[idzebra-moved-to-github.git] / perl / lib / IDZebra / Session.pm

# $Id: Session.pm,v 1.14 2003-03-12 17:08:53 pop Exp $
# 
# Zebra perl API header
# =============================================================================
package IDZebra::Session;

use strict;
use warnings;

BEGIN {
    use IDZebra;
    use Scalar::Util;
    use IDZebra::Logger qw(:flags :calls);
    use IDZebra::Resultset;
    use IDZebra::ScanList;
    use IDZebra::RetrievalRecord;
    require Exporter;
    our $VERSION = do { my @r = (q$Revision: 1.14 $ =~ /\d+/g); sprintf "%d."."%02d" x $#r, @r }; 
    our @ISA = qw(IDZebra::Logger Exporter);
    our @EXPORT = qw (TRANS_RW TRANS_RO);
}

use constant TRANS_RW => 1;
use constant TRANS_RO => 0;

1;
# -----------------------------------------------------------------------------
# Class constructors, destructor
# -----------------------------------------------------------------------------
sub new {
    my ($proto, %args) = @_;
    my $class = ref($proto) || $proto;
    my $self = {};
    $self->{args} = \%args;
    
    bless ($self, $class);
    $self->{cql_ct} = undef;
    $self->{cql_mapfile} = "";
    return ($self);

    $self->{databases} = {};
}

sub start_service {
    my ($self, %args) = @_;

    my $zs;
    unless (defined($self->{zs})) {
        if (defined($args{'configFile'})) {
            $self->{zs} = IDZebra::start($args{'configFile'});
        } else {
            $self->{zs} = IDZebra::start("zebra.cfg");
        }
    }
}

sub stop_service {
    my ($self) = @_;
    if (defined($self->{zs})) {
        IDZebra::stop($self->{zs}) if ($self->{zs});    
        $self->{zs} = undef;
    }
}


sub open {
    my ($proto,%args) = @_;
    my $self = {};

    if (ref($proto)) { $self = $proto; } else { 
        $self = $proto->new(%args);
    }

    unless (%args) {
        %args = %{$self->{args}};
    }

    $self->start_service(%args);

    unless (defined($self->{zs})) {
        croak ("Falied to open zebra service");
    }    

    unless (defined($self->{zh})) {
        $self->{zh}=IDZebra::open($self->{zs}); 
    }   

    # Reset result set counter
    $self->{rscount} = 0;

    # This is needed in order to somehow initialize the service
    $self->databases("Default");

    # Load the default configuration
    $self->group(%args);


    # Set shadow usage
    my $shadow = defined($args{shadow}) ? $args{shadow} : 0;
    $self->shadow($shadow);
    
    $self->{odr_input} = IDZebra::odr_createmem($IDZebra::ODR_DECODE);
    $self->{odr_output} = IDZebra::odr_createmem($IDZebra::ODR_ENCODE);

    return ($self);
}

sub checkzh {
    my ($self) = @_;
    unless (defined($self->{zh})) {
        croak ("Zebra session is not opened");
    }
}

sub close {
    my ($self) = @_;

    if ($self->{zh}) {

        my $stats = 0; 
        # Delete all resulsets
        my $r = IDZebra::deleteResultSet($self->{zh},
                                         1, #Z_DeleteRequest_all,
                                         0,[],
                                         $stats);

        while (IDZebra::trans_no($self->{zh}) > 0) {
            logf (LOG_WARN,"Explicitly closing transaction with session");
            $self->end_trans;
        }

        IDZebra::close($self->{zh});
        $self->{zh} = undef;
    }
    
    if ($self->{odr_input}) {
        IDZebra::odr_reset($self->{odr_input});
        IDZebra::odr_destroy($self->{odr_input});
        $self->{odr_input} = undef;  
    }

    if ($self->{odr_output}) {
        IDZebra::odr_reset($self->{odr_output});
        IDZebra::odr_destroy($self->{odr_output});
        $self->{odr_output} = undef;  
    }

    $self->stop_service;
}

sub DESTROY {
    my ($self) = @_;
    logf (LOG_LOG,"DESTROY $self");
    $self->close; 

    if (defined ($self->{cql_ct})) {
      IDZebra::cql_transform_close($self->{cql_ct});
    }

