[BACK]Return to searchcat.pl CVS log [TXT] [DIR] Up to [LON-CAPA] / loncom / metadata_database

Diff for /loncom/metadata_database/searchcat.pl between versions 1.27 and 1.35

version 1.27, 2003/01/04 19:23:31 version 1.35, 2003/06/25 14:42:00
Line 26 Line 26
 #  #
 # http://www.lon-capa.org/  # http://www.lon-capa.org/
 #  #
 # YEAR=2001  
 # 04/14/2001, 04/16/2001 Scott Harrison  
 #  
 # YEAR=2002  
 # 05/11/2002 Scott Harrison  
 #  
 ###  ###
   
 # This script goes through a LON-CAPA resource  =pod
 # directory and gathers metadata.  
 # The metadata is entered into a SQL database.  =head1 NAME
   
   B<searchcat.pl> - put authoritative filesystem data into sql database.
   
   =head1 SYNOPSIS
   
   Ordinarily this script is to be called from a loncapa cron job
   (CVS source location: F<loncapa/loncom/cron/loncapa>; typical
   filesystem installation location: F</etc/cron.d/loncapa>).
   
   Here is the cron job entry.
   
   C<# Repopulate and refresh the metadata database used for the search catalog.>
   C<10 1 * * 7    www    /home/httpd/perl/searchcat.pl>
   
   This script only allows itself to be run as the user C<www>.
   
   =head1 DESCRIPTION
   
   This script goes through a loncapa resource directory and gathers metadata.
   The metadata is entered into a SQL database.
   
   This script also does general database maintenance such as reformatting
   the C<loncapa:metadata> table if it is deprecated.
   
   This script evaluates dynamic metadata from the authors'
   F<nohist_resevaldata.db> database file in order to store it in MySQL, as
   well as to compress the filesize (add up all "count"-type metadata).
   
   This script is playing an increasingly important role for a loncapa
   library server.  The proper operation of this script is critical for a smooth
   and correct user experience.
   
   =cut
   
 use lib '/home/httpd/lib/perl/';  use lib '/home/httpd/lib/perl/';
 use LONCAPA::Configuration;  use LONCAPA::Configuration;
Line 150  sub dynamicmeta { Line 177  sub dynamicmeta {
 require "find.pl";  require "find.pl";
 sub wanted {  sub wanted {
     (($dev,$ino,$mode,$nlink,$uid,$gid) = lstat($_)) &&      (($dev,$ino,$mode,$nlink,$uid,$gid) = lstat($_)) &&
     -f _ &&          -f _ &&
     /^.*\.meta$/ && !/^.+\.\d+\.[^\.]+\.meta$/ &&          /^.*\.meta$/ && !/^.+\.\d+\.[^\.]+\.meta$/ &&
     push(@metalist,"$dir/$_");          push(@metalist,"$dir/$_");
 }  }
   
 # ---------------  Read loncapa_apache.conf and loncapa.conf and get variables  # ---------------  Read loncapa_apache.conf and loncapa.conf and get variables
Line 168  exit unless $perlvar{'lonRole'} eq 'libr Line 195  exit unless $perlvar{'lonRole'} eq 'libr
   
 my $wwwid=getpwnam('www');  my $wwwid=getpwnam('www');
 if ($wwwid!=$<) {  if ($wwwid!=$<) {
    $emailto="$perlvar{'lonAdmEMail'},$perlvar{'lonSysEMail'}";      $emailto="$perlvar{'lonAdmEMail'},$perlvar{'lonSysEMail'}";
    $subj="LON: $perlvar{'lonHostID'} User ID mismatch";      $subj="LON: $perlvar{'lonHostID'} User ID mismatch";
    system("echo 'User ID mismatch. searchcat.pl must be run as user www.' |\      system("echo 'User ID mismatch. searchcat.pl must be run as user www.' |\
  mailto $emailto -s '$subj' > /dev/null");   mailto $emailto -s '$subj' > /dev/null");
    exit 1;      exit 1;
 }  }
   
   
Line 206  my $dbh; Line 233  my $dbh;
   
 # ------------------------------------------------------------- get .meta files  # ------------------------------------------------------------- get .meta files
 opendir(RESOURCES,"$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}");  opendir(RESOURCES,"$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}");
 my @homeusers=grep  my @homeusers = grep {
           {&ishome("$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$_")}      &ishome("$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$_")
           grep {!/^\.\.?$/} readdir(RESOURCES);      } grep {!/^\.\.?$/} readdir(RESOURCES);
 closedir RESOURCES;  closedir RESOURCES;
   
   #
   # Create the statement handlers we need
   my $delete_sth = $dbh->prepare
       ("DELETE FROM metadata WHERE url LIKE BINARY ?");
   
   my $insert_sth = $dbh->prepare
       ("INSERT INTO metadata VALUES (".
        "?,".   # title
        "?,".   # author
        "?,".   # subject
        "?,".   # m2???
        "?,".   # version
        "?,".   # current
        "?,".   # notes
        "?,".   # abstract
        "?,".   # mime
        "?,".   # language
        "?,".   # creationdate
        "?,".   # revisiondate
        "?,".   # owner
        "?)"    # copyright
        );
   
 foreach my $user (@homeusers) {  foreach my $user (@homeusers) {
     print LOG "\n=== User: ".$user."\n\n";      print LOG "\n=== User: ".$user."\n\n";
 # Remove left-over db-files from potentially crashed searchcat run      # Remove left-over db-files from potentially crashed searchcat run
     my $prodir=&propath($perlvar{'lonDefDomain'},$user);      my $prodir=&propath($perlvar{'lonDefDomain'},$user);
     unlink($prodir.'/nohist_new_resevaldata.db');      unlink($prodir.'/nohist_new_resevaldata.db');
 # Use find.pl      # Use find.pl
     undef @metalist;      undef @metalist;
     @metalist=();      @metalist=();
     &find("$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$user");      &find("$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$user");
       # -- process each file to get metadata and put into search catalog SQL
 # -- process each file to get metadata and put into search catalog SQL database      # database.  Also, check to see if already there.
 # Also, check to see if already there.      # I could just delete (without searching first), but this works for now.
 # I could just delete (without searching first), but this works for now.      foreach my $m (@metalist) {
 foreach my $m (@metalist) {          print LOG "- ".$m."\n";
     print LOG "- ".$m."\n";          my $ref=&metadata($m);
     my $ref=&metadata($m);          my $m2='/res/'.&declutter($m);
     my $m2='/res/'.&declutter($m);          $m2=~s/\.meta$//;
     $m2=~s/\.meta$//;          &dynamicmeta($m2);
     &dynamicmeta($m2);          $delete_sth->execute($m2);
     my $q2="select * from metadata where url like binary '$m2'";          $insert_sth->execute($ref->{'title'},
     my $sth = $dbh->prepare($q2);                               $ref->{'author'},
     $sth->execute();                               $ref->{'subject'},
     my $r1=$sth->fetchall_arrayref;                               $m2,
     if (@$r1) {                               $ref->{'keywords'},
  $sth=$dbh->prepare("delete from metadata where url like binary '$m2'");                               'current',
         $sth->execute();                               $ref->{'notes'},
                                $ref->{'abstract'},
                                $ref->{'mime'},
                                $ref->{'language'},
                                sqltime($ref->{'creationdate'}),
                                sqltime($ref->{'lastrevisiondate'}),
                                $ref->{'owner'},
                                $ref->{'copyright'});
   #        if ($dbh->err()) {
   #            print STDERR "Error:".$dbh->errstr()."\n";
   #        }
           $ref = undef;
     }      }
     $sth=$dbh->prepare('insert into metadata values ('.      
   '"'.delete($ref->{'title'}).'"'.','.      # --------------------------------------------------- Clean up database
   '"'.delete($ref->{'author'}).'"'.','.      # Need to, perhaps, remove stale SQL database records.
   '"'.delete($ref->{'subject'}).'"'.','.      # ... not yet implemented
   '"'.$m2.'"'.','.          
   '"'.delete($ref->{'keywords'}).'"'.','.      # ------------------------------------------- Copy over the new db-files
   '"'.'current'.'"'.','.  
   '"'.delete($ref->{'notes'}).'"'.','.  
   '"'.delete($ref->{'abstract'}).'"'.','.  
   '"'.delete($ref->{'mime'}).'"'.','.  
   '"'.delete($ref->{'language'}).'"'.','.  
   '"'.sqltime(delete($ref->{'creationdate'})).'"'.','.  
   '"'.sqltime(delete($ref->{'lastrevisiondate'})).'"'.','.  
   '"'.delete($ref->{'owner'}).'"'.','.  
   '"'.delete($ref->{'copyright'}).'"'.')');  
     $sth->execute();  
 }  
   
 # ----------------------------------------------------------- Clean up database  
 # Need to, perhaps, remove stale SQL database records.  
 # ... not yet implemented  
   
   
 # -------------------------------------------------- Copy over the new db-files  
     system('mv '.$prodir.'/nohist_new_resevaldata.db '.      system('mv '.$prodir.'/nohist_new_resevaldata.db '.
          $prodir.'/nohist_resevaldata.db');             $prodir.'/nohist_resevaldata.db');
 }  }
 # --------------------------------------------------- Close database connection  # --------------------------------------------------- Close database connection
 $dbh->disconnect;  $dbh->disconnect;
 print LOG "\n==== Searchcat completed ".localtime()." ====\n";  print LOG "\n==== Searchcat completed ".localtime()." ====\n";
 close(LOG);  close(LOG);
 exit 0;  exit 0;
   
