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 1.15 by ahart, Fri Nov 30 18:20:59 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 ("$ENV{'USER'}\@$ENV{'HOSTNAME'}", \@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 144 | Line 173 | dbUpdate
173   {
174    my $id = shift;
175    my $dataset = shift;
176 <  my $user = shift;
176 >  my $userAndHost = shift;
177    my $format = shift;
178    my $location = shift;
179    my $fileListName = shift;
# Line 156 | Line 185 | dbUpdate
185    my $lumiSummaryName = shift;
186    my $release = shift;
187  
188 +  my $user = $userAndHost;
189 +  $user =~ s/@.*$//g;
190 +
191    my $fileList;
192    my $pset;
193    my $crabCfg;
# Line 235 | Line 267 | dbUpdate
267          }
268      }
269    ($nFiles, $size) = sizeOfDataset ($dataset, $location, $fileList);
270 <  my $fullLocation = "$ENV{'PWD'}/$location";
270 >  my $fullLocation = $location;
271 >  $fullLocation = "$ENV{'PWD'}/$location" if !($location =~ m/^\//);
272  
273    $dataset = addSlashes ($dataset);
274    $user = addSlashes ($user);
# Line 261 | Line 294 | dbUpdate
294        my $id = 1;
295        $id = $row[0] + 1 if $results->numrows ();
296  
297 <      $query = "insert into ntuple (id, dataset, creationTime, lastUpdateTime, user, format, location, fileList, nFiles, sizeInGB, status, comment, pset, crabCfg, jsonFile, lumiSummary, version) values ($id, '$dataset', now(), now(), '$user', '$format', '$fullLocation', '$fileList', $nFiles, $size, '$status', '$comment', '$pset', '$crabCfg', '$jsonFile', '$lumiSummary', '$release')";
297 >      $query = "insert into ntuple (id, dataset, creationTime, lastUpdateTime, lastUpdateUser, user, format, location, fileList, nFiles, sizeInGB, status, comment, pset, crabCfg, jsonFile, lumiSummary, version) values ($id, '$dataset', now(), now(), '$userAndHost', '$user', '$format', '$fullLocation', '$fileList', $nFiles, $size, '$status', '$comment', '$pset', '$crabCfg', '$jsonFile', '$lumiSummary', '$release')";
298      }
299    if ($id > 0)
300      {
# Line 270 | Line 303 | dbUpdate
303  
304        $values .= ", dataset='$dataset'" if $dataset;
305        $values .= ", lastUpdateTime=now()";
306 +      $values .= ", lastUpdateUser='$user'";
307        $values .= ", format='$format'" if $format;
308        $values .= ", location='$fullLocation'" if $location;
309        $values .= ", fileList='$fileList'" if $fileList;
# Line 292 | Line 326 | dbUpdate
326   }
327  
328   sub
329 + dbUpdateOther
330 + {
331 +  my $id = shift;
332 +  my $comment = shift;
333 +  my $user = shift;
334 +
335 +  $comment = addSlashes ($comment);
336 +
337 +  my $query = "update other set lastUpdateTime=now(), lastUpdateUser='$user', comment='$comment'  where id=$id";
338 +  $db->selectdb ("ntuple");
339 +  my $results = $db->query ($query);
340 +
341 +  return $results;
342 + }
343 +
344 + sub
345   dbDelete
346   {
347    my $id = shift;
348  
299  return if $id < 0;
349    my $query = "delete from ntuple where id=$id";
350    $db->selectdb ("ntuple");
351    my $results = $db->query ($query);
352   }
353  
354   sub
355 + dbDeleteOther
356 + {
357 +  my $id = shift;
358 +
359 +  my $query = "delete from other where id=$id";
360 +  $db->selectdb ("ntuple");
361 +  my $results = $db->query ($query);
362 + }
363 +
364 + sub
365   printHelp
366   {
367    my $command = shift;
# Line 326 | Line 385 | printHelp
385        printf "%-29s%s\n", "  -r, --release NAME", "ntuple release used to produce ntuples";
386        printf "%-29s%s\n", "  -s, --lumiSummary FILE", "lumiSummary.json reported by CRAB";
387      }
388 +  elsif ($command eq "createOther")
389 +    {
390 +      print "Usage: $exeName [OPTION]... createOther DIRECTORIES_AND_FILES\n";
