#!/usr/bin/perl -w
#	get_possible_epics.perl [filename]
#
# 	by: Jaideep Singh
#
#	date: 12/17/2005 
#
#	about:	reads a raw data file and looks for all possible EPICS variables
#		creates a file that contains a list of unique possible EPICS variables names and their count
#		skips over the first 500 lines of a raw data file (i don't know how this script handles a split data file...)
#		
#	EPICS info is injected into the data stream at certain times intervals.
#	More important stuff is logged more often.
#	Immediately preceding one log cycle, a timestamp is grabbed
#	Each variable is recorded in ASCII format on its own line: 
#
#	[variable name, upto 30 characters][at least one blank space][upto 20 characters for value] 
#	
#	for example:
#
#	IPM1H04B.XPOS                  -0.0408508
#	000000000111111111122222222223333333333444444444455
#	123456789012345678901234567890123456789012345678901
#
#	Variable names can have ".", ":","+","_" as well as alphanumeric characters


#getting name of file to search + ARGUMENT ERROR HANDLING
$filename = $ARGV[0];
print "Uh, what file do you want me to look at dude?: Too few arguments\n" and die unless ($filename);
print "Scared and confused: Too many arguments\n" and die unless (($#ARGV+1) < 2);
open(FILE,$filename) or print "doh! ($filename): File not found\n" and die;
print "\nSearching raw data file:\t$filename\n";

#initializing variables
$count = 0;
$line_count = 0;
$max_size_variable_name = 0;
$max_number_of_spaces   = 0;
$max_size_value         = 0;
$min_size_variable_name = 9999;
$min_number_of_spaces   = 9999;
$min_size_value         = 9999;
my %variable_count;

#searching line by line, grabbing the timestamp before and after searching
print "\nstart\t";
print scalar localtime;
print "\n";
while(<FILE>)
{
 $line_count++;
 if (/([\w\-.:+]{5,30})(\s{1,26})([\w+\-.]{1}[\w+\-.\s:]{0,19})/)
 {
  if ((length($_) < 52) && ($line_count > 500))
  { 
   chomp; 
   $count++; 
#   print "$line_count : $_\n";
   $max_size_variable_name = length($1) if (length($1) > $max_size_variable_name);
   $max_number_of_spaces   = length($2) if (length($2) > $max_number_of_spaces);
   $max_size_value         = length($3) if (length($3) > $max_size_value);
   $min_size_variable_name = length($1) if (length($1) < $min_size_variable_name);
   $min_number_of_spaces   = length($2) if (length($2) < $min_number_of_spaces);
   $min_size_value         = length($3) if (length($3) < $min_size_value);
   if ($variable_count{$1}) { $variable_count{$1}++ } else { $variable_count{$1} = 1}
  }
 }
};
close(FILE);
print "end\t";
print scalar localtime;
print "\n\nFound $count total instances of possible EPICS variables\n\n";
print "minimum size of variable name : $min_size_variable_name\n";
print "maximum size of variable name : $max_size_variable_name\n";
print "minimum number of spaces      : $min_number_of_spaces\n";
print "maximum number of spaces      : $max_number_of_spaces\n";
print "minimum size of value         : $min_size_value\n";
print "maximum size of value         : $max_size_value\n\n";

#setting up output file
@file_name_parts = split(/\./,$filename);
$file_out = $file_name_parts[0]."_possible_epics_var.txt";
print "OUTPUT filename : $file_out\n\n";
open(OUTPUT,">$file_out");
#$total_time_sec = 220;
foreach $variable_name (sort keys %variable_count) 
{ 
#	$update_time = int(100*$total_time_sec/$variable_count{$variable_name})/100;
#	print "$variable_name\t\t\t$update_time\n" if ($update_time < 60)  
#	print OUTPUT "$variable_name\t\t\t$variable_count{$variable_name}\n" if ($variable_count{$variable_name} > 2)  
	print OUTPUT "$variable_name\n" if ($variable_count{$variable_name} > 1)  
}
close(OUTPUT);