#!/usr/bin/perl

#/media/yiguanas/Sam3/ubuntuSW/MT4/tester/history/AJnicify.pl

#INPUT:
#USDJPY.txt: (is the original, SORTED file as extracted from the zip archive)
#<TICKER>,<DTYYYYMMDD>,<TIME>,<OPEN>,<HIGH>,<LOW>,<CLOSE>,<VOL>
#USDJPY,20010102,230300,114.43,114.43,114.43,114.43,4
#USDJPY,20010102,230400,114.44,114.44,114.44,114.44,4
#USDJPY,20010102,230500,114.44,114.44,114.44,114.44,4

#OUTPUT format: remove header, symbol and comma; add hyphens and colon; remove 00
#2001-11-09,20:19,120.29,120.30,120.29,120.30,4
#2001-11-09,20:20,120.30,120.30,120.30,120.30,4
#2001-11-09,20:21,120.30,120.31,120.30,120.31,4
#2001-11-09,20:22,120.31,120.31,120.31,120.31,4
#2001-11-09,20:23,120.31,120.31,120.30,120.30,4

#TO EXECUTE THIS PROGRAM:
# $ cat ./USDJPY.txt | perl ./AJnicify.pl > xxx.txt #OR:
# $ cat ./USDJPY.txt | perl ./AJnicify.pl 20011109,2019 20011109,2022
#2001-11-09,20:20,120.30,120.30,120.30,120.30,4
#2001-11-09,20:21,120.30,120.31,120.30,120.31,4
#2001-11-09,20:22,120.31,120.31,120.31,120.31,4

#XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

sub no_args_sub {

   <>;    #first line (=unwanted header) gets lost intentionally

   while ( defined($_ = <>) )    #reads from stdin (& theoretically also command line args)
   {
      # s/USDJPY,//;      #works
      s/([A-Z].....,)(.*)/2/;    #remove first 6 capital letters
      s/([0-9]...)([0-9].)([0-9].),([0-9].)([0-9].)([0-9].)/1-2-3,4:5/;
      print $_;
   } #while

} #sub no_args_sub

#XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

sub two_args_sub {
   open STDERR, '>/dev/null';    #naughty!

#my distinction of stdin / command line args isn't perfect. Want to avoid error msgs:
#Can't open 20011109,2019: No such file or directory at ./AJnicify.pl line 69.
#Can't open 20011109,2022: No such file or directory at ./AJnicify.pl line 69.

   if (!($firstarg le $secondarg)) {
      die "1st arg must be < 2nd arg and of form 20010102,2346n";
   }

   <>;    #get command line out of the way of the stream
   <>;    #first line (=unwanted header) gets lost intentionally

   my $continue = 1;
   do # I use DO UNTIL instead of usual WHILE to get that extra match at the start
   {
      s/.{6},//;    #works
      s/d+,d+(?=.*)/$&/;
      if ($& ge $firstarg) { $continue = 0; }
   } until (($continue == 0) || (!defined($_ = <>)));

   $continue = 1;
   while ( (defined($_ = <>)) && ($continue == 1) )
   #: reads from stdin (& theoretically also command line args)
   {
      my $tmp = $_;
      $tmp =~ s/d+,d+(?=.*)/$&/;
      if ($& gt $secondarg) { $continue = 0; }
      s/([A-Z].....,)(.*)/2/;    #remove first 6 capital letters
      s/([0-9]...)([0-9].)([0-9].),([0-9].)([0-9].)([0-9].)/1-2-3,4:5/;
      print $_;
   } # while

   close STDERR;

} # two_args_sub

#XXXXXXXXXXXXXXXXXXXXXXX "MAIN" XXXXXXXXXXXXXXXXXXXXXXX

my $CL_len = $#ARGV + 1;    #GLOBAL VARS; must do this BEFORE <>

$firstarg  = $ARGV[0];      #print "firstarg=$firstargn";
$secondarg = $ARGV[1];      #print "secondarg=$secondargn";

if    ($CL_len == 2) { two_args_sub; }
elsif ($CL_len == 0) { no_args_sub; }
else {
   die "I quit - I needed no args or 2 args of form: 20010102,2346 (=2nd Jan 2001,23:46)n";
}