From f20476d31f60bd1e8c83e91379f414bdc0cfdd60 Mon Sep 17 00:00:00 2001 From: Ralph Amissah Date: Sat, 13 Dec 2008 14:08:19 -0500 Subject: mostly additional metadata fields additional metadata fields css minor touch help minor info touch --- lib/sisu/v0/css.rb | 2 +- lib/sisu/v0/db_columns.rb | 41 ++++++++------ lib/sisu/v0/db_create.rb | 120 +++++++++++++++++++++++------------------ lib/sisu/v0/db_tests.rb | 78 ++++++++++++++------------- lib/sisu/v0/harvest_authors.rb | 27 ++++++---- lib/sisu/v0/param.rb | 41 ++++++++++---- 6 files changed, 182 insertions(+), 127 deletions(-) (limited to 'lib') diff --git a/lib/sisu/v0/css.rb b/lib/sisu/v0/css.rb index 2094134f..927a253f 100644 --- a/lib/sisu/v0/css.rb +++ b/lib/sisu/v0/css.rb @@ -302,7 +302,7 @@ WOK position: absolute; top: 80px; left: 0; - margin-left: 0; + margin-left: 1%; width: 20%; } #column_center { diff --git a/lib/sisu/v0/db_columns.rb b/lib/sisu/v0/db_columns.rb index 6c9778ab..1ea16c95 100644 --- a/lib/sisu/v0/db_columns.rb +++ b/lib/sisu/v0/db_columns.rb @@ -66,9 +66,15 @@ module SiSU_DB_columns def lt_subtitle 600 end - def lt_creator + def lt_author 600 end + def lt_author_title + 100 + end + def lt_author_nationality + 100 + end def lt_illustrator 600 end @@ -87,21 +93,6 @@ module SiSU_DB_columns def lt_date 10 end - def lt_date_created - 10 - end - def lt_date_issued - 10 - end - def lt_date_available - 10 - end - def lt_date_modified - 10 - end - def lt_date_valid - 10 - end def lt_type 600 end @@ -198,6 +189,24 @@ module SiSU_DB_columns def lt_subj 64 end + def lt_orig_pub + 400 + end + def lt_orig_pub_date + 400 + end + def lt_orig_pub_institution + 200 + end + def lt_orig_pub_nationality + 200 + end + def lt_writing_focus_nationality + 100 + end + def lt_topic_register + 2000 + end def document_clean # restriction not necessary #12000 #2712 60000 end diff --git a/lib/sisu/v0/db_create.rb b/lib/sisu/v0/db_create.rb index 189b7acf..df1e91df 100644 --- a/lib/sisu/v0/db_create.rb +++ b/lib/sisu/v0/db_create.rb @@ -100,59 +100,73 @@ module SiSU_DB_create } unless @opt.cmd =~/q/ @conn.execute(%{ CREATE TABLE metadata ( - tid BIGINT PRIMARY KEY, - title VARCHAR(#{lt_title}) NULL, - subtitle VARCHAR(#{lt_subtitle}) NULL, - creator VARCHAR(#{lt_creator}) NULL, - illustrator VARCHAR(#{lt_illustrator}) NULL, - translator VARCHAR(#{lt_translator}) NULL, - subject VARCHAR(#{lt_subject}) NULL, - date VARCHAR(#{lt_date}) NULL, - date_created VARCHAR(#{lt_date_created}) NULL, - date_issued VARCHAR(#{lt_date_issued}) NULL, - date_available VARCHAR(#{lt_date_available}) NULL, - date_valid VARCHAR(#{lt_date_valid}) NULL, - date_modified VARCHAR(#{lt_date_modified}) NULL, -/* date DATE, */ -/* date_created DATE, */ -/* date_issued DATE, */ -/* date_available DATE, */ -/* date_valid DATE, */ -/* date_modified DATE, */ - type VARCHAR(#{lt_type}) NULL, - description VARCHAR(#{lt_description}) NULL, - publisher VARCHAR(#{lt_publisher}) NULL, - contributor VARCHAR(#{lt_contributor}) NULL, - prepared_by VARCHAR(#{lt_prepared_by}) NULL, - digitized_by VARCHAR(#{lt_digitized_by}) NULL, - format VARCHAR(#{lt_format}) NULL, - identifier VARCHAR(#{lt_identifier}) NULL, - source VARCHAR(#{lt_source}) NULL, - language VARCHAR(#{lt_language}) NULL, - language_original VARCHAR(#{lt_language_original}) NULL, - relation VARCHAR(#{lt_relation}) NULL, - coverage VARCHAR(#{lt_coverage}) NULL, - rights VARCHAR(#{lt_rights}) NULL, - copyright VARCHAR(#{lt_copyright}) NULL, - owner VARCHAR(#{lt_owner}) NULL, - keywords VARCHAR(#{lt_keywords}) NULL, - comment VARCHAR(#{lt_comment}) NULL, - loc VARCHAR(#{lt_loc}) NULL, - dewey VARCHAR(#{lt_dewey}) NULL, - isbn VARCHAR(#{lt_isbn}) NULL, - pg VARCHAR(#{lt_pg}) NULL, - abstract VARCHAR(#{lt_abstract}) NULL, - prefix_a TEXT NULL, - prefix_b TEXT NULL, - skin VARCHAR(#{lt_skin}) NULL, - markup VARCHAR(#{lt_markup}) NULL, - links VARCHAR(#{lt_links}) NULL, - information VARCHAR(#{lt_information}) NULL, - contact VARCHAR(#{lt_contact}) NULL, - suffix VARCHAR(#{lt_suffix}) NULL, - filename VARCHAR(#{lt_filename}) NULL UNIQUE, - types CHAR(#{lt_types}) NULL, - subj