[ViewVC] Annotation of: cvsroot/UserCode/OSUT3Analysis/DBTools/scripts/osudb

#!/usr/bin/env perl

use strict;
use Mysql;
use Getopt::Long;
use File::Copy;

sub getDataset;
sub dbUpdate;
sub dbUpdateOther;
sub dbDelete;
sub dbDeleteOther;
sub addSlashes;
sub uploadRelease;
sub downloadRelease;
sub datasetExists;

our $db = Mysql->connect ("cmshead.mps.ohio-state.edu", "ntuple", "osuT3User") or die "Failed to connect to Tier 3, stopped";

my %opt;
Getopt::Long::Configure ("bundling");
GetOptions (\%opt, "comment|c=s", "format|f=s", "pyConfig|p=s", "crabCfg|b=s", "jsonFile|j=s", "lumiSummary|s=s", "location|l=s", "fileList|t=s", "release|r=s", "other|o", "help|h");
my $argc = @ARGV;

printHelp ($ARGV[0]) if $opt{"help"};
printHelp () if $argc != 2 && $ARGV[0] ne "createOther";
printHelp () if $ARGV[0] ne "create" && $ARGV[0] ne "createOther" && $ARGV[0] ne "update" && $ARGV[0] ne "finish" && $ARGV[0] ne "deprecate" && $ARGV[0] ne "uploadRelease" && $ARGV[0] ne "downloadRelease" && $ARGV[0] ne "deleteEntry";
if (($ARGV[0] eq "create" || $ARGV[0] eq "finish") && !$opt{"location"})
  {
    print "The directory containing the ntuples must be given!\n";
    exit;
  }
if ($ARGV[0] eq "uploadRelease" && !$opt{"pyConfig"} && !$opt{"release"})
  {
    print "Both the Python and the CMSSW release must be given!\n";
    exit;
  }
if ($ARGV[0] eq "uploadRelease")
  {
    my $parentDir = "./$opt{'release'}";
    $parentDir =~ s/CMSSW_[^\/]*//g;
    $parentDir =~ s/\/\/*/\//g;
    my $cmsswRelease = $opt{'release'};
    $cmsswRelease =~ s/^.*CMSSW_([^\/]*).*$/CMSSW_$1/;
    `tar -C $parentDir -czf $ARGV[1].tar.gz $opt{"release"}`;
    $opt{"format"} = "BEAN" if !$opt{"format"};
    uploadRelease ($opt{"format"}, $cmsswRelease, "$ARGV[1].tar.gz", $opt{"pyConfig"}, $ARGV[1], "$ENV{'USER'}\@$ENV{'HOSTNAME'}", $opt{"comment"});
  }
elsif ($ARGV[0] eq "downloadRelease")
  {
    downloadRelease ($ARGV[1], $opt{"pyConfig"});
  }
elsif ($ARGV[0] eq "createOther")
  {
    my @listOfFiles = @ARGV;
    @listOfFiles = reverse (@listOfFiles);
    pop (@listOfFiles);
    @listOfFiles = reverse (@listOfFiles);
    dbOtherCreate ("$ENV{'USER'}\@$ENV{'HOSTNAME'}", \@listOfFiles, $opt{"comment"});
  }
else
  {
    my $id = -1;
    my $fullDataset;
    ($id, $fullDataset) = getDataset ($ARGV[0], $ARGV[1]) if $ARGV[0] ne "create" && ($ARGV[0] ne "update" || !$opt{"other"}) && ($ARGV[0] ne "deleteEntry" || !$opt{"other"});
    $id = $ARGV[1] if ($ARGV[0] eq "update" && $opt{"other"}) || ($ARGV[0] eq "deleteEntry" && $opt{"other"});
    $fullDataset = $ARGV[1] if $id < 0;
    $ARGV[0] = "create" if $id < 0;
    my $status = "present";
    $status = "" if $ARGV[0] eq "update";
    $status = "deprecated" if $ARGV[0] eq "deprecate";
    $opt{"format"} = "BEAN" if $ARGV[0] eq "create" && !$opt{"format"};
    dbUpdate ($id, $fullDataset, "$ENV{'USER'}\@$ENV{'HOSTNAME'}", $opt{"format"}, $opt{"location"}, $opt{"fileList"}, $status, $opt{"comment"}, $opt{"pyConfig"}, $opt{"crabCfg"}, $opt{"jsonFile"}, $opt{"lumiSummary"}, $opt{"release"}) if !$opt{"other"};
    dbUpdateOther ($id, $opt{"comment"}, "$ENV{'USER'}\@$ENV{'HOSTNAME'}") if $opt{"other"};
    dbDelete ($id) if $ARGV[0] eq "deleteEntry" && !$opt{"other"};
    dbDeleteOther ($id) if $ARGV[0] eq "deleteEntry" && $opt{"other"};
  }

