date-sanitize

#!/usr/bin/env bash
set -euo pipefail

fatal() {
	printf 'date-sanitize: %s\n' "$1" >&2
	exit 1
}

usage() {
	cat <<'EOF'
Usage: date-sanitize [OPTIONS] [PATH ...]

Options:
  -r, --recursive             Recurse into directories
  -e, --ext EXT1,EXT2,...     Comma-separated extensions (default: jpg,jpeg,heic,tif,tiff,png)
      --tz ZONE               Fallback timezone for naive timestamps
      --min-year YYYY         Ignore candidates before this year (default: 2000)
      --set-create-too        Also update EXIF:CreateDate (default)
      --no-set-create-too     Skip EXIF:CreateDate updates
      --quiet                 Suppress per-file status lines
      --debug                 Print every accepted candidate
      --deep                  Include embedded/maker-note streams (slower)
      --no-deep               Disable deep scan if previously enabled
  -h, --help                  Show this help
EOF
}

trim() {
	local s="$1"
	s="${s#${s%%[!$' \t\n\r']*}}"
	s="${s%${s##*[!$' \t\n\r']} }"
	printf '%s' "$s"
}

print_status() {
	local status="$1" file="$2" from="$3" to="$4" source="$5" count="${6:-}"
	file=${file##*/}
	[[ -n $from ]] || from=NONE
	[[ -n $to ]] || to=NONE
	[[ -n $source ]] || source=NONE
	local line="$status  $file  from=$from  to=$to  source=$source"
	[[ -n $count ]] && line+="  parsed=$count"
	if [[ $status == ERROR ]]; then
		printf '%s\n' "$line" >&2
	elif (( QUIET == 0 )); then
		printf '%s\n' "$line"
	fi
}

exiftool_wrapper() {
	if [[ -n $FALLBACK_TZ ]]; then
		TZ="$FALLBACK_TZ" exiftool "$@"
	else
		exiftool "$@"
	fi
}

collect_files() {
	FILES=()
	for target in "${TARGETS[@]}"; do
		if [[ -f $target ]]; then
			has_extension "$target" && FILES+=("$target")
			continue
		fi
		if [[ -d $target ]]; then
			local depth=()
			(( RECURSIVE )) || depth=(-mindepth 1 -maxdepth 1)
			while IFS= read -r -d '' path; do
				has_extension "$path" && FILES+=("$path")
			done < <(find "$target" "${depth[@]}" -type f -print0)
			continue
		fi
		printf 'WARN  %s  reason=skipped (not a file)\n' "$target" >&2
	done
}

has_extension() {
	local file="$1"
	local ext="${file##*.}"
	[[ $file == *.* ]] || return 1
	ext=${ext,,}
	for allow in "${EXTENSIONS[@]}"; do
		[[ $ext == $allow ]] && return 0
	done
	return 1
}

min_epoch_for_year() {
	perl -MTime::Local -e 'print timegm(0,0,0,1,0,$ARGV[0])' "$1"
}

process_file() {
	local file="$1"
	local current_dto current_subsec current_epoch
	current_dto=$(exiftool_wrapper -s3 -DateTimeOriginal "$file" 2>/dev/null || true)
	current_subsec=$(exiftool_wrapper -s3 -SubSecTimeOriginal "$file" 2>/dev/null || true)
	current_epoch=$(exiftool_wrapper -s3 -DateTimeOriginal -d %s "$file" 2>/dev/null || true)

	local -a args=(
		-time:all -a -u -U -G1
		-api RequestAll=3
		-api QuickTimeUTC=1
		-api DateFormat=%s\|%Y:%m:%d\ %H:%M:%S\|%z\|%f
		-tab
	)
	(( DEEP )) && args+=(-ee -ee2 -ee3)

	local dump
	if ! dump=$(exiftool_wrapper "${args[@]}" "$file" 2>/dev/null); then
		print_status ERROR "$file" "$current_dto" NONE NONE
		return 1
	fi

	mapfile -t lines <<<"$dump"
	local -a tags epochs displays offsets subsecs
	for entry in "${lines[@]}"; do
		[[ -n $entry ]] || continue
		IFS=$'\t' read -r group tag payload <<<"$entry"
		if [[ -z $tag ]]; then
			IFS=':' read -r group tag <<<"${group:-:}"
			payload=${tag:-}
		fi
		[[ -n $group && -n $tag ]] || continue
		[[ $group == GPS* || $group == MacOS* ]] && continue
		IFS='|' read -r epoch display offset subsec <<<"${payload:-}"
		epoch=$(trim "${epoch:-}")
		[[ $epoch =~ ^[0-9]{9,}$ ]] || continue
		display=$(trim "${display:-}")
		offset=$(trim "${offset:-}")
		subsec=$(trim "${subsec:-}")
		tags+=("$group:$tag")
		epochs+=("$epoch")
		displays+=("$display")
		offsets+=("$offset")
		subsecs+=("$subsec")
		(( DEBUG )) && printf 'DEBUG %s %s\n' "$group" "$display" >&2
	done

