bpl_mods_mapper.rb

# Note: this is all pseudocode to mock up a DSL at this point.
# THIS IS A VERY INCOMPLETE EXAMPLE; much work to do for mods.

bpl_mods_mapper = Krikri::Mapper.new

bpl_mods_mapper.build do
  input_format          :xml # :json, :csv, :tsv others?
  selector              type: :xpath, "//record"

  # select an XPath relative to `selector` as the default parent for mappings
  default_parent        xpath: "metadata/mods:mods"

  aggregation do

    provider do
      # Set "literals" and "identifiers" just as strings
      prefLabel         "Digital Commonwealth"
      providedLabel     prefLabel   # refer to a previously set value
      uri               "http://jux/wuz"
    end

    # specify a particular XML element via XPath
    dataProvider        xpath: "mods:location/mods:physicalLocation"
    # specify using XPath when attributes have specific values
    isShownAt           xpath: "mods:location/mods:url[@usage='primary' and @access='object in context']"
    preview             xpath: "mods:location/mods:url[@access='preview']"

    aggregatedCHO do

      collection do
        title           xpath: "dc:collection"
        # run a function based on a parsed value
        uri             generate_opaque_uri(xpath: selector + "/header/setspec")
      end

      multiple_map(:contributor, :creator) do |record|
        :creator = record.xpath("mods:name[mods:role/mods:roleTerm='creator']")
        :contributor = record.xpath("mods:name")
        if :creator.nil?
          :creator = :contributor[0]
          :contributor = :contributor[1..-1]
        end
        :contributor.delete(:creator) if :creator in :contributor
        # remove <affiliation>, <displayForm>, <description>, and <role>.
      end 

      date do
        providedLabel   xpath: "mods:originInfo/*[self::dateCreated|self::dateIssued|self::dateOther|self::copyrightDate][@encoding='w3cdtf' and @keyDate='yes']"
      end

      description       xpath: "*[self::abstract|self::note]"
      extent            xpath: "mods:physicalDescription/mods:extent"
      format            xpath: "mods:genre"
      identifier        do |record|
        id = record.xpath("mods:identifier")
        id_value = id.xpath("text()")
        id_type = id.xpath("@type").titleize
        return "#{id_type}: #{id_value}"
      end

      language do
        providedLabel   xpath: "mods:language/mods:languageTerm"
        uri             xpath: "mods:language/mods:languageTerm[@valueURI]"
      end

      publisher         xpath: "mods:originInfo/mods:place/mods:placeTerm[@type='text']|mods:originInfo/mods:publisher"

      # relation do
      # <mods:location><mods:physicalLocation>
      # CONCATENATED with ". "
      # <mods:location><mods:holdingSimple><mods:copyInformation><mods:subLocation>
      # CONCATENATED with ". "
      # <mods:relatedItem type="host"><mods:titleInfo><mods:title>
      # CONCATENATED with ". "
      # <mods:relatedItem type=""series""><mods:titleInfo><mods:title>
      ### (e.g., Boston Public Library. Leslie Jones photograph collection)
      # end

      rights            xpath: "mods:accessCondition"

      #spatial do
        # <mods:subject><mods:hierarchicalGeographic>
        # <mods:subject><mods:geographic>
        # <mods:subject><mods:cartographics><mods:coordinates>
      #end

      subject do
        providedLabel   xpath: "mods:subject/*[not(self::hierarchicalGeographic|self::geographic|self::cartographics)]"
      end

      temporal do
        providedLabel   xpath: "mods:subject/mods:temporal"
      end 

      # "<titleInfo> has two subelements:
      # <title> <nonSort> AND <subTitle>
      # (<partNumber> and <partName> are not currently supported).
      # Supported title types are: 
      # <mods:titleInfo usage=""primary"">,
      # <mods:titleInfo type=""alternative"">,
      # <mods:titleInfo type=""translated"">,
      # <mods:titleInfo type=""uniform"">

      title             xpath: "mods:titleInfo/mods:title/mods:nonSort|mods:titleInfo/mods:subTitle"
      type              xpath: "mods:typeOfResource"
    end
  end
end

harvester_inferface_sketch.rb

# Note: this is all pseudocode to mock up a DSL at this point.

# This presents an interface slightly different from the other files,
# focused on the external interactions from Harvester -> OriginalRecord ->
# DPLA::MAP::Aggregation

harvester = Krikri::Harvesters::OAIHarvester.new(:endpoint => endpoint_url)

Krikri::Mapper.define :uiuc_marcxml do
  input_format  :xml
  selector      type: :xpath, "//record/metadata/record"
  
