folder '$new_folder'"
		if [[ "$DRY_RUN" == "false" ]]; then
			mkdir -p "$new_folder"
		fi
	fi

	if [[ "$SYMLINK_ONLY" == true ]]; then
		decho "Symlinking file '$current_path' to '$new_path'..."
		if [[ "$DRY_RUN" == "false" ]]; then
			ln -s "$(realpath "$current_path")" "$new_path"
		fi
	else
		decho "Moving file '$current_path' to '$new_path'..."
		if [[ "$DRY_RUN" == "false" ]]; then
			mv --no-clobber "$current_path" "$new_path"
		fi
	fi
}


# Arguments: new_folder, current_ebook_path, current_metadata_path
move_or_link_ebook_file_and_metadata() {
	local new_folder="$1" current_ebook_path="$2" current_metadata_path="$3" line
	declare -A d=( ["EXT"]="${current_ebook_path##*.}" ) # metadata and the file extension

	while IFS='' read -r line || [[ -n "$line" ]]; do
		#TODO: fix this properly
		d["$(echo "${line%%:*}" | sed -e 's/[ \t]*$//' -e 's/ /_/g' -e 's/[^a-zA-Z0-9_]//g' -e 's/\(.*\)/\U\1/')"]="$(echo "${line#*: }" | sed -e 's/[\\/\*\?<>\|\x01-\x1F\x7F\x22\x24\x60]/_/g' | cut -c 1-100 )"
	done < "$current_metadata_path"

	decho "Variables that will be used for the new filename construction:"
	local key
	for key in "${!d[@]}"; do
		echo "${d[${key}]}" | debug_prefixer "    ${key}" 25
	done

	local new_name
	new_name="$(eval echo "$OUTPUT_FILENAME_TEMPLATE")"
	decho "The new file name of the book file/link '$current_ebook_path' will be: '$new_name'"

	local new_path
	new_path="$(unique_filename "${new_folder%/}" "$new_name")"
	basename=$(echo "$new_path" | cut -f 1 -d '.')
	echo -n "$new_path"

	move_or_link_file "$current_ebook_path" "$new_path"

	if [[ "$KEEP_METADATA" != true ]]; then
		decho "Removing metadata file '$current_metadata_path'..."
		rm "$current_metadata_path"
	else
		decho "Moving metadata file '$current_metadata_path' to '${basename}.${OUTPUT_METADATA_EXTENSION}'..."
		if [[ "$DRY_RUN" != true ]]; then
			mv "$current_metadata_path" "${basename}.${OUTPUT_METADATA_EXTENSION}"
		else
			rm "$current_metadata_path"
		fi
	fi
}


# Uses Calibre's fetch-ebook-metadata CLI tool to download metadata from
# online sources. The first parameter is the debug prefix, the second is the
# coma-separated list of allowed plugins and the rest are passed directly
# to fetch-ebook-metadata
fetch_metadata() {
	local isbn_sources
	IFS=, read -ra isbn_sources <<< "$2"

	local isbn_source="" args=()
	for isbn_source in "${isbn_sources[@]:-}"; do
		args+=("${isbn_source:+--allowed-plugin=$isbn_source}")
	done

	decho "Calling fetch-ebook-metadata --verbose" "${args[*]}" "${@:3}"
	fetch-ebook-metadata --verbose ${args[@]:+"${args[@]}"} "${@:3}" 2> >(debug_prefixer "[$1] " 0 --width=100 -s) | grep -E '[a-zA-Z()]+ +: .*'
}