uwdc 0.0.1

data/lib/uwdc/mods.rb

@@ -0,0 +1,206 @@
+require 'ostruct'
+
+module UWDC
+  # Public: Obtain the MODS metadata for a UWDC METS object
+  #
+  # Example
+  #
+  # @mods = UWDC::Mods.new('ECJ6ZXEYWUE7B8W')
+  # # => MODS object fetched from Fedora
+  # 
+  # @mods = UWDC::Mods.new('ECJ6ZXEYWUE7B8W', File.read('../file.xml'))
+  # # => object constructed from XML file
+  class Mods < Mets
+    
+    # Public: standardized, supported MODS attributes
+    def self.attributes
+      [:titles, :names, :dates, :forms, :abstracts, :subjects, :related_items]
+    end
+
+    # Public: Access the XML nodes of the MODS XML
+    #
+    # Example
+    #
+    # @mods.nodes
+    # # => Nokogiri::XML::NodeSet
+    #
+    # Returns the Nokogiri::XML::NodeSet for the parsed MODS XML
+    def nodes
+      @xml.nodes.xpath("//dmdSec[contains(@ID,'#{@id}')]//mods[1]")
+    end
+    
+    # Public: Access the XML nodes of the METS file
+    #
+    # Example
+    #
+    # @mods.metadata
+    # # => {:title => ['A life idyl', ...], ...}
+    #
+    # Returns Hash
+    def metadata
+      attributes = UWDC::Mods.attributes.inject({}) do |result, method|
+        result[method] = self.send(method)
+        result
+      end
+      attributes[:access_conditions] = self.access_conditions
+      attributes
+    end
+    
+    # Public: Array of title data
+    #
+    # Example
+    #
+    # @mods.titles
+    # # => ['A life ldyl', ...]
+    # 
+    # Returns Array
+    def titles
+      clean_nodes(nodes.xpath("//mods/titleInfo//title"))
+    end
+    
+    # Public: Array of roles and nameParts
+    #
+    # Example
+    #
+    # @mods.names
+    # # => [#<OpenStruct name="Foo", role="Bar">, ...]
+    #
+    # @mods.names.first.name
+    # # => "Foo"
+    #
+    # Returns Array of OpenStructs
+    def names
+      nodes.xpath("//mods/name").inject([]){|arr,node| arr << capture_name(node); arr}
+    end
+    
+    # Public: Array of dates
+    #
+    # Example
+    #
+    # @mods.dates
+    # # => ["1985", ...]
+    #
+    # Returns Array of Strings
+    def dates
+      clean_nodes(nodes.xpath("//mods/originInfo//dateIssued"))
+    end
+  
+    # Public: Array of forms
+    # Example
+    #
+    # @mods.forms
+    # # => ["StillImage", ...]
+    #
+    # Returns Array of Strings
+    def forms
+      clean_nodes(nodes.xpath("//mods/physicalDescription//form"))
+    end
+
+    # Public: Array of abstracts (sometimes translated)
+    # @TODO: support lang here
+    #
+    # Example
+    #
+    # @mods.abstracts
+    # # => ["The man is an entertainer who captured the wild hyena...", ...]
+    #
+    # Returns Array of Strings
+    def abstracts
+      clean_nodes(nodes.xpath("//mods//abstract"))
+    end
+  
+    # Public: Array of subject data
+    #
+    # Example
+    #
+    # @mods.subjects
+    # # => ["Delehanty, James", "Animals", ...]
+    #
+    # Returns Array of Strings
+    def subjects
+      clean_nodes(nodes.xpath("//mods/subject//topic"))
+    end
+    
+    # Public: Array of geographic subject data
+    # @TODO: subjects_heirarchical_geographic
+    #
+    # Example
+    # @mods.subjects_heirarchical_geographic
+    # # => ["Foo", "Bar", "Baz"]
+    def subjects_heirarchical_geographic
+    end
+    
+    # Public: OpenStruct - Terms of Use and Reuse Rights
+    #
+    # Example
+    # 
+    # @mods.access_conditions
+    # # => #<OpenStruct rights=["Delehanty, Jim"], reuse=["Delehanty, Jim"]>
+    #
+    # @mods.access_conditions.rights
+    # # => "Delehanty, Jim"
+    #
+    # Retuns an OpenStruct
+    def access_conditions
+      OpenStruct.new(rights: rights, reuse: reuse)
+    end
+    
+    # Public: Array of Structs - Related items' label and value
+    #
+    # Example
+    #
+    # @mods.related_items
+    # # => [#<OpenStruct label="Part of", name="Africa Focus">, ...]
+    #
+    # @mods.related_items.first.name
+    # # => "Africa Focus"
+    #
+    # Retuns Array of OpenStruct
+    def related_items
+      nodes.xpath("//mods/relatedItem").inject([]){|arr,node| arr << capture_relation(node) ; arr }
+    end
+  
+    # Public: Check MODS for validity
+    # @TODO: Use the UWDC MODS Schema
+    #
+    # Example
+    #
+    # @mods.valid?
+    # # => true
+    #
+    # Returns Boolean
+    def valid?
+      response = http_client.get("http://www.loc.gov/standards/mods/mods.xsd")
+      xsd = Nokogiri::XML::Schema.new(response.body)
+      xml = Nokogiri::XML(nodes.to_xml)
+      xsd.valid?(xml)
+    end
+    
+    private
+
+    # Internal: Rights/Ownership String    
+    def rights
+      clean_nodes(nodes.xpath("//accessCondition[@type='rightsOwnership']"))
+    end
+
+    # Internal: Reuse Rights String    
+    def reuse
+      clean_nodes(nodes.xpath("//accessCondition[@type='useAndReproduction']"))
+    end
+    
+    # Internal: Related Label String
+    def related_label(node)
+      node['displayLabel'] ? node['displayLabel'] : "Related item"
+    end
+    
+    # Internal: Capture relation metadata
+    def capture_relation(node)
+      OpenStruct.new(label: related_label(node), name: node.xpath('.//name | .//title').text)
+    end
+
+    # Internal: Capture name metadata
+    def capture_name(node)
+      OpenStruct.new(name: node.xpath('.//namePart').text, role: node.xpath('.//role/roleTerm').text)
+    end
+  end
+end

data/lib/uwdc/origin.rb

@@ -0,0 +1,12 @@
+module UWDC
+  # Return the UWDC/PREMIS-ish submitters of a METS file
+  class Origin < Mets
+    def nodes
+      @xml.nodes.xpath("//amdSec[contains(@ID,'#{@id}')]//origin[1]")
+    end
+
+    def submitters
+      clean_nodes(nodes.xpath(".//agentName"))
+    end
+  end
+end

data/lib/uwdc/rels_ext.rb

@@ -0,0 +1,15 @@
+module UWDC
+  # Return the CMODELS related to METS file objects
+  class RelsExt < Mets
+    def nodes
+      @xml.nodes.xpath("//amdSec[contains(@ID,'#{@id}')]//RDF[1]")
+    end
+    
+    def models
+      nodes.xpath("//amdSec[contains(@ID,'RELS-EXT')]").inject({}) do |result, node|
+        result["#{node.attribute('ID')}"] = pick_attribute(node.xpath(".//hasModel"), 'resource')
+        result
+      end
+    end
+  end
+end

data/lib/uwdc/struct_map.rb

@@ -0,0 +1,12 @@
+module UWDC
+  # Return the structure of a METS file
+  class StructMap < Mets
+    def nodes
+      @xml.nodes.xpath("//structMap[contains(@ID,'#{@id}')]")
+    end
+  
+    def structure
+      nodes.xpath('div').inject([]){|result, div| result << Div.new(div)}
+    end
+  end
+end

data/lib/uwdc/version.rb

@@ -0,0 +1,3 @@
+module UWDC
+  VERSION = "0.0.1"
+end

data/lib/uwdc/xml.rb

@@ -0,0 +1,38 @@
+module UWDC
+  # Read or Fetch METS XML files 
+  class XML
+    extend HTTPClient::IncludeClient
+    include_http_client
+    
+    attr_accessor :xml, :nodes, :status
+   
+    def initialize(id, xml=nil, status=nil)
+      @id ||= id
+      @xml,@status = http_xml(xml,status)
+    end
+    
+    def http_xml(xml,status)
+      unless xml
+        begin
+          # http://depot.library.wisc.edu:9090/fedora/objects/1711.dl:33QOBSVPJLWEM8S/methods/1711.dl:SDefineFirstClassObject/viewMets
+          #response = http_client.get("http://depot.library.wisc.edu/uwdcutils/METS/1711.dl:#{@id}")
+          response = http_client.get("http://depot.library.wisc.edu:9090/fedora/objects/1711.dl:#{@id}/methods/1711.dl:SDefineFirstClassObject/viewMets")
+          xml = response.body
+          status = response.status
+        rescue TimeoutError, HTTPClient::ConfigurationError, HTTPClient::BadResponseError, Nokogiri::SyntaxError => error
+          exception = error
+        end
+      end
+      [xml,status]
+    end
+    
+    def nodes
+      nokogiri_parse(@xml)
+    end
+
+    def nokogiri_parse(xml)
+      nodes = Nokogiri::XML.parse(xml)
+      nodes.remove_namespaces!
+    end
+  end
+end

data/spec/fixtures/africa_focus_mets.xml

@@ -0,0 +1,178 @@
+<?xml version="1.0" encoding="utf-8" standalone="no"?>
+<mets OBJID="1711.dl:ECJ6ZXEYWUE7B8W" LABEL="Hyena Wrestler with Muzzled Hyena" xmlns="http://www.loc.gov/METS/" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.loc.gov/METS/ http://depot.library.wisc.edu:9090/fedora/get/1711.dl:XMLSchema-METS-1.7/XSD">
+ <metsHdr/>
+ <dmdSec ID="x1711-dl_ECJ6ZXEYWUE7B8W-BIB0">
+  <mdWrap MIMETYPE="text/xml" MDTYPE="MODS" LABEL="MODS">
+   <xmlData>
+    <mods:mods xmlns:mods="http://www.loc.gov/mods/v3">
+  <mods:titleInfo>
+    <mods:title>Hyena Wrestler with Muzzled Hyena</mods:title>
+  </mods:titleInfo>
+  <mods:identifier type="hdl">1711.dl/ECJ6ZXEYWUE7B8W</mods:identifier>
+  <mods:name>
+    <mods:namePart>Delehanty, James</mods:namePart>
+    <mods:role>
+      <mods:roleTerm type="text">Photographer</mods:roleTerm>
+    </mods:role>
+  </mods:name>
+  <mods:subject>
+    <mods:topic>Delehanty, James</mods:topic>
+  </mods:subject>
+  <mods:subject>
+    <mods:topic>Animals</mods:topic>
+  </mods:subject>
+  <mods:subject>
+    <mods:topic>Hyenas</mods:topic>
+  </mods:subject>
+  <mods:subject>
+    <mods:topic>Portraits</mods:topic>
+  </mods:subject>
+  <mods:subject>
+    <mods:topic>Sports</mods:topic>
+  </mods:subject>
+  <mods:subject>
+    <mods:topic>Wrestling</mods:topic>
+  </mods:subject>
+  <mods:subject>
+    <mods:topic>West Africa</mods:topic>
+  </mods:subject>
+  <mods:abstract> The man is an entertainer who captured the wild hyena when it was a cub. He earns his living by wandering from village to village in Central Niger wrestling the hyena for the amusement of the crowd.</mods:abstract>
+  <mods:originInfo>
+    <mods:dateIssued encoding="iso8601">1985</mods:dateIssued>
+  </mods:originInfo>
+  <mods:typeOfResource>still image</mods:typeOfResource>
+  <mods:physicalDescription>
+    <mods:form>StillImage</mods:form>
+  </mods:physicalDescription>
+  <mods:relatedItem displayLabel="Part of" type="host">
+    <mods:titleInfo>
+      <mods:title>Africa Focus</mods:title>
+    </mods:titleInfo>
+  </mods:relatedItem>
+  <mods:accessCondition type="useAndReproduction">Delehanty, Jim</mods:accessCondition>
+  <mods:accessCondition type="rightsOwnership">Delehanty, Jim</mods:accessCondition>
+  <mods:recordInfo>
+    <mods:recordIdentifier>1711.dl:ECJ6ZXEYWUE7B8W.BIB0</mods:recordIdentifier>
+    <mods:recordOrigin>UWDCC</mods:recordOrigin>
+    <mods:recordChangeDate encoding="w3cdtf">2004-08-02</mods:recordChangeDate>
+    <mods:languageOfCataloging>
+      <mods:languageTerm authority="iso639-2b">eng</mods:languageTerm>
+    </mods:languageOfCataloging>
+  </mods:recordInfo>
+<mods:relatedItem displayLabel="Collection" type="host">
+      <mods:identifier>1711.dl:CollectionAfricaFocus</mods:identifier>
+      <mods:name>Africa Focus: Sights and Sounds of a Continent</mods:name>
+     </mods:relatedItem>
+    </mods:mods>
+   </xmlData>
+  </mdWrap>
+ </dmdSec>
+ <dmdSec ID="x1711-dl_S2OGIJADZCOBF83-DC">
+  <mdWrap MIMETYPE="text/xml" MDTYPE="DC" LABEL="Dublin Core Record for this object">
+   <xmlData>
+    <oai_dc:dc xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/ http://www.openarchives.org/OAI/2.0/oai_dc.xsd">
+  <dc:identifier>1711.dl:S2OGIJADZCOBF83</dc:identifier>
+</oai_dc:dc>
+   </xmlData>
+  </mdWrap>
+ </dmdSec>
+ <amdSec ID="x1711-dl_ECJ6ZXEYWUE7B8W-ORIGIN">
+  <digiprovMD ID="x1711-dl_ECJ6ZXEYWUE7B8W-ORIGIN-PROV0">
+   <mdWrap MIMETYPE="text/xml" MDTYPE="OTHER" OTHERMDTYPE="UNSPECIFIED" LABEL="ORIGIN">
+    <xmlData>
+     <uwdcProv:origin xmlns:premis="info:lc/xmlns/premis-v2" xmlns:uwdcProv="http://digital.library.wisc.edu/1711.dl/UWDCProv-1.0">
+  <uwdcProv:projectID>AfricaFocus</uwdcProv:projectID>
+  <uwdcProv:localID>2913jd27</uwdcProv:localID>
+  <uwdcProv:submitter>
+    <premis:agent>
+      <premis:agentIdentifier>
+        <premis:agentIdentifierType>UWDC Agent ID</premis:agentIdentifierType>
+        <premis:agentIdentifierValue>UWMadCLSASP</premis:agentIdentifierValue>
+      </premis:agentIdentifier>
+      <premis:agentName>University of Wisconsin--Madison. African Studies Program.</premis:agentName>
+    </premis:agent>
+  </uwdcProv:submitter>
+</uwdcProv:origin>
+    </xmlData>
+   </mdWrap>
+  </digiprovMD>
+ </amdSec>
+ <amdSec ID="x1711-dl_ECJ6ZXEYWUE7B8W-RELS-EXT">
+  <techMD ID="x1711-dl_ECJ6ZXEYWUE7B8W-RELS-EXT-TECHMD">
+   <mdWrap MIMETYPE="application/rdf+xml" MDTYPE="OTHER" OTHERMDTYPE="UNSPECIFIED" LABEL="Relationships">
+    <xmlData>
+     <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
+
+<rdf:Description rdf:about="info:fedora/1711.dl:ECJ6ZXEYWUE7B8W">
+	<hasMember xmlns="info:fedora/fedora-system:def/relations-external#" rdf:resource="info:fedora/1711.dl:S2OGIJADZCOBF83"/>
+	<hasModel xmlns="info:fedora/fedora-system:def/model#" rdf:resource="info:fedora/1711.dl:CModelUWDCObject"/>
+	<hasModel xmlns="info:fedora/fedora-system:def/model#" rdf:resource="info:fedora/1711.dl:CModelFirstClassObject"/>
+	<hasModel xmlns="info:fedora/fedora-system:def/model#" rdf:resource="info:fedora/1711.dl:CModelCompositeObject"/>
+	<isMemberOfProject xmlns="http://digital.library.wisc.edu/1711.dl/rdf/1.0/relations#" rdf:resource="hdl:1711.dl/AfricaFocus"/>
+	<isMemberOfCollection xmlns="info:fedora/fedora-system:def/relations-external#" rdf:resource="info:fedora/1711.dl:CollectionAfricaFocus"/>
+</rdf:Description>
+
+</rdf:RDF>
+    </xmlData>
+   </mdWrap>
+  </techMD>
+ </amdSec>
+ <amdSec ID="x1711-dl_S2OGIJADZCOBF83-RELS-EXT">
+  <techMD ID="x1711-dl_S2OGIJADZCOBF83-RELS-EXT-TECHMD">
+   <mdWrap MIMETYPE="application/rdf+xml" MDTYPE="OTHER" OTHERMDTYPE="UNSPECIFIED" LABEL="Relationships">
+    <xmlData>
+     <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
+
+<rdf:Description rdf:about="info:fedora/1711.dl:S2OGIJADZCOBF83">
+	<hasModel xmlns="info:fedora/fedora-system:def/model#" rdf:resource="info:fedora/1711.dl:CModelUWDCObject"/>
+	<hasModel xmlns="info:fedora/fedora-system:def/model#" rdf:resource="info:fedora/1711.dl:CModelImageWithDefaultRes"/>
+	<isMemberOfProject xmlns="http://digital.library.wisc.edu/1711.dl/rdf/1.0/relations#" rdf:resource="hdl:1711.dl/AfricaFocus"/>
+	<isMemberOf xmlns="info:fedora/fedora-system:def/relations-external#" rdf:resource="info:fedora/1711.dl:ECJ6ZXEYWUE7B8W"/>
+</rdf:Description>
+
+</rdf:RDF>
+    </xmlData>
+   </mdWrap>
+  </techMD>
+ </amdSec>
+ <fileSec>
+  <fileGrp ID="x1711-dl_ECJ6ZXEYWUE7B8W-DISSEMINATORS">
+   <file ID="x1711-dl_ECJ6ZXEYWUE7B8W-title" MIMETYPE="text/plain;charset=UTF-8" USE="title">
+    <FLocat LOCTYPE="URL" xlink:type="simple" xlink:href="http://depot.library.wisc.edu/repository/fedora/1711.dl:ECJ6ZXEYWUE7B8W/methods/1711.dl:SDefineUWDCObject/getTitle"/>
+   </file>
+   <file ID="x1711-dl_ECJ6ZXEYWUE7B8W-visualrepresentation" MIMETYPE="image/jpeg" USE="visualrepresentation">
+    <FLocat LOCTYPE="URL" xlink:type="simple" xlink:href="http://depot.library.wisc.edu/repository/fedora/1711.dl:ECJ6ZXEYWUE7B8W/methods/1711.dl:SDefineCompositeObject/getVisualRepresentation"/>
+   </file>
+  </fileGrp>
+  <fileGrp ID="x1711-dl_S2OGIJADZCOBF83-DISSEMINATORS">
+   <file ID="x1711-dl_S2OGIJADZCOBF83-title" MIMETYPE="text/plain;charset=UTF-8" USE="title">
+    <FLocat LOCTYPE="URL" xlink:type="simple" xlink:href="http://depot.library.wisc.edu/repository/fedora/1711.dl:S2OGIJADZCOBF83/methods/1711.dl:SDefineUWDCObject/getTitle"/>
+   </file>
+   <file ID="x1711-dl_S2OGIJADZCOBF83-thumb" MIMETYPE="image/jpeg" USE="thumb">
+    <FLocat LOCTYPE="URL" xlink:type="simple" xlink:href="http://depot.library.wisc.edu/repository/fedora/1711.dl:S2OGIJADZCOBF83/methods/1711.dl:SDefineImageWithDefaultRes/getThumb"/>
+   </file>
+   <file ID="x1711-dl_S2OGIJADZCOBF83-reference" MIMETYPE="image/jpeg" USE="reference">
+    <FLocat LOCTYPE="URL" xlink:type="simple" xlink:href="http://depot.library.wisc.edu/repository/fedora/1711.dl:S2OGIJADZCOBF83/methods/1711.dl:SDefineImageWithDefaultRes/getReference"/>
+   </file>
+   <file ID="x1711-dl_S2OGIJADZCOBF83-large" MIMETYPE="image/jpeg" USE="large">
+    <FLocat LOCTYPE="URL" xlink:type="simple" xlink:href="http://depot.library.wisc.edu/repository/fedora/1711.dl:S2OGIJADZCOBF83/methods/1711.dl:SDefineImageWithDefaultRes/getLarge"/>
+   </file>
+   <file ID="x1711-dl_S2OGIJADZCOBF83-icon" MIMETYPE="image/jpeg" USE="icon">
+    <FLocat LOCTYPE="URL" xlink:type="simple" xlink:href="http://depot.library.wisc.edu/repository/fedora/1711.dl:S2OGIJADZCOBF83/methods/1711.dl:SDefineImageWithIcon/getIcon"/>
+   </file>
+  </fileGrp>
+ </fileSec>
+ <structMap ID="x1711-dl_ECJ6ZXEYWUE7B8W-STRUCTMAP">
+  <div ID="x1711-dl_ECJ6ZXEYWUE7B8W-STRUCT" DMDID="x1711-dl_ECJ6ZXEYWUE7B8W-BIB0" ADMID="x1711-dl_ECJ6ZXEYWUE7B8W-RELS-EXT x1711-dl_ECJ6ZXEYWUE7B8W-ORIGIN">
+   <fptr FILEID="x1711-dl_ECJ6ZXEYWUE7B8W-title"/>
+   <fptr FILEID="x1711-dl_ECJ6ZXEYWUE7B8W-visualrepresentation"/>
+   <div ID="x1711-dl_S2OGIJADZCOBF83-STRUCT" ORDER="1" DMDID="x1711-dl_S2OGIJADZCOBF83-DC" ADMID="x1711-dl_S2OGIJADZCOBF83-RELS-EXT" CONTENTIDS="info:fedora/1711.dl:S2OGIJADZCOBF83">
+    <fptr FILEID="x1711-dl_S2OGIJADZCOBF83-title"/>
+    <fptr FILEID="x1711-dl_S2OGIJADZCOBF83-thumb"/>
+    <fptr FILEID="x1711-dl_S2OGIJADZCOBF83-reference"/>
+    <fptr FILEID="x1711-dl_S2OGIJADZCOBF83-large"/>
+    <fptr FILEID="x1711-dl_S2OGIJADZCOBF83-icon"/>
+   </div>
+  </div>
+ </structMap>
+</mets>