ViewVC Help
View File | Revision Log | Show Annotations | Root Listing
root/cvsroot/UserCode/OSUT3Analysis/DBTools/scripts/osudb
(Generate patch)

Comparing UserCode/OSUT3Analysis/DBTools/scripts/osudb (file contents):
Revision 1.6 by ahart, Wed Aug 1 14:15:20 2012 UTC vs.
Revision 1.14 by ahart, Mon Oct 22 14:45:07 2012 UTC

# Line 7 | Line 7 | use File::Copy;
7  
8   sub getDataset;
9   sub dbUpdate;
10 + sub dbUpdateOther;
11   sub dbDelete;
12 + sub dbDeleteOther;
13   sub addSlashes;
14   sub uploadRelease;
15 + sub downloadRelease;
16 + sub datasetExists;
17 + sub uploadConfig;
18  
19   our $db = Mysql->connect ("cmshead.mps.ohio-state.edu", "ntuple", "osuT3User") or die "Failed to connect to Tier 3, stopped";
20  
21   my %opt;
22   Getopt::Long::Configure ("bundling");
23 < 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", "help|h");
23 > 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", "recipe|e=s", "help|h");
24   my $argc = @ARGV;
25  
26   printHelp ($ARGV[0]) if $opt{"help"};
27 < printHelp () if $argc != 2;
28 < printHelp () if $ARGV[0] ne "create" && $ARGV[0] ne "update" && $ARGV[0] ne "finish" && $ARGV[0] ne "deprecate" && $ARGV[0] ne "uploadRelease" && $ARGV[0] ne "deleteEntry";
27 > printHelp () if $argc != 2 && $ARGV[0] ne "createOther";
28 > 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" && $ARGV[0] ne "uploadConfig";
29   if (($ARGV[0] eq "create" || $ARGV[0] eq "finish") && !$opt{"location"})
30    {
31      print "The directory containing the ntuples must be given!\n";
32      exit;
33    }
34 < if ($ARGV[0] eq "uploadRelease" && !$opt{"pyConfig"} && !$opt{"release"})
34 > if ($ARGV[0] eq "uploadRelease" && (!$opt{"pyConfig"} || !$opt{"release"} || !$opt{"recipe"}))
35    {
36 <    print "Both the Python and the CMSSW release must be given!\n";
36 >    print "The Python config, CMSSW release directory, and recipe file must be given!\n";
37      exit;
38    }
39 < if ($ARGV[0] ne "uploadRelease")
39 > if ($ARGV[0] eq "uploadConfig" && (!$opt{"pyConfig"} || !$opt{"release"}))
40    {
41 <    my $id = -1;
42 <    my $fullDataset;
38 <    ($id, $fullDataset) = getDataset ($ARGV[0], $ARGV[1]) if $ARGV[0] ne "create";
39 <    $fullDataset = $ARGV[1] if $id < 0;
40 <    $ARGV[0] = "create" if $id < 0;
41 <    my $status = "present";
42 <    $status = "" if $ARGV[0] eq "update";
43 <    $status = "deprecated" if $ARGV[0] eq "deprecate";
44 <    $opt{"format"} = "BEAN" if $ARGV[0] eq "create" && !$opt{"format"};
45 <    dbUpdate ($id, $fullDataset, $ENV{"USER"}, $opt{"format"}, $opt{"location"}, $opt{"fileList"}, $status, $opt{"comment"}, $opt{"pyConfig"}, $opt{"crabCfg"}, $opt{"jsonFile"}, $opt{"lumiSummary"}, $opt{"release"});
46 <    dbDelete ($id) if $ARGV[0] eq "deleteEntry";
41 >    print "The Python config and ntuple release must be given!\n";
42 >    exit;
43    }
44 < else
44 > if ($ARGV[0] eq "uploadRelease")
45    {
46      my $parentDir = "./$opt{'release'}";
47      $parentDir =~ s/CMSSW_[^\/]*//g;
# Line 54 | Line 50 | else
50      $cmsswRelease =~ s/^.*CMSSW_([^\/]*).*$/CMSSW_$1/;
51      `tar -C $parentDir -czf $ARGV[1].tar.gz $opt{"release"}`;
52      $opt{"format"} = "BEAN" if !$opt{"format"};
53 <    uploadRelease ($opt{"format"}, $cmsswRelease, "$ARGV[1].tar.gz", $opt{"pyConfig"}, $ARGV[1], $ENV{"USER"}, $opt{"comment"});
