diff --git a/eprints1.pl b/eprints1.pl
index f888cc56541004185e87d6b74b315fc282d62dbf..e28851a26871b37e8f653a9802b9dd472b62f597 100755
--- a/eprints1.pl
+++ b/eprints1.pl
@@ -114,8 +114,8 @@ my $op_mode= 'unknown';
# ======================================================================
# BEGIN OT2UT: Othes to Utheses
# my $ot2ut_context= 'ot2ut-entw'; # TODO: parametrize
-my $ot2ut_context= 'ot2ut-test'; # TODO: parametrize
-# my $ot2ut_context= 'ot2ut-prod'; # TODO: parametrize
+# my $ot2ut_context= 'ot2ut-test'; # TODO: parametrize
+my $ot2ut_context= 'ot2ut-prod'; # TODO: parametrize
my $oma_sleep_time= 10;
my %map_ot2ut_roles=
@@ -141,8 +141,11 @@ my %map_ot2ut_json_columns=
ubw_intern_2 => 'utheses_note_2',
);
-my $fnm_utheses_faculty_map= '/home/gg/work/gitlab.uv/zid-services/utheses-admin-tools/faculty/univie.json';
+# my $fnm_utheses_faculty_map_old= '/home/gg/work/gitlab.uv/zid-services/utheses-admin-tools/faculty/univie.json';
+my $fnm_utheses_departments_map_new= '/home/gg/work/gitlab.uv/zid-services/utheses/uthesesAPI/UthesesAPI/Resources/Departments.json';
+
my $utheses_faculty_map;
+my $utheses_faculty_map_loaded= 0;
# NOTE: alternative module (currently without persistent department identifier)
# https://gitlab.uv.univie.ac.at/zid-services/hopla/blob/dev-utheses/lib/utheses_data.pm
@@ -2805,17 +2808,6 @@ sub ot2ut
my $col_att= $db_ot2ut->get_collection('attachments');
my $col_policy_utheses= $db_ot2ut->get_collection('policy.utheses');
- unless (defined ($utheses_faculty_map))
- {
- my $utheses_faculty_list= Util::JSON::read_json_file($fnm_utheses_faculty_map);
- # print __LINE__, " utheses_faculty_list: ", Dumper($utheses_faculty_list);
- my %utheses_faculty_map= map { $_->{oracle_id} => $_ } @$utheses_faculty_list;
- $utheses_faculty_map= \%utheses_faculty_map;
-
- # Util::JSON::write_json_file('@facultymap.json', $utheses_faculty_map);
- # print __LINE__, " utheses_faculty_map: ", Dumper($utheses_faculty_map); exit;
- }
-
# find items to upload
if (!@eprint_ids)
{
@@ -3255,8 +3247,13 @@ sub generate_utheses_metadata
my ($eprint_status, $fileinfo, $dir)= map { $row->{$_} } qw(eprint_status fileinfo dir);
- my $docs= get_documents($epr_db, $eprintid, $dir);
+ my ($docs, $doc_errors)= get_documents($epr_db, $eprintid, $dir);
print __LINE__, " docs: ", Dumper($docs);
+ if (@$doc_errors)
+ {
+ print __LINE__, " doc_errors: ", Dumper($doc_errors);
+ push (@errors, @$doc_errors);
+ }
my ($lang_pdf, $files)= analyze_files($eprintid, $fileinfo, $dir);
print __LINE__, " lang_pdf=[$lang_pdf] files: ", Dumper($files);
@@ -3330,28 +3327,56 @@ sub generate_utheses_metadata
$utp->{studies}->{program_designation}= $s;
}
+ unless ($utheses_faculty_map_loaded)
+ {
+
+=begin comment
+
+the structure of json file has slightly changed
+ my $utheses_faculty_list= Util::JSON::read_json_file($fnm_utheses_faculty_map_old);
+ # print __LINE__, " utheses_faculty_list: ", Dumper($utheses_faculty_list);
+ my %utheses_faculty_map= map { $_->{oracle_id} => $_ } @$utheses_faculty_list;
+
+=end comment
+=cut
+
+ # 2020-09-21: new structure; this includes historic faculty codes
+ my $utheses_departments_list= Util::JSON::read_json_file($fnm_utheses_departments_map_new);
+ # print __LINE__, " utheses_faculty_list: ", Dumper($utheses_faculty_list);
+ my %utheses_faculty_map= map { $_->{oracle_id} => $_ } @{$utheses_departments_list->{departments}};
+
+ $utheses_faculty_map= \%utheses_faculty_map;
+ Util::JSON::write_json_file('@facultymap.json', $utheses_faculty_map);
+ # print __LINE__, " utheses_faculty_map: ", Dumper($utheses_faculty_map); exit;
+ $utheses_faculty_map_loaded= 1;
+ }
+
{
my $faculty_code_str= $row->{fakultaet}; # like A32 etc..., but also NULL and ''
+ $faculty_code_str= 'A108' if ($faculty_code_str eq 'PGC');
+
+ my $faculty_info;
if ($faculty_code_str =~ m#^A(\d+)$#)
{
my $faculty_code_num= $1;
- my $faculty_info= $utheses_faculty_map->{$faculty_code_num};
- if (defined ($faculty_info))
- {
- $utp->{studies}->{department}=
- {
- name => $faculty_info->{de},
- id => $faculty_info->{id},
- };
- }
- else
+ $faculty_info= $utheses_faculty_map->{$faculty_code_num};
+ }
+ elsif ($faculty_code_str =~ m#^[B-H]\d*$#) # e.g.: B, D2, F, F2, G, G2, H, H2
+ { # historic codes, prepend 'hist-' to the string; this is how the reference file stores it.