VARCHAR(#{lt_subj}) NULL + tid BIGINT PRIMARY KEY, + title VARCHAR(#{lt_title}) NULL, + subtitle VARCHAR(#{lt_subtitle}) NULL, + author VARCHAR(#{lt_author}) NULL, +/* plan to replace creator field, currently used, with author field */ + creator VARCHAR(#{lt_author}) NULL, + author_title VARCHAR(#{lt_author_title}) NULL, + author_nationality VARCHAR(#{lt_author_nationality}) NULL, + illustrator VARCHAR(#{lt_illustrator}) NULL, + translator VARCHAR(#{lt_translator}) NULL, + subject VARCHAR(#{lt_subject}) NULL, + date VARCHAR(#{lt_date}) NULL, + date_added_to_site VARCHAR(#{lt_date}) NULL, + date_created VARCHAR(#{lt_date}) NULL, + date_issued VARCHAR(#{lt_date}) NULL, + date_available VARCHAR(#{lt_date}) NULL, + date_valid VARCHAR(#{lt_date}) NULL, + date_modified VARCHAR(#{lt_date}) NULL, + date_translated VARCHAR(#{lt_date}) NULL, +/* date DATE, */ +/* date_added_to_site DATE, */ +/* date_created DATE, */ +/* date_issued DATE, */ +/* date_available DATE, */ +/* date_valid DATE, */ +/* date_modified DATE, */ +/* date_translated DATE, */ + type VARCHAR(#{lt_type}) NULL, + description VARCHAR(#{lt_description}) NULL, + publisher VARCHAR(#{lt_publisher}) NULL, + contributor VARCHAR(#{lt_contributor}) NULL, + prepared_by VARCHAR(#{lt_prepared_by}) NULL, + digitized_by VARCHAR(#{lt_digitized_by}) NULL, + format VARCHAR(#{lt_format}) NULL, + identifier VARCHAR(#{lt_identifier}) NULL, + source VARCHAR(#{lt_source}) NULL, + language VARCHAR(#{lt_language}) NULL, + language_original VARCHAR(#{lt_language_original}) NULL, + relation VARCHAR(#{lt_relation}) NULL, + coverage VARCHAR(#{lt_coverage}) NULL, + rights VARCHAR(#{lt_rights}) NULL, + copyright VARCHAR(#{lt_copyright}) NULL, + owner VARCHAR(#{lt_owner}) NULL, + keywords VARCHAR(#{lt_keywords}) NULL, + comment VARCHAR(#{lt_comment}) NULL, + loc VARCHAR(#{lt_loc}) NULL, + dewey VARCHAR(#{lt_dewey}) NULL, + isbn VARCHAR(#{lt_isbn}) NULL, + pg VARCHAR(#{lt_pg}) NULL, + abstract VARCHAR(#{lt_abstract}) NULL, + prefix_a TEXT NULL, + prefix_b TEXT NULL, + skin VARCHAR(#{lt_skin}) NULL, + markup VARCHAR(#{lt_markup}) NULL, + links VARCHAR(#{lt_links}) NULL, + information VARCHAR(#{lt_information}) NULL, + contact VARCHAR(#{lt_contact}) NULL, + suffix VARCHAR(#{lt_suffix}) NULL, + filename VARCHAR(#{lt_filename}) NULL UNIQUE, + types CHAR(#{lt_types}) NULL, + subj VARCHAR(#{lt_subj}) NULL, + original_publication VARCHAR(#{lt_orig_pub}) NULL, + original_publication_date VARCHAR(#{lt_orig_pub_date}) NULL, + original_publication_institution VARCHAR(#{lt_orig_pub_institution}) NULL, + original_publication_nationality VARCHAR(#{lt_orig_pub_nationality}) NULL, + writing_focus_nationality VARCHAR(#{lt_writing_focus_nationality}) NULL, + topic_register VARCHAR(#{lt_topic_register}) NULL ); #{@comment.metadata} }) diff --git a/lib/sisu/v0/db_tests.rb b/lib/sisu/v0/db_tests.rb index 4a18f694..86e7ca25 100644 --- a/lib/sisu/v0/db_tests.rb +++ b/lib/sisu/v0/db_tests.rb @@ -70,45 +70,49 @@ module SiSU_DB_tests end def verify unless @opt.cmd =~/q/ - puts @ck.tp[:fns].length.to_s + ' checklength ' + @ck.tp[:fns] if @ck.tp[:fns] and @ck.tp[:fns].length >@ck.lt_filename - puts @ck.tp[:title].length.to_s + ' checklength ' + @ck.tp[:title] if @ck.tp[:title] and @ck.tp[:title].length >@ck.lt_title - puts @ck.tp[:subtitle].length.to_s + ' checklength ' + @ck.tp[:subtitle] if @ck.tp[:subtitle] and @ck.tp[:subtitle].length >@ck.lt_subtitle - puts @ck.tp[:creator].length.to_s + ' checklength ' + @ck.tp[:creator] if @ck.tp[:creator] and @ck.tp[:creator].length >@ck.lt_creator - puts @ck.tp[:illustrator].length.to_s + ' checklength ' + @ck.tp[:illustrator] if @ck.tp[:illustrator] and @ck.tp[:illustrator].length >@ck.lt_illustrator - puts @ck.tp[:translator].length.to_s + ' checklength ' + @ck.tp[:translator] if @ck.tp[:translator] and @ck.tp[:translator].length >@ck.lt_translator - puts @ck.tp[:prepared_by].length.to_s + ' checklength ' + @ck.tp[:prepared_by] if @ck.tp[:prepared_by] and @ck.tp[:prepared_by].length >@ck.lt_prepared_by - puts @ck.tp[:digitized_by].length.to_s + ' checklength ' + @ck.tp[:digitized_by] if @ck.tp[:digitized_by] and @ck.tp[:digitized_by].length >@ck.lt_digitized_by - puts @ck.tp[:subject].length.to_s + ' checklength ' + @ck.tp[:subject] if @ck.tp[:subject] and @ck.tp[:subject].length >@ck.lt_subject - puts @ck.tp[:description].length.to_s + ' checklength ' + @ck.tp[:description] if @ck.tp[:description] and @ck.tp[:description].length >@ck.lt_description - puts @ck.tp[:publisher].length.to_s + ' checklength ' + @ck.tp[:publisher] if @ck.tp[:publisher] and @ck.tp[:publisher].length >@ck.lt_publisher - puts @ck.tp[:contributor].length.to_s + ' checklength ' + @ck.tp[:contributor] if @ck.tp[:contributor] and @ck.tp[:contributor].length >@ck.lt_contributor - puts @ck.tp[:date].length.to_s + ' checklength ' + @ck.tp[:date] if @ck.tp[:date] and @ck.tp[:date].length >@ck.lt_date - puts @ck.tp[:date_created].length.to_s + ' checklength ' + @ck.tp[:date_created] if @ck.tp[:date_created] and @ck.tp[:date_created].length >@ck.lt_date_created - puts @ck.tp[:date_issued].length.to_s + ' checklength ' + @ck.tp[:date_issued] if @ck.tp[:date_issued] and @ck.tp[:date_issued].length >@ck.lt_date_issued - puts @ck.tp[:date_valid].length.to_s + ' checklength ' + @ck.tp[:date_valid] if @ck.tp[:date_valid] and @ck.tp[:date_valid].length >@ck.lt_date_valid - puts @ck.tp[:date_available].length.to_s + ' checklength ' + @ck.tp[:date_available] if @ck.tp[:date_available] and @ck.tp[:date_available].length >@ck.lt_date_available - puts @ck.tp[:date_modified].length.to_s + ' checklength ' + @ck.tp[:date_modified] if @ck.tp[:date_modified] and @ck.tp[:date_modified].length >@ck.lt_date_modified - puts @ck.tp[:type].length.to_s + ' checklength ' + @ck.tp[:type] if @ck.tp[:type] and @ck.tp[:type].length >@ck.lt_type - puts @ck.tp[:format].length.to_s + ' checklength ' + @ck.tp[:format] if @ck.tp[:format] and @ck.tp[:format].length >@ck.lt_format - puts @ck.tp[:identifier].length.to_s + ' checklength ' + @ck.tp[:identifier] if @ck.tp[:identifier] and @ck.tp[:identifier].length >@ck.lt_identifier - puts @ck.tp[:source].length.to_s + ' checklength ' + @ck.tp[:source] if @ck.tp[:source] and @ck.tp[:source].length >@ck.lt_source - puts @ck.tp[:language].length.to_s + ' checklength ' + @ck.tp[:language] if @ck.tp[:language] and @ck.tp[:language].length >@ck.lt_language - puts @ck.tp[:language_original].length.to_s + ' checklength ' + @ck.tp[:language_original] if @ck.tp[:language_original] and @ck.tp[:language_original].length >@ck.lt_language_original + puts @ck.tp[:fns].length.to_s + ' checklength ' + @ck.tp[:fns] if @ck.tp[:fns] and @ck.tp[:fns].length >@ck.lt_filename + puts @ck.tp[:title].length.to_s + ' checklength ' + @ck.tp[:title] if @ck.tp[:title] and @ck.tp[:title].length >@ck.lt_title + puts @ck.tp[:subtitle].length.to_s + ' checklength ' + @ck.tp[:subtitle] if @ck.tp[:subtitle] and @ck.tp[:subtitle].length >@ck.lt_subtitle + puts @ck.tp[:creator].length.to_s + ' checklength ' + @ck.tp[:creator] if @ck.tp[:creator] and @ck.tp[:creator].length >@ck.lt_creator + puts @ck.tp[:author_title].length.to_s + ' checklength ' + @ck.tp[:author_title] if @ck.tp[:author_title] and @ck.tp[:author_title].length >@ck.lt_author_title + puts @ck.tp[:illustrator].length.to_s + ' checklength ' + @ck.tp[:illustrator] if @ck.tp[:illustrator] and @ck.tp[:illustrator].length >@ck.lt_illustrator + puts @ck.tp[:translator].length.to_s + ' checklength ' + @ck.tp[:translator] if @ck.tp[:translator] and @ck.tp[:translator].length >@ck.lt_translator + puts @ck.tp[:prepared_by].length.to_s + ' checklength ' + @ck.tp[:prepared_by] if @ck.tp[:prepared_by] and @ck.tp[:prepared_by].length >@ck.lt_prepared_by + puts @ck.tp[:digitized_by].length.to_s + ' checklength ' + @ck.tp[:digitized_by] if @ck.tp[:digitized_by] and @ck.tp[:digitized_by].length >@ck.lt_digitized_by + puts @ck.tp[:subject].length.to_s + ' checklength ' + @ck.tp[:subject] if @ck.tp[:subject] and @ck.tp[:subject].length >@ck.lt_subject + puts @ck.tp[:description].length.to_s + ' checklength ' + @ck.tp[:description] if @ck.tp[:description] and @ck.tp[:description].length >@ck.lt_description + puts @ck.tp[:publisher].length.to_s + ' checklength ' + @ck.tp[:publisher] if @ck.tp[:publisher] and @ck.tp[:publisher].length >@ck.lt_publisher + puts @ck.tp[:contributor].length.to_s + ' checklength ' + @ck.tp[:contributor] if @ck.tp[:contributor] and @ck.tp[:contributor].length >@ck.lt_contributor + puts @ck.tp[:date].length.to_s + ' checklength ' + @ck.tp[:date] if @ck.tp[:date] and @ck.tp[:date].length >@ck.lt_date + puts @ck.tp[:date_created].length.to_s + ' checklength ' + @ck.tp[:date_created] if @ck.tp[:date_created] and @ck.tp[:date_created].length >@ck.lt_date + puts @ck.tp[:date_issued].length.to_s + ' checklength ' + @ck.tp[:date_issued] if @ck.tp[:date_issued] and @ck.tp[:date_issued].length >@ck.lt_date + puts @ck.tp[:date_valid].length.to_s + ' checklength ' + @ck.tp[:date_valid] if @ck.tp[:date_valid] and @ck.tp[:date_valid].length >@ck.lt_date + puts @ck.tp[:date_available].length.to_s + ' checklength ' + @ck.tp[:date_available] if @ck.tp[:date_available] and @ck.tp[:date_available].length >@ck.lt_date + puts @ck.tp[:date_modified].length.to_s + ' checklength ' + @ck.tp[:date_modified] if @ck.tp[:date_modified] and @ck.tp[:date_modified].length >@ck.lt_date + puts @ck.tp[:date_translated].length.to_s + ' checklength ' + @ck.tp[:date_translated] if @ck.tp[:date_translated] and @ck.tp[:date_translated].length >@ck.lt_date + puts @ck.tp[:date_added_to_site].length.to_s + ' checklength ' + @ck.tp[:date_added_to_site] if @ck.tp[:date_added_to_site] and @ck.tp[:date_added_to_site].length >@ck.lt_date + puts @ck.tp[:type].length.to_s + ' checklength ' + @ck.tp[:type] if @ck.tp[:type] and @ck.tp[:type].length >@ck.lt_type + puts @ck.tp[:format].length.to_s + ' checklength ' + @ck.tp[:format] if @ck.tp[:format] and @ck.tp[:format].length >@ck.lt_format + puts @ck.tp[:identifier].length.to_s + ' checklength ' + @ck.tp[:identifier] if @ck.tp[:identifier] and @ck.tp[:identifier].length >@ck.lt_identifier + puts @ck.tp[:source].length.to_s + ' checklength ' + @ck.tp[:source] if @ck.tp[:source] and @ck.tp[:source].length >@ck.lt_source + puts @ck.tp[:language].length.to_s + ' checklength ' + @ck.tp[:language] if @ck.tp[:language] and @ck.tp[:language].length >@ck.lt_language + puts @ck.tp[:language_original].length.to_s + ' checklength ' + @ck.tp[:language_original] if @ck.tp[:language_original] and @ck.tp[:language_original].length >@ck.lt_language_original #puts @ck.tp[:language_char].length.to_s + ' checklength ' + @ck.tp[:language_char] if @ck.tp[:language_char] and @ck.tp[:language_char].length >@ck.lt_language_char #puts @ck.tp[:language_original_char].length.to_s + ' checklength ' + @ck.tp[:language_original_char] if @ck.tp[:language_original_char] and @ck.tp[:language_original_char].length >@ck.lt_language_original_char - puts @ck.tp[:relation].length.to_s + ' checklength ' + @ck.tp[:relation] if @ck.tp[:relation] and @ck.tp[:relation].length >@ck.lt_relation - puts @ck.tp[:coverage].length.to_s + ' checklength ' + @ck.tp[:coverage] if @ck.tp[:coverage] and @ck.tp[:coverage].length >@ck.lt_coverage - puts @ck.tp[:rights].length.to_s + ' checklength ' + @ck.tp[:rights] if @ck.tp[:rights] and @ck.tp[:rights].length >@ck.lt_rights - puts @ck.tp[:copyright].length.to_s + ' checklength ' + @ck.tp[:copyright] if @ck.tp[:copyright] and @ck.tp[:copyright].length >@ck.lt_copyright - puts @ck.tp[:owner].length.to_s + ' checklength ' + @ck.tp[:owner] if @ck.tp[:owner] and @ck.tp[:owner].length >@ck.lt_owner - puts @ck.tp[:keywords].length.to_s + ' checklength ' + @ck.tp[:keywords] if @ck.tp[:keywords] and @ck.tp[:keywords].length >@ck.lt_keywords - puts @ck.tp[:abstract].length.to_s + ' checklength ' + @ck.tp[:abstract] if @ck.tp[:abstract] and @ck.tp[:abstract].length >@ck.lt_abstract - puts @ck.tp[:comment].length.to_s + ' checklength ' + @ck.tp[:comment] if @ck.tp[:comment] and @ck.tp[:comment].length >@ck.lt_comment - puts @ck.tp[:loc].length.to_s + ' checklength ' + @ck.tp[:loc] if @ck.tp[:loc] and @ck.tp[:loc].length >@ck.lt_loc - puts @ck.tp[:dewey].length.to_s + ' checklength ' + @ck.tp[:dewey] if @ck.tp[:dewey] and @ck.tp[:dewey].length >@ck.lt_dewey - puts @ck.tp[:isbn].length.to_s + ' checklength ' + @ck.tp[:isbn] if @ck.tp[:isbn] and @ck.tp[:isbn].length >@ck.lt_isbn - puts @ck.tp[:pg].length.to_s + ' checklength ' + @ck.tp[:pg] if @ck.tp[:pg] and @ck.tp[:pg].length >@ck.lt_pg - puts @ck.tp[:date] if @ck.tp[:date] and not @ck.tp[:date].empty? and @ck.tp[:date] !