sub
addSlashes
{
  my $string = shift;

  $string =~ s/\\/\\\\/g;
  $string =~ s/'/\\'/g;
  $string =~ s/"/\\"/g;
  $string =~ s/\\0/\\\\0/g;

  return $string;
}

sub
getDataset
{
  my $command = shift;
  my $dataset = shift;

  my $results;
  my $queryDataset = $dataset;
  $queryDataset =~ s/\*/%/g;
  $queryDataset =~ s/(.*)/%$1%/g;
  my $query = "select id,dataset,user,creationTime from ntuple where dataset like '$queryDataset' order by creationTime";
  $db->selectdb ("ntuple");
  $results = $db->query ($query);
  if ($results->numrows () == 1)
    {
      my @row = $results->fetchrow ();
      return ($row[0], $row[1]);
    }
  if ($results->numrows () == 0)
    {
      if ($command ne "deleteEntry" && $command ne "deprecate")
        {
          print "Database entry does not exist. Create it? (Y/n): ";
          my $response = <STDIN>;
          $response =~ s/\n//g;
          $response = "y" if !$response;
          exit if substr (lc ($response), 0, 1) ne 'y';
          return (-1, "");
        }
      else
        {
          print "Database entry does not exist.\n";
          exit;
        }
    }
  my %id;
  my %fullDataset;
  print "Found multiple database entries matching\n";
  print "\"$dataset\":\n";
  print "( 0) new\n" if $command ne "deleteEntry" && $command ne "deprecate";
  for (my $i = 1; $i <= $results->numrows (); $i++)
    {
      my @row = $results->fetchrow ();
      $id{"$i"} = $row[0];
      $fullDataset{"$i"} = $row[1];
      printf "(%2d) $row[1]\n", $i;
      print "     created by $row[2] on $row[3]\n";
    }
  print "\nWhich entry would you like to modify?";
  if ($command ne "deleteEntry" && $command ne "deprecate")
    {
      print " (Select 0 to create a new entry): "
    }
  else
    {
      print ": ";
    }
  my $response = <STDIN>;
  $response =~ s/[ \t\n]//g;
  return (-1, "") if $response == 0 && $command ne "deleteEntry" && $command ne "deprecate";
  if (!(exists $id{$response}))
    {
      print "Your selection \"$response\" was not a valid option! Quitting.\n";
      exit;
    }

  return ($id{$response}, $fullDataset{$response});
}

sub
dbUpdate
{
  my $id = shift;
  my $dataset = shift;
  my $user = shift;
  my $format = shift;
  my $location = shift;
  my $fileListName = shift;
  my $status = shift;
  my $comment = shift;
  my $psetName = shift;
  my $crabCfgName = shift;
  my $jsonFileName = shift;
  my $lumiSummaryName = shift;
  my $release = shift;

