- init (adapted from Submission Application)

lib/SLUB/LZA/TA/Archivematica/Elasticsearch.pm

+package SLUB::LZA::TA::Archivematica::Elasticsearch;
+# contains stuff related to Artefactuals Archivematica
+# Andreas Romeyke (romeyke@slub-dresden.de)
+# HINT: the server is v6.x.y of elasticsearch, therefore this code is using old client
+use v5.36;
+use namespace::autoclean -except => qr{import};
+use Search::Elasticsearch 6.81; # install with: cpanm Search::Elasticsearch~"<7.0"
+use Search::Elasticsearch::Client::6_0;
+
+use Exporter 'import';
+our @EXPORT = qw(
+    version_elasticsearch
+    query_elasticsearch_count
+    query_elasticsearch
+    prepare_query
+);
+our @EXPORT_OK = qw();
+
+#$Search::Elasticsearch::Error::DEBUG=2;
+
+sub _instantiate($protocol, $host, $port) {
+    return Search::Elasticsearch->new(
+        nodes    => "$protocol://$host:$port",
+        #trace_to => 'Stderr',
+        #cnx_pool => 'Sniff',
+        cxn_pool => 'Static::NoPing',
+        client => '6_0::Direct'
+    );
+}
+
+sub version_elasticsearch($protocol, $host, $port) {
+    my $e = _instantiate($protocol, $host, $port);
+    my $info = $e->info;
+    return $info->{version}->{number};
+}
+
+sub query_elasticsearch($protocol, $host, $port, $index_name, $query_hash, $options= undef) {
+    # HINT: filter_paths not well supported
+
+    # next lines to deep copy hash, to avoid sideeffects in caller
+    my $local_query_hash;
+    foreach my $key (keys %{$query_hash} ) {
+        $local_query_hash->{$key} = $query_hash->{$key}
+    }
+    if ((defined $options) and ($options->{no_source} == 1)) {
+        $local_query_hash->{_source}="false"; # avoids transfering complete source tree in results
+    }
+    my $e = _instantiate($protocol, $host, $port);
+    my $res = $e->search(
+        index => $index_name,
+        body  => $local_query_hash, #hashref
+    );
+    return $res;
+}
+
+sub query_elasticsearch_count($protocol, $host, $port, $index_name, $query_hash) {
+    my $res = query_elasticsearch($protocol, $host, $port, $index_name, $query_hash, {no_source => 1});
+    return $res->{hits}->{total};
+}
+
+sub prepare_query ($opt) {
+    my @must;
+    my @should;
+    # example query:
+    # query => {
+    #                 bool => {
+    #                     must   =>
+    #                         [
+    #                             {
+    #                                 match => {
+    #                                     "transferMetadata.SLUBArchiv-externalWorkflow", "testcases",
+    #                                 },
+    #                             },
+    #                         ],
+    #                     should => [
+    #                         {
+    #                             match => {
+    #                                 "transferMetadata.SLUBArchiv-externalId", "test-sip_2020-07-17_13-40-17_96152",
+    #                             },
+    #                         }
+    #                     ]
+    #                 },
+    #             }
+
+    # Is a SLUB-specific ID in the archive?
+    #     curl -XGET -H 'Content-Type: application/json'
+    #     'localhost:9200/aips/_search?pretty=true' -d '
+    #     {
+    #        "query": {
+    #           "term": {
+    #               "transferMetadata.SLUBArchiv-lzaId.keyword":
+    #               "SLUB:LZA:testworkflow:testcases:test-sip_2020-07-17_13-40-17_96152"
+    #           }
+    #        }
+    #     }'
+    #
+    #     What is the "SLUBArchiv-exportToArchiveDate" value of all matches
+    #     curl -XGET -H 'Content-Type: application/json'
+    #     'localhost:9200/aips/_search?pretty=true' -d '
+    #     {
+    #         "query": {
+    #             "match_all": {}
+    #         },
+    #         "_source": [
+    #             "transferMetadata.SLUBArchiv-lzaId",
+    #             "transferMetadata.SLUBArchiv-exportToArchiveDate"
+    #         ]
+    #     }'
+    #
+    #     Find all AIPs with a SLUB-specific rights status
+    #     curl -XGET -H 'Content-Type: application/json'
+    # 'localhost:9200/aips/_search?pretty=true' -d '
+    # {
+    #   "query": {
+    #     "term": {
+    # "transferMetadata.slubarchiv:rightsRecord_dict.slubarchiv:copyrightStatus.keyw
+    # ord": "copyrighted"
+    #     }
+    #   },
+    #   "_source": ["uuid"]
+    # }'
+    #
+    #   Reporting for a given date range: AIP count,
+    # number of archived files, size of archived data, etc.