~/\d\d-\d\d-\d\d/ + puts @ck.tp[:relation].length.to_s + ' checklength ' + @ck.tp[:relation] if @ck.tp[:relation] and @ck.tp[:relation].length >@ck.lt_relation + puts @ck.tp[:coverage].length.to_s + ' checklength ' + @ck.tp[:coverage] if @ck.tp[:coverage] and @ck.tp[:coverage].length >@ck.lt_coverage + puts @ck.tp[:rights].length.to_s + ' checklength ' + @ck.tp[:rights] if @ck.tp[:rights] and @ck.tp[:rights].length >@ck.lt_rights + puts @ck.tp[:copyright].length.to_s + ' checklength ' + @ck.tp[:copyright] if @ck.tp[:copyright] and @ck.tp[:copyright].length >@ck.lt_copyright + puts @ck.tp[:owner].length.to_s + ' checklength ' + @ck.tp[:owner] if @ck.tp[:owner] and @ck.tp[:owner].length >@ck.lt_owner + puts @ck.tp[:keywords].length.to_s + ' checklength ' + @ck.tp[:keywords] if @ck.tp[:keywords] and @ck.tp[:keywords].length >@ck.lt_keywords + puts @ck.tp[:abstract].length.to_s + ' checklength ' + @ck.tp[:abstract] if @ck.tp[:abstract] and @ck.tp[:abstract].length >@ck.lt_abstract + puts @ck.tp[:comment].length.to_s + ' checklength ' + @ck.tp[:comment] if @ck.tp[:comment] and @ck.tp[:comment].length >@ck.lt_comment + puts @ck.tp[:loc].length.to_s + ' checklength ' + @ck.tp[:loc] if @ck.tp[:loc] and @ck.tp[:loc].length >@ck.lt_loc + puts @ck.tp[:dewey].length.to_s + ' checklength ' + @ck.tp[:dewey] if @ck.tp[:dewey] and @ck.tp[:dewey].length >@ck.lt_dewey + puts @ck.tp[:isbn].length.to_s + ' checklength ' + @ck.tp[:isbn] if @ck.tp[:isbn] and @ck.tp[:isbn].length >@ck.lt_isbn + puts @ck.tp[:pg].length.to_s + ' checklength ' + @ck.tp[:pg] if @ck.tp[:pg] and @ck.tp[:pg].length >@ck.lt_pg + puts @ck.tp[:topic_register].length.to_s + ' checklength ' + @ck.tp[:topic_register] if @ck.tp[:topic_register] and @ck.tp[:topci_register].length >@ck.lt_topic_register + puts @ck.tp[:date] if @ck.tp[:date] and not @ck.tp[:date].empty? and @ck.tp[:date] !~/\d\d-\d\d-\d\d/ end end end diff --git a/lib/sisu/v0/harvest_authors.rb b/lib/sisu/v0/harvest_authors.rb index 9586fcc3..9678e4f7 100644 --- a/lib/sisu/v0/harvest_authors.rb +++ b/lib/sisu/v0/harvest_authors.rb @@ -102,16 +102,18 @@ module HARVEST_authors end def extract_harvest data,filename,idx_array=@data,@filename,@idx_array - @publication_details,@title,@subtitle,@fulltitle,@author,@author_format=nil,nil,nil,nil,nil,nil + @orig_pub,@title,@subtitle,@fulltitle,@author,@author_format=nil,nil,nil,nil,nil,nil @authors=[] rgx={} rgx[:author]=/^@(?:author|creator):\s+(.+)/ rgx[:title]=/^@title:\s+(.+)/ rgx[:subtitle]=/^@subtitle:\s+(.+)/ - rgx[:publication_details]=/^@original_publication_details:\s+(.+)/ + rgx[:date]=/^@subtitle:\s+(.+)/ + rgx[:date]=/^@date:\s+(\d{4})/ + rgx[:orig_pub]=/^@original_publication:\s+(.+)/ data.each do |para| - if para=~ rgx[:publication_details] - @publication_details=rgx[:publication_details].match(para)[1] + if para=~ rgx[:orig_pub] + @orig_pub=rgx[:orig_pub].match(para)[1] end if para=~ rgx[:title] @title=rgx[:title].match(para)[1] @@ -122,19 +124,22 @@ module HARVEST_authors if para=~ rgx[:author] @author_format=rgx[:author].match(para)[1] end - break if @title and @subtitle and @author and @publication_details + if para=~ rgx[:date] + @date=rgx[:date].match(para)[1] + end + break if @title and @subtitle and @author and @date and @orig_pub end @fulltitle=if @subtitle @title + ' - ' + @subtitle else @title end - if @title and @author_format #and @publication_details + if @title and @author_format #and @orig_pub (publication details) creator=FORMAT::Author.new(@author_format.strip).author_details @authors,@authorship=creator[:authors],creator[:authorship] file=filename.sub(/\.ss[mt]$/,'') - idx_array <<= { :filename => filename, :file => file, :publication_details => @publication_details, :title => @fulltitle, :author => creator } + idx_array <<= { :filename => filename, :file => file, :orig_pub => @orig_pub, :date => @date, :title => @fulltitle, :author => creator } else - #p "missing author field: #@filename title: #@title; author: #@author_format; idx: #@publication_details" + #p "missing author field: #@filename title: #@title; author: #@author_format; idx: #@orig_pub" end idx_array.flatten! idx_array @@ -156,7 +161,7 @@ module HARVEST_authors if @@the_idx_authors[author].class==NilClass @@the_idx_authors[author]={:md => []} end - @@the_idx_authors[author][:md] << { :filename => idx[:filename], :file => idx[:file], :author => idx[:author], :title => idx[:title] } + @@the_idx_authors[author][:md] << { :filename => idx[:filename], :file => idx[:file], :author => idx[:author], :title => idx[:title], :date => idx[:date] } end end @the_idx=@@the_idx_authors @@ -290,9 +295,9 @@ WOK @output[:html] << x a[1][:md].each do |x| if @output[:html_mnt].class == File - @output[:html_mnt] << %{

[src]  #{x[:title]}, #{x[:author][:authors_s]} -- [#{x[:file]}.sst]

} + @output[:html_mnt] << %{

[src]  #{x[:date]} #{x[:title]}, #{x[:author][:authors_s]} -- [#{x[:file]}.sst]

} end - @output[:html] << %{

#{x[:title]}, #{x[:author][:authors_s]}

} + @output[:html] << %{

#{x[:date]} #{x[:title]}, #{x[:author][:authors_s]}

} end end end diff --git a/lib/sisu/v0/param.rb b/lib/sisu/v0/param.rb index 0d2095dd..ec80e0dd 100644 --- a/lib/sisu/v0/param.rb +++ b/lib/sisu/v0/param.rb @@ -118,9 +118,9 @@ module SiSU_Param @doc={ :lv=>[] } @doc[:fns],@doc[:fnb],@doc[:scr_suffix]='','','' @@publisher='SiSU scribe' - attr_accessor :cmd,:mod,:env,:fn,:fns,:fnb,:fnn,:fnt,:fnl,:flv,:fnz,:fnstex,:ocn,:sfx_src,:sfx,:pdf,:file_type,:dir_out,:dir_tex,:dir_lout,:txt_path,:site_skin,:sisu,:sisu_version,:ruby_version,:title,:dc_title,:html_title,:subtitle,:subtitle_tex,:creator_home,:dc_creator,:authors,:authorship,:translator,:illustrator,:prepared_by,:digitized_by,:dc_subject,:dc_description,:dc_publisher,:dc_contributor,:dc_date,:dc_date_created,:dc_date_issued,:dc_date_available,:dc_date_valid,:dc_date_modified,:date_scheme,:date_created_scheme,:date_issued_scheme,:date_available_scheme,:date_valid_scheme,:date_modified_scheme,:dc_type,:dc_format,:dc_identifier,:dc_source,:dc_language,:language_original,:dc_relation,:dc_coverage,:dc_rights,:keywords,:comments,:abstract,:cls_loc,:cls_dewey,:cls_pg,:cls_isbn,:papersize,:papersize_array,:toc,:lv1,:lv2,:lv3,:lv4,:lv5,:lv6,:pagenew,:pagebreak,:num_top,:toc_lev_limit,:flag_endnotes,:flag_auto_endnotes,:flag_separate_endnotes,:flag_separate_endnotes_make,:flag_auto_heading_num,:markup,:markup_instruction,:markup_version,:markup_declared,:make_bold,:make_italic,:flag_tables,:vocabulary,:doc_skin,:doc_css,:yaml,:lnk,:prefix_a,:prefix_b,:suffix,:information,:contact,:icon,:image,:ad_url,:ad_png,:ad_alt,:ad_began,:flag_promo,:promo,:ad_home,:stmp,:stmpd,:sc_filename,:sc_number,:sc_date,:sc_time,:sc_info,:yamladdr,:locale,:wc_lines,:wc_words,:wc_bytes,:file_encoding,:file_size,:user,:home,:hostname,:pwd,:firstseg,:programs,:creator_copymark,:lang,:en,:dgst,:dgst_skin,:generated,:tags,:tag_array,:concord_make,:seg_names,:seg_autoname_safe,:set_header_title,:set_heading_top,:set_heading_seg,:heading_seg_first,:heading_seg_first_flag,:base_program,:man_section,:man_name,:man_synopsis,:ec,:opt,:sem_tag,:book_idx,:topic_register,:topic_register_array,:original_publication_details + attr_accessor :cmd,:mod,:env,:fn,:fns,:fnb,:fnn,:fnt,:fnl,:flv,:fnz,:fnstex,:ocn,:sfx_src,:sfx,:pdf,:file_type,:dir_out,:dir_tex,:dir_lout,:txt_path,:site_skin,:sisu,:sisu_version,:ruby_version,:title,:dc_title,:html_title,:subtitle,:subtitle_tex,:creator_home,:dc_creator,:author_title,:author_nationality,:authors,:authorship,:translator,:illustrator,:prepared_by,:digitized_by,:dc_subject,:dc_description,:dc_publisher,:dc_contributor,:dc_date,:dc_date_created,:dc_date_issued,:dc_date_available,:dc_date_valid,:dc_date_modified,:date_translated,:date_added_to_site,:date_scheme,:date_created_scheme,:date_issued_scheme,:date_available_scheme,:date_valid