mods 0.0.2 → 0.0.4

data/README.rdoc

@@ -1,10 +1,12 @@
 = Mods
 
-A Ruby gem to parse MODS (Metadata Object Description Schema) records.  More information about MODS can be found at http://www.loc.gov/standards/mods/registry.php.
+{<img src="https://secure.travis-ci.org/sul-dlss/mods.png?branch=master" alt="Build Status" />}[http://travis-ci.org/sul-dlss/mods]
+
+A Gem to parse MODS (Metadata Object Description Schema) records.  More information about MODS can be found at http://www.loc.gov/standards/mods/registry.php.
 
 Source code at https://github.com/sul-dlss/mods/
 
-Generated API docs at  http://rubydoc.info/gems/mods/
+Generated API docs at  http://rubydoc.info/github/sul-dlss/mods/
 
 == Installation
 
@@ -35,5 +37,7 @@ TODO: Write usage instructions here
 
 == Releases
 
+0.0.4 implement language, location, origin_info, physical_description
+0.0.3 use nom-xml gem and make this more nokogiri-ish; implement name, title, and simple top level elements with no subelements
 0.0.2 Set up rake tasks, publishing rdoc, and continuous integration.
 0.0.1 Grab the name

data/lib/mods.rb

@@ -1,5 +1,15 @@
-require "mods/version"
+require 'mods/constants'
+require 'mods/marc_relator_codes'
+require 'mods/name'
+require 'mods/nom_terminology'
+require 'mods/reader'
+require 'mods/record'
+require 'mods/title_info'
+require 'mods/version'
+
+require 'nokogiri'
+require 'nom/xml'
 
 module Mods
-  # Your code goes here...
+    
 end

data/lib/mods/constants.rb

@@ -0,0 +1,60 @@
+module Mods
+  # the version of MODS supported by this gem
+  MODS_VERSION = '3.4'
+  
+  MODS_NS_V3 = "http://www.loc.gov/mods/v3"
+  MODS_NS = MODS_NS_V3
+  MODS_XSD = "http://www.loc.gov/standards/mods/mods.xsd"
+  
+  DOC_URL = "http://www.loc.gov/standards/mods/"
+
+  # top level elements that cannot have subelement children
+  TOP_LEVEL_ELEMENTS_SIMPLE = [
+    'abstract',
+    'accessCondition',
+    'classification',
+    'extension',
+    'genre',
+    'identifier',
+    'note',
+    'tableOfContents',
+    'targetAudience',
+    'typeOfResource',
+    ]
+
+  # top level elements that can have subelement children
+  TOP_LEVEL_ELEMENTS_COMPLEX = [
+    'language',
+    'location',
+    'name',
+    'originInfo',
+    'part',
+    'physicalDescription',
+    'recordInfo',
+    'relatedItem',
+    'subject',
+    'titleInfo' ]
+
+  TOP_LEVEL_ELEMENTS = Array.new(TOP_LEVEL_ELEMENTS_SIMPLE).concat(TOP_LEVEL_ELEMENTS_COMPLEX)
+
+  # enumerated attribute values
+  TITLE_INFO_TYPES = ['abbreviated', 'translated', 'alternative', 'uniform']
+  RELATED_ITEM_TYPES = [
+    'preceding', 'succeeding', 'original', 'host', 'constituent', 'series', 
+    'otherVersion', 'otherFormat', 'isReferencedBy', 'references', 'reviewOf'
+    ]
+
+  # enumerated values
+  TYPE_OF_RESOURCE_VALUES = [
+    'text', 'cartographic', 'notated music', 'sound recording-musical', 'sound recording-nonmusical',
+  'sound recording',
+  'still image',
+  'moving image',
+  'three dimensional object',
+  'software', 
+  'multimedia',
+  'mixed material']
+  
+  ORIGIN_INFO_DATE_ELEMENTS = ['dateIssued', 'dateCreated', 'dateCaptured', 'dateValid', 'dateModified', 'copyrightDate', 'dateOther']
+
+end

data/lib/mods/marc_relator_codes.rb

@@ -0,0 +1,228 @@
+# Represents the Marc Relator Codes mapped to Relators, from http://www.loc.gov/marc/relators/relacode.html
+#  key - Marc Relator code
+#  value - Marc Relator term
+MARC_RELATOR = {
+  'acp' => 'Art copyist',
+  'act' => 'Actor',
+  'adp' => 'Adapter',
+  'aft' => 'Author of afterword, colophon, etc.',
+  'anl' => 'Analyst',
+  'anm' => 'Animator',
+  'ann' => 'Annotator',
+  'ant' => 'Bibliographic antecedent',
+  'app' => 'Applicant',
+  'aqt' => 'Author in quotations or text abstracts',
+  'arc' => 'Architect',
+  'ard' => 'Artistic director',
+  'arr' => 'Arranger',
+  'art' => 'Artist',
+  'asg' => 'Assignee',
+  'asn' => 'Associated name',
+  'att' => 'Attributed name',
+  'auc' => 'Auctioneer',
+  'aud' => 'Author of dialog',
+  'aui' => 'Author of introduction',
+  'aus' => 'Author of screenplay',
+  'aut' => 'Author',
+  'bdd' => 'Binding designer',
+  'bjd' => 'Bookjacket designer',
+  'bkd' => 'Book designer',
+  'bkp' => 'Book producer',
+  'blw' => 'Blurb writer',
+  'bnd' => 'Binder',
+  'bpd' => 'Bookplate designer',
+  'bsl' => 'Bookseller',
+  'ccp' => 'Conceptor',
+  'chr' => 'Choreographer',
+  'clb' => 'Collaborator',
+  'cli' => 'Client',
+  'cll' => 'Calligrapher',
+  'clr' => 'Colorist',
+  'clt' => 'Collotyper',
+  'cmm' => 'Commentator',
+  'cmp' => 'Composer',
+  'cmt' => 'Compositor',
+  'cng' => 'Cinematographer',
+  'cnd' => 'Conductor',
+  'cns' => 'Censor',
+  'coe' => 'Contestant -appellee',
+  'col' => 'Collector',
+  'com' => 'Compiler',
+  'con' => 'Conservator',
+  'cos' => 'Contestant',
+  'cot' => 'Contestant -appellant',
+  'cov' => 'Cover designer',
+  'cpc' => 'Copyright claimant',
+  'cpe' => 'Complainant-appellee',
+  'cph' => 'Copyright holder',
+  'cpl' => 'Complainant',
+  'cpt' => 'Complainant-appellant',
+  'cre' => 'Creator',
+  'crp' => 'Correspondent',
+  'crr' => 'Corrector',
+  'csl' => 'Consultant',
+  'csp' => 'Consultant to a project',
+  'cst' => 'Costume designer',
+  'ctb' => 'Contributor',
+  'cte' => 'Contestee-appellee',
+  'ctg' => 'Cartographer',
+  'ctr' => 'Contractor',
+  'cts' => 'Contestee',
+  'ctt' => 'Contestee-appellant',
+  'cur' => 'Curator',
+  'cwt' => 'Commentator for written text',
+  'dfd' => 'Defendant',
+  'dfe' => 'Defendant-appellee',
+  'dft' => 'Defendant-appellant',
+  'dgg' => 'Degree grantor',
+  'dis' => 'Dissertant',
+  'dln' => 'Delineator',
+  'dnc' => 'Dancer',
+  'dnr' => 'Donor',
+  'dpb' => 'Distribution place',
+  'dpc' => 'Depicted',
+  'dpt' => 'Depositor',
+  'drm' => 'Draftsman',
+  'drt' => 'Director',
+  'dsr' => 'Designer',
+  'dst' => 'Distributor',
+  'dtc' => 'Data contributor',
+  'dte' => 'Dedicatee',
+  'dtm' => 'Data manager',
+  'dto' => 'Dedicator',
+  'dub' => 'Dubious author',
+  'edt' => 'Editor',
+  'egr' => 'Engraver',
+  'elg' => 'Electrician',
+  'elt' => 'Electrotyper',
+  'eng' => 'Engineer',
+  'etr' => 'Etcher',
+  'evp' => 'Event place',
+  'exp' => 'Expert',
+  'fac' => 'Facsimilist',
+  'fld' => 'Field director',
+  'flm' => 'Film editor',
+  'fmo' => 'Former owner',
+  'fpy' => 'First party',
+  'fnd' => 'Funder',
+  'frg' => 'Forger',
+  'gis' => 'Geographic information specialist',
+  'grt' => 'Graphic technician',
+  'hnr' => 'Honoree',
+  'hst' => 'Host',
+  'ill' => 'Illustrator',
+  'ilu' => 'Illuminator',
+  'ins' => 'Inscriber',
+  'inv' => 'Inventor',
+  'itr' => 'Instrumentalist',
+  'ive' => 'Interviewee',
+  'ivr' => 'Interviewer',
+  'lbr' => 'Laboratory',
+  'lbt' => 'Librettist',
+  'ldr' => 'Laboratory director',
+  'led' => 'Lead',
+  'lee' => 'Libelee-appellee',
+  'lel' => 'Libelee',
+  'len' => 'Lender',
+  'let' => 'Libelee-appellant',
+  'lgd' => 'Lighting designer',
+  'lie' => 'Libelant-appellee',
+  'lil' => 'Libelant',
+  'lit' => 'Libelant-appellant',
+  'lsa' => 'Landscape architect',
+  'lse' => 'Licensee',
+  'lso' => 'Licensor',
+  'ltg' => 'Lithographer',
+  'lyr' => 'Lyricist',
+  'mcp' => 'Music copyist',
+  'mfp' => 'Manufacture place',
+  'mfr' => 'Manufacturer',
+  'mdc' => 'Metadata contact',
+  'mod' => 'Moderator',
+  'mon' => 'Monitor',
+  'mrb' => 'Marbler',
+  'mrk' => 'Markup editor',
+  'msd' => 'Musical director',
+  'mte' => 'Metal-engraver',
+  'mus' => 'Musician',
+  'nrt' => 'Narrator',
+  'opn' => 'Opponent',
+  'org' => 'Originator',
+  'orm' => 'Organizer of meeting',
+  'oth' => 'Other',
+  'own' => 'Owner',
+  'pat' => 'Patron',
+  'pbd' => 'Publishing director',
+  'pbl' => 'Publisher',
+  'pdr' => 'Project director',
+  'pfr' => 'Proofreader',
+  'pht' => 'Photographer',
+  'plt' => 'Platemaker',
+  'pma' => 'Permitting agency',
+  'pmn' => 'Production manager',
+  'pop' => 'Printer of plates',
+  'ppm' => 'Papermaker',
+  'ppt' => 'Puppeteer',
+  'prc' => 'Process contact',
+  'prd' => 'Production personnel',
+  'prf' => 'Performer',
+  'prg' => 'Programmer',
+  'prm' => 'Printmaker',
+  'pro' => 'Producer',
+  'prp' => 'Production place',
+  'prt' => 'Printer',
+  'pta' => 'Patent applicant',
+  'pte' => 'Plaintiff -appellee',
+  'ptf' => 'Plaintiff',
+  'pth' => 'Patent holder',
+  'ptt' => 'Plaintiff-appellant',
+  'pup' => 'Publication place',
+  'rbr' => 'Rubricator',
+  'rce' => 'Recording engineer',
+  'rcp' => 'Recipient',
+  'red' => 'Redactor',
+  'ren' => 'Renderer',
+  'res' => 'Researcher',
+  'rev' => 'Reviewer',
+  'rps' => 'Repository',
+  'rpt' => 'Reporter',
+  'rpy' => 'Responsible party',
+  'rse' => 'Respondent-appellee',
+  'rsg' => 'Restager',
+  'rsp' => 'Respondent',
+  'rst' => 'Respondent-appellant',
+  'rth' => 'Research team head',
+  'rtm' => 'Research team member',
+  'sad' => 'Scientific advisor',
+  'sce' => 'Scenarist',
+  'scl' => 'Sculptor',
+  'scr' => 'Scribe',
+  'sds' => 'Sound designer',
+  'sec' => 'Secretary',
+  'sgn' => 'Signer',
+  'sht' => 'Supporting host',
+  'sng' => 'Singer',
+  'spk' => 'Speaker',
+  'spn' => 'Sponsor',
+  'spy' => 'Second party',
+  'srv' => 'Surveyor',
+  'std' => 'Set designer',
+  'stl' => 'Storyteller',
+  'stm' => 'Stage manager',
+  'stn' => 'Standards body',
+  'str' => 'Stereotyper',
+  'tcd' => 'Technical director',
+  'tch' => 'Teacher',
+  'ths' => 'Thesis advisor',
+  'trc' => 'Transcriber',
+  'trl' => 'Translator',
+  'tyd' => 'Type designer',
+  'tyg' => 'Typographer',
+  'uvp' => 'University place',
+  'vdg' => 'Videographer',
+  'voc' => 'Vocalist',
+  'wam' => 'Writer of accompanying material',
+  'wdc' => 'Woodcutter',
+  'wde' => 'Wood-engraver',
+  'wit' => 'Witness'
+}

