#!/usr/bin/env perl

$g_version = "1.0 20221022";
$g_auther_name = "Aikinsetsu Yuaiho";
$g_auther_url = "https://yuaiho.com/";

##### ========================================
### Load Modules
##### ----------------------------------------

# Below modules may be installed.

## Module of getting current Directory
use Cwd;

# Bellow modules may be not installed as default.
use Time::Local 'timelocal';
use TimeDate;

##### ========================================


##### ========================================
### Configurations and preference
##### ----------------------------------------

#### Preferences
### ** is Program Dfault. 

### Timezone offset [sec] between UTC/GMT and final destination server location.
### This program does not support DST/daylight saving time processing.
## ** $g_pref_tz_sec = 9*3600*(1); 
## Note: Default parameter means shipment default. Dose not mean recommended.
## Setting examples
#$g_pref_tz_sec = 9*3600*(1); #UTC+9/JST(Japan)
#$g_pref_tz_sec = 8*3600*(-1); #UTC-9/PST(USA)
#$g_pref_tz_sec = 6.5*3600*(1); #UTC+6:30/MMT(Myanmar)
$g_pref_tz_sec = 9*3600*(1); #UTC+9/JST(Japan)


### Match header
##    0: Use Received: header only. Cannot use for "Sent" / "Draft" / "Sent mesages".
##    1: Use Date: header only. 
## ** 2: Use both. Use Received: header if it's found, however use Data: header.
$g_pref_mode_header = 2;

### Parse header section only.
## This parameter available only $g_pref_mode_header is set 0 or 2.
##    0: Find time stamp across mail file.
## ** 1: Find only mailheader. (Parse stops when blank line(^$) was found.)
$g_pref_mode_parse_headeronly = 1;


#### Shell program paths
# Path of touch command. 
# ** "/usr/bin/touch -d"
# Note: Default value is FreeBSD default. -d option is required.
$g_path_prog_touch = "/usr/bin/touch -d";

# Path of echo command
# ** "/bin/echo"
# Note: Default value is FreeBSD default.
$g_path_prog_echo = "/bin/echo";


##### ========================================

##### ========================================
### how to use
##### ----------------------------------------
# This program expect that user inputs mail file list.
# Notice: Add 
#
# Run sample
# [user@host ~/]$ cat <mailfile list> | <thisprogram>
# [user@host mailboxdir]$ find ./cur -type f | <thisprogram>
##### ========================================

##### ========================================
### Arguments routines
##### ========================================

@args = @ARGV;
for ($i=0; $i<=$#ARGV; $i++){
	if ($args[$i] eq "--debug"){
		$g_mode_debug = 1;
	}elsif($args[$i] eq "--commit"){
		$g_mode_commit = 1;
	}elsif($args[$i] eq "--noinfo"){
		$g_mode_noinfo = 1;
	}elsif($args[$i] eq "--absolute"){
		$g_mode_absolute = 1;
	}elsif( ($args[$i] eq "--help") || ($args[$i] eq "--usage") || ($args[$i] eq "-h") ){
		&show_usage;
		exit;
	}else{
		print STDERR "ERROR: Argument \"$args[$i]\" is not implemented.\n";
		exit;
	}
}


## English month to numberical month table
%MMchr2num = (
	Jan => 1,
	Feb => 2,
	Mar => 3,
	Apr => 4,
	May => 5,
	Jun => 6,
	Jul => 7,
	Aug => 8,
	Sep => 9,
	Oct => 10,
	Nov => 11,
	Dec => 12);

## Get Current Dir
$g_dir_cur = Cwd::getcwd();

## Get filelist from Standerd In.
@files = <STDIN>;

foreach $file (@files){
	
	chomp($file);
	my $mailfile;
	if($g_mode_absolute){
		$mailfile = "$file";
	} else {
		$mailfile = "${g_dir_cur}/$file";
	}
	
	## Initialize flags and parameters
	$flag_valid_header = undef;
	$flag_received_found = undef;
	$YYYY = undef;
	$MM = undef;
	$DD = undef;
	$hhmmss = undef;
	$offset = undef;
	
	## Open mail file with read only mode
	if( -f $mailfile){
		open(IN, "<", "$mailfile");
	} elsif (-d $mailfile){
		next;
	} else {
		print STDERR "File ${mailfile} cannot be opened.\n";
	}

	## Large files consume a lot of memory.
	## If you know that in advance, consider the following loop method.
	## foreach $line (@lines) { => while ($line = <IN>) {
	## Memo: If you choose the above loop methods, do not forget the following...
	## close(<IN>) moves end of loop. Delete @lines = <IN>;
	@lines = <IN>;
	close(IN);

	## Start file execute
	&print_info("-----\n");
	&print_info("INFO: Executing file \"$mailfile\"\n");
	
	foreach $line (@lines) {

		if ($g_pref_mode_parse_headeronly){
			if($line =~ /^$/){
				last;
			}
		}

		if ( ($g_pref_mode_header == 0) || ($g_pref_mode_header == 2) ){
			if($line =~ /^Received:/){
				$flag_received_found = 1;
			}
		}
		
		my $check_timestamp = &extract_timestamp($line);
		if ( ($check_timestamp == undef ) || ($check_timestamp == 0) ){
			next;
		}else{
			## This file has valid header
			$flag_valid_header = 1;

			## Crop Unnecessary strings.
			if ($line =~ /^Date:.+/){
				## Date header
				$input = (split(/Date:/,$line))[1];
			} elsif ($line =~ /;/){
				## Timestamp in Received header usually include ";". 
				$input = (split(/;/,$line))[1];
			} else {
				$input = $line;
			}

			if ($g_mode_debug){
				print STDERR "DEBUG: \$line  $line";
				print STDERR "DEBUG: \$input $input\n";
			}

			chomp($input);

			### Start Parse

			## Remove Week of day
			if($input =~ /(Mon|Tue|Wed|Th[ur]|Fri|Sat|Sun)/){
				$input =~ s/$1//eg;
			}

			## Extract Year
			if($input =~ /([1-9][0-9]{3})/){
				$YYYY = $1;
				$input =~ s/$1//eg;
			}
	
			## Extract timestamp of hh:mm:ss
			if($input =~ /([0-2][0-9]:[0-5][0-9]:[0-5][0-9])/){
				$hhmmss = $1;
				$input =~ s/$1//eg;
			}
			
			## Extract month.
			if($input =~ /([JFMASOND][aepuco][nbrylgptvc])/){
				$MM = $MMchr2num{"$1"};
				$input =~ s/$1//eg;
			}

			## Extract Timezone(Hours difference between UTC and localtime)
			if($input =~ /([+-][01][0-9][0-5][035])/){
				$header_tz_diff = $1;
				$input =~ tr/$1//;
			}

			## Extract day of month
			if($input =~ /(0?[0-9]|[1-2][0-9]|3[0-1])/){
				$DD = $1;
				$input =~ s/$1//eg;
			}
			
			## Get UNIX epoch sec from header timestamp
			my $header_year = $YYYY - 1900;
			my $header_month = $MM - 1;
			my $header_day = $DD;
			my ($header_hour, $header_min, $header_sec) = split(/:/, $hhmmss);
			my $header_epoch_sec_local = timelocal($header_sec, $header_min, $header_hour, $header_day, $header_month, $header_year);
			my $header_timestamp = sprintf("%04d-%02d-%02d %s%s",$YYYY, $MM, $DD, $hhmmss, $header_tz_diff);

			## Get hour and minutes from TZ value "+/-hhmm" (Differnce between UTC and Localtime)
			my $header_tz_diff_symbol = substr($header_tz_diff, 0, 1); # + in +0900
			my $header_tz_diff_hours = substr($header_tz_diff, 1, 2); # 09 in +0900
			my $header_tz_diff_minutes = substr($header_tz_diff, 3, 2); # 00 in +0900

			
			## Invert TZ offset symbol change
			if( ($header_tz_diff_symbol eq "-" ) ){
				$header_tz_diff_symbol = +1;
			} else {
				$header_tz_diff_symbol = -1;
			}
			
			## Get TZ seconds difference of between UTC and localtime.
			$header_tz_diff_sec = $header_tz_diff_symbol * $header_tz_diff_hours * 3600 + $header_tz_diff_minutes * 60;

			## Get UTC from header time stamp
			$header_epoch_sec_utc = $header_epoch_sec_local + $header_tz_diff_sec;

			## Final destination arrvied time
			$arrived_epoch_sec = $header_epoch_sec_utc + $g_pref_tz_sec;

			if($g_mode_debug){
				## Header local time
				my ($s, $mi, $h, $d, $mo, $y) = localtime($header_epoch_sec_local);
				$y += 1900;
				$mo++;
				$test_reconv_local = sprintf("%04d-%02d-%02d %02d:%02d:%02d", $y, $mo, $d, $h, $mi, $s);

				## Header UTC time
				my ($s, $mi, $h, $d, $mo, $y) = localtime($header_epoch_sec_utc);
				$y += 1900;
				$mo++;
				$test_reconv_utc = sprintf("%04d-%02d-%02d %02d:%02d:%02d", $y, $mo, $d, $h, $mi, $s);
			}

			## Final destination arrived local time
			my ($s, $mi, $h, $d, $mo, $y) = localtime($arrived_epoch_sec);
			$y += 1900;
			$mo++;
			$test_reconv_arrival = sprintf("%04d-%02d-%02d %02d:%02d:%02d", $y, $mo, $d, $h, $mi, $s);

			## Make arguments of touch -d command.
			$touch_mod_timestamp = sprintf("%04d-%02d-%02dT%02d:%02d:%02d", $y, $mo, $d, $h, $mi, $s);

			## Output
	
			if($g_mode_debug){
				print "DEBUG: LocalTime(Header_Read):A=$header_timestamp\n";
				print "DEBUG: Localtime_Epoch(Header_Read):B=timelocal(A)=${header_epoch_sec_local} localtime(timelocal(B)): $test_reconv_local Inverted_TZ_sec:${header_tz_diff_sec}\n";
				print "DEBUG: UTC_Epoch(Header_Read):C=B+Inverted_TZ_sec=${header_epoch_sec_local}+(${header_tz_diff_sec})=${header_epoch_sec_utc} UTC(Header_Read)=timelocal(C):D=${test_reconv_utc}\n";
				print "DEBUG: LocalTime_Epoch(Final_Dest):D=(C)+\$g_pref_tz_sec=${header_epoch_sec_utc}+(${g_pref_tz_sec})=$arrived_epoch_sec LocalTime(Final_Dest)=timelocal(D):$test_reconv_arrival\n";
			} else {
				&print_info("INFO: Header read: $header_timestamp -> Final destinaiton time: $touch_mod_timestamp \n");
			}

			## Make shell command
			$exec_shell_touch = "$g_path_prog_touch ${touch_mod_timestamp} \"${mailfile}\" >> /dev/null";

			if ($g_mode_commit){
				my $exec_shell_touch_return = `$exec_shell_touch`;
				my $exec_shell_touch_reuslt = `$g_path_prog_echo $?`;
				chomp($exec_shell_touch_reuslt);
				if ($exec_shell_touch_reuslt == 0){
					&print_info("INFO: Touch success : \"${mailfile}\"\n");
				} else {
					print "ERROR: Touch failed  : \"${mailfile}\"\n";
				}
			} else {
				print "$exec_shell_touch\n";
			}

			### Last: Matched
			last;
		}
	}

	## After reached EOF
	if( $flag_valid_header eq undef ){
		print "ERROR: File \"${mailfile}\" has no valid hedear.\n";
	}
	&print_info("-----\n");

}

sub print_info {
	my $msg = $_[0];
	if (!$g_mode_noinfo){
		print $msg;
	}
}

sub print_debug {
	my $msg = $_[0];
	if ($g_mode_debug){
		print $msg;
	}
}

sub show_usage {
	print <<"EOSU";
Timstamp parser for touch command application.
Version: ${g_version}

<Show usage>
\$ ./dateparse.pl [-h|--usage|--help]
-h		Show usage
--usage		Show usage
--hlep		Show usage

<Execute mail files>
\$ cat <Filelist> | ./dateparse.pl [--debug] [--commit] [--noinfo] [--absolute]
\$ find ./ --type f | ./dateparse.pl [--debug] [--commit] [--noinfo] [--absolute]
--dubug		Show DEBUG: messages.
--commit	Run touch -d command internally to change file timestamp. If omit, show command only.
--noinfo	Don't show INFO: messages.
--absolute	<Filelist> path is written Full path. If omit, add current directory path to <Filelist> as prefix.

Copyright 2022 ${g_auther_name}
${g_auther_url}
EOSU
}

sub extract_timestamp {
	my $l = $_[0];
	if ($g_pref_mode_header == 0) {
		## Received: header timestamp match
		if ( ($l !~ /^Date:.+/) && ($l =~ /[1-9][0-9]{3}/) && ($l =~ /[0-2][0-9]:[0-5][0-9]:[0-5][0-9]/) ){
			if ($flag_received_found){
				return 1;
			} else {
				return 0;
			}
		} else {
			return 0;
		}
	} elsif ($g_pref_mode_header == 1 ) {
		## Date: header timestamp match
		if ($l =~ /^Date:.+/){
			return 1;
		} else {
			return 0;
		}
	} elsif ($g_pref_mode_header == 2) {
		## Use both.
		if ( ($l !~ /^Date:.+/) && ($l =~ /[1-9][0-9]{3}/) && ($l =~ /[0-2][0-9]:[0-5][0-9]:[0-5][0-9]/) ){
			if ($flag_received_found){
				return 1;
			} else {
				return 0;
			}
		} elsif ($l =~ /^Date:.+/) {
			return 1;
		} else {
			return 0;
		}
	} else {
		## Not supported parameter
		print STDERR "\$g_pref_mode_header = $g_pref_mode_header is not supported\n";
		exit;
	}
}