_scheme,:date_modified_scheme,:dc_type,:dc_format,:dc_identifier,:dc_source,:dc_language,:language_original,:dc_relation,:dc_coverage,:dc_rights,:keywords,:comments,:abstract,:cls_loc,:cls_dewey,:cls_pg,:cls_isbn,:papersize,:papersize_array,:toc,:lv1,:lv2,:lv3,:lv4,:lv5,:lv6,:pagenew,:pagebreak,:num_top,:toc_lev_limit,:flag_endnotes,:flag_auto_endnotes,:flag_separate_endnotes,:flag_separate_endnotes_make,:flag_auto_heading_num,:markup,:markup_instruction,:markup_version,:markup_declared,:make_bold,:make_italic,:flag_tables,:vocabulary,:doc_skin,:doc_css,:yaml,:lnk,:prefix_a,:prefix_b,:suffix,:information,:contact,:icon,:image,:ad_url,:ad_png,:ad_alt,:ad_began,:flag_promo,:promo,:ad_home,:stmp,:stmpd,:sc_filename,:sc_number,:sc_date,:sc_time,:sc_info,:yamladdr,:locale,:wc_lines,:wc_words,:wc_bytes,:file_encoding,:file_size,:user,:home,:hostname,:pwd,:firstseg,:programs,:creator_copymark,:lang,:en,:dgst,:dgst_skin,:generated,:tags,:tag_array,:concord_make,:seg_names,:seg_autoname_safe,:set_header_title,:set_heading_top,:set_heading_seg,:heading_seg_first,:heading_seg_first_flag,:base_program,:man_section,:man_name,:man_synopsis,:ec,:opt,:sem_tag,:book_idx,:topic_register,:topic_register_array,:original_publication,:original_publication_date,:original_publication_nationality,:original_publication_institution,:writing_focus def initialize(fns_array,opt) - @env=@fn=@fns=@fnb=@fnn=@fnt=@fnl=@flv=@fnz=@fnstex=@ocn=@sfx_src=@sfx=@pdf=@file_type=@dir_out=@dir_tex=@dir_lout=@txt_path=@flag_endnotes=@flag_auto_endnotes=@flag_separate_endnotes=@flag_separate_endnotes_make=@site_skin=@sisu=@sisu_version=@ruby_version=@title=@dc_title=@html_title=@subtitle=@subtitle_tex=@creator_home=@dc_creator=@translator=@illustrator=@prepared_by=@digitized_by=@dc_subject=@dc_description=@dc_publisher=@dc_contributor=@dc_date=@dc_date_created=@dc_date_issued=@dc_date_available=@dc_date_valid=@dc_date_modified=@date_scheme=@date_created_scheme=@date_issued_scheme=@date_available_scheme=@date_valid_scheme=@date_modified_scheme=@dc_type=@dc_format=@dc_identifier=@dc_source=@dc_language=@language_original=@dc_relation=@dc_coverage=@dc_rights=@keywords=@comments=@abstract=@cls_loc=@cls_dewey=@cls_pg=@cls_isbn=@papersize=@toc=@lv1=@lv2=@lv3=@lv4=@lv5=@lv6=@pagenew=@pagebreak=@num_top=@toc_lev_limit=@flag_auto_heading_num=@make_bold=@make_italic=@flag_tables=@vocabulary=@doc_skin=@doc_css=@yaml=@lnk=@prefix_a=@prefix_b=@suffix=@information=@contact=@icon=@ad_url=@ad_png=@ad_alt=@ad_began=@promo=@ad_home=@stmp=@stmpd=@sc_filename=@sc_number=@sc_date=@sc_time=@sc_info=@yamladdr=@locale=@wc_lines=@wc_words=@wc_bytes=@file_encoding=@file_size=@firstseg=@programs=@creator_copymark=@lang=@en=@dgst=@dgst_skin=@generated=@heading_seg_first=@base_program=@man_synopsis=@topic_register=@original_publication_details=nil + @env=@fn=@fns=@fnb=@fnn=@fnt=@fnl=@flv=@fnz=@fnstex=@ocn=@sfx_src=@sfx=@pdf=@file_type=@dir_out=@dir_tex=@dir_lout=@txt_path=@flag_endnotes=@flag_auto_endnotes=@flag_separate_endnotes=@flag_separate_endnotes_make=@site_skin=@sisu=@sisu_version=@ruby_version=@title=@dc_title=@html_title=@subtitle=@subtitle_tex=@creator_home=@dc_creator=@author_title=@author_nationality=@translator=@illustrator=@prepared_by=@digitized_by=@dc_subject=@dc_description=@dc_publisher=@dc_contributor=@dc_date=@dc_date_created=@dc_date_issued=@dc_date_available=@dc_date_valid=@dc_date_modified=@date_translated=@date_added_to_site=@date_scheme=@date_created_scheme=@date_issued_scheme=@date_available_scheme=@date_valid_scheme=@date_modified_scheme=@dc_type=@dc_format=@dc_identifier=@dc_source=@dc_language=@language_original=@dc_relation=@dc_coverage=@dc_rights=@keywords=@comments=@abstract=@cls_loc=@cls_dewey=@cls_pg=@cls_isbn=@papersize=@toc=@lv1=@lv2=@lv3=@lv4=@lv5=@lv6=@pagenew=@pagebreak=@num_top=@toc_lev_limit=@flag_auto_heading_num=@make_bold=@make_italic=@flag_tables=@vocabulary=@doc_skin=@doc_css=@yaml=@lnk=@prefix_a=@prefix_b=@suffix=@information=@contact=@icon=@ad_url=@ad_png=@ad_alt=@ad_began=@promo=@ad_home=@stmp=@stmpd=@sc_filename=@sc_number=@sc_date=@sc_time=@sc_info=@yamladdr=@locale=@wc_lines=@wc_words=@wc_bytes=@file_encoding=@file_size=@firstseg=@programs=@creator_copymark=@lang=@en=@dgst=@dgst_skin=@generated=@heading_seg_first=@base_program=@man_synopsis=@topic_register=@original_publication_details=@original_publication=@original_publication_date=@original_publication_nationality=@original_publication_institution=@writing_focus=nil @man_section=1 @man_name='man page "name/whatis" information not provided, set in header @man: name=[whatis information]' @data,@fns,@cmd,@mod,@opt=fns_array,opt.fns,opt.cmd,opt.mod,opt #@data used as data @@ -324,6 +324,8 @@ module SiSU_Param names=name_format($1) @authorship=@dc_creator=names[:name_str] @authors=names[:name_a_h] + when /^@(?:creator|author)\.title:\s+(.+?)$/; @author_title=$1 # Prof. Dr. etc. + when /^@(?:creator|author)\.nationality:\s+(.+?)$/; @author_nationality=$1 when /^(?:@(?:translator|translated_by):|0~(?:translator|translated_by))\s+(.+?)$/m #% metainfo names=name_format($1) @translator=names[:name_str] @@ -336,18 +338,20 @@ module SiSU_Param when /^(?:@digitized_by:|0~digitized_by)\s+(.+?)$/m #% metainfo DC names=name_format($1) @digitized_by=names[:name_str] - when /^(?:@subject:|0~subject)\s+(.+?)$/m #% metainfo DC - @dc_subject=$1 - when /^(?:@description:|0~description)\s+(.+?)$/m #% metainfo DC & rss feed - @dc_description=$1 + when /^(?:@subject:|0~subject)\s+(.+?)$/m; @dc_subject=$1 #% metainfo DC + when /^(?:@description:|0~description)\s+(.+?)$/m; @dc_description=$1 #% metainfo DC & rss feed when /^(?:@contributor:|0~contributor)\s+(.+?)$/m #% metainfo DC names=name_format($1) @dc_contributor=names[:name_str] - when /^(?:@publisher:|0~publisher)\s+(.+?)$/m #% metainfo DC - @dc_publisher=$1 + when /^(?:@publisher:|0~publisher)\s+(.+?)$/m; @dc_publisher=$1 #% metainfo DC + when /^@original_publication:\s+(.+?)$/m; @original_publication=$1 + when /^@original_publication\.date:\s+(.+?)$/; @original_publication_date=$1 + when /^@original_publication\.nationality:\s+(.+?)$/; @original_publication_nationality=$1 + when /^@original_publication\.institution:\s+(.+?)$/; @original_publication_institution=$1 + when /^@writing_focus\.nationality:\s+(.+?)$/; @writing_focus=$1 # e.g. Finland (where and article on Finnish law) when /^(?:@|0~)date.+?$/m #% metainfo DC if para =~/(?:@date:|0~date)\s+(.+?)$/m - @dc_date=$1.strip + @dc_date=$1.strip #% original publication date unless the substantive text is updated/modified, then date of update if @dc_date !~regx_date \ and not @dc_date.empty? tell=SiSU_Screen::Ansi.new(@cmd,'Date Format should be','YYYY-MM-DD','please correct document','Date','field, current value:',@dc_date) @@ -355,6 +359,15 @@ module SiSU_Param end @date_scheme='scheme="ISO-8601"' if @dc_date =~/\d{4}-\d{2}-\d{2}/ end + if para =~/(?:@date\.added_to_site:|0~date\.added_to_site)\s+(.+?)$/m + @date_added_to_site=$1.strip + if @date_added_to_site !~regx_date \ + and not @date_added_to_site.empty? + tell=SiSU_Screen::Ansi.new(@cmd,'Date Format should be','YYYY-MM-DD','please correct document','Date','field, current value:',@date_added_to_site) + tell.instruct if @cmd =~/v/ + end + @date_scheme='scheme="ISO-8601"' if @date_added_to_site =~/\d{4}-\d{2}-\d{2}/ + end if para =~/(?:@date\.created:|0~date\.created)\s*(.+?)$/m date=$1.strip if date !~regx_date \ @@ -405,6 +418,16 @@ module SiSU_Param @dc_date_modified=date @date_modified_scheme='scheme="ISO-8601"' if date =~/\d{4}-\d{2}-\d{2}/ end + if para =~/^(?:@date\.translated:|0~date\.translated)\s*(.+?)$/m + date=$1.strip + if date !~regx_date \ + and not date.empty? + tell=SiSU_Screen::Ansi.new(@cmd,'Date Format should be','YYYY-MM-DD','please correct document','Date','field, current value:',date) + tell.instruct if @cmd =~/v/ + end + @date_translated=date + @date_translated_scheme='scheme="ISO-8601"' if date =~/\d{4}-\d{2}-\d{2}/ + end when /^(?:@type:|0~type)\s+(.+?)$/m; @dc_type=$1 #% metainfo DC when /^(?:@format:|0~format)\s+(.+?)$/m; @dc_format=$1 #% metainfo DC #when /^(?:@identifier:|0~identifier)\s+(.+?)$/m; @dc_identifier=$1 #% metainfo DC -- cgit v1.2.3