391 +      print "Creates an entry in the database for non-ntuple data.\n";
392 +      print "\n";
393 +      print "Mandatory arguments to long options are mandatory for short options too.\n";
394 +      printf "%-29s%s\n", "  -c, --comment COMMENT", "comment for the database entry";
395 +    }
396    elsif ($command eq "update")
397      {
398        print "Usage: $exeName [OPTION]... update NAME\n";
# Line 337 | Line 404 | printHelp
404        printf "%-29s%s\n", "  -f, --format FORMAT", "ntuple format";
405        printf "%-29s%s\n", "  -j, --jsonFile FILE", "JSON file used for this dataset";
406        printf "%-29s%s\n", "  -l, --location DIRECTORY", "directory containing the ntuples";
407 +      printf "%-29s%s\n", "  -o, --other", "update an entry in the non-ntuple database";
408        printf "%-29s%s\n", "  -p, --pyConfig FILE", "Python config used to produce ntuples";
409        printf "%-29s%s\n", "  -r, --release NAME", "ntuple release used to produce ntuples";
410        printf "%-29s%s\n", "  -s, --lumiSummary FILE", "lumiSummary.json reported by CRAB";
# Line 344 | Line 412 | printHelp
412    elsif ($command eq "finish")
413      {
414        print "Usage: $exeName -l DIRECTORY finish NAME\n";
347
415        print "Finalizes the database entry for dataset NAME, changing its status to\n";
416        print "\"present\".  This is intended to be the final step in command-line based ntuple\n";
417        print "production.\n";
# Line 354 | Line 421 | printHelp
421      }
422    elsif ($command eq "deleteEntry")
423      {
424 <      print "Usage: $exeName deleteEntry NAME\n";
424 >      print "Usage: $exeName [OPTION]... deleteEntry NAME\n";
425        print "Deletes the database entry for dataset NAME. This is intended primarily for\n";
426        print "mistaken database entries. If you wish to actually delete a set of ntuples,\n";
427        print "please use the \"deprecate\" command instead.\n";
428 +      print "\n";
429 +      printf "%-29s%s\n", "  -o, --other", "update an entry in the non-ntuple database";
430      }
431    elsif ($command eq "deprecate")
432      {
# Line 368 | Line 437 | printHelp
437      }
438    elsif ($command eq "uploadRelease")
439      {
440 <      print "Usage: $exeName -p FILE -r DIRECTORY [OPTION]... uploadRelease NAME\n";
440 >      print "Usage: $exeName -e FILE -p FILE -r DIRECTORY [OPTION]... uploadRelease NAME\n";
441        print "Copies an ntuple release to the appropriate area on the Tier 3, and creates a\n";
442        print "database entry for it, with NAME being the name of the release. An ntuple\n";
443        print "release is a CMSSW release with all the necessary packages added for creating\n";
# Line 376 | Line 445 | printHelp
445        print "\n";
446        print "Mandatory arguments to long options are mandatory for short options too.\n";
447        printf "%-29s%s\n", "  -c, --comment COMMENT", "comment for the database entry";
448 +      printf "%-29s%s\n", "  -e, --recipe FILE", "file containing the recipe for the release";
449        printf "%-29s%s\n", "  -f, --format FORMAT", "ntuple format (default: BEAN)";
450        printf "%-29s%s\n", "  -p, --pyConfig FILE", "default Python config for the release";
451        printf "%-29s%s\n", "  -r, --release DIRECTORY", "CMSSW release containing ntuple packages";
452      }
453 +  elsif ($command eq "downloadRelease")
454 +    {
455 +      print "Usage: $exeName [OPTION]... downloadRelease NAME\n";
456 +      print "Copies an ntuple release to the current directory on the Tier 3. Optionally\n";
457 +      print "copies the corresponding Python configuration file registered in the database.\n";
458 +      print "\n";
