#!/usr/bin/perl

use strict;
use POSIX;

die "Usage: ./checkLogs.pl input_dir log_dir missed_dir chunk_size\n" 
  unless @ARGV == 4;

my $inputDir = $ARGV[0];
my $outputDir = $ARGV[1];
my $missedDir = $ARGV[2];
my $numFilesPerChunk = $ARGV[3];

my $DEBUG_ME = 1;


#count original files in inputdir, calculate how many log files

my @fileList = `ls $inputDir`;

my @logList = `ls $outputDir/*.log`;

if(1 == $DEBUG_ME)
{
  print STDERR "filelist is @fileList...\n";
  print STDERR "loglist is @logList...\n";
}


#convert logList to logHash

my %logHash;

for( my $i = 0; $i < $#logList + 1; $i++)
{
  my $file = $logList[$i];
  chomp $file;

  $logHash{$file} = 1;
}

my $numFiles = ($#fileList + 1.0);

my $expectedNumLogs =  ceil($numFiles / $numFilesPerChunk);

if(1 == $DEBUG_ME)
{
  print STDERR "Num files: $numFiles; numFilesPerChunk: $numFilesPerChunk;\n";
  print STDERR "Expected number of log files: $expectedNumLogs\n";
}

for( my $i = 0; $i < $expectedNumLogs; $i++)
{
#check file exists
#very important to coordinate log filename pattern in processAcquaintData 
#format is "subset$subsetIndex.log"
  
  my $fName = "$outputDir/subset$i.log";

  if(1 == $DEBUG_ME)
  {
    print STDERR "checking for log file $fName...\n";
  }
  
  if(1 != $logHash{$fName})
  {
    print "MISSED CHUNK $i.\n";

    &copyMissedFiles($i);
  }
  else
  {
    #scan log file for errors
    &checkLogForErrors($fName);
  }
  
}



sub copyMissedFiles
{
  my $index = $_[0];

  my $start = $index * $numFilesPerChunk;

  my $end = $start + $numFilesPerChunk;

  for( my $i = $start; $i < $end; $i++ )
  {
    my $file = $fileList[$i];

    if("" ne $file) #may be fewer files in last chunk
    {
      chomp $file;

      if(1 == $DEBUG_ME)
      {
	print STDERR "copy source name is $file...\n";
	
	print STDERR "Calling cp $inputDir/$file $missedDir...\n";
      }
      `cp $inputDir/$file $missedDir`;
    }
  }

}


sub checkLogForErrors
{
#  my $fileName = $outputDir."/".$_[0];

  my $fileName = $_[0];

  open (IN, "<$fileName") or die "Can't open $fileName: $!\n";

  my $numSentenceError = "false";
  my $emptySentenceError = "false";

  while(<IN>)
  {
    if($_ =~ m/null sentence/)
    {
      $emptySentenceError = "true";
    }
    elsif($_ =~ m/mismatched number/)
    {
      $numSentenceError = "true";
    }
  }

  if("true" eq $emptySentenceError)
  {
    print "ERROR reported in $fileName: empty sentence.\n";
  }

  if("true" eq $numSentenceError)
  {
    print "ERROR reported in $fileName:";
    print " number of sentences in vs. out doesn't match.\n";
  }

}