  # ...
  # the rest of the internals here; extrapolate from other examples
  # ...
end

original_rec = harvester.get_records.first

# map a single record
agg = Krikri::Mapper.map(:uiuc_marcxml, original_rec)
# => #<DPLA::MAP::Aggregation:0x3f93c7bf4efc(default)>
agg.persisted? # => true

# allow #map to run on #each if passed an enumerator
# ? this is maybe a long process if called on a typical harvester
Krikri::Mapper.map(:uiuc_marcxml, harvester.get_records)
# => [#<DPLA::MAP::Aggregation:0x3f93c7bf4efc(default), #<DPLA::MAP::Aggregation:0x3f93c7be595c(default)>, ...]

mdl_oai_qdc_mapper.rb

# Note: this is all pseudocode to mock up a DSL at this point.

mdl_oai_qdc_mapper = Krikri::Mapper.new

mdl_oai_qdc_mapper.build do
  input_format          :xml # :json, :csv, :tsv others?
  selector              type: :xpath, "//record"

  # select an XPath relative to `selector` as the default parent for mappings
  default_parent        xpath: "metadata/oai_qdc:qualifieddc"

  aggregation do

    provider do
      # Set "literals" and "identifiers" just as strings
      prefLabel         "Minnesota Digital Library"
      providedLabel     prefLabel   # refer to a previously set value
      uri               "http://foo/bar"
    end

    # specify a particular XML element via XPath
    dataProvider        xpath: "dc:publisher"
    # select the last dc:identifier
    isShownAt           xpath: "dc:identifier[last()]"

    # Use a function defined elsewhere (but where?)
    object              contentdm_map_object(isShownAt)

    # Use an inline function passed as a block; "with" ensures
    # that `preview` will only get set if `isShownAt` is not nil
    preview             with: isShownAt do
      if isShownAt.include? "cdm/ref"
        isShownAt.gsub("cdm/ref", "utils/getthumbnail")
      elsif if isShownAt.include? "u?"
        url_parts = isShownAt.split "u?"
        base_url = url_parts[0]
        rest = url_parts[1].split ","
        root = rest[0]
        pointer = rest[1]
        "#{base_url}cgi-bin/thumbnail.exe?CISOROOT=#{root}&CISOPTR=#{pointer}"
      else
        nil
      end
    end

    aggregatedCHO do

      collection do
        title           xpath: "dc:collection"
        # run a function based on a parsed value
        uri             generate_opaque_uri(xpath: selector + "/header/setspec")
      end

      contributor do
        providedLabel   xpath: "dc:contributor"
      end 

      creator do
        providedLabel   xpath: "dc:creator"
      end

      date do
        providedLabel   xpath: "dc:created"
      end

      description       xpath: "dc:description"
      extent            xpath: "dc:extent"
      format            xpath: "dc:medium"

      genre do
        providedLabel   xpath: "dcterms:format"
      end

      identifier        xpath: "dc:identifier"

      language do
        providedLabel   xpath: "dc:language"
      end 

      publisher         xpath: "dc:source"
      relation          xpath: "dc:isPartOf"
      #isReplacedBy
      #replaces
      rights            xpath: "dc:rights"

      spatial do
        providedLabel   xpath: "dcterms:spatial"
      end

      subject do
        # split on semicolons? here or later?
        providedLabel   xpath: "dc:subject"
      end

      temporal do
        providedLabel   xpath: "dcterms:temporal"
      end 

      title             xpath: "dc:title"
      type              map_type_from_literal(xpath: "dc:type")
    end
  end
end

parser_dsl.rb

uiuc_marcxml_mapper.build do
  # this specifies the parser
  # `selector`, `mapper_helper`, and other possible parser configurations would be 
  # encapsulated at that level.
  input_format          :marc # :xml, :json, :csv, :tsv, :modsxml, :rdf, others?
  
  aggregation do
    provider do
      # Set "literals" and "identifiers" just as strings
      prefLabel         "University of Illinois Urbana Champaign, University Library"
      providedLabel     prefLabel   # refer to a previously set value
                                    # #method_missing doing a very different thing in this 
                                    # context. Here be dragons?
      uri               "http://baz/quux"
    end
    
    dataProvider        provider.prefLabel # #method_missing magic
    isShownAt           record['856u']
    
    aggregatedCHO do
      contributor do
        # TODO: exclude 720 when $e is "aut" or "cre"
        # tom: figure this out!
        providedLabel   traject: "700:710:711:720" do |fld|
          if fld.tag == "720" && (fld['e'] == "aut" || fld['e'] == "cre")
            nil
          else
            extract_marc(fld)
          end
        end
      end 

      creator do
        providedLabel   record['100'], record['110'], record['111']
      end

      date do
        providedLabel   record['260']['c']  
      end

      # all 5xx fields, except 538
      # tom: figure this out!
      description       traject: "5.." do |fld| # note: this is a marcspecism
        extract_marc(fld) unless fld.tag == "538"
      end

      extent            record['300']['ac'], record['340']['b']