459 +      print "Mandatory arguments to long options are mandatory for short options too.\n";
460 +      printf "%-29s%s\n", "  -p, --pyConfig FILE", "Python configuration file name";
461 +    }
462 +  elsif ($command eq "uploadConfig")
463 +    {
464 +      print "Usage: $exeName -p FILE -r DIRECTORY [OPTION]... uploadConfig NAME\n";
465 +      print "Creates an entry in the database for a customized Python configuration file.\n";
466 +      print "\n";
467 +      print "Mandatory arguments to long options are mandatory for short options too.\n";
468 +      printf "%-29s%s\n", "  -c, --comment COMMENT", "comment for the database entry";
469 +      printf "%-29s%s\n", "  -p, --pyConfig FILE", "Python config file";
470 +      printf "%-29s%s\n", "  -r, --release NAME", "ntuple release with which to use this config";
471 +    }
472    else
473      {
474        print "Usage: $exeName [OPTION]... COMMAND NAME\n";
# Line 391 | Line 480 | printHelp
480        print "\n";
481        print "COMMAND may be one of the following:\n";
482        printf "%-29s%s\n", "  create", "creates the entry";
483 +      printf "%-29s%s\n", "  createOther", "creates an entry for non-ntuple data";
484        printf "%-29s%s\n", "  update", "updates the entry";
485        printf "%-29s%s\n", "  finish", "finalizes the database entry";
486        printf "%-29s%s\n", "  deleteEntry", "removes the database entry";
487        printf "%-29s%s\n", "  deprecate", "marks the dataset for deletion";
488        printf "%-29s%s\n", "  uploadRelease", "upload an ntuple release";
489 +      printf "%-29s%s\n", "  downloadRelease", "download an ntuple release";
490 +      printf "%-29s%s\n", "  uploadConfig", "upload an ntuple config";
491      }
492  
493    exit;
# Line 449 | Line 541 | uploadRelease
541    my $name = shift;
542    my $user = shift;
543    my $comment = shift;
544 +  my $recipeName = shift;
545  
453  move ($release, "/home/hart/public_html/releases/$name.tar.gz") or die "Ntuple releases may only be uploaded on the Tier 3, stopped";
546    if (!(-e $psetName))
547      {
548        print "$psetName does not exist!\n";
549        exit;
550      }
551 +  if (!(-e $recipeName))
552 +    {
553 +      print "$recipeName does not exist!\n";
554 +      exit;
555 +    }
556 +  move ($release, "/home/hart/public_html/releases/$name.tar.gz") or die "Ntuple releases may only be uploaded on the Tier 3, stopped";
557    open (PY_CONFIG, "<$psetName");
558    my @pset = <PY_CONFIG>;
559    close (PY_CONFIG);
560    my $pset = join ("", @pset);
561 +  open (RECIPE, "<$recipeName");
562 +  my @recipe = <RECIPE>;
563 +  close (RECIPE);
564 +  my $recipe = join ("\n", @recipe);
565  
566    my $query = "select max(id) from ntupleRelease";
567    $db->selectdb ("ntuple");
# Line 474 | Line 576 | uploadRelease
576    $format = addSlashes ($format);
577    $cmsswRelease = addSlashes ($cmsswRelease);
578    $comment = addSlashes ($comment);
579 +  $recipe = addSlashes ($recipe);
580 +
581 +  my $query = "insert into ntupleRelease (id, name, pset, user, pending, format, cmsswRelease, comment, recipe) values ($id, '$name', '$pset', '$user', 1, '$format', '$cmsswRelease', '$comment', '$recipe')";
582 +  $db->selectdb ("ntuple");
583 +  my $results = $db->query ($query);
584 + }
585 +
586 + sub
587 + downloadRelease
588 + {
589 +  my $release = shift;
590 +  my $pyConfigName = shift;
591 +
592 +  my $query = "select id,pset from ntupleRelease where name='$release'";
593 +  $db->selectdb ("ntuple");
594 +  my $results = $db->query ($query);
595 +  if ($results->numrows () != 1)
596 +    {
597 +      print "Ntuple release \"$release\" not found!\n";
598 +      exit;
599 +    }
600 +  if (!(-e "/home/hart/public_html/releases/$release.tar.gz"))
601 +    {