+ $faculty_info= $utheses_faculty_map->{'hist-' . $faculty_code_str};
+ }
+
+ if (defined ($faculty_info))
+ {
+ $utp->{studies}->{department}=
{
- push (@mapping_errors, { error => 'faculty_no_match', fakultaet => $faculty_code_str });
- }
+ name => $faculty_info->{de},
+ id => $faculty_info->{id},
+ };
}
else
{
- push (@mapping_errors, { error => 'faculty_unknown', fakultaet => $faculty_code_str });
+ push (@mapping_errors, { error => 'faculty_no_match', fakultaet => $faculty_code_str });
}
}
@@ -3432,6 +3457,8 @@ sub generate_utheses_metadata
if (@mapping_errors)
{
$utp->{import}->{info}->{mapping_errors}= \@mapping_errors;
+ push (@errors, @mapping_errors);
+ # push (@warnings, @mapping_errors);
}
my ($creators_xml, $creators_json)= $epr->get_creators($eprintid);
@@ -3486,6 +3513,8 @@ sub get_documents
my $eprint_id= shift;
my $dir= shift;
+ my @errors= ();
+
my $document_rows= $epr_db->get_all_x('document', ['eprintid=?', $eprint_id]);
# print __LINE__, " document_rows: ", Dumper($document_rows);
@@ -3527,6 +3556,12 @@ sub get_documents
my (@stat_doc)= stat($path_doc);
$row->{path_doc}= $path_doc;
$row->{stat_doc}= \@stat_doc;
+ unless (@stat_doc)
+ {
+ my $ec= { error => 'doc_missing', path_doc => $path_doc, rev_dir => $rev_dir, dir => $dir };
+ push (@errors, $ec);
+ $row->{error}= $ec;
+ }
if (defined ($row->{placement}) && $row->{pos} != $row->{placement})
{
@@ -3588,7 +3623,7 @@ sub get_documents
$res->{documents}= \@docs2,
- $res;
+ return ($res, \@errors);
}
sub get_history
@@ -4114,7 +4149,7 @@ sub analyze_files
$format= 'pdf' if ($icon eq '/style/images/fileicons/pdf.png' || $icon eq '/style/images/fileicons/application_pdf.png');
$filepath =~ s#%([\dA-Fa-f]{2})#chr(hex($1))#ge; # filenames are URL encoded, see 19072 for an example
- print __LINE__, " vor utf8::decode filepath=[$filepath]\n";
+ print __LINE__, " vor utf8::decode filepath=[$filepath]\n";
utf8::decode($filepath);
print __LINE__, " nach utf8::decode filepath=[$filepath]\n";
@@ -4500,7 +4535,7 @@ sub update_policies
$db_ot2ut= IRMA::db::get_any_db($cnf, 'ot2ut_database') unless (defined ($db_ot2ut));
my $col_utp= $db_ot2ut->get_collection('utheses.policies');
- if (defined ($refresh_oldest_policies) && $refresh_oldest_policies > 0 && $refresh_oldest_policies <= 1000)
+ if (defined ($refresh_oldest_policies) && $refresh_oldest_policies > 0 && $refresh_oldest_policies <= 3000)
{
my $c1= $col_utp->find( {}, { '_id' => 1, eprint_id => 1, generated => 1 });
$c1->sort( { generated => 1 } );
@@ -4601,8 +4636,9 @@ sub update_policies
print __LINE__, ' ', '='x70, "\n";
}
- my @removed_from_mysql= $col_utp->find( { eprint_id => { '$nin' => \@eprint_ids_in_mysql } }, { '_id' => 1, eprint_id => 1 })->all();
- print __LINE__, " removed from mysql ", Dumper(\@removed_from_mysql);
+ my @removed_from_othes= $col_utp->find( { eprint_id => { '$nin' => \@eprint_ids_in_mysql } }, { '_id' => 1, eprint_id => 1 })->all();
+ print __LINE__, " removed from othes ", Dumper(\@removed_from_othes);
+ my @eprint_id_removed_from_othes= map { $_->{eprint_id} } @removed_from_othes;
my %stats=
(
@@ -4616,6 +4652,7 @@ sub update_policies
lst_inserted => \@lst_inserted,
lst_updated => \@lst_updated,