}
# -----------------------------------------------------------------------------
# Record group selection  This is a bit nasty... but used at many places 
# -----------------------------------------------------------------------------
sub group {
    my ($self,%args) = @_;
    $self->checkzh;
    if ($#_ > 0) {
        $self->{rg} = $self->_makeRecordGroup(%args);
        $self->_selectRecordGroup($self->{rg});
    }
    return($self->{rg});
}

sub selectRecordGroup {
    my ($self, $groupName) = @_;
    $self->checkzh;
    $self->{rg} = $self->_getRecordGroup($groupName);
    $self->_selectRecordGroup($self->{rg});
}

sub _displayRecordGroup {
    my ($self, $rg) = @_;
    print STDERR "-----\n";
    foreach my $key qw (groupName 
                        databaseName 
                        path recordId 
                        recordType 
                        flagStoreData 
                        flagStoreKeys 
                        flagRw 
                        fileVerboseLimit 
                        databaseNamePath 
                        explainDatabase 
                        followLinks) {
        print STDERR "$key:",$rg->{$key},"\n";
    }
}

sub _cloneRecordGroup {
    my ($self, $orig) = @_;
    my $rg = IDZebra::recordGroup->new();
    my $r = IDZebra::init_recordGroup($rg);
    foreach my $key qw (groupName 
                        databaseName 
                        path 
                        recordId 
                        recordType 
                        flagStoreData 
                        flagStoreKeys 
                        flagRw 
                        fileVerboseLimit 
                        databaseNamePath 
                        explainDatabase 
                        followLinks) {
        $rg->{$key} = $orig->{$key} if ($orig->{$key});
    }
    return ($rg);
}

sub _getRecordGroup {
    my ($self, $groupName, $ext) = @_;
    my $rg = IDZebra::recordGroup->new();
    my $r = IDZebra::init_recordGroup($rg);
    $rg->{groupName} = $groupName if ($groupName ne "");  
    $ext = "" unless ($ext);
    $r = IDZebra::res_get_recordGroup($self->{zh}, $rg, $ext);
    return ($rg);
}

sub _makeRecordGroup {
    my ($self, %args) = @_;
    my $rg;

    my @keys = keys(%args);
    unless ($#keys >= 0) {
        return ($self->{rg});
    }

    if ($args{groupName}) {
        $rg = $self->_getRecordGroup($args{groupName});
    } else {
        $rg = $self->_cloneRecordGroup($self->{rg});
    }
    $self->_setRecordGroupOptions($rg, %args);
    return ($rg);
}

sub _setRecordGroupOptions {
    my ($self, $rg, %args) = @_;

    foreach my $key qw (databaseName 
                        path 
                        recordId 
                        recordType 
                        flagStoreData 
                        flagStoreKeys 
                        flagRw 
                        fileVerboseLimit 
                        databaseNamePath 
                        explainDatabase 
                        followLinks) {
        if (defined ($args{$key})) {
            $rg->{$key} = $args{$key};
        }
    }
}
sub _selectRecordGroup {
    my ($self, $rg) = @_;

    my $r = IDZebra::set_group($self->{zh}, $rg);
    my $dbName;
    unless ($dbName = $rg->{databaseName}) {
        $dbName = 'Default';
    }
    unless ($self->databases($dbName)) {
        croak("Fatal error selecting database $dbName");
    }
}
# -----------------------------------------------------------------------------
# Selecting databases for search (and also for updating - internally)
# -----------------------------------------------------------------------------
sub databases {
    my ($self, @databases) = @_;

    $self->checkzh;

    unless ($#_ >0) {
        return (keys(%{$self->{databases}}));
    }

    my %tmp;
    my $changed = 0;
    foreach my $db (@databases) {
        $tmp{$db}++;
        next if ($self->{databases}{$db});
        $changed++;
    }

    foreach my $db (keys (%{$self->{databases}})) {
        $changed++ unless ($tmp{$db});
    }

    if ($changed) {

        delete ($self->{databases});
        foreach my $db (@databases) {
            $self->{databases}{$db}++;
        }

        if (IDZebra::select_databases($self->{zh}, 
                                                ($#databases + 1), 
                                                \@databases)) {
            logf(LOG_FATAL, 
                 "Could not select database(s) %s errCode=%d",
                 join(",",@databases),
                 $self->errCode());
            return (0);
        } else {
            logf(LOG_LOG,"Database(s) selected: %s",join(",",@databases));
        }
    }
    return (keys(%{$self->{databases}}));
}

# -----------------------------------------------------------------------------
# Error handling
# -----------------------------------------------------------------------------
sub errCode {
    my ($self) = @_;
    return(IDZebra::errCode($self->{zh}));
}

sub errString {
    my ($self) = @_;
    return(IDZebra::errString($self->{zh}));
}

sub errAdd {
    my ($self) = @_;
    return(IDZebra::errAdd($self->{zh}));
}

# -----------------------------------------------------------------------------
# Transaction stuff
# -----------------------------------------------------------------------------
sub begin_trans {
    my ($self, $m) = @_;
    $m = TRANS_RW unless (defined ($m));
    if (my $err = IDZebra::begin_trans($self->{zh},$m)) {
        if ($self->errCode == 2) {
            croak ("TRANS_RW not allowed within TRANS_RO");
        } else {
            croak("Error starting transaction; code:".
                  $self->errCode . " message: " . $self->errString);
        }
    }
}

sub end_trans {
    my ($self) = @_;
    $self->checkzh;
    my $stat = IDZebra::ZebraTransactionStatus->new();
    IDZebra::end_trans($self->{zh}, $stat);
    return ($stat);
}

sub shadow {
    my ($self, $value) = @_;
    $self->checkzh;
    if ($#_ > 0) { 
        $value = 0 unless (defined($value));
        my $r =IDZebra::set_shadow_enable($self->{zh},$value); 
    }
    return (IDZebra::get_shadow_enable($self->{zh}));
}

sub commit {
    my ($self) = @_;
    $self->checkzh;
    if ($self->shadow) {
        return(IDZebra::commit($self->{zh}));
    }
}

# -----------------------------------------------------------------------------
# We don't really need that...
# -----------------------------------------------------------------------------
sub odr_reset {
    my ($self, $name) = @_;
    if ($name !~/^(input|output)$/) {
        croak("Undefined ODR '$name'");
    }
  IDZebra::odr_reset($self->{"odr_$name"});
}

# -----------------------------------------------------------------------------
# Init/compact
# -----------------------------------------------------------------------------
sub init {
    my ($self) = @_;
    $self->checkzh;
    return(IDZebra::init($self->{zh}));
}

sub compact {
    my ($self) = @_;
    $self->checkzh;
    return(IDZebra::compact($self->{zh}));
}

sub update {
    my ($self, %args) = @_;
    $self->checkzh;
    my $rg = $self->_update_args(%args);
    $self->_selectRecordGroup($rg);
    $self->begin_trans;
    IDZebra::repository_update($self->{zh});
    $self->_selectRecordGroup($self->{rg});
    $self->end_trans;
}

sub delete {
    my ($self, %args) = @_;
    $self->checkzh;
    my $rg = $self->_update_args(%args);
    $self->_selectRecordGroup($rg);
    $self->begin_trans;
    IDZebra::repository_delete($self->{zh});
    $self->_selectRecordGroup($self->{rg});
    $self->end_trans;
}

sub show {
    my ($self, %args) = @_;
    $self->checkzh;
    my $rg = $self->_update_args(%args);
    $self->_selectRecordGroup($rg);
    $self->begin_trans;
    IDZebra::repository_show($self->{zh});
    $self->_selectRecordGroup($self->{rg});
    $self->end_trans;
}

sub _update_args {
    my ($self, %args) = @_;
    my $rg = $self->_makeRecordGroup(%args);
    $self->_selectRecordGroup($rg);
    return ($rg);
}

# -----------------------------------------------------------------------------
# Per record update
# -----------------------------------------------------------------------------

sub update_record {
    my ($self, %args) = @_;
    $self->checkzh;
    return(IDZebra::update_record($self->{zh},
                                  $self->_record_update_args(%args)));
}

sub delete_record {
    my ($self, %args) = @_;
    $self->checkzh;
    return(IDZebra::delete_record($self->{zh},
                                  $self->_record_update_args(%args)));
}

sub _record_update_args {
    my ($self, %args) = @_;

    my $sysno   = $args{sysno}      ? $args{sysno}      : 0;
    my $match   = $args{match}      ? $args{match}      : "";
    my $rectype = $args{recordType} ? $args{recordType} : "";
    my $fname   = $args{file}       ? $args{file}       : "<no file>";
    my $force   = $args{force}      ? $args{force}      : 0;

    my $buff;

    if ($args{data}) {
        $buff = $args{data};
    } 
    elsif ($args{file}) {
        CORE::open (F, $args{file}) || warn ("Cannot open $args{file}");
        $buff = join('',(<F>));
        CORE::close (F);
    }
    my $len = length($buff);

    delete ($args{sysno});
    delete ($args{match});
    delete ($args{recordType});
    delete ($args{file});
    delete ($args{data});
    delete ($args{force});

    my $rg = $self->_makeRecordGroup(%args);

