From 7424460be5718c82f35b3257ae4459ecdb656fa7 Mon Sep 17 00:00:00 2001 From: Thomas Klausner Date: Wed, 15 Nov 2023 22:16:23 +0100 Subject: [PATCH] Bug 35345: Add --where option to rebuild_elasticsearch.pl MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Sometimes we need to only re-index a subset of our bibliographic data or authorities. Currently this is only possible by enumerating all id (-bn or -ai), which does not work well when indexing eg 100.000 items of a 2.000.000 DB. Re-indexing everything is also overkill. This patch adds an `--where` flag to misc/search_tools/rebuild_elasticsearch.pl which can take arbitrary SQL (that of course has to match the respective tables) and adds it as an additional param to the resultset to index To test, start koha-testing-docker with ElasticSearch enabled, for example via `ktd --es7 up Before applying the patch, rebuild_elasticsearch will index all data: Biblios: $ misc/search_tools/rebuild_elasticsearch.pl -b -v [12387] Checking state of biblios index [12387] Indexing biblios [12387] Committing final records... [12387] Total 435 records indexed (there might be a waring regarding a broken biblio, which can be ignored) Auth: $ misc/search_tools/rebuild_elasticsearch.pl -a -v [12546] Checking state of authorities index [12546] Indexing authorities [12546] 1000 records processed [12546] Committing final records... [12546] Total 1706 records indexed Now apply the patch Biblio, limit by range of biblioid: $ misc/search_tools/rebuild_elasticsearch.pl -b -v --where "biblionumber between 100 and 150" [12765] Checking state of biblios index [12765] Indexing biblios [12765] Committing final records... [12765] Total 50 records indexed Note that only 50 records where indexed (instead of the whole set of 435 records) Auth, limit by authtypecode: $ misc/search_tools/rebuild_elasticsearch.pl -a -v --where "authtypecode = 'GEOGR_NAME'" [12848] Checking state of authorities index [12848] Indexing authorities [12848] Committing final records... [12848] Total 142 records indexed Again, only 142 have been indexed. Note that this patch also includes perltidied versions of the affected files (because the qa script complained) Sponsored-by: Steiermärkische Landesbibliothek --- Koha/BiblioUtils.pm | 43 +++++---- Koha/MetadataRecord/Authority.pm | 99 +++++++++++-------- misc/search_tools/rebuild_elasticsearch.pl | 105 ++++++++++++--------- 3 files changed, 148 insertions(+), 99 deletions(-) diff --git a/Koha/BiblioUtils.pm b/Koha/BiblioUtils.pm index cb7baed0db..d28dbf7db3 100644 --- a/Koha/BiblioUtils.pm +++ b/Koha/BiblioUtils.pm @@ -34,7 +34,6 @@ use Koha::MetadataIterator; use Koha::Database; use Modern::Perl; - use base qw(Koha::MetadataRecord); __PACKAGE__->mk_accessors(qw( record schema id datatype )); @@ -51,8 +50,8 @@ the biblionumber can be provided too. =cut sub new { - my $class = shift; - my $record = shift; + my $class = shift; + my $record = shift; my $biblionumber = shift; my $self = $class->SUPER::new( @@ -89,10 +88,10 @@ It will return C if the biblio doesn't exist. =cut sub get_from_biblionumber { - my ($class, $bibnum, %options) = @_; + my ( $class, $bibnum, %options ) = @_; - my $marc = $class->get_marc_biblio($bibnum, %options); - return $class->new($marc, $bibnum); + my $marc = $class->get_marc_biblio( $bibnum, %options ); + return $class->new( $marc, $bibnum ); } =head2 get_all_biblios_iterator @@ -120,34 +119,41 @@ records instead of all. =cut sub get_all_biblios_iterator { - my ($class, %options) = @_; + my ( $class, %options ) = @_; my $search_terms = {}; - my ($slice_modulo, $slice_count); - if ($options{slice}) { - $slice_count = $options{slice}->{count}; + my ( $slice_modulo, $slice_count ); + if ( $options{slice} ) { + $slice_count = $options{slice}->{count}; $slice_modulo = $options{slice}->{index}; $search_terms = \[ 'mod(biblionumber, ?) = ?', $slice_count, $slice_modulo ]; } my $search_options = { columns => [qw/ biblionumber /] }; - if ( $options{desc} ){ - $search_options->{order_by} = { -desc => 'biblionumber' }; + if ( $options{desc} ) { + $search_options->{order_by} = { -desc => 'biblionumber' }; } my $database = Koha::Database->new(); my $schema = $database->schema(); - my $rs = Koha::Biblios->search( + my $rs = Koha::Biblios->search( $search_terms, - $search_options ); + $search_options + ); + + if ( my $sql = $options{where} ) { + $rs = $rs->search( \[$sql] ); + } + my $next_func = sub { + # Warn and skip bad records, otherwise we break the loop while (1) { my $row = $rs->next(); return if !$row; my $next = eval { - my $marc = $row->metadata->record({ embed_items => 1 }); - $class->new($marc, $row->biblionumber); + my $marc = $row->metadata->record( { embed_items => 1 } ); + $class->new( $marc, $row->biblionumber ); }; if ($@) { warn sprintf "Something went wrong reading record for biblio %s: %s\n", $row->biblionumber, $@; @@ -180,10 +186,9 @@ If set to true, item data is embedded in the record. Default is to not do this. =cut sub get_marc_biblio { - my ($class, $bibnum, %options) = @_; + my ( $class, $bibnum, %options ) = @_; - my $record = Koha::Biblios->find($bibnum) - ->metadata->record( { $options{item_data} ? ( embed_items => 1 ) : () } ); + my $record = Koha::Biblios->find($bibnum)->metadata->record( { $options{item_data} ? ( embed_items => 1 ) : () } ); return $record; } diff --git a/Koha/MetadataRecord/Authority.pm b/Koha/MetadataRecord/Authority.pm index d16970a828..e9a8466c43 100644 --- a/Koha/MetadataRecord/Authority.pm +++ b/Koha/MetadataRecord/Authority.pm @@ -67,7 +67,6 @@ sub new { return $self; } - =head2 get_from_authid my $auth = Koha::MetadataRecord::Authority->get_from_authid($authid); @@ -80,23 +79,35 @@ unfortunate but unavoidable fact. =cut sub get_from_authid { - my $class = shift; - my $authid = shift; + my $class = shift; + my $authid = shift; my $marcflavour = lc C4::Context->preference("marcflavour"); - my $dbh=C4::Context->dbh; - my $sth=$dbh->prepare("select authtypecode, marcxml from auth_header where authid=?"); + my $dbh = C4::Context->dbh; + my $sth = $dbh->prepare("select authtypecode, marcxml from auth_header where authid=?"); $sth->execute($authid); - my ($authtypecode, $marcxml) = $sth->fetchrow; - my $record=eval {MARC::Record->new_from_xml(StripNonXmlChars($marcxml),'UTF-8', - (C4::Context->preference("marcflavour") eq "UNIMARC"?"UNIMARCAUTH":C4::Context->preference("marcflavour")))}; + my ( $authtypecode, $marcxml ) = $sth->fetchrow; + my $record = eval { + MARC::Record->new_from_xml( + StripNonXmlChars($marcxml), 'UTF-8', + ( + C4::Context->preference("marcflavour") eq "UNIMARC" + ? "UNIMARCAUTH" + : C4::Context->preference("marcflavour") + ) + ); + }; return if ($@); $record->encoding('UTF-8'); - my $self = $class->SUPER::new( { authid => $authid, - authtypecode => $authtypecode, - schema => $marcflavour, - record => $record }); + my $self = $class->SUPER::new( + { + authid => $authid, + authtypecode => $authtypecode, + schema => $marcflavour, + record => $record + } + ); bless $self, $class; return $self; @@ -111,16 +122,24 @@ Create the Koha::MetadataRecord::Authority object associated with the provided a =cut sub get_from_breeding { - my $class = shift; + my $class = shift; my $import_record_id = shift; - my $marcflavour = lc C4::Context->preference("marcflavour"); + my $marcflavour = lc C4::Context->preference("marcflavour"); - my $dbh=C4::Context->dbh; - my $sth=$dbh->prepare("select marcxml from import_records where import_record_id=? and record_type='auth';"); + my $dbh = C4::Context->dbh; + my $sth = $dbh->prepare("select marcxml from import_records where import_record_id=? and record_type='auth';"); $sth->execute($import_record_id); my $marcxml = $sth->fetchrow; - my $record=eval {MARC::Record->new_from_xml(StripNonXmlChars($marcxml),'UTF-8', - (C4::Context->preference("marcflavour") eq "UNIMARC"?"UNIMARCAUTH":C4::Context->preference("marcflavour")))}; + my $record = eval { + MARC::Record->new_from_xml( + StripNonXmlChars($marcxml), 'UTF-8', + ( + C4::Context->preference("marcflavour") eq "UNIMARC" + ? "UNIMARCAUTH" + : C4::Context->preference("marcflavour") + ) + ); + }; return if ($@); $record->encoding('UTF-8'); @@ -131,10 +150,13 @@ sub get_from_breeding { require C4::AuthoritiesMarc; my $authtypecode = C4::AuthoritiesMarc::GuessAuthTypeCode($record); - my $self = $class->SUPER::new( { - schema => $marcflavour, - authtypecode => $authtypecode, - record => $record }); + my $self = $class->SUPER::new( + { + schema => $marcflavour, + authtypecode => $authtypecode, + record => $record + } + ); bless $self, $class; return $self; @@ -142,8 +164,8 @@ sub get_from_breeding { sub authorized_heading { my ($self) = @_; - if ($self->schema =~ m/marc/) { - return Koha::Util::MARC::getAuthorityAuthorizedHeading($self->record, $self->schema); + if ( $self->schema =~ m/marc/ ) { + return Koha::Util::MARC::getAuthorityAuthorizedHeading( $self->record, $self->schema ); } return; } @@ -173,14 +195,12 @@ records instead of all. =cut sub get_all_authorities_iterator { - my ($self, %options) = @_; + my ( $self, %options ) = @_; - my $search_terms = { - marcxml => { '!=', undef } - }; - my ($slice_modulo, $slice_count); - if ($options{slice}) { - $slice_count = $options{slice}->{count}; + my $search_terms = { marcxml => { '!=', undef } }; + my ( $slice_modulo, $slice_count ); + if ( $options{slice} ) { + $slice_count = $options{slice}->{count}; $slice_modulo = $options{slice}->{index}; $search_terms = { '-and' => [ @@ -191,24 +211,29 @@ sub get_all_authorities_iterator { } my $search_options->{columns} = [qw/ authid /]; - if ($options{desc}) { + if ( $options{desc} ) { $search_options->{order_by} = { -desc => 'authid' }; } my $database = Koha::Database->new(); my $schema = $database->schema(); - my $rs = - $schema->resultset('AuthHeader')->search( + my $rs = $schema->resultset('AuthHeader')->search( $search_terms, - $search_options); + $search_options + ); + + if ( my $sql = $options{where} ) { + $rs = $rs->search( \[$sql] ); + } my $next_func = sub { + # Warn and skip bad records, otherwise we break the loop while (1) { my $row = $rs->next(); return if !$row; - my $auth = __PACKAGE__->get_from_authid($row->authid); - if (!$auth) { + my $auth = __PACKAGE__->get_from_authid( $row->authid ); + if ( !$auth ) { warn "Something went wrong reading record for authority $row->authid: $@\n"; next; } diff --git a/misc/search_tools/rebuild_elasticsearch.pl b/misc/search_tools/rebuild_elasticsearch.pl index 337ac19dd2..ffb7301238 100755 --- a/misc/search_tools/rebuild_elasticsearch.pl +++ b/misc/search_tools/rebuild_elasticsearch.pl @@ -34,6 +34,7 @@ B [B<--desc>] [B<-bn|--bnumber>] [B<-ai|--authid>] +[B<-w|--where SQL>] [B<-p|--processes>] [B<-v|--verbose>] [B<-h|--help>] @@ -87,6 +88,10 @@ repeated. Only index the supplied authority id, mostly for testing purposes. May be repeated. +=item B<-w|--where> + +Pass some additional SQL to limit the records to be indexed. + =item B<-p|--processes> Number of processes to use for indexing. This can be used to do more indexing @@ -125,13 +130,13 @@ use Pod::Usage qw( pod2usage ); use Try::Tiny qw( catch try ); my $verbose = 0; -my $commit = 5000; -my ($delete, $reset, $help, $man, $processes); -my ($index_biblios, $index_authorities); -my (@biblionumbers,@authids); +my $commit = 5000; +my ( $delete, $reset, $help, $man, $processes ); +my ( $index_biblios, $index_authorities ); +my ( @biblionumbers, @authids, $where ); my $desc; -$|=1; # flushes output +$| = 1; # flushes output GetOptions( 'c|commit=i' => \$commit, @@ -142,6 +147,7 @@ GetOptions( 'desc' => \$desc, 'bn|bnumber=i' => \@biblionumbers, 'ai|authid=i' => \@authids, + 'w|where=s' => \$where, 'p|processes=i' => \$processes, 'v|verbose+' => \$verbose, 'h|help' => \$help, @@ -149,62 +155,69 @@ GetOptions( ); # Default is to do both -unless ($index_authorities || $index_biblios) { +unless ( $index_authorities || $index_biblios ) { $index_authorities = $index_biblios = 1; } -if ($processes && ( @biblionumbers || @authids) ) { +if ( $processes && ( @biblionumbers || @authids ) ) { die "Argument p|processes cannot be combined with bn|bnumber or ai|authid"; } -pod2usage(1) if $help; +pod2usage(1) if $help; pod2usage( -exitstatus => 0, -verbose => 2 ) if $man; _sanity_check(); -if ($reset){ +if ($reset) { Koha::SearchEngine::Elasticsearch->reset_elasticsearch_mappings; $delete = 1; } -_verify_index_state($Koha::SearchEngine::Elasticsearch::BIBLIOS_INDEX, $delete) if ($index_biblios); -_verify_index_state($Koha::SearchEngine::Elasticsearch::AUTHORITIES_INDEX, $delete) if ($index_authorities); +_verify_index_state( $Koha::SearchEngine::Elasticsearch::BIBLIOS_INDEX, $delete ) if ($index_biblios); +_verify_index_state( $Koha::SearchEngine::Elasticsearch::AUTHORITIES_INDEX, $delete ) if ($index_authorities); my $slice_index = 0; my $slice_count = ( $processes //= 1 ); my %iterator_options; -if ($slice_count > 1) { +if ( $slice_count > 1 ) { + # Fire up child processes for processing slices from 2 on. This main process will handle slice 1. $slice_index = 0; - for (my $proc = 1; $proc < $slice_count; $proc++) { + for ( my $proc = 1 ; $proc < $slice_count ; $proc++ ) { my $pid = fork(); die "Failed to fork a child process\n" unless defined $pid; - if ($pid == 0) { + if ( $pid == 0 ) { + # Child process, give it a slice to process $slice_index = $proc; last; } } + # Fudge the commit count a bit to spread out the Elasticsearch commits $commit *= 1 + 0.10 * $slice_index; - $commit = int( $commit ); - _log(1, "Processing slice @{[$slice_index + 1]} of $slice_count\n"); + $commit = int($commit); + _log( 1, "Processing slice @{[$slice_index + 1]} of $slice_count\n" ); $iterator_options{slice} = { index => $slice_index, count => $slice_count }; } -if( $desc ){ +if ($desc) { $iterator_options{desc} = 1; } +if ($where) { + $iterator_options{where} = $where; +} + my $next; if ($index_biblios) { - _log(1, "Indexing biblios\n"); + _log( 1, "Indexing biblios\n" ); if (@biblionumbers) { $next = sub { my $r = shift @biblionumbers; return () unless defined $r; - return ($r, Koha::BiblioUtils->get_from_biblionumber($r, item_data => 1 )); + return ( $r, Koha::BiblioUtils->get_from_biblionumber( $r, item_data => 1 ) ); }; } else { my $records = Koha::BiblioUtils->get_all_biblios_iterator(%iterator_options); @@ -212,16 +225,16 @@ if ($index_biblios) { $records->next(); } } - _do_reindex($next, $Koha::SearchEngine::Elasticsearch::BIBLIOS_INDEX); + _do_reindex( $next, $Koha::SearchEngine::Elasticsearch::BIBLIOS_INDEX ); } if ($index_authorities) { - _log(1, "Indexing authorities\n"); + _log( 1, "Indexing authorities\n" ); if (@authids) { $next = sub { my $r = shift @authids; return () unless defined $r; my $a = Koha::MetadataRecord::Authority->get_from_authid($r); - return ($r, $a); + return ( $r, $a ); }; } else { my $records = Koha::MetadataRecord::Authority->get_all_authorities_iterator(%iterator_options); @@ -229,12 +242,13 @@ if ($index_authorities) { $records->next(); } } - _do_reindex($next, $Koha::SearchEngine::Elasticsearch::AUTHORITIES_INDEX); + _do_reindex( $next, $Koha::SearchEngine::Elasticsearch::AUTHORITIES_INDEX ); } -if ($slice_index == 0) { +if ( $slice_index == 0 ) { + # Main process, wait for children - for (my $proc = 1; $proc < $processes; $proc++) { + for ( my $proc = 1 ; $proc < $processes ; $proc++ ) { wait(); } } @@ -252,21 +266,22 @@ Checks the index state and recreates it if requested. sub _verify_index_state { my ( $index_name, $recreate ) = @_; - _log(1, "Checking state of $index_name index\n"); + _log( 1, "Checking state of $index_name index\n" ); my $indexer = Koha::SearchEngine::Elasticsearch::Indexer->new( { index => $index_name } ); if ($recreate) { - _log(1, "Dropping and recreating $index_name index\n"); + _log( 1, "Dropping and recreating $index_name index\n" ); $indexer->drop_index() if $indexer->index_exists(); $indexer->create_index(); - } - elsif (!$indexer->index_exists) { + } elsif ( !$indexer->index_exists ) { + # Create index if does not exist $indexer->create_index(); - } elsif ($indexer->is_index_status_ok) { + } elsif ( $indexer->is_index_status_ok ) { + # Update mapping unless index is some kind of problematic state $indexer->update_mappings(); - } elsif ($indexer->is_index_status_recreate_required) { + } elsif ( $indexer->is_index_status_recreate_required ) { warn qq/Index "$index_name" has status "recreate required", suggesting it should be recreated/; } } @@ -293,7 +308,7 @@ sub _do_reindex { my $record = $record->record; $count++; if ( $verbose == 1 ) { - _log( 1, "$count records processed\n" ) if ( $count % 1000 == 0); + _log( 1, "$count records processed\n" ) if ( $count % 1000 == 0 ); } else { _log( 2, "$id\n" ); } @@ -303,13 +318,13 @@ sub _do_reindex { if ( !( --$commit_count ) ) { _log( 1, "Committing $commit records...\n" ); my $response; - try{ + try { $response = $indexer->update_index( \@id_buffer, \@commit_buffer ); _handle_response($response); _log( 1, "Commit complete\n" ); } catch { - _log(1,"Elasticsearch exception thrown: ".$_->type."\n"); - _log(2,"Details: ".$_->details."\n"); + _log( 1, "Elasticsearch exception thrown: " . $_->type . "\n" ); + _log( 2, "Details: " . $_->details . "\n" ); }; $commit_count = $commit; @id_buffer = (); @@ -333,6 +348,7 @@ Checks some basic stuff to ensure that it's sane before we start. =cut sub _sanity_check { + # Do we have an elasticsearch block defined? my $conf = C4::Context->config('elasticsearch'); die "No 'elasticsearch' block is defined in koha-conf.xml.\n" if ( !$conf ); @@ -346,14 +362,17 @@ Parse the return from update_index and display errors depending on verbosity of sub _handle_response { my ($response) = @_; - if( $response->{errors} eq 'true' ){ + if ( $response->{errors} eq 'true' ) { _log( 1, "There were errors during indexing\n" ); - if ( $verbose > 1 ){ - foreach my $item (@{$response->{items}}){ + if ( $verbose > 1 ) { + foreach my $item ( @{ $response->{items} } ) { next unless defined $item->{index}->{error}; - print "Record #" . $item->{index}->{_id} . " " . - $item->{index}->{error}->{reason} . " (" . $item->{index}->{error}->{type} . ") : " . - $item->{index}->{error}->{caused_by}->{type} . " (" . $item->{index}->{error}->{caused_by}->{reason} . ")\n"; + print "Record #" + . $item->{index}->{_id} . " " + . $item->{index}->{error}->{reason} . " (" + . $item->{index}->{error}->{type} . ") : " + . $item->{index}->{error}->{caused_by}->{type} . " (" + . $item->{index}->{error}->{caused_by}->{reason} . ")\n"; } } } @@ -371,7 +390,7 @@ include a trailing newline automatically. =cut sub _log { - my ($level, $msg) = @_; + my ( $level, $msg ) = @_; - print "[$$] $msg" if ($verbose >= $level); + print "[$$] $msg" if ( $verbose >= $level ); } -- 2.34.1