data/lib/mods/name.rb

@@ -0,0 +1,18 @@
+module Mods
+
+  class Name
+
+    NS_HASH = {'m' => MODS_NS_V3}
+    SUBELEMENTS = ['namePart', 'displayForm', 'affiliation', 'role', 'description']
+        
+    # attributes on name node
+    ATTRIBUTES = ['type', 'authority', 'authorityURI', 'valueURI', 'displayLabel', 'usage', 'altRepGroup', 'nameTitleGroup']
+
+    # valid values for type attribute on name node <name type="val"/>
+    TYPES = ['personal', 'corporate', 'conference', 'family']
+    # valid values for type attribute on namePart node <name><namePart type="val"/></name>
+    NAME_PART_TYPES = ['date', 'family', 'given', 'termsOfAddress']
+  
+  end
+  
+end

data/lib/mods/nom_terminology.rb

@@ -0,0 +1,221 @@
+module Mods
+  
+  class Record
+    
+    # set the NOM terminology;  do NOT use namespaces
+    # @param mods_ng_xml a Nokogiri::Xml::Document object containing MODS (without namespaces)
+    def set_terminology_no_ns(mods_ng_xml)
+      mods_ng_xml.set_terminology() do |t|
+
+# FIXME: may want to deal with camelcase vs. underscore in method_missing 
+ 
+        # These elements have no subelements - w00t!
+        Mods::TOP_LEVEL_ELEMENTS_SIMPLE.each { |elname|
+          t.send elname, :path => "/mods/#{elname}"
+        }
+
+        # TITLE_INFO ----------------------------------------------------------------------------
+        # note - titleInfo can be a top level element or a sub-element of relatedItem 
+        #   (<titleInfo> as subelement of <subject> is not part of the MODS namespace)
+        
+        t.title_info :path => '/mods/titleInfo' do |n|
+          n.type_at :path => '@type'
+          n.title :path => 'title'
+          n.subTitle :path => 'subTitle'
+          n.nonSort :path => 'nonSort'
+          n.partNumber :path => 'partNumber'
+          n.partName :path => 'partName'
+          n.sort_title :path => '.', :accessor => lambda { |node| 
+            if node.type_at.text != "alternative" || (node.type_at.text == "alternative" && mods_ng_xml.xpath('/mods/titleInfo').size == 1)
+              node.title.text + (!node.subTitle.text.empty? ? "#{@title_delimiter}#{node.subTitle.text}" : "" ) 
+            end
+          }
+          n.full_title :path => '.', :accessor => lambda { |node| 
+             (!node.nonSort.text.empty? ? "#{node.nonSort.text} " : "" ) + 
+             node.title.text + 
+             (!node.subTitle.text.empty? ? "#{@title_delimiter}#{node.subTitle.text}" : "" ) 
+          }
+          n.short_title :path => '.', :accessor => lambda { |node|  
+            if node.type_at.text != "alternative"
+              (!node.nonSort.text.empty? ? "#{node.nonSort.text} " : "" ) + 
+              node.title.text
+            end
+          }
+          n.alternative_title :path => '.', :accessor => lambda { |node|  
+            if node.type_at.text == "alternative"
+              (!node.nonSort.text.empty? ? "#{node.nonSort.text} " : "" ) + 
+              node.title.text
+            end
+          }
+        end
+        
+        # current way to do short_title correctly
+#        t.short_title :path => '/mods/titleInfo[not(@type=alternative)]', :accessor => lambda { |node|  
+#            (!node.nonSort.text.empty? ? "#{node.nonSort.text} " : "" ) + 
+#            node.title.text
+#          }
+
+
+        # NAME ------------------------------------------------------------------------------------
+
+        t.plain_name :path => '/mods/name' do |n|
+          