    # If no record type is given, then try to find it out from the
    # file extension;
    unless ($rectype) {
        if (my ($ext) = $fname =~ /\.(\w+)$/) {
            my $rg2 = $self->_getRecordGroup($rg->{groupName},$ext);
            $rectype = $rg2->{recordType};
        } 
    }

    $rg->{databaseName} = "Default" unless ($rg->{databaseName});

    unless ($rectype) {
        $rectype="";
    }
    return ($rg, $rectype, $sysno, $match, $fname, $buff, $len, $force);
}

# -----------------------------------------------------------------------------
# CQL stuff
sub cqlmap {
    my ($self,$mapfile) = @_;
    if ($#_ > 0) {
        if ($self->{cql_mapfile} ne $mapfile) {
            unless (-f $mapfile) {
                croak("Cannot find $mapfile");
            }
            if (defined ($self->{cql_ct})) {
              IDZebra::cql_transform_close($self->{cql_ct});
            }
            $self->{cql_ct} = IDZebra::cql_transform_open_fname($mapfile);
            $self->{cql_mapfile} = $mapfile;
        }
    }
    return ($self->{cql_mapfile});
}

sub cql2pqf {
    my ($self, $cqlquery) = @_;
    unless (defined($self->{cql_ct})) {
        croak("CQL map file is not specified yet.");
    }
    my $res = "\0" x 2048;
    my $r = IDZebra::cql2pqf($self->{cql_ct}, $cqlquery, $res, 2048);
    if ($r) {
        carp ("Error transforming CQL query: '$cqlquery', status:$r");
    }
    $res=~s/\0.+$//g;
    return ($res,$r); 
}


# -----------------------------------------------------------------------------
# Search 
# -----------------------------------------------------------------------------
sub search {
    my ($self, %args) = @_;

    $self->checkzh;

    if ($args{cqlmap}) { $self->cqlmap($args{cqlmap}); }

    my $query;
    if ($args{pqf}) {
        $query = $args{pqf};
    }
    elsif ($args{cql}) {
        my $cqlstat;
        ($query, $cqlstat) =  $self->cql2pqf($args{cql});
        unless ($query) {
            croak ("Failed to transform query: '$args{cql}', ".
                   "status: ($cqlstat)");
        }
    }
    unless ($query) {
        croak ("No query given to search");
    }

    my @origdbs;

    if ($args{databases}) {
        @origdbs = $self->databases;
        $self->databases(@{$args{databases}});
    }

    my $rsname = $args{rsname} ? $args{rsname} : $self->_new_setname;

    my $rs = $self->_search_pqf($query, $rsname);

    if ($args{databases}) {
        $self->databases(@origdbs);
    }

    if ($args{sort}) {
        if ($rs->errCode) {
            carp("Sort skipped due to search error: ".
                 $rs->errCode);
        } else {
            $rs->sort($args{sort});
        }
    }

    return ($rs);
}

sub _new_setname {
    my ($self) = @_;
    return ("set_".$self->{rscount}++);
}

sub _search_pqf {
    my ($self, $query, $setname) = @_;

    my $hits = IDZebra::search_PQF($self->{zh},
                                   $self->{odr_input},
                                   $self->{odr_output},
                                   $query,
                                   $setname);

    my $rs  = IDZebra::Resultset->new($self,
                                      name        => $setname,
                                      recordCount => $hits,
                                      errCode     => $self->errCode,
                                      errString   => $self->errString);
    return($rs);
}

# -----------------------------------------------------------------------------
# Sort
#
# Sorting of multiple result sets is not supported by zebra...
# -----------------------------------------------------------------------------

sub sortResultsets {
    my ($self, $sortspec, $setname, @sets) = @_;

    $self->checkzh;

    if ($#sets > 0) {
        croak ("Sorting/merging of multiple resultsets is not supported now");
    }

    my @setnames;
    my $count = 0;
    foreach my $rs (@sets) {
        push (@setnames, $rs->{name});
        $count += $rs->{recordCount};  # is this really sure ??? It doesn't 
                                       # matter now...
    }

    my $status = IDZebra::sort($self->{zh},
                               $self->{odr_output},
                               $sortspec,
                               $setname,
                               \@setnames);

    my $errCode = $self->errCode;
    my $errString = $self->errString;

    logf (LOG_LOG, "Sort status $setname: %d, errCode: %d, errString: %s", 
          $status, $errCode, $errString);

    if ($status || $errCode) {$count = 0;}

    my $rs  = IDZebra::Resultset->new($self,
                                      name        => $setname,
                                      recordCount => $count,
                                      errCode     => $errCode,
                                      errString   => $errString);
    
    return ($rs);
}
# -----------------------------------------------------------------------------
# Scan
# -----------------------------------------------------------------------------
sub scan {
    my ($self, %args) = @_;

    $self->checkzh;

    unless ($args{expression}) {
        croak ("No scan expression given");
    }

    my $sl = IDZebra::ScanList->new($self,%args);

    return ($sl);
}

# ============================================================================

__END__

=head1 NAME

IDZebra::Session - A Zebra database server session for update and retrieval

=head1 SYNOPSIS

  $sess = IDZebra::Session->new(configFile => 'demo/zebra.cfg');
  $sess->open();

  $sess = IDZebra::Session->open(configFile => 'demo/zebra.cfg',
                                 groupName  => 'demo1');

  $sess->group(groupName => 'demo2');

  $sess->init();

  $sess->begin_trans;

  $sess->update(path      =>  'lib');

  my $s1=$sess->update_record(data       => $rec1,
                              recordType => 'grs.perl.pod',
                              groupName  => "demo1",
                              );

  my $stat = $sess->end_trans;

  $sess->databases('demo1','demo2');

  my $rs1 = $sess->search(cqlmap    => 'demo/cql.map',
                          cql       => 'dc.title=IDZebra',
                          databases => [qw(demo1 demo2)]);
  $sess->close;

=head1 DESCRIPTION

Zebra is a high-performance, general-purpose structured text indexing and retrieval engine. It reads structured records in a variety of input formats (eg. email, XML, MARC) and allows access to them through exact boolean search expressions and relevance-ranked free-text queries. 

Zebra supports large databases (more than ten gigabytes of data, tens of millions of records). It supports incremental, safe database updates on live systems. You can access data stored in Zebra using a variety of Index Data tools (eg. YAZ and PHP/YAZ) as well as commercial and freeware Z39.50 clients and toolkits. 

=head1 OPENING AND CLOSING A ZEBRA SESSIONS

For the time beeing only local database services are supported, the same way as calling zebraidx or zebrasrv from the command shell. In order to open a local Zebra database, with a specific configuration file, use

  $sess = IDZebra::Session->new(configFile => 'demo/zebra.cfg');
  $sess->open();

or

  $sess = IDZebra::Session->open(configFile => 'demo/zebra.cfg');

where $sess is going to be the object representing a Zebra Session. Whenever this variable gets out of scope, the session is closed, together with all active transactions, etc... Anyway, if you'd like to close the session, just say:

  $sess->close();

This will
  - close all transactions
  - destroy all result sets and scan lists 
  - close the session

Note, that if I<shadow registers> are enabled, the changes will not be committed automatically.

In the future different database access methods are going to be available, 
like:

  $sess = IDZebra::Session->open(server => 'ostrich.technomat.hu:9999');

You can also use the B<record group> arguments described below directly when calling the constructor, or the open method:

  $sess = IDZebra::Session->open(configFile => 'demo/zebra.cfg',
                                 groupName  => 'demo');


=head1 RECORD GROUPS 

If you manage different sets of records that share common characteristics, you can organize the configuration settings for each type into "groups". See the Zebra manual on the configuration file (zebra.cfg). 

For each open session a default record group is assigned. You can configure it in the constructor, or by the B<group> method:

  $sess->group(groupName => ..., ...)

The following options are available:

=over 4

=item B<groupName>

This will select the named record group, and load the corresponding settings from the configuration file. All subsequent values will overwrite those...

=item B<databaseName>

The name of the (logical) database the updated records will belong to. 

=item B<path>

This path is used for directory updates (B<update>, B<delete> methods);
 
=item B<recordId>

This option determines how to identify your records. See I<Zebra manual: Locating Records>

=item B<recordType>

The record type used for indexing. 

=item B<flagStoreData> 

Specifies whether the records should be stored internally in the Zebra system files. If you want to maintain the raw records yourself, this option should be false (0). If you want Zebra to take care of the records for you, it should be true(1). 

=item B<flagStoreKeys>

Specifies whether key information should be saved for a given group of records. If you plan to update/delete this type of records later this should be specified as 1; otherwise it should be 0 (default), to save register space. 

=item B<flagRw>

?

=item B<fileVerboseLimit>

Skip log messages, when doing a directory update, and the specified number of files are processed...

=item B<databaseNamePath>

?

=item B<explainDatabase>

The name of the explain database to be used

=item B<followLinks>              

Follow links when doing directory update.

=back

You can use the same parameters calling all update methods.

=head1 TRANSACTIONS (READ / WRITE LOCKS)

A transaction is a block of record update (insert / modify / delete) or retrieval procedures. So, all call to such function will implicitly start a transaction, unless one is already started by

  $sess->begin_trans;

or 

  $sess->begin_trans(TRANS_RW)

(these two are equivalents). The effect of this call is a kind of lock: if you call is a write lock is put on the registers, so other processes trying to update the database will be blocked. If there is already an RW (Read-Write) transaction opened by another process, the I<begin_trans> call will be blocked.

You can also use

  $sess->begin_trans(TRANS_RO),

if you would like to put on a "read lock". This one is B<deprecated>, as while you have explicitly opened a transaction for read, you can't open another one for update. For example:

  $sess->begin_trans(TRANS_RO);
  $sess->begin_tran(TRANS_RW); # invalid, die here
  $sess->end_trans;
  $sess->end_trans;

is invalid, but

  $sess->begin_tran(TRANS_RW); 
  $sess->begin_trans(TRANS_RO);
  $sess->end_trans;
  $sess->end_trans;

is valid, but probably useless. Note again, that for each retrieval call, an RO transaction is opened. I<TRANS_RW> and I<TRANS_RO> are exported by default by IDZebra::Session.pm.

For multiple per-record I<updates> it's efficient to start transactions explicitly: otherwise registers (system files, vocabularies, etc..) are updated one by one. After finishing all requested updates, use

  $stat = $sess->end_trans;

The return value is a ZebraTransactionStatus object, containing the following members as a hash reference:

  $stat->{processed} # Number of records processed
  $stat->{updated}   # Number of records processed
  $stat->{deleted}   # Number of records processed
  $stat->{inserted}  # Number of records processed
  $stat->{stime}     # System time used
  $stat->{utime}     # User time used

Normally, if the perl code dies due to some runtime error, or the session is closed, then the API attempts to close all pending transactions.

=head1 THE SHADOW REGISTERS

The Zebra server supports updating of the index structures. That is, you can add, modify, or remove records from databases managed by Zebra without rebuilding the entire index. Since this process involves modifying structured files with various references between blocks of data in the files, the update process is inherently sensitive to system crashes, or to process interruptions: Anything but a successfully completed update process will leave the register files in an unknown state, and you will essentially have no recourse but to re-index everything, or to restore the register files from a backup medium. Further, while the update process is active, users cannot be allowed to access the system, as the contents of the register files may change unpredictably. 

You can solve these problems by enabling the shadow register system in Zebra. During the updating procedure, zebraidx will temporarily write changes to the involved files in a set of "shadow files", without modifying the files that are accessed by the active server processes. If the update procedure is interrupted by a system crash or a signal, you simply repeat the procedure - the register files have not been changed or damaged, and the partially written shadow files are automatically deleted before the new updating procedure commences. 

At the end of the updating procedure (or in a separate operation, if you so desire), the system enters a "commit mode". First, any active server processes are forced to access those blocks that have been changed from the shadow files rather than from the main register files; the unmodified blocks are still accessed at their normal location (the shadow files are not a complete copy of the register files - they only contain those parts that have actually been modified). If the commit process is interrupted at any point during the commit process, the server processes will continue to access the shadow files until you can repeat the commit procedure and complete the writing of data to the main register files. You can perform multiple update operations to the registers before you commit the changes to the system files, or you can execute the commit operation at the end of each update operation. When the commit phase has completed successfully, any running server processes are instructed to switch their operations to the new, operational register, and the temporary shadow files are deleted. 

By default, (in the API !) the use of shadow registers is disabled. If zebra is configured that way (there is a "shadow" entry in zebra.cfg), then the shadow system can be enabled by calling:

 $sess->shadow(1);

or disabled by

 $sess->shadow(0);

If shadow system is enabled, then you have to commit changes you did, by calling:
 
 $sess->commit;

Note, that you can also determine shadow usage in the session constructor:

 $sess = IDZebra::Session->open(configFile => 'demo/zebra.cfg',
                                shadow    => 1);
 
Changes to I<shadow> will not have effect, within a I<transaction> (ie.: a transaction is started either with shadow enabled or disabled). For more details, read Zebra documentation: I<Safe Updating - Using Shadow Registers>.

=head1 UPDATING DATA

There are two ways to update data in a Zebra database using the perl API. You can update an entire directory structure just the way it's done by zebraidx:

  $sess->update(path      =>  'lib');

This will update the database with the files in directory "lib", according to the current record group settings.

  $sess->update();

This will update the database with the files, specified by the default record group setting. I<path> has to be specified there...

  $sess->update(groupName => 'demo1',
                path      =>  'lib');

Update the database with files in "lib" according to the settings of group "demo1"

  $sess->delete(groupName => 'demo1',
                path      =>  'lib');

Delete the records derived from the files in directory "lib", according to the "demo1" group settings. Sounds complex? Read zebra documentation about identifying records.

You can also update records one by one, even directly from the memory:

  $sysno = $sess->update_record(data       => $rec1,
                                recordType => 'grs.perl.pod',
                                groupName  => "demo1");

This will update the database with the given record buffer. Note, that in this case recordType is explicitly specified, as there is no filename given, and for the demo1 group, no default record type is specified. The return value is the system assigned id of the record.

You can also index a single file:

  $sysno = $sess->update_record(file => "lib/IDZebra/Data1.pm");

Or, provide a buffer, and a filename (where filename will only be used to identify the record, if configured that way, and possibly to find out it's record type):

  $sysno = $sess->update_record(data => $rec1,
                                file => "lib/IDZebra/Data1.pm");

And some crazy stuff:

  $sysno = $sess->delete_record(sysno => $sysno);

where sysno in itself is sufficient to identify the record

  $sysno = $sess->delete_record(data => $rec1,
                                recordType => 'grs.perl.pod',
                                groupName  => "demo1");

This case the record is extracted, and if already exists, located in the database, then deleted... 

  $sysno = $sess->delete_record(data       => $rec1,
                                match      => $myid,
                                recordType => 'grs.perl.pod',
                                groupName  => "demo1");

Don't try this at home! This case, the record identifier string (which is normally generated according to the rules set in recordId directive of zebra.cfg) is provided directly....


B<Important:> Note, that one record can be updated only once within a transaction - all subsequent updates are skipped. 

=head1 DATABASE SELECTION

Within a zebra repository you can define logical databases. You can either do this by record groups, or by providing the databaseName argument for update methods. For each record the database name it belongs to is stored. 

For searching, you can select databases by calling:

  $sess->databases('db1','db2');

This will not do anything if the given and only the given databases are already selected. You can get the list of the actually selected databases, by calling:
  
  @dblist = $sess->databases();

=head1 SEARCHING

It's nice to be able to store data in your repository... But it's useful to reach it as well. So this is how to do searching:

  $rs = $sess->search(databases => [qw(demo1,demo2)], # optional
                      pqf       => '@attr 1=4 computer');

This is going to execute a search in databases demo1 and demo2, for title 'com,puter'. This is a PQF (Prefix Query Format) search, see YAZ documentation for details. The database selection is optional: if it's provided, the given list of databases is selected for this particular search, then the original selection is restored.

=head2 CCL searching

Not all users enjoy typing in prefix query structures and numerical attribute values, even in a minimalistic test client. In the library world, the more intuitive Common Command Language (or ISO 8777) has enjoyed some popularity - especially before the widespread availability of graphical interfaces. It is still useful in applications where you for some reason or other need to provide a symbolic language for expressing boolean query structures. 

The CCL searching is not currently supported by this API.

=head2 CQL searching

CQL - Common Query Language - was defined for the SRW protocol. In many ways CQL has a similar syntax to CCL. The objective of CQL is different. Where CCL aims to be an end-user language, CQL is the protocol query language for SRW. 

In order to map CQL queries to Zebra internal search structures, you have to define a mapping, the way it is described in YAZ documentation: I<Specification of CQL to RPN mapping>. The mapping is interpreted by the method:

  $sess->cqlmap($mapfile);

Or, you can directly provide the I<mapfile> parameter for the search:

  $rs = $sess->search(cqlmap    => 'demo/cql.map',
                      cql       => 'dc.title=IDZebra');

As you see, CQL searching is so simple: just give the query in the I<cql> parameter.

=head2 Sorting

If you'd like the search results to be sorted, use the I<sort> parameter:

  $rs = $sess->search(cql       => 'IDZebra',
                      sort      => '1=4 ia');
  
Note, that B<currently> this is (almost) equivalent to

  $rs = $sess->search(cql       => 'IDZebra');
  $rs->sort('1=4 ia');
  
but in the further versions of Zebra and this API a single phase search and sort will take place, optimizing performance. For more details on sorting, see I<IDZebra::ResultSet> manpage.

=head1 RESULTSETS

As you have seen, the result of the search request is a I<Resultset> object.
It contains number of hits, and search status, and can be used to sort and retrieve the resulting records.

  $count = $rs->count;

  printf ("RS Status is %d (%s)\n", $rs->errCode, $rs->errString);

I<$rs-E<gt>errCode> is 0, if there were no errors during search. Read the I<IDZebra::Resultset> manpage for more details.

=head1 SCANNING

Zebra supports scanning index values. The result of the 

  $sl = $sess->scan(expression => "a");

call is an I<IDZebra::ScanList> object, what you can use to list the values. The scan expression has to be provided in a PQF like format. Examples:

B< a> (scan trough words of "default", "Any" indexes)


B< @attr 1=1016 a> (same effect)


B< @attr 1=4 @attr 6=2 a>  (scan trough titles as phrases)

An illegal scan expression will cause your code to die. If you'd like to select databases just for the scan call, you can optionally use the I<databases> parameter:

  $sl = $sess->scan(expression => "a",
                    databases  => [qw(demo1 demo2)]);
  
You can use the I<IDZebra::ScanList> object returned by the i<scan> method, to reach the result. Check I<IDZebra::ScanList> manpage for more details.

=head1 SESSION STATUS AND ERRORS

Most of the API calls causes die, if an error occures. You avoid this, by using eval {} blocks. The following methods are available to get the status of Zebra service:

=over 4

=item B<errCode>

The Zebra provided error code... (for the result of the last call);

=item B<errString>

Error string corresponding to the message

=item B<errAdd>

Additional information for the status

=back

This functionality may change, see TODO.

=head1 LOGGING AND MISC. FUNCTIONS

Zebra provides logging facility for the internal events, and also for application developers trough the API. See manpage I<IDZebra::Logger> for details.

=over 4

=item B<IDZebra::LogFile($filename)>

Will set the output file for logging. By default it's STDERR;

=item B<IDZebra::LogLevel(15)>

Set log level. 0 for no logs. See IDZebra::Logger for usable flags.

=back

Some other functions

=over 4

=item B<$sess-E<gt>init>

Initialize, and clean registers. This will remove all data!

=item B<$sess-E<gt>compact>

Compact the registers (? does this work)

=item B<$sess-E<gt>show>

Doesn't have too much meaning. Don't try :)

=back

=head1 TODO

=over 4

=item B<Clean up error handling>

By default all zebra errors should cause die. (such situations could be avoided by using eval {}), and then check for errCode, errString... An optional flag or package variable should be introduced to override this, and skip zebra errors, to let the user decide what to do.

=item B<Make the package self-distributable>

Build and link with installed header and library files

=item B<Testing>

Test shadow system, unicode...

=item B<C API>

Cleanup, arrange, remove redundancy

=back

=head1 COPYRIGHT

Fill in

=head1 AUTHOR

Peter Popovics, pop@technomat.hu

=head1 SEE ALSO

Zebra documentation, Zebra::ResultSet, Zebra::ScanList, Zebra::Logger manpages

=cut