	local parsed=${#epochs[@]}
	(( parsed )) || { print_status SKIPPED "$file" "$current_dto" NONE NONE 0; return 0; }

	local best=-1 best_epoch=0
	for i in "${!epochs[@]}"; do
		local epoch=${epochs[i]}
		(( epoch < MIN_EPOCH || epoch > MAX_EPOCH )) && continue
		if (( best == -1 || epoch < best_epoch )); then
			best=$i
			best_epoch=$epoch
		fi
	done
	(( best != -1 )) || { print_status SKIPPED "$file" "$current_dto" NONE NONE "$parsed"; return 0; }

	local winner_tag=${tags[best]}
	local winner_display=${displays[best]}
	local winner_offset=${offsets[best]}
	local winner_subsec=${subsecs[best]}

	local write_wall=${winner_display%%+(*}
	write_wall=$(trim "$write_wall")
	[[ -n $write_wall ]] || write_wall=$(perl -e 'print scalar localtime($ARGV[0])' "$best_epoch")

	local final_subsec="$winner_subsec"
	[[ -z $final_subsec ]] && final_subsec="$current_subsec"
	local final_offset="$winner_offset"
	if [[ -z $final_offset && -n $FALLBACK_TZ ]]; then
		final_offset=$(TZ="$FALLBACK_TZ" date +%z)
	fi

	local target_string="$write_wall"
	[[ -n $final_subsec ]] && target_string+=".$final_subsec"
	[[ -n $final_offset ]] && target_string+="$final_offset"

	local same=0
	if [[ $current_epoch =~ ^[0-9]+$ && $current_epoch == $best_epoch ]]; then
		local existing_subsec=$(trim "$current_subsec")
		[[ -z $winner_subsec || $winner_subsec == $existing_subsec ]] && same=1
	fi

	if (( same )); then
		print_status UNCHANGED "$file" "$current_dto" "$target_string" "$winner_tag" "$parsed"
		return 0
	fi

	local -a edits=(-P "-EXIF:DateTimeOriginal=$write_wall")
	if [[ -n $final_offset ]]; then
		edits+=("-EXIF:OffsetTimeOriginal=${final_offset:0:3}:${final_offset:3:2}")
	fi
	if (( SET_CREATE_TOO )); then
		edits+=("-EXIF:CreateDate=$write_wall")
		[[ -n $final_offset ]] && edits+=("-EXIF:OffsetTimeDigitized=${final_offset:0:3}:${final_offset:3:2}")
	fi
	[[ -n $final_subsec ]] && edits+=("-EXIF:SubSecTimeOriginal=$final_subsec")
	if (( SET_CREATE_TOO )) && [[ -n $final_subsec ]]; then
		edits+=("-EXIF:SubSecTimeDigitized=$final_subsec")
	fi

	if ! exiftool_wrapper "${edits[@]}" "$file" >/dev/null; then
		print_status ERROR "$file" "$current_dto" NONE "$winner_tag" "$parsed"
		return 1
	fi

	print_status APPLIED "$file" "$current_dto" "$target_string" "$winner_tag" "$parsed"
	return 0
}

# Defaults
RECURSIVE=0
QUIET=0
SET_CREATE_TOO=1
DEBUG=0
DEEP=0
MIN_YEAR=2000
FALLBACK_TZ=""
EXTENSIONS=(jpg jpeg heic tif tiff png)
TARGETS=()

# Args
while (($#)); do
	case "$1" in
		-r|--recursive) RECURSIVE=1 ;;
		-e|--ext)
			[[ $# -ge 2 ]] || fatal "--ext requires a value"
			IFS=',' read -r -a EXTENSIONS <<<"${2,,}"
			shift
			;;
		--tz) [[ $# -ge 2 ]] || fatal "--tz requires a value"; FALLBACK_TZ="$2"; shift ;;
		--min-year) [[ $2 =~ ^[0-9]{4}$ ]] || fatal "--min-year must be YYYY"; MIN_YEAR=$2; shift ;;
		--set-create-too) SET_CREATE_TOO=1 ;;
		--no-set-create-too) SET_CREATE_TOO=0 ;;
		--quiet) QUIET=1 ;;
		--debug) DEBUG=1 ;;
		--deep) DEEP=1 ;;
		--no-deep) DEEP=0 ;;
		-h|--help) usage; exit 0 ;;
		--) shift; break ;;
		-*) fatal "unknown option: $1" ;;
		*) TARGETS+=("$1") ;;
	esac
	shift
done

while (($#)); do
	TARGETS+=("$1")
	shift
done

(( ${#TARGETS[@]} )) || TARGETS=(".")
[[ -n $FALLBACK_TZ ]] || FALLBACK_TZ=${TZ:-}
MIN_EPOCH=$(min_epoch_for_year "$MIN_YEAR")
MAX_EPOCH=$(( $(date +%s) + 86400 ))

collect_files
(( ${#FILES[@]} )) || { (( QUIET )) || printf 'No matching files.\n'; exit 0; }

failures=0
for file in "${FILES[@]}"; do
	process_file "$file" || (( failures++ ))
done

(( failures )) && exit 2
exit 0