+          Mods::Name::ATTRIBUTES.each { |attr_name|
+            if attr_name != 'type'
+              n.send attr_name, :path => "@#{attr_name}", :accessor => lambda { |a| a.text }
+            else
+              n.type_at :path => "@#{attr_name}", :accessor => lambda { |a| a.text }
+            end
+          }
+          
+          n.namePart :path => 'namePart' do |np|
+            np.type_at :path => '@type'
+          end
+          n.displayForm :path => 'displayForm'
+          n.affiliation :path => 'affiliation'
+          n.description_el :path => 'description' # description is used by Nokogiri
+          n.role :path => 'role/roleTerm' do |r|
+            r.type_at :path => "@type", :accessor => lambda { |a| a.text }
+            r.authority :path => "@authority", :accessor => lambda { |a| a.text }
+          end
+        end
+
+        t.personal_name :path => '/mods/name[@type="personal"]' do |n|
+          n.family_name :path => 'namePart[@type="family"]'
+          n.given_name :path => 'namePart[@type="given"]'
+          n.termsOfAddress :path => 'namePart[@type="termsOfAddress"]'
+          n.date :path => 'namePart[@type="date"]'
+        end
+        
+        t.corporate_name :path => '/mods/name[@type="corporate"]'
+        t.conference_name :path => '/mods/name[@type="conference"]'
+
+        # LANGUAGE -------------------------------------------------------------------------------
+
+        t.language :path => '/mods/language' do |n|
+          n.languageTerm :path => 'languageTerm' do |lt|
+            lt.type_at :path => '@type', :accessor => lambda { |a| a.text }
+            lt.authority :path => '@authority', :accessor => lambda { |a| a.text }
+          end
+          n.code_term :path => 'languageTerm[@type="code"]'
+          n.text_term :path => 'languageTerm[@type="text"]'
+          n.scriptTerm :path => 'scriptTerm'
+        end
+
+        # PHYSICAL_DESCRIPTION -------------------------------------------------------------------
+        t.physical_description :path => '/mods/physicalDescription' do |n|
+          n.digitalOrigin :path => 'digitalOrigin'
+          n.extent :path => 'extent'
+          n.form :path => 'form' do |f|
+            f.authority :path => '@authority', :accessor => lambda { |a| a.text }
+            f.type_at :path => '@type', :accessor => lambda { |a| a.text }
+          end
+          n.internetMediaType :path => 'internetMediaType'
+          n.note :path => 'note' do |nn|
+            nn.displayLabel :path => '@displayLabel', :accessor => lambda { |a| a.text }
+            nn.type_at :path => '@type', :accessor => lambda { |a| a.text }
+          end
+          n.reformattingQuality :path => 'reformattingQuality'
+        end
+        
+        # LOCATION -------------------------------------------------------------------------------
+        t.location :path => '/mods/location' do |n|
+          n.physicalLocation :path => 'physicalLocation' do |e|
+            e.authority :path => '@authority', :accessor => lambda { |a| a.text }
+            e.displayLabel :path => '@displayLabel', :accessor => lambda { |a| a.text }
+          end
+          n.shelfLocator :path => 'shelfLocator'
+          n.url :path => 'url' do |e|
+            e.dateLastAccessed :path => '@dateLastAccessed', :accessor => lambda { |a| a.text }
+            e.displayLabel :path => '@displayLabel', :accessor => lambda { |a| a.text }
+            e.note :path => '@note', :accessor => lambda { |a| a.text }
+            e.access :path => '@access', :accessor => lambda { |a| a.text }
+            e.usage :path => '@usage', :accessor => lambda { |a| a.text }
+          end
+          n.holdingSimple :path => 'holdingSimple'
+          n.holdingExternal :path => 'holdingExternal'
+        end
+        
+        # ORIGIN_INFO --------------------------------------------------------------------------
+        t.origin_info :path => '/mods/originInfo' do |n|
+          n.place :path => 'place' do |e|
+            e.placeTerm :path => 'placeTerm' do |ee|
+              ee.type_at :path => '@type', :accessor => lambda { |a| a.text }
+              ee.authority :path => '@authority', :accessor => lambda { |a| a.text }
+            end
+          end
+          n.publisher :path => 'publisher'
+          Mods::ORIGIN_INFO_DATE_ELEMENTS.each { |date_el|
+            n.send date_el, :path => "#{date_el}" do |d|
+              d.encoding :path => '@encoding', :accessor => lambda { |a| a.text }
+              d.point :path => '@point', :accessor => lambda { |a| a.text }
+              d.keyDate :path => '@keyDate', :accessor => lambda { |a| a.text }
+              d.qualifier :path => '@qualifier', :accessor => lambda { |a| a.text }
+              if date_el == 'dateOther'
+                d.type_at :path => '@type', :accessor => lambda { |a| a.text }
+              end
+            end
+          }
+          n.edition :path => 'edition'
+          n.issuance :path => 'issuance'
+          n.frequency :path => 'frequency' do |f|
+            f.authority :path => '@authority', :accessor => lambda { |a| a.text }
+          end
+        end
+        
+      end # terminology
+
+      mods_ng_xml.nom!
+
+      mods_ng_xml
+    end # set_terminology_no_ns
+
+# TODO: common top level element attributes:  ID; xlink; lang; xml:lang; script; transliteration 
+#       authority, authorityURI, valueURI
+#       displayLabel, usage altRepGroup
+#       type
+# TODO:   common subelement attributes:   lang, xml:lang, script, transliteration
+# TODO: other common attribute:  supplied
+    
+    
+    # set the NOM terminology, with namespaces
+    # @param mods_ng_xml a Nokogiri::Xml::Document object containing MODS (with namespaces)
+    def set_terminology_ns(mods_ng_xml)
+      mods_ng_xml.set_terminology(:namespaces => { 'm' => Mods::MODS_NS}) do |t|
+
+        # note - titleInfo can be a top level element or a sub-element of subject and relatedItem
+        t.title_info :path => '//m:titleInfo' do |n|
+          n.title :path => 'm:title'
+          n.subTitle :path => 'm:subTitle'
+          n.nonSort :path => 'm:nonSort'
+          n.partNumber :path => 'm:partNumber'
+          n.partName :path => 'm:partName'
+          n.type_at :path => '@type'
+        end
+
+        t.author :path => '//m:name' do |n|
+          n.valueURI :path => '@valueURI'
+          n.namePart :path => 'm:namePart', :single => true
+        end
+
+        t.corporate_authors :path => '//m:name[@type="corporate"]'
+        t.personal_authors :path => 'm:name[@type="personal"]' do |n|
+          n.roleTerm :path => 'm:role/m:roleTerm'
+          n.name_role_pair :path => '.', :accessor => lambda { |node| node.roleTerm.text + ": " + node.namePart.text }
+        end
+
+        t.language :path => 'm:language' do |n|
+          n.value :path => 'm:languageTerm', :accessor => :text
+        end
+
+      end
+      
+      mods_ng_xml.nom!
+      
+      mods_ng_xml
+    end # set_terminology_ns
+    
+  end
+  
+end
+