   
   
 # =============================================================================  # =============================================================================
   
 # ---------------------------------------------------------------- Get metadata  # ---------------------------------------------------------------- Get metadata
Line 286  sub metadata { Line 333  sub metadata {
         my $parser=HTML::TokeParser->new(\$metastring);          my $parser=HTML::TokeParser->new(\$metastring);
         my $token;          my $token;
         while ($token=$parser->get_token) {          while ($token=$parser->get_token) {
            if ($token->[0] eq 'S') {              if ($token->[0] eq 'S') {
       my $entry=$token->[1];                  my $entry=$token->[1];
               my $unikey=$entry;                  my $unikey=$entry;
               if (defined($token->[2]->{'part'})) {                   if (defined($token->[2]->{'part'})) { 
                  $unikey.='_'.$token->[2]->{'part'};                       $unikey.='_'.$token->[2]->{'part'}; 
       }                  }
               if (defined($token->[2]->{'name'})) {                   if (defined($token->[2]->{'name'})) { 
                  $unikey.='_'.$token->[2]->{'name'};                       $unikey.='_'.$token->[2]->{'name'}; 
       }                  }
               if ($metacache{$uri.'keys'}) {                  if ($metacache{$uri.'keys'}) {
                  $metacache{$uri.'keys'}.=','.$unikey;                      $metacache{$uri.'keys'}.=','.$unikey;
               } else {                  } else {
                  $metacache{$uri.'keys'}=$unikey;                      $metacache{$uri.'keys'}=$unikey;
       }                  }
               map {                  map {
   $metacache{$uri.''.$unikey.'.'.$_}=$token->[2]->{$_};                      $metacache{$uri.''.$unikey.'.'.$_}=$token->[2]->{$_};
               } @{$token->[3]};                  } @{$token->[3]};
               unless (                  unless (
                  $metacache{$uri.''.$unikey}=$parser->get_text('/'.$entry)                          $metacache{$uri.''.$unikey}=$parser->get_text('/'.$entry)
       ) { $metacache{$uri.''.$unikey}=                          ) { $metacache{$uri.''.$unikey}=
       $metacache{$uri.''.$unikey.'.default'};                                  $metacache{$uri.''.$unikey.'.default'};
       }                          }
           }              }
        }          }
     }      }
     return \%metacache;      return \%metacache;
 }  }
Line 317  sub metadata { Line 364  sub metadata {
 # ------------------------------------------------------------ Serves up a file  # ------------------------------------------------------------ Serves up a file
 # returns either the contents of the file or a -1  # returns either the contents of the file or a -1
 sub getfile {  sub getfile {
   my $file=shift;      my $file=shift;
   if (! -e $file ) { return -1; };      if (! -e $file ) { return -1; };
   my $fh=IO::File->new($file);      my $fh=IO::File->new($file);
   my $a='';      my $a='';
   while (<$fh>) { $a .=$_; }      while (<$fh>) { $a .=$_; }
   return $a      return $a;
 }  }
   
 # ------------------------------------------------------------- Declutters URLs  # ------------------------------------------------------------- Declutters URLs
Line 370  sub sqltime { Line 417  sub sqltime {
   
 sub maketime {  sub maketime {
     my %th=@_;      my %th=@_;
     return POSIX::mktime(      return POSIX::mktime(($th{'seconds'},$th{'minutes'},$th{'hours'},
         ($th{'seconds'},$th{'minutes'},$th{'hours'},                            $th{'day'},$th{'month'}-1,
          $th{'day'},$th{'month'}-1,$th{'year'}-1900,0,0,$th{'dlsav'}));                            $th{'year'}-1900,0,0,$th{'dlsav'}));
 }  }
   
   
Line 383  sub maketime { Line 430  sub maketime {
 sub unsqltime {  sub unsqltime {
     my $timestamp=shift;      my $timestamp=shift;
     if ($timestamp=~/^(\d+)\-(\d+)\-(\d+)\s+(\d+)\:(\d+)\:(\d+)$/) {      if ($timestamp=~/^(\d+)\-(\d+)\-(\d+)\s+(\d+)\:(\d+)\:(\d+)$/) {
        $timestamp=&maketime(          $timestamp=&maketime('year'=>$1,'month'=>$2,'day'=>$3,
    'year'=>$1,'month'=>$2,'day'=>$3,                               'hours'=>$4,'minutes'=>$5,'seconds'=>$6);
            'hours'=>$4,'minutes'=>$5,'seconds'=>$6);  
     }      }
     return $timestamp;      return $timestamp;
 }  }
Removed from v.1.27  
changed lines
  Added in v.1.35

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>