602 +      print "Release is in the database but no package exists!\n";
603 +      exit;
604 +    }
605 +  if ($pyConfigName)
606 +    {
607 +      my @row = $results->fetchrow ();
608 +      open (PY_CONFIG, ">$pyConfigName");
609 +      print PY_CONFIG $row[1];
610 +      close (PY_CONFIG);
611 +    }
612 +  copy ("/home/hart/public_html/releases/$release.tar.gz", "$release.tar.gz");
613 + }
614 +
615 + sub
616 + dbOtherCreate
617 + {
618 +  my $userAndHost = shift;
619 +  my $listOfFiles = shift;
620 +  my $comment = shift;
621 +
622 +  my $user = $userAndHost;
623 +  $user =~ s/@.*$//g;
624 +
625 +  my $size = 0.0;
626 +  my $nFiles = 0.0;
627 +  for (my $i = 0; $i < @$listOfFiles; $i++)
628 +    {
629 +      if (!(-e $$listOfFiles[$i]))
630 +        {
631 +          print "$$listOfFiles[$i] does not exist!\n";
632 +          exit;
633 +        }
634 +      $nFiles += `ls -R $$listOfFiles[$i] | grep -v ':\$' | grep -v '^\$' | wc -l`;
635 +      my $fileSize = `du -s $$listOfFiles[$i]`;
636 +      $fileSize =~ s/([^ ]*) .*/$1/;
637 +      $fileSize /= 1024 * 1024;
638 +      $size += $fileSize;
639 +      $$listOfFiles[$i] = "$ENV{'PWD'}/$$listOfFiles[$i]" if !($$listOfFiles[$i] =~ m/^\//);
640 +    }
641 +  $size = sprintf "%.2f", $size;
642 +  my $location = join ("<br />", @$listOfFiles);
643 +
644 +  my $query = "select max(id) from other";
645 +  $db->selectdb ("ntuple");
646 +  my $results = $db->query ($query);
647 +  my @row = $results->fetchrow ();
648 +  my $id = 1;
649 +  $id = $row[0] + 1 if $results->numrows ();
650 +
651 +  $user = addSlashes ($user);
652 +  $location = addSlashes ($location);
653 +  $comment = addSlashes ($comment);
654  
655 <  my $query = "insert into ntupleRelease (id, name, pset, user, pending, format, cmsswRelease, comment) values ($id, '$name', '$pset', '$user', 1, '$format', '$cmsswRelease', '$comment')";
655 >  $query = "insert into other (id, creationTime, lastUpdateTime, lastUpdateUser, user, location, nFiles, sizeInGB, comment) values ($id, now(), now(), '$userAndHost', '$user', '$location', $nFiles, $size, '$comment')";
656 >  $db->selectdb ("ntuple");
657 >  $results = $db->query ($query);
658 >
659 >  print "ID $id\n";
660 > }
661 >
662 > sub
663 > uploadConfig
664 > {
665 >  my $name = shift;
666 >  my $user = shift;
667 >  my $psetName = shift;
668 >  my $release = shift;
669 >  my $comment = shift;
670 >
671 >  if (!(-e $psetName))
672 >    {
673 >      print "$psetName does not exist!\n";
674 >      exit;
675 >    }
676 >  if ($release)
677 >    {
678 >      my $query = "select id from ntupleRelease where name='$release'";
679 >      $db->selectdb ("ntuple");
680 >      my $results = $db->query ($query);
681 >      if ($results->numrows () != 1)
682 >        {
683 >          print "Ntuple release \"$release\" not found!\n";
684 >          exit;
685 >        }
686 >    }
687 >  open (PSET, "<$psetName");
688 >  my @pset = <PSET>;
689 >  close (PSET);
690 >  my $pset = join ("", @pset);
691 >
692 >  my $query = "select max(id) from ntupleConfig";
693    $db->selectdb ("ntuple");
694    my $results = $db->query ($query);
695 +  my @row = $results->fetchrow ();
696 +  my $id = 1;
697 +  $id = $row[0] + 1 if $results->numrows ();
698 +
699 +  $name = addSlashes ($name);
700 +  $user = addSlashes ($user);
701 +  $pset = addSlashes ($pset);
702 +  $release = addSlashes ($release);
703 +  $comment = addSlashes ($comment);
704 +
705 +  my $query = "insert into ntupleConfig (id, name, pset, user, pending, ntupleRelease, comment) values ($id, '$name', '$pset', '$user', 1, '$release', '$comment')";
706 +  $db->selectdb ("ntuple");
707 +  $results = $db->query ($query);
708   }

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines