bibframe_ruby 0.1.2 → 0.2.0

data/lib/bibframe_ruby/marc_converter.rb

@@ -0,0 +1,67 @@
+# frozen_string_literal: true
+
+# BibframeRuby::MarcConverter: Converts MARC records to BIBFRAME RDF/XML via XSLT
+
+require "nokogiri"
+require "marc"
+require "stringio"
+
+module BibframeRuby
+  # Converts MARC (binary or MARCXML) to BIBFRAME RDF/XML using the vendored marc2bibframe2 XSLT.
+  class MarcConverter
+    XSL_DIR = File.expand_path("../../vendor/marc2bibframe2/xsl", __dir__)
+    XSL_PATH = File.join(XSL_DIR, "marc2bibframe2.xsl")
+    # Mutex to serialise Dir.chdir calls — Dir.chdir is process-global and not thread-safe.
+    CHDIR_MUTEX = Mutex.new
+
+    def initialize(input, baseuri: "http://example.org/", idsource: "http://id.loc.gov/vocabulary/organizations/dlc")
+      @input = input
+      @baseuri = baseuri
+      @idsource = idsource
+    end
+
+    def convert
+      xml = self.class.marcxml?(@input) ? @input : binary_to_marcxml(@input)
+      doc = Nokogiri::XML(xml)
+      rdfxml = transform(doc)
+      rdfxml.to_xml
+    end
+
+    def self.marcxml?(input)
+      input.lstrip.start_with?("<")
+    end
+
+    private
+
+    def binary_to_marcxml(input)
+      reader = MARC::Reader.new(StringIO.new(input))
+      output = StringIO.new
+      writer = MARC::XMLWriter.new(output)
+      reader.each { |record| writer.write(record) }
+      writer.close
+      output.string
+    end
+
+    def transform(doc)
+      params = ["baseuri", "\"#{@baseuri}\""]
+      params += ["idsource", "\"#{@idsource}\""] if @idsource
+
+      # Ensure the stylesheet is loaded (and its Dir.chdir released) before we
+      # acquire CHDIR_MUTEX again for the transform call.
+      sheet = self.class.stylesheet
+      CHDIR_MUTEX.synchronize do
+        Dir.chdir(XSL_DIR) do
+          sheet.transform(doc, params)
+        end
+      end
+    end
+
+    def self.stylesheet
+      @stylesheet ||= CHDIR_MUTEX.synchronize do
+        Dir.chdir(XSL_DIR) do
+          Nokogiri::XSLT(File.read(XSL_PATH))
+        end
+      end
+    end
+  end
+end

data/lib/bibframe_ruby/parser.rb

@@ -3,10 +3,11 @@
 require "json/ld"
 require "rdf"
 require "rdf/turtle"
+require "rdf/rdfxml"
 
 # BibframeRuby::Parser: Converts serialized RDF input into an RDF graph
 module BibframeRuby
-  # Parses serialized RDF input (JSON-LD, Turtle) into an RDF::Graph using the appropriate reader.
+  # Parses serialized RDF input (JSON-LD, Turtle, RDF/XML) into an RDF::Graph using the appropriate reader.
   class Parser
     EXTENSION_MAP = {
       ".jsonld" => :jsonld,
@@ -16,7 +17,8 @@ module BibframeRuby
 
     READER_MAP = {
       jsonld: JSON::LD::Reader,
-      turtle: RDF::Turtle::Reader
+      turtle: RDF::Turtle::Reader,
+      rdfxml: RDF::RDFXML::Reader
     }.freeze
 
     def initialize(input, format: :jsonld)

data/lib/bibframe_ruby/version.rb

@@ -2,5 +2,5 @@
 
 # BibframeRuby::VERSION: Gem version constant
 module BibframeRuby
-  VERSION = "0.1.2"
+  VERSION = "0.2.0"
 end

data/lib/bibframe_ruby.rb

@@ -17,6 +17,7 @@ require_relative "bibframe_ruby/models/hub"
 require_relative "bibframe_ruby/parser"
 require_relative "bibframe_ruby/graph_builder"
 require_relative "bibframe_ruby/graph"
+require_relative "bibframe_ruby/marc_converter"
 
 # BibframeRuby: Main entry point for parsing BIBFRAME data into Ruby objects
 module BibframeRuby
@@ -34,6 +35,12 @@ module BibframeRuby
     parse(input, format: format)
   end
 
+  def self.convert_marc(path, baseuri: "http://example.org/", idsource: "http://id.loc.gov/vocabulary/organizations/dlc")
+    input = File.read(path)
+    rdfxml = MarcConverter.new(input, baseuri: baseuri, idsource: idsource).convert
+    parse(rdfxml, format: :rdfxml)
+  end
+
   def self.parse_uri(uri, redirect_limit: 5)
     raise Error, "Too many redirects" if redirect_limit.zero?