  my $fileList;
  my $pset;
  my $crabCfg;
  my $jsonFile;
  my $lumiSummary;
  my $size = 0;
  my $nFiles = 0;
  if ($fileListName)
    {
      if (!(-e $fileListName))
        {
          print "$fileListName does not exist!\n";
          exit;
        }
      open (FILE_LIST, "<$fileListName");
      my @fileList = <FILE_LIST>;
      close (FILE_LIST);
      $fileList = join ("", @fileList);
    }
  if ($psetName)
    {
      if (!(-e $psetName))
        {
          print "$psetName does not exist!\n";
          exit;
        }
      open (PY_CONFIG, "<$psetName");
      my @pset = <PY_CONFIG>;
      close (PY_CONFIG);
      $pset = join ("", @pset);
    }
  if ($crabCfgName)
    {
      if (!(-e $crabCfgName))
        {
          print "$crabCfgName does not exist!\n";
          exit;
        }
      open (CRAB_CONFIG, "<$crabCfgName");
      my @crabCfg = <CRAB_CONFIG>;
      close (CRAB_CONFIG);
      $crabCfg = join ("", @crabCfg);
    }
  if ($jsonFileName)
    {
      if (!(-e $jsonFileName))
        {
          print "$jsonFileName does not exist!\n";
          exit;
        }
      open (JSON_FILE, "<$jsonFileName");
      my @jsonFile = <JSON_FILE>;
      close (JSON_FILE);
      $jsonFile = join ("", @jsonFile);
    }
  if ($lumiSummaryName)
    {
      if (!(-e $lumiSummaryName))
        {
          print "$lumiSummaryName does not exist!\n";
          exit;
        }
      open (LUMI_SUMMARY, "<$lumiSummaryName");
      my @lumiSummary = <LUMI_SUMMARY>;
      close (LUMI_SUMMARY);
      $lumiSummary = join ("", @lumiSummary);
    }
  if ($release)
    {
      my $query = "select id from ntupleRelease where name='$release'";
      $db->selectdb ("ntuple");
      my $results = $db->query ($query);
      if ($results->numrows () != 1)
        {
          print "Ntuple release \"$release\" not found!\n";
          exit;
        }
    }
  ($nFiles, $size) = sizeOfDataset ($dataset, $location, $fileList);
  my $fullLocation = $location;
  $fullLocation = "$ENV{'PWD'}/$location" if !($location =~ m/^\//);

  $dataset = addSlashes ($dataset);
  $user = addSlashes ($user);
  $format = addSlashes ($format);
  $pset = addSlashes ($pset);
  $crabCfg = addSlashes ($crabCfg);
  $jsonFile = addSlashes ($jsonFile);
  $lumiSummary = addSlashes ($lumiSummary);
  $fullLocation = addSlashes ($fullLocation);
  $fileList = addSlashes ($fileList);
  $status = addSlashes ($status);
  $comment = addSlashes ($comment);
  $release = addSlashes ($release);

  my $results;
  my $query;
  if ($id < 0)
    {
      $query = "select max(id) from ntuple";
      $db->selectdb ("ntuple");
      $results = $db->query ($query);
      my @row = $results->fetchrow ();
      my $id = 1;
      $id = $row[0] + 1 if $results->numrows ();

      $query = "insert into ntuple (id, dataset, creationTime, lastUpdateTime, lastUpdateUser, user, format, location, fileList, nFiles, sizeInGB, status, comment, pset, crabCfg, jsonFile, lumiSummary, version) values ($id, '$dataset', now(), now(), '$user', '$user', '$format', '$fullLocation', '$fileList', $nFiles, $size, '$status', '$comment', '$pset', '$crabCfg', '$jsonFile', '$lumiSummary', '$release')";
    }
  if ($id > 0)
    {
      my $fields;
      my $values;

      $values .= ", dataset='$dataset'" if $dataset;
      $values .= ", lastUpdateTime=now()";
      $values .= ", lastUpdateUser='$user'";
      $values .= ", format='$format'" if $format;
      $values .= ", location='$fullLocation'" if $location;
      $values .= ", fileList='$fileList'" if $fileList;
      $values .= ", nFiles=$nFiles" if $nFiles;
      $values .= ", sizeInGB=$size" if $size;
      $values .= ", status='$status'" if $status;
      $values .= ", comment='$comment'" if $comment;
      $values .= ", pset='$pset'" if $pset;
      $values .= ", crabCfg='$crabCfg'" if $crabCfg;
      $values .= ", jsonFile='$jsonFile'" if $jsonFile;
      $values .= ", lumiSummary='$lumiSummary'" if $lumiSummary;
      $values .= ", version='$release'" if $release;

      $values =~ s/^, //;
      $query = "update ntuple set $values where id=$id";
    }
  $results = $db->query ($query);

  return $results;
}

sub
dbUpdateOther
{
  my $id = shift;
  my $comment = shift;
  my $user = shift;

  $comment = addSlashes ($comment);

  my $query = "update other set lastUpdateTime=now(), lastUpdateUser='$user', comment='$comment'  where id=$id";
  $db->selectdb ("ntuple");
  my $results = $db->query ($query);

  return $results;
}

sub
dbDelete
{
  my $id = shift;

  my $query = "delete from ntuple where id=$id";
  $db->selectdb ("ntuple");
  my $results = $db->query ($query);
}

sub
dbDeleteOther
{
  my $id = shift;

  my $query = "delete from other where id=$id";
  $db->selectdb ("ntuple");
  my $results = $db->query ($query);
}

sub
printHelp
{
  my $command = shift;

  my $exeName = $0;
  $exeName =~ s/^.*\/([^\/]*)$/$1/;

  if ($command eq "create")
    {
      print "Usage: $exeName -l DIRECTORY [OPTION]... create NAME\n";
      print "Creates an entry in the database for dataset NAME. The ntuples are assumed to\n";
      print "be present on the Tier 3, with the status being set accordingly.\n";
      print "\n";
      print "Mandatory arguments to long options are mandatory for short options too.\n";
      printf "%-29s%s\n", "  -b, --crabCfg FILE", "CRAB config used to submit ntuple jobs";
      printf "%-29s%s\n", "  -c, --comment COMMENT", "comment for the database entry";
      printf "%-29s%s\n", "  -f, --format FORMAT", "ntuple format (default: BEAN)";
      printf "%-29s%s\n", "  -j, --jsonFile FILE", "JSON file used for this dataset";
      printf "%-29s%s\n", "  -l, --location DIRECTORY", "directory containing the ntuples";
      printf "%-29s%s\n", "  -p, --pyConfig FILE", "Python config used to produce ntuples";
      printf "%-29s%s\n", "  -r, --release NAME", "ntuple release used to produce ntuples";
      printf "%-29s%s\n", "  -s, --lumiSummary FILE", "lumiSummary.json reported by CRAB";
    }
  elsif ($command eq "createOther")
    {
      print "Usage: $exeName [OPTION]... createOther DIRECTORIES_AND_FILES\n";
      print "Creates an entry in the database for non-ntuple data.\n";
      print "\n";
      print "Mandatory arguments to long options are mandatory for short options too.\n";
      printf "%-29s%s\n", "  -c, --comment COMMENT", "comment for the database entry";
    }
  elsif ($command eq "update")
    {
      print "Usage: $exeName [OPTION]... update NAME\n";
      print "Updates an existing database entry for dataset NAME.\n";
      print "\n";
      print "Mandatory arguments to long options are mandatory for short options too.\n";
      printf "%-29s%s\n", "  -b, --crabCfg FILE", "CRAB config used to submit ntuple jobs";
      printf "%-29s%s\n", "  -c, --comment COMMENT", "comment for the database entry";
      printf "%-29s%s\n", "  -f, --format FORMAT", "ntuple format";
      printf "%-29s%s\n", "  -j, --jsonFile FILE", "JSON file used for this dataset";
      printf "%-29s%s\n", "  -l, --location DIRECTORY", "directory containing the ntuples";
      printf "%-29s%s\n", "  -o, --other", "update an entry in the non-ntuple database";
      printf "%-29s%s\n", "  -p, --pyConfig FILE", "Python config used to produce ntuples";
      printf "%-29s%s\n", "  -r, --release NAME", "ntuple release used to produce ntuples";
      printf "%-29s%s\n", "  -s, --lumiSummary FILE", "lumiSummary.json reported by CRAB";
    }
  elsif ($command eq "finish")
    {
      print "Usage: $exeName -l DIRECTORY finish NAME\n";

