From a620b41e621259df64a5c2dcad46a30e3dafddac Mon Sep 17 00:00:00 2001
From: Andreas Romeyke <andreas.romeyke@slub-dresden.de>
Date: Mon, 14 Apr 2025 19:06:48 +0200
Subject: [PATCH] - fixed bug, query twice, first one to get total counts,
second one to set pagination (disable it)
---
lib/SLUB/LZA/TA/Command/sample_testing.pm | 20 +++++++++++++++++++-
1 file changed, 19 insertions(+), 1 deletion(-)
diff --git a/lib/SLUB/LZA/TA/Command/sample_testing.pm b/lib/SLUB/LZA/TA/Command/sample_testing.pm
index fe1ea1d..015d1c0 100644
--- a/lib/SLUB/LZA/TA/Command/sample_testing.pm
+++ b/lib/SLUB/LZA/TA/Command/sample_testing.pm
@@ -201,7 +201,7 @@ sub _execute {
# next lines extend query with reporting
$aips_query->{'_source'} = {'includes' =>
['uuid', 'filePath', 'transferMetadata.bim:bag-info_dict.bim:SLUBArchiv-lzaId', 'created']};
- #p($aips_query);
+ $aips_query->{size} = 1; # we only need to find total size
$aips_response = SLUB::LZA::TA::Archivematica::Elasticsearch::query_elasticsearch(
$SLUB::LZA::TA::config{elasticsearch_protocol},
$SLUB::LZA::TA::config{elasticsearch_host},
@@ -218,6 +218,24 @@ sub _execute {
if ($opt->{debug}) {
say STDERR "found $max_found_aips AIPs, use only a $opt->{factor} subsample of $sample_size AIPs";
}
+
+ $aips_query->{size} = $max_found_aips;
+ # query again
+ $aips_response = SLUB::LZA::TA::Archivematica::Elasticsearch::query_elasticsearch(
+ $SLUB::LZA::TA::config{elasticsearch_protocol},
+ $SLUB::LZA::TA::config{elasticsearch_host},
+ $SLUB::LZA::TA::config{elasticsearch_port},
+ 'aips', # indexname
+ $aips_query, # query_hash ref
+ {
+ debug => $opt->{debug},
+
+ }
+ );
+ my $max_found_aips2 = $aips_response->{hits}->{total};
+ my $max_found_aips2_selected = scalar( @{ $aips_response->{hits}->{hits} } );
+ die "different result set between two consecutive calls of ElasticSearch query" if ($max_found_aips != $max_found_aips2);
+ die "limited size in response of ElasticSearch query, expecting $max_found_aips, got $max_found_aips2_selected" if ($max_found_aips != $max_found_aips2_selected);
# select first sample size aips
my @sample_set = sort {$a->{uuid} cmp $b->{uuid}} List::Util::sample $sample_size, map {$_->{'_source'} } @{ $aips_response->{hits}->{hits} };
my @result = map { _check_aip($_, $opt) } @sample_set;
--
GitLab