--- foxridge-archiver/makemeta-vlp.pl 2006/12/14 15:50:05 1.7 +++ foxridge-archiver/makemeta-vlp.pl 2017/03/16 17:00:43 1.10 @@ -1,4 +1,4 @@ -#!/usr/local/bin/perl -w +#!/usr/bin/perl -w use strict; use XML::LibXML; @@ -10,7 +10,7 @@ use MPIWGStor; $|=1; # program version -my $version = "0.2.5 (14.12.2006 ROC)"; +my $version = "0.2.7 (27.8.2010 ROC)"; my $help = "use: makemeta-vlp [options] file.xml options: @@ -20,6 +20,7 @@ options: -online-mode mode for creating online/permanent files -archive-mode mode for creating archive/data files -access=free adds free access tag for online-mode + -texttool adds texttool tag for online-mode "; logger("INFO", "makemeta-vlp $version"); @@ -120,6 +121,16 @@ my %subtype_map = ( 'Pages' => 'meta/bib/number-of-pages', '#Cover pages only, articles have been extracted' => 'meta/bib/comment' }, + 'Journal' => { + '_name' => 'report', + 'Title' => 'meta/bib/title', + 'SecondaryTitle' => 'meta/bib/institution', + 'Author' => 'meta/bib/author', + 'Place_Published' => 'meta/bib/city', + 'Year' => 'meta/bib/year', + 'Date' => 'meta/bib/date', + 'Pages' => 'meta/bib/pages', + }, 'Magazine Article' => { '_name' => 'magazine-article', 'Author' => 'meta/bib/author', @@ -130,6 +141,15 @@ my %subtype_map = ( 'Date' => 'meta/bib/issue-date', 'Pages' => 'meta/bib/pages' }, + 'Newspaper Article' => { + '_name' => 'newspaper-article', + 'Author' => 'meta/bib/author', + 'Title' => 'meta/bib/title', + 'Year' => 'meta/bib/year', + 'Secondary_Title' => 'meta/bib/newspaper', + 'Date' => 'meta/bib/issue-date', + 'Pages' => 'meta/bib/pages' + }, 'Report' => { '_name' => 'report', 'Author' => 'meta/bib/author', @@ -183,10 +203,14 @@ my %lang_map = ( 'Japanese' => 'ja', 'Dutch' => 'nl', 'Spanish' => 'es', - 'Swedish' => 'sv' + 'Swedish' => 'sv', + 'Russian' => 'ru', + 'Polish' => 'pl', + 'Greek' => 'el' ); # storage fields my $arch_id_field = 'ID'; +my $access_free_field = 'online'; ####################################################### # internal parameters @@ -222,6 +246,12 @@ logger('DEBUG', "online_mode: $online_mo my $archive_mode = (exists $$args{'archive-mode'}) ? $$args{'archive-mode'} : 0; logger('DEBUG', "archive_mode: $archive_mode"); +# create texttool tag (online mode only) +my $texttool = (exists $$args{'texttool'}) ? $$args{'texttool'} : 1; +logger('DEBUG', "texttool: $texttool"); +# image dir for texttool +my $texttool_img_dir = "pages"; + # access type my $access_type = (exists $$args{'access'}) ? $$args{'access'} : ""; @@ -280,7 +310,11 @@ sub find_permanent_dir { return; } my $dir = "$online_base/lit$dest_id"; - return $dir; + if (-d $dir) { + logger('DEBUG', "directory $dir exists"); + return $dir; + } + return; } @@ -425,6 +459,25 @@ sub process_fm_entry { my $acc_tag = create_element_path('meta/access-conditions/access@type=institution', $index_root, $namespace); create_text_path('name', $access_type, $acc_tag, $namespace); } + } elsif ($online_mode) { + # read access conditions from "online" field in DB dump + my $online = sstrip($input_node->findvalue("fm:$access_free_field")); + if ($online) { + create_element_path('meta/access-conditions/access@type=free', $index_root, $namespace); + } else { + my $acc_tag = create_element_path('meta/access-conditions/access@type=institution', $index_root, $namespace); + create_text_path('name', 'mpiwg', $acc_tag, $namespace); + } + } + + # texttool tag with image dir + if ($online_mode && $texttool) { + if ( -d "$doc_dir/$texttool_img_dir" ) { + create_text_path('meta/texttool/image', $texttool_img_dir,$index_root, $namespace); + } else { + logger('WARNING', "page image directory missing!"); + $warncnt++; + } } # convert bib entries