      print "Finalizes the database entry for dataset NAME, changing its status to\n";
      print "\"present\".  This is intended to be the final step in command-line based ntuple\n";
      print "production.\n";
      print "\n";
      print "Mandatory arguments to long options are mandatory for short options too.\n";
      printf "%-29s%s\n", "  -l, --location DIRECTORY", "directory containing the ntuples";
    }
  elsif ($command eq "deleteEntry")
    {
      print "Usage: $exeName [OPTION]... deleteEntry NAME\n";
      print "Deletes the database entry for dataset NAME. This is intended primarily for\n";
      print "mistaken database entries. If you wish to actually delete a set of ntuples,\n";
      print "please use the \"deprecate\" command instead.\n";
      print "\n";
      printf "%-29s%s\n", "  -o, --other", "update an entry in the non-ntuple database";
    }
  elsif ($command eq "deprecate")
    {
      print "Usage: $exeName deprecate NAME\n";
      print "Marks the dataset NAME for deletion. WARNING: The dataset will be deleted from\n";
      print "the Tier 3 within one week of this action. If you wish to simply delete a\n";
      print "mistaken database entry, use the \"deleteEntry\" command instead.\n";
    }
  elsif ($command eq "uploadRelease")
    {
      print "Usage: $exeName -p FILE -r DIRECTORY [OPTION]... uploadRelease NAME\n";
      print "Copies an ntuple release to the appropriate area on the Tier 3, and creates a\n";
      print "database entry for it, with NAME being the name of the release. An ntuple\n";
      print "release is a CMSSW release with all the necessary packages added for creating\n";
      print "ntuples, along with a default Python config file.\n";
      print "\n";
      print "Mandatory arguments to long options are mandatory for short options too.\n";
      printf "%-29s%s\n", "  -c, --comment COMMENT", "comment for the database entry";
      printf "%-29s%s\n", "  -f, --format FORMAT", "ntuple format (default: BEAN)";
      printf "%-29s%s\n", "  -p, --pyConfig FILE", "default Python config for the release";
      printf "%-29s%s\n", "  -r, --release DIRECTORY", "CMSSW release containing ntuple packages";
    }
  elsif ($command eq "downloadRelease")
    {
      print "Usage: $exeName [OPTION]... downloadRelease NAME\n";
      print "Copies an ntuple release to the current directory on the Tier 3. Optionally\n";
      print "copies the corresponding Python configuration file registered in the database.\n";
      print "\n";
      print "Mandatory arguments to long options are mandatory for short options too.\n";
      printf "%-29s%s\n", "  -p, --pyConfig FILE", "Python configuration file name";
    }
  else
    {
      print "Usage: $exeName [OPTION]... COMMAND NAME\n";
      print "Manipulates entry in the OSU Tier 3 ntuple database given by NAME.\n";
      print "\n";
      print "Mandatory arguments to long options are mandatory for short options too.\n";
      printf "%-29s%s\n", "  -h, --help", "print help. If COMMAND is present, print help";
      printf "%-29s%s\n", " ", "specific to COMMAND.";
      print "\n";
      print "COMMAND may be one of the following:\n";
      printf "%-29s%s\n", "  create", "creates the entry";
      printf "%-29s%s\n", "  createOther", "creates an entry for non-ntuple data";
      printf "%-29s%s\n", "  update", "updates the entry";
      printf "%-29s%s\n", "  finish", "finalizes the database entry";
      printf "%-29s%s\n", "  deleteEntry", "removes the database entry";
      printf "%-29s%s\n", "  deprecate", "marks the dataset for deletion";
      printf "%-29s%s\n", "  uploadRelease", "upload an ntuple release";
      printf "%-29s%s\n", "  downloadRelease", "download an ntuple release";
    }

  exit;
}

sub
sizeOfDataset
{
  my $dataset = shift;
  my $location = shift;
  my $fileList = shift;

  my $size = 0.0;
  my $nFiles = 0;
  if ($location && !$fileList)
    {
      if (!(-e $location))
        {
          print "$location does not exist!\n";
          exit;
        }
      $nFiles = `ls $location | wc -l`;
      $size = `du -s $location`;
      $size =~ s/([^ ]*) .*/$1/;
      $size /= 1024 * 1024;
      $size = sprintf "%.2f", $size;
    }
  elsif ($fileList)
    {
      foreach my $file (split (/\n/, $fileList))
        {
          $nFiles++;
          my $fileSize = `du -s $file`;
          $fileSize =~ s/([^ ]*) .*/$1/;
          $size += $fileSize;
        }
      $size /= 1024 * 1024;
      $size = sprintf "%.2f", $size;
    }

  return ($nFiles, $size);
}

sub
uploadRelease
{
  my $format = shift;
  my $cmsswRelease = shift;
  my $release = shift;
  my $psetName = shift;
  my $name = shift;
  my $user = shift;
  my $comment = shift;

  move ($release, "/home/hart/public_html/releases/$name.tar.gz") or die "Ntuple releases may only be uploaded on the Tier 3, stopped";
  if (!(-e $psetName))
    {
      print "$psetName does not exist!\n";
      exit;
    }
  open (PY_CONFIG, "<$psetName");
  my @pset = <PY_CONFIG>;
  close (PY_CONFIG);
  my $pset = join ("", @pset);

  my $query = "select max(id) from ntupleRelease";
  $db->selectdb ("ntuple");
  my $results = $db->query ($query);
  my @row = $results->fetchrow ();
  my $id = 1;
  $id = $row[0] + 1 if $results->numrows ();

  $name = addSlashes ($name);
  $pset = addSlashes ($pset);
  $user = addSlashes ($user);
  $format = addSlashes ($format);
  $cmsswRelease = addSlashes ($cmsswRelease);
  $comment = addSlashes ($comment);

  my $query = "insert into ntupleRelease (id, name, pset, user, pending, format, cmsswRelease, comment) values ($id, '$name', '$pset', '$user', 1, '$format', '$cmsswRelease', '$comment')";
  $db->selectdb ("ntuple");
  my $results = $db->query ($query);
}

sub
downloadRelease
{
  my $release = shift;
  my $pyConfigName = shift;

  my $query = "select id,pset from ntupleRelease where name='$release'";
  $db->selectdb ("ntuple");
  my $results = $db->query ($query);
  if ($results->numrows () != 1)
    {
      print "Ntuple release \"$release\" not found!\n";
      exit;
    }
  if (!(-e "/home/hart/public_html/releases/$release.tar.gz"))
    {
      print "Release is in the database but no package exists!\n";
      exit;
    }
  if ($pyConfigName)
    {
      my @row = $results->fetchrow ();
      open (PY_CONFIG, ">$pyConfigName");
      print PY_CONFIG $row[1];
      close (PY_CONFIG);
    }
  copy ("/home/hart/public_html/releases/$release.tar.gz", "$release.tar.gz");
}

sub
dbOtherCreate
{
  my $user = shift;
  my $listOfFiles = shift;
  my $comment = shift;

  my $size = 0.0;
  my $nFiles = 0.0;
  for (my $i = 0; $i < @$listOfFiles; $i++)
    {
      if (!(-e $$listOfFiles[$i]))
        {
          print "$$listOfFiles[$i] does not exist!\n";
          exit;
        }
      $nFiles += `ls -R $$listOfFiles[$i] | grep -v ':\$' | grep -v '^\$' | wc -l`;
      my $fileSize = `du -s $$listOfFiles[$i]`;
      $fileSize =~ s/([^ ]*) .*/$1/;
      $fileSize /= 1024 * 1024;
      $size += $fileSize;
      $$listOfFiles[$i] = "$ENV{'PWD'}/$$listOfFiles[$i]" if !($$listOfFiles[$i] =~ m/^\//);
    }
  $size = sprintf "%.2f", $size;
  my $location = join ("<br />", @$listOfFiles);

  my $query = "select max(id) from other";
  $db->selectdb ("ntuple");
  my $results = $db->query ($query);
  my @row = $results->fetchrow ();
  my $id = 1;
  $id = $row[0] + 1 if $results->numrows ();

  $user = addSlashes ($user);
  $location = addSlashes ($location);
  $comment = addSlashes ($comment);

  my $query = "insert into other (id, creationTime, lastUpdateTime, lastUpdateUser, user, location, nFiles, sizeInGB, comment) values ($id, now(), now(), '$user', '$user', '$location', $nFiles, $size, '$comment')";
  $db->selectdb ("ntuple");
  my $results = $db->query ($query);

  print "ID $id\n";
}
Revision:	1.11
Committed:	Wed Oct 17 19:52:03 2012 UTC (12 years, 6 months ago) by ahart
Branch:	MAIN
Changes since 1.10:	+46 -2 lines
Log Message:	Add a downloadRelease command.