53 >    uploadRelease ($opt{"format"}, $cmsswRelease, "$ARGV[1].tar.gz", $opt{"pyConfig"}, $ARGV[1], "$ENV{'USER'}", $opt{"comment"}, $opt{"recipe"});
54 >  }
55 > elsif ($ARGV[0] eq "downloadRelease")
56 >  {
57 >    downloadRelease ($ARGV[1], $opt{"pyConfig"});
58 >  }
59 > elsif ($ARGV[0] eq "uploadConfig")
60 >  {
61 >    uploadConfig ($ARGV[1], $ENV{"USER"}, $opt{"pyConfig"}, $opt{"release"}, $opt{"comment"});
62 >  }
63 > elsif ($ARGV[0] eq "createOther")
64 >  {
65 >    my @listOfFiles = @ARGV;
66 >    @listOfFiles = reverse (@listOfFiles);
67 >    pop (@listOfFiles);
68 >    @listOfFiles = reverse (@listOfFiles);
69 >    dbOtherCreate ("$ENV{'USER'}", \@listOfFiles, $opt{"comment"});
70 >  }
71 > else
72 >  {
73 >    my $id = -1;
74 >    my $fullDataset;
75 >    ($id, $fullDataset) = getDataset ($ARGV[0], $ARGV[1]) if $ARGV[0] ne "create" && ($ARGV[0] ne "update" || !$opt{"other"}) && ($ARGV[0] ne "deleteEntry" || !$opt{"other"});
76 >    $id = $ARGV[1] if ($ARGV[0] eq "update" && $opt{"other"}) || ($ARGV[0] eq "deleteEntry" && $opt{"other"});
77 >    $fullDataset = $ARGV[1] if $id < 0;
78 >    $ARGV[0] = "create" if $id < 0;
79 >    my $status = "present";
80 >    $status = "" if $ARGV[0] eq "update";
81 >    $status = "deprecated" if $ARGV[0] eq "deprecate";
82 >    $opt{"format"} = "BEAN" if $ARGV[0] eq "create" && !$opt{"format"};
83 >    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"};
84 >    dbUpdateOther ($id, $opt{"comment"}, "$ENV{'USER'}\@$ENV{'HOSTNAME'}") if $opt{"other"};
85 >    dbDelete ($id) if $ARGV[0] eq "deleteEntry" && !$opt{"other"};
86 >    dbDeleteOther ($id) if $ARGV[0] eq "deleteEntry" && $opt{"other"};
87    }
88  
89   sub
# Line 80 | Line 109 | getDataset
109    my $queryDataset = $dataset;
110    $queryDataset =~ s/\*/%/g;
111    $queryDataset =~ s/(.*)/%$1%/g;
112 <  my $query = "select id,dataset,user,creationTime from ntuple where dataset like '$queryDataset' order by lastUpdateTime";
112 >  my $query = "select id,dataset,user,creationTime from ntuple where dataset like '$queryDataset' order by creationTime";
113    $db->selectdb ("ntuple");
114    $results = $db->query ($query);
115    if ($results->numrows () == 1)
# Line 235 | Line 264 | dbUpdate
264          }
265      }
266    ($nFiles, $size) = sizeOfDataset ($dataset, $location, $fileList);
267 <  my $fullLocation = "$ENV{'PWD'}/$location";
267 >  my $fullLocation = $location;
268 >  $fullLocation = "$ENV{'PWD'}/$location" if !($location =~ m/^\//);
269  
270    $dataset = addSlashes ($dataset);
271    $user = addSlashes ($user);
# Line 261 | Line 291 | dbUpdate
291        my $id = 1;
292        $id = $row[0] + 1 if $results->numrows ();
293  
294 <      $query = "insert into ntuple (id, dataset, creationTime, lastUpdateTime, user, format, location, fileList, nFiles, sizeInGB, status, comment, pset, crabCfg, jsonFile, lumiSummary, version) values ($id, '$dataset', now(), now(), '$user', '$format', '$fullLocation', '$fileList', $nFiles, $size, '$status', '$comment', '$pset', '$crabCfg', '$jsonFile', '$lumiSummary', '$release')";
294 >      $query = "insert into ntuple (id, dataset, creationTime, lastUpdateTime, lastUpdateUser, user, format, location, fileList, nFiles, sizeInGB, status, comment, pset, crabCfg, jsonFile, lumiSummary, version) values ($id, '$dataset', now(), now(), '$user', '$user', '$format', '$fullLocation', '$fileList', $nFiles, $size, '$status', '$comment', '$pset', '$crabCfg', '$jsonFile', '$lumiSummary', '$release')";
295      }
296    if ($id > 0)
297      {
# Line 270 | Line 300 | dbUpdate
300  
301        $values .= ", dataset='$dataset'" if $dataset;
302        $values .= ", lastUpdateTime=now()";
303 +      $values .= ", lastUpdateUser='$user'";
304        $values .= ", format='$format'" if $format;
305        $values .= ", location='$fullLocation'" if $location;
306        $values .= ", fileList='$fileList'" if $fileList;
# Line 292 | Line 323 | dbUpdate
323   }
324  
325   sub
326 + dbUpdateOther
327 + {
328 +  my $id = shift;
329 +  my $comment = shift;
330 +  my $user = shift;
331 +
332 +  $comment = addSlashes ($comment);
333 +
334 +  my $query = "update other set lastUpdateTime=now(), lastUpdateUser='$user', comment='$comment'  where id=$id";
335 +  $db->selectdb ("ntuple");
336 +  my $results = $db->query ($query);
337 +
338 +  return $results;
339 + }
340 +
341 + sub
342   dbDelete
343   {
344    my $id = shift;
345  
299  return if $id < 0;
346    my $query = "delete from ntuple where id=$id";
347    $db->selectdb ("ntuple");
348    my $results = $db->query ($query);
349   }
350  
351   sub
352 + dbDeleteOther
353 + {
354 +  my $id = shift;
355 +
356 +  my $query = "delete from other where id=$id";
357 +  $db->selectdb ("ntuple");
358 +  my $results = $db->query ($query);
359 + }
360 +
361 + sub
362   printHelp
363   {
364    my $command = shift;
# Line 326 | Line 382 | printHelp
382        printf "%-29s%s\n", "  -r, --release NAME", "ntuple release used to produce ntuples";
383        printf "%-29s%s\n", "  -s, --lumiSummary FILE", "lumiSummary.json reported by CRAB";
384      }
385 +  elsif ($command eq "createOther")
386 +    {
387 +      print "Usage: $exeName [OPTION]... createOther DIRECTORIES_AND_FILES\n";
388 +      print "Creates an entry in the database for non-ntuple data.\n";
389 +      print "\n";
390 +      print "Mandatory arguments to long options are mandatory for short options too.\n";
391 +      printf "%-29s%s\n", "  -c, --comment COMMENT", "comment for the database entry";
392 +    }
393    elsif ($command eq "update")
394      {
395        print "Usage: $exeName [OPTION]... update NAME\n";
# Line 337 | Line 401 | printHelp
401        printf "%-29s%s\n", "  -f, --format FORMAT", "ntuple format";
402        printf "%-29s%s\n", "  -j, --jsonFile FILE", "JSON file used for this dataset";
403        printf "%-29s%s\n", "  -l, --location DIRECTORY", "directory containing the ntuples";
404 +      printf "%-29s%s\n", "  -o, --other", "update an entry in the non-ntuple database";
405        printf "%-29s%s\n", "  -p, --pyConfig FILE", "Python config used to produce ntuples";
406        printf "%-29s%s\n", "  -r, --release NAME", "ntuple release used to produce ntuples";
407        printf "%-29s%s\n", "  -s, --lumiSummary FILE", "lumiSummary.json reported by CRAB";
# Line 344 | Line 409 | printHelp
409    elsif ($command eq "finish")
410      {
411        print "Usage: $exeName -l DIRECTORY finish NAME\n";
347
412        print "Finalizes the database entry for dataset NAME, changing its status to\n";
413        print "\"present\".  This is intended to be the final step in command-line based ntuple\n";
414        print "production.\n";
# Line 354 | Line 418 | printHelp
418      }
419    elsif ($command eq "deleteEntry")
420      {
421 <      print "Usage: $exeName deleteEntry NAME\n";
421 >      print "Usage: $exeName [OPTION]... deleteEntry NAME\n";
422        print "Deletes the database entry for dataset NAME. This is intended primarily for\n";
423        print "mistaken database entries. If you wish to actually delete a set of ntuples,\n";
424        print "please use the \"deprecate\" command instead.\n";
425 +      print "\n";
426 +      printf "%-29s%s\n", "  -o, --other", "update an entry in the non-ntuple database";
427      }
428    elsif ($command eq "deprecate")
429      {
# Line 368 | Line 434 | printHelp
434      }
435    elsif ($command eq "uploadRelease")
436      {
437 <      print "Usage: $exeName -p FILE -r DIRECTORY [OPTION]... uploadRelease NAME\n";
437 >      print "Usage: $exeName -e FILE -p FILE -r DIRECTORY [OPTION]... uploadRelease NAME\n";
438        print "Copies an ntuple release to the appropriate area on the Tier 3, and creates a\n";
439        print "database entry for it, with NAME being the name of the release. An ntuple\n";
440        print "release is a CMSSW release with all the necessary packages added for creating\n";
# Line 376 | Line 442 | printHelp
442        print "\n";
443        print "Mandatory arguments to long options are mandatory for short options too.\n";
444        printf "%-29s%s\n", "  -c, --comment COMMENT", "comment for the database entry";
445 +      printf "%-29s%s\n", "  -e, --recipe FILE", "file containing the recipe for the release";
446        printf "%-29s%s\n", "  -f, --format FORMAT", "ntuple format (default: BEAN)";
447        printf "%-29s%s\n", "  -p, --pyConfig FILE", "default Python config for the release";
448        printf "%-29s%s\n", "  -r, --release DIRECTORY", "CMSSW release containing ntuple packages";
449      }
450 +  elsif ($command eq "downloadRelease")
451 +    {
452 +      print "Usage: $exeName [OPTION]... downloadRelease NAME\n";
453 +      print "Copies an ntuple release to the current directory on the Tier 3. Optionally\n";
454 +      print "copies the corresponding Python configuration file registered in the database.\n";
455 +      print "\n";
456 +      print "Mandatory arguments to long options are mandatory for short options too.\n";
457 +      printf "%-29s%s\n", "  -p, --pyConfig FILE", "Python configuration file name";
458 +    }
459 +  elsif ($command eq "uploadConfig")
460 +    {
461 +      print "Usage: $exeName -p FILE -r DIRECTORY [OPTION]... uploadConfig NAME\n";
462 +      print "Creates an entry in the database for a customized Python configuration file.\n";
463 +      print "\n";
464 +      print "Mandatory arguments to long options are mandatory for short options too.\n";
465 +      printf "%-29s%s\n", "  -c, --comment COMMENT", "comment for the database entry";
466 +      printf "%-29s%s\n", "  -p, --pyConfig FILE", "Python config file";
467 +      printf "%-29s%s\n", "  -r, --release NAME", "ntuple release with which to use this config";
468 +    }
469    else
470      {