+    #   curl -XGET -H 'Content-Type: application/json'
+    # 'localhost:9200/aips/_search?pretty=true' -d '
+    # {
+    #   "query": {
+    #     "range": {
+    #       "created": {
+    #         "gte": 1638313200,
+    #         "lt": 1640991600
+    #       }
+    #     }
+    #   },
+    #   "aggs": {
+    #     "total_size": {"sum": {"field": "size"}},
+    #     "total_file_count": {"sum": {"field": "file_count"}}
+    #   },
+    #   "size": 0
+    #  }'
+
+    #    Free Search - Find AIPs based on catalog data
+    # included by producers
+    #    curl -XGET -H 'Content-Type: application/json'
+    # 'localhost:9200/aips/_search?pretty=true' -d '
+    # {
+    #"query": {
+    #    "query_string": {
+    #        "fields": [
+    #            "transferMetadata.record_dict.datafield_dict.subfield",
+    #            "transferMetadata.mods:mods_dict.mods:titleInfo_dict.mods:title"
+    #        ],
+    #            "query": "Riesaer Tageblatt und Anzeiger"
+    #    }
+    #},
+    #"_source": ["uuid"]
+    #}'
+
+
+    if (exists $opt->{source}) {
+        #push @queries, "IE.sourceMD.content=\"$opt->{source}\"";
+    }
+    if (exists $opt->{lzaid}) {
+        push @must, {
+            "term" => {
+                "transferMetadata.SLUBArchiv-lzaId.keyword" =>
+                    "$opt->{lzaid}"
+            }
+        };
+    }
+    if (exists $opt->{descriptive}) {
+        push @should => {
+            match => {
+                "transferMetadata.dc" => "$opt->{descriptive}",
+            }
+        }
+    }
+    if (exists $opt->{creationdate}) {
+        push @should => {
+            "created" => $opt->{creationdate}
+        };
+    }
+    if (exists $opt->{modificationdate}) {
+        #push @queries, "FILE.objectCharacteristics.modificationDate==$opt->{modificationdate}";
+    }
+    if (exists $opt->{'with_format'}) {
+        #push @queries, "FILE.generalFileCharacteristics.formatLibraryId==$opt->{'with_format'}";
+    }
+    # unsupported by Rosetta <= 7.0
+    #if (exists $opt->{'without_format'}) {
+    #    push @queries, "FILE.generalFileCharacteristics.formatLibraryId!=$opt->{'without_format'}";
+    #}
+    if (exists $opt->{'with_valid_files'}) {
+        #push @queries, "FILE.fileValidation.isValid==true";
+    }
+    if (exists $opt->{'with_invalid_files'}) {
+        #push @queries, "FILE.fileValidation.isValid==false";
+    }
+    if (exists $opt->{'with_passed_viruschecks'}) {
+        #push @queries, "FILE.fileVirusCheck.status==true";
+    }
+    if (exists $opt->{'with_failed_viruschecks'}) {
+        #push @queries, "FILE.fileVirusCheck.status==false";
+    }
+    if (exists $opt->{'with_missed_viruschecks'}) {
+        #push @queries, "FILE.fileVirusCheck.status==\"\"";
+    }
+    #my $query = join(" and ", @queries);
+    my $query;
+   # my $must_s;
+    if (scalar @must > 0) {
+#        $must_s= {
+ #           must => [ @must ]
+  #      };
+        $query->{query}->{bool}->{must} = [ @must ];
+    }
+    my $should_s;
+    if (scalar @should > 0) {
+    #    $should_s={
+     #       should => [ @should ]
+      #  };
+        $query->{query}->{bool}->{should} = [ @should ];
+    }
+
+    if (exists $opt->{debug}) {
+        use Data::Printer;
+        say  "opt=", np( $opt);
+        say "\@must=",np(@must);
+        say "\@should=",np(@should);
+        say "QUERY=",np( $query);
+    }
+    return $query;
+}
+
+
+# vim: set tabstop=4
+1;