lst_unchanged => \@lst_unchanged,
+ removed_from_othes => \@eprint_id_removed_from_othes,
);
my $col_stats= $db_ot2ut->get_collection('statistics');
@@ -4849,7 +4886,7 @@ sub policies_stats
my $b2b= $cctab->bucket(0, @s2b);
if (defined ($b2b))
{
- $b2b->{annotation}= { bgcolor => 'pink', note => 'abstract und keywords locked, aber pdf public' };
+ $b2b->{annotation}= { bgcolor => 'pink', note => 'abstract und keywords locked, aber pdf public; bekommt keine DOI' };
# print __LINE__, " special bucket found: ", Dumper($b2b);
}
@@ -4859,7 +4896,7 @@ sub policies_stats
my $b2c= $cctab->bucket(0, @s2c);
if (defined ($b2c))
{
- $b2c->{annotation}= { bgcolor => 'pink', note => 'abstract und keywords locked, aber pdf public' };
+ $b2c->{annotation}= { bgcolor => 'pink', note => 'abstract und keywords locked, aber pdf public; bekommt keine DOI' };
# print __LINE__, " special bucket found: ", Dumper($b2c);
}
# ZZZZ
@@ -5029,7 +5066,7 @@ EOX
my $ck1= ($pct == 100.0) ? 'lightgreen' : 'lightblue';
my $ck2= ($e == 0) ? 'lightgreen' : 'lightpink';
- printf IDX (" <td bgcolor=\"$ck1\">%d</td><td bgcolor=\"$ck1\">%5.2f %%</td><td bgcolor=\"$ck2\">%d</td>\n", $c, $pct, $e);
+ printf IDX (" <td bgcolor=\"$ck1\">%d</td><td bgcolor=\"$ck1\">%7.4f %%</td><td bgcolor=\"$ck2\">%d</td>\n", $c, $pct, $e);
}
print IDX "</tr>\n";
@@ -5076,7 +5113,7 @@ EOX
my $pct= $c*100.0/$b_othes;
my $ck1= ($pct == 100.0) ? 'lightgreen' : 'lightblue';
my $ck2= ($e == 0) ? 'lightgreen' : 'pink';
- printf MB (" <td bgcolor=\"$ck1\">%d</td><td bgcolor=\"$ck1\">%5.2f %%</td><td bgcolor=\"$ck2\">%d</td>\n", $c, $pct, $e);
+ printf MB (" <td bgcolor=\"$ck1\">%d</td><td bgcolor=\"$ck1\">%7.4f %%</td><td bgcolor=\"$ck2\">%d</td>\n", $c, $pct, $e);
}
print MB "</tr>\n";
@@ -5125,9 +5162,10 @@ EOX
{
my $link;
- if ($context eq 'test') { $link= 'https://utheses-frontend.ctest.univie.ac.at/client/?#/view/document/utheses/' . $utheses_id; }
- elsif ($context eq 'entw') { $link= 'https://utheses-frontend-entw-utheses.ctest.univie.ac.at/?#/view/document/utheses/' . $utheses_id; }
- elsif ($context eq 'prod') { $link= 'unknown'; }
+ if ($context eq 'test') { $link= 'https://utheses-frontend.ctest.univie.ac.at/client/#/detail/' . $utheses_id . '/'; }
+ elsif ($context eq 'entw') { $link= 'https://utheses-frontend-entw-utheses.ctest.univie.ac.at/#/detail/' . $utheses_id . '/'; }
+ # elsif ($context eq 'prod') { $link= 'https://utheses-admin-ui-utheses-prod.cprod.univie.ac.at/client#/view/document/utheses/' . $utheses_id . '/'; }
+ elsif ($context eq 'prod') { $link= 'https://utheses-admin-ui-utheses-prod.cprod.univie.ac.at/client#/detail/' . $utheses_id . '/'; }
print BLOCK " <td bgcolor=\"lightgreen\"><a href=\"$link\" target=\"$context\">$utheses_id</a></td>\n";
}
@@ -5166,7 +5204,7 @@ EOX
my $pct= ($total_othes == 0) ? 0 : $c*100.0/$total_othes;
my $ck1= ($pct == 100.0) ? 'lightgreen' : 'lightblue';
my $ck2= ($e == 0) ? 'lightgreen' : 'lightpink';
- printf IDX (" <td bgcolor=\"$ck1\">%d</td><td bgcolor=\"$ck1\">%5.2f %%</td><td bgcolor=\"$ck2\">%d</td>\n", $c, $pct, $e);
+ printf IDX (" <td bgcolor=\"$ck1\">%d</td><td bgcolor=\"$ck1\">%7.4f %%</td><td bgcolor=\"$ck2\">%d</td>\n", $c, $pct, $e);
}
print IDX "</tr>\n";