471        print "Usage: $exeName [OPTION]... COMMAND NAME\n";
# Line 391 | Line 477 | printHelp
477        print "\n";
478        print "COMMAND may be one of the following:\n";
479        printf "%-29s%s\n", "  create", "creates the entry";
480 +      printf "%-29s%s\n", "  createOther", "creates an entry for non-ntuple data";
481        printf "%-29s%s\n", "  update", "updates the entry";
482        printf "%-29s%s\n", "  finish", "finalizes the database entry";
483        printf "%-29s%s\n", "  deleteEntry", "removes the database entry";
484        printf "%-29s%s\n", "  deprecate", "marks the dataset for deletion";
485        printf "%-29s%s\n", "  uploadRelease", "upload an ntuple release";
486 +      printf "%-29s%s\n", "  downloadRelease", "download an ntuple release";
487 +      printf "%-29s%s\n", "  uploadConfig", "upload an ntuple config";
488      }
489  
490    exit;
# Line 449 | Line 538 | uploadRelease
538    my $name = shift;
539    my $user = shift;
540    my $comment = shift;
541 +  my $recipeName = shift;
542  
453  move ($release, "/home/hart/public_html/releases/$name.tar.gz") or die "Ntuple releases may only be uploaded on the Tier 3, stopped";
543    if (!(-e $psetName))
544      {
545        print "$psetName does not exist!\n";
546        exit;
547      }
548 +  if (!(-e $recipeName))
549 +    {
550 +      print "$recipeName does not exist!\n";
551 +      exit;
552 +    }
553 +  move ($release, "/home/hart/public_html/releases/$name.tar.gz") or die "Ntuple releases may only be uploaded on the Tier 3, stopped";
554    open (PY_CONFIG, "<$psetName");
555    my @pset = <PY_CONFIG>;
556    close (PY_CONFIG);
557    my $pset = join ("", @pset);
558 +  open (RECIPE, "<$recipeName");
559 +  my @recipe = <RECIPE>;
560 +  close (RECIPE);
561 +  my $recipe = join ("\n", @recipe);
562  
563    my $query = "select max(id) from ntupleRelease";
564    $db->selectdb ("ntuple");
# Line 474 | Line 573 | uploadRelease
573    $format = addSlashes ($format);
574    $cmsswRelease = addSlashes ($cmsswRelease);
575    $comment = addSlashes ($comment);
576 +  $recipe = addSlashes ($recipe);
577 +
578 +  my $query = "insert into ntupleRelease (id, name, pset, user, pending, format, cmsswRelease, comment, recipe) values ($id, '$name', '$pset', '$user', 1, '$format', '$cmsswRelease', '$comment', '$recipe')";
579 +  $db->selectdb ("ntuple");
580 +  my $results = $db->query ($query);
581 + }
582 +
583 + sub
584 + downloadRelease
585 + {
586 +  my $release = shift;
587 +  my $pyConfigName = shift;
588 +
589 +  my $query = "select id,pset from ntupleRelease where name='$release'";
590 +  $db->selectdb ("ntuple");
591 +  my $results = $db->query ($query);
592 +  if ($results->numrows () != 1)
593 +    {
594 +      print "Ntuple release \"$release\" not found!\n";
595 +      exit;
596 +    }
597 +  if (!(-e "/home/hart/public_html/releases/$release.tar.gz"))
598 +    {
599 +      print "Release is in the database but no package exists!\n";
600 +      exit;
601 +    }
602 +  if ($pyConfigName)
603 +    {
604 +      my @row = $results->fetchrow ();
605 +      open (PY_CONFIG, ">$pyConfigName");
606 +      print PY_CONFIG $row[1];
607 +      close (PY_CONFIG);
608 +    }
609 +  copy ("/home/hart/public_html/releases/$release.tar.gz", "$release.tar.gz");