      # again, this is taking marcspec+traject syntax just as a demonstration
      format            traject: "LDR:007:337:338:340" do |fld|
        if fld.tag[0] == "3"
          extract(fld['a'])
        elsif field.tag == "LDR"
          # return LDR/6
          fld[6]
        else
          # Return 007/0
          fld[0]
        end 
      end

      #genre do
      #  providedLabel   #external mapping
      #end

      identifier        # 001; 020 [prefix ="ISBN: "]; 022$a [prefix ="ISSN: "]; 035$a; 050$a$b [prefix ="LC call number: "]

      language do
        providedLabel   # 041$a [$2 ids source, i.e. iso689-1]; OR 008 (positions 35-37)
      end 

      publisher         traject: extract_marc("260ab")
      relation          # 760-787
      #isReplacedBy
      #replaces
      rights            traject: extract_marc("506:540")

      spatial do
        providedLabel   traject: extract_marc("650z:651a:662")
      end

      subject do
        providedLabel   traject: extract_marc("600:61.:650:651:653:654:655:656:657:658:69.")
      end

      temporal do
        providedLabel   traject: extract_marc("648")
      end 

      title             traject: extract_marc("245:242:240") # don't use 245c
      type              traject: extract_marc("337a")
    end
  end
end

uiuc_marcxml_mapper.rb

# Note: this is all pseudocode to mock up a DSL at this point.

uiuc_marcxml_mapper = Krikri::Mapper.new

uiuc_marcxml_mapper.build do
  input_format          :xml # :json, :csv, :tsv others?
  selector              type: :xpath, "//record/metadata/record"

  # could expose convenience methods or other tools that assist with mapping
  # e.g. marcspec: http://cklee.github.io/marc-spec/marc-spec.html
  mapper_helper         :traject

  aggregation do

    provider do
      # Set "literals" and "identifiers" just as strings
      prefLabel         "University of Illinois Urbana Champaign, University Library"
      providedLabel     prefLabel   # refer to a previously set value
      uri               "http://baz/quux"
    end

    # specify a particular XML element via XPath
    dataProvider        provider.prefLabel
    isShownAt           traject: extract_marc("856u")

    aggregatedCHO do

      contributor do
        # TODO: exclude 720 when $e is "aut" or "cre"
        providedLabel   traject: "700:710:711:720" do |fld|
          if fld.tag == "720" && (fld['e'] == "aut" || fld['e'] == "cre")
            nil
          else
            extract_marc(fld)
          end
        end
      end 

      creator do
        providedLabel   traject: extract("100:110:111")
      end

      date do
        providedLabel   traject: extract_marc("260c")  
      end

      # all 5xx fields, except 538
      description       traject: "5.." do |fld| # note: this is a marcspecism
        extract_marc(fld) unless fld.tag == "538"
      end

      extent            traject: extract_marc("300ac:340b")

      # again, this is taking marcspec+traject syntax just as a demonstration
      format            traject: "LDR:007:337:338:340" do |fld|
        if fld.tag[0] == "3"
          extract(fld['a'])
        elsif field.tag == "LDR"
          # return LDR/6
          fld[6]
        else
          # Return 007/0
          fld[0]
        end 
      end

      #genre do
      #  providedLabel   #external mapping
      #end

      identifier        # 001; 020 [prefix ="ISBN: "]; 022$a [prefix ="ISSN: "]; 035$a; 050$a$b [prefix ="LC call number: "]

      language do
        providedLabel   # 041$a [$2 ids source, i.e. iso689-1]; OR 008 (positions 35-37)
      end 

      publisher         traject: extract_marc("260ab")
      relation          # 760-787
      #isReplacedBy
      #replaces
      rights            traject: extract_marc("506:540")

      spatial do
        providedLabel   traject: extract_marc("650z:651a:662")
      end

      subject do
        providedLabel   traject: extract_marc("600:61.:650:651:653:654:655:656:657:658:69.")
      end

      temporal do
        providedLabel   traject: extract_marc("648")
      end 

      title             traject: extract_marc("245:242:240") # don't use 245c
      type              traject: extract_marc("337a")
    end
  end
end