610 + }
611 +
612 + sub
613 + dbOtherCreate
614 + {
615 +  my $user = shift;
616 +  my $listOfFiles = shift;
617 +  my $comment = shift;
618 +
619 +  my $size = 0.0;
620 +  my $nFiles = 0.0;
621 +  for (my $i = 0; $i < @$listOfFiles; $i++)
622 +    {
623 +      if (!(-e $$listOfFiles[$i]))
624 +        {
625 +          print "$$listOfFiles[$i] does not exist!\n";
626 +          exit;
627 +        }
628 +      $nFiles += `ls -R $$listOfFiles[$i] | grep -v ':\$' | grep -v '^\$' | wc -l`;
629 +      my $fileSize = `du -s $$listOfFiles[$i]`;
630 +      $fileSize =~ s/([^ ]*) .*/$1/;
631 +      $fileSize /= 1024 * 1024;
632 +      $size += $fileSize;
633 +      $$listOfFiles[$i] = "$ENV{'PWD'}/$$listOfFiles[$i]" if !($$listOfFiles[$i] =~ m/^\//);
634 +    }
635 +  $size = sprintf "%.2f", $size;
636 +  my $location = join ("<br />", @$listOfFiles);
637 +
638 +  my $query = "select max(id) from other";
639 +  $db->selectdb ("ntuple");
640 +  my $results = $db->query ($query);
641 +  my @row = $results->fetchrow ();
642 +  my $id = 1;
643 +  $id = $row[0] + 1 if $results->numrows ();
644 +
645 +  $user = addSlashes ($user);
646 +  $location = addSlashes ($location);
647 +  $comment = addSlashes ($comment);
648 +
649 +  $query = "insert into other (id, creationTime, lastUpdateTime, lastUpdateUser, user, location, nFiles, sizeInGB, comment) values ($id, now(), now(), '$user', '$user', '$location', $nFiles, $size, '$comment')";
650 +  $db->selectdb ("ntuple");
651 +  $results = $db->query ($query);
652 +
653 +  print "ID $id\n";
654 + }
655 +
656 + sub
657 + uploadConfig
658 + {
659 +  my $name = shift;
660 +  my $user = shift;
661 +  my $psetName = shift;
662 +  my $release = shift;
663 +  my $comment = shift;
664  
665 <  my $query = "insert into ntupleRelease (id, name, pset, user, pending, format, cmsswRelease, comment) values ($id, '$name', '$pset', '$user', 1, '$format', '$cmsswRelease', '$comment')";
665 >  if (!(-e $psetName))
666 >    {
667 >      print "$psetName does not exist!\n";
668 >      exit;
669 >    }
670 >  if ($release)
671 >    {
672 >      my $query = "select id from ntupleRelease where name='$release'";
673 >      $db->selectdb ("ntuple");
674 >      my $results = $db->query ($query);
675 >      if ($results->numrows () != 1)
676 >        {
677 >          print "Ntuple release \"$release\" not found!\n";
678 >          exit;
679 >        }
680 >    }
681 >  open (PSET, "<$psetName");
682 >  my @pset = <PSET>;
683 >  close (PSET);
684 >  my $pset = join ("", @pset);
685 >
686 >  my $query = "select max(id) from ntupleConfig";
687    $db->selectdb ("ntuple");
688    my $results = $db->query ($query);
689 +  my @row = $results->fetchrow ();
690 +  my $id = 1;
691 +  $id = $row[0] + 1 if $results->numrows ();
692 +
693 +  $name = addSlashes ($name);
694 +  $user = addSlashes ($user);
695 +  $pset = addSlashes ($pset);
696 +  $release = addSlashes ($release);
697 +  $comment = addSlashes ($comment);
698 +
699 +  my $query = "insert into ntupleConfig (id, name, pset, user, pending, ntupleRelease, comment) values ($id, '$name', '$pset', '$user', 1, '$release', '$comment')";
700 +  $db->selectdb ("ntuple");
701 +  $results = $db->query ($query);
702   }

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines