marc4j4r 0.1.0

data/lib/marc4j4r.rb

@@ -0,0 +1,345 @@
+unless defined? JRUBY_VERSION
+  raise "Only works under JRUBY"
+end
+
+begin
+  include_class Java::org.marc4j.marc.impl.RecordImpl
+rescue NameError  => e
+  jardir = File.join(File.dirname(__FILE__), '..', 'jars')
+  require "#{jardir}/marc4j.jar"
+end
+
+require 'set'
+
+module MARC4J4R
+
+  # Add some sugar to the MarcReader interface
+  #
+  # Adjust the interface so that a #new call to any implementations that
+  # implement it can take a java.io.InputStream, ruby IO obejct, or String
+  # (that will be interpreted as a filename) without complaining.
+  #
+  # The mechanism -- running module_eval on a string-representation of the 
+  # new method in each of the hard-coded implementations of MarcReader
+  # (MarcStreamReader,MarcPermissiveStreamReader,MarcXmlReader) -- is ugly
+  # and deeply unsettling.
+  #
+  # @author Bill Dueber
+  #
+  # A string used to override the initializer for each stream reader
+  # Need to do it this ugly way because of the way java and ruby interact;
+  # can't just add it to the MarcReader interface the way I wanted to.
+  
+  NEWINIT = <<-ENDBINDER
+    alias_method :oldinit, :initialize
+    def initialize(fromwhere)
+      stream = nil
+      if fromwhere.is_a? Java::JavaIO::InputStream
+        stream = fromwhere
+      elsif fromwhere.is_a? IO
+        stream = fromwhere.to_inputstream
+      else
+        stream = java.io.FileInputStream.new(fromwhere.to_java_string)
+      end
+      if self.class == Java::org.marc4j.MarcPermissiveStreamReader
+        self.oldinit(stream, true, true)
+      else
+        self.oldinit(stream)
+      end
+    end
+  ENDBINDER
+
+  Java::org.marc4j.MarcStreamReader.module_eval(NEWINIT)
+  Java::org.marc4j.MarcPermissiveStreamReader.module_eval(NEWINIT)
+  Java::org.marc4j.MarcXmlReader.module_eval(NEWINIT)
+  
+  
+  # Get a marc reader of the appropriate type
+  # @param [String, IO, java.io.InputStream] input The IO stream (or filename) from which you want to read
+  # @param [:strictmarc, :permissivemarc, :marcxml] The type of MARC reader you want. 
+  # @return [MarcReader] A MarcReader object with the syntactic sugar added in this file (e.g, each)
+  #
+  # @example Get a strict binary MARC reader for the file 'test.mrc'
+  #  reader = MARC4J4R.reader('test.mrc')
+  #
+  # @example Get a permissive binary MARC reader
+  #  reader = MARC4J4R.reader('test.mrc', :permissivemarc)
+  #
+  # @example Get a reader for an xml file
+  #  reader = MARC4J4R.reader('test.xml', :marcxml)
+  #
+  # @example Get a reader based on an existing IO object
+  #   require 'open-uri'
+  #   infile = open('http://my.machine.com/test.mrc')
+  #   reader = MARC4J4R.reader(infile)
+  
+  def reader(input, type  = :strictmarc)
+    case type
+    when :strictmarc
+      return Java::org.marc4j.MarcStreamReader.new(input)
+    when :permissivemarc
+      return Java::org.marc4j.MarcPermissiveStreamReader.new(input)
+    when :marcxml
+      return Java::org.marc4j.MarcXmlReader.new(input)
+    else
+      raise ArgumentError, "Reader type must be :strictmarc, :permissivemarc, or :marcxml"
+    end
+  end
+  module_function :reader
+
+end
+
+
+# Re-open the MarcReader interface, define #each and include Enumerable
+#
+# We also automatically call #hashify on the records that stream through
+# #each in order to speed up RecordImpl#[] when (a) doing many operations on a single
+# record, and (b) we're not worried about interleaved tags (e.g., a 520 followed by a 510 followed
+# by another 520)
+
+module Java::OrgMarc4j::MarcReader
+  include Enumerable
+  
+  # Return the next record, after calling #hashify on it
+  def each
+    while self.hasNext
+      r = self.next
+      r.hashify
+      yield r
+    end
+  end  
+end
+
+
+
+include_class Java::org.marc4j.marc.impl::RecordImpl
+include_class Java::org.marc4j.marc.impl::ControlFieldImpl
+include_class Java::org.marc4j.marc.impl::DataFieldImpl
+include_class Java::org.marc4j.marc.impl::SubfieldImpl
+
+# Open up RecordImpl to add some sugar, including Enumberable as well
+# @author Bill Dueber
+
+class RecordImpl
+  include Enumerable
+
+  # Create a local hash by tag number; makes some stuff faster
+  # Called automatically if you use reader.each
+  
+  def hashify 
+    return if @hashedtags # don't do it more than once
+    @hashedtags = {}
+    self.getVariableFields.each do |f|
+      @hashedtags[f.tag] ||= []
+      @hashedtags[f.tag].push f
+    end
+  end
+  
+  # Create a nice string of the record
+  def to_s
+    arr = ['LEADER ' + self.leader]
+    self.each do |f|
+      arr.push f.to_s
+    end
+    return arr.join("\n")
+  end
+
+  # Get the leader as a string (marc4j would otherwise return Leader object)
+  def leader
+    self.get_leader.toString
+  end
+
+  # Cycle through the fields in the order the appear in the record
+  def each
+    self.getVariableFields.each do |f|
+      yield f
+    end
+  end
+
+  # Get the first field associated with a tag
+  # @param [String] tag The tag
+  # @return [Field] The first matching field, or nil if none. Note that
+  # to mirror ruby-marc, this returns a single field 
+  
+  def [] tag
+    if defined? @hashedtags
+      return @hashedtags[tag][0]
+    else    
+      return self.getVariableField(tag)
+    end
+  end
+
+
+  # Get a (possibly empty) list of fields with the given tag(s)
+  #
+  # @param [String, Array<String>] tags A string (or Array of strings) with the tags you're interested in
+  # @param [Boolean] originalorder Whether or not results should be presented in the original order within the
+  # record or with a two-column sort of (a) Order of the tag in the list of tags sent, (b) order within that tag
+  # in the record
+  # @return [Array<Field>] Either an empty list or a list of one or more matched fields will be returned. 
+  #
+  # originalorder == false will use an internal hash and be faster in many cases (see #hashify)
+  #
+  # @example originalorder == false
+  #  # Given a record that looks like 
+  #  # 010    $a 68027371 
+  #  # 035    $a (RLIN)MIUG0001728-B 
+  #  # 035    $a (CaOTULAS)159818044 
+  #  # 035    $a (OCoLC)ocm00001728
+  #
+  #  r.find_by_tag(['035', '010']).each {|f| puts f.to_s}
+  #  # 035    $a (RLIN)MIUG0001728-B 
+  #  # 035    $a (CaOTULAS)159818044 
+  #  # 035    $a (OCoLC)ocm00001728
+  #  # 010    $a 68027371 
+  #
+  # # The results are ordered first by tag as passed in, then by original order within the tag
+  #
+  # @example originalorder == true
+  #  r.find_by_tag(['035', '010'], true).each {|f| puts f.to_s}
+  #  # 010    $a 68027371 
+  #  # 035    $a (RLIN)MIUG0001728-B 
+  #  # 035    $a (CaOTULAS)159818044 
+  #  # 035    $a (OCoLC)ocm00001728
+
+
+  def find_by_tag(tags, originalorder = false)
+    if !tags.is_a? Array
+      tags = [tags]
+    end
+    if (originalorder == false and @hashedtags == nil)
+      self.hashify
+    end
+    if originalorder
+      return self.find_all {|f| tags.include? f.tag}
+    else
+      # puts "Tags is #{tags}: got #{@hashedtags.values_at(*tags)}"
+      return @hashedtags.values_at(*tags).flatten.compact
+    end
+  end
+  
+
+  # Return the record as valid MARC-XML
+  # @return String A MARC-XML representation of the record, including the XML header
+  def to_xml
+    return @xml if @xml
+    @xml = java.io.StringWriter.new
+    res = javax.xml.transform.stream.StreamResult.new(@xml)
+    writer =  org.marc4j.MarcXmlWriter.new(res)
+    writer.write(self)
+    return @xml.toString
+  end
+
+end
+
+class ControlFieldImpl
+  def value
+    return self.data
+  end
+  
+  # Pretty-print
+  # @param [String] joiner What string to use to join the subfields
+  # @param [String] The pretty string
+  def to_s
+    return self.tag + "    " + self.value
+  end
+
+end
+
+class DataFieldImpl
+  include Enumerable
+
+  # Pretty-print
+  # @param [String] joiner What string to use to join the subfields
+  # @param [String] The pretty string
+  def to_s (joiner = ' ')
+    arr =  [self.tag + ' ' + self.indicator1 + self.indicator2]
+    self.each do |s|
+      arr.push s.to_s
+    end
+    return arr.join(joiner)
+  end
+
+  # Get the value of the first subfield of this field with the given code
+  # @param [String] code 1-character string of the subfield code
+  # @return [String] The value of the first matched subfield
+  def [] code
+    raise ArgumentError, "Code must be a one-character string, not #{code}" unless code.is_a? String and code.size == 1
+    # note that code[0] is just converting the one-character string into an integer 
+    # char value that the underlying java can deal with
+    self.getSubfield(code[0]).getData
+  end
+
+
+  # Get all values from the subfields for the given code or array of codes
+  # @param [String, Array<String>] code (Array of?) 1-character string(s) of the subfield code
+  # @param [Boolean] myorder Use the order of subfields that I gave instead of the order they're in the record
+  # @return [Array<String>] A possibly-empty array of Strings made up of the values in the subfields whose
+  # code is included in the given codes. If myorder == true, use the order in which they are passed in; if a code is repeated
+  # (ocassionally legal) subfield values will appear first ordered by the passed array, then by order within
+  # the document.
+  #
+  # If myorder is false, just return the values for matching subfields in the order they appear in the field.
+  #
+  # @example Quick examples:
+  #  # 260    $a New York, $b Van Nostrand Reinhold Co. $c 1969
+  # rec['260'].sub_values('a') #=> ["New York,"]
+  # rec['260'].sub_values(['a', 'c']) #=> ["New York,", "1969"]
+  # rec['260'].sub_values(['c', 'a']) #=> ["New York,", "1969"]
+  # rec['260'].sub_values(['c', 'a'], true) #=> ["1969", "New York"]
+  
+  def sub_values(code, myorder = false)
+    unless [Set, Array].include? code.class
+      code = [code] 
+      # puts "Arrayified for code #{code} / #{code.class}"
+    end
+    if myorder
+      subs = []
+      code.each do |c|
+        subs << self.find_all {|s| c == s.code}
+      end
+      return subs.flatten.map {|s| s.data}
+    else
+      return self.find_all{|s| code.include? s.code}.map {|s| s.data}
+    end
+  end
+
+  # Get first indicator as a one-character string
+  def indicator1
+    return self.getIndicator1.chr
+  end
+
+  # Get second indicator as a one-character string
+  def indicator2
+    return self.getIndicator2.chr
+  end
+
+  # Iterate over the subfields
+  def each
+    self.getSubfields.each do |s|
+      yield s
+    end
+  end
+
+  # Get the concatentated values of the subfields in order the appear in the field
+  # @param [String] joiner The string used to join the subfield values
+  def value joiner=' '
+    data = self.getSubfields.map {|s| s.data}
+    return data.join(joiner)
+  end  
+end
+
+class SubfieldImpl
+  
+  def value
+    return self.data
+  end
+  
+  def code
+    return self.getCode.chr
+  end
+
+  def to_s
+    return '$' + self.code + " " + self.data
+  end
+end

data/test/helper.rb

@@ -0,0 +1,10 @@
+require 'rubygems'
+require 'test/unit'
+require 'shoulda'
+
+$LOAD_PATH.unshift(File.join(File.dirname(__FILE__), '..', 'lib'))
+$LOAD_PATH.unshift(File.dirname(__FILE__))
+require 'marc4j4r'
+
+class Test::Unit::TestCase
+end

data/test/one.dat

@@ -0,0 +1 @@
+00755cam  22002414a 4500001001300000003000600013005001700019008004100036010001700077020004300094040001800137042000800155050002600163082001700189100003100206245005400237260004200291300007200333500003300405650003700438630002500475630001300500fol05731351 IMchF20000613133448.0000107s2000    nyua          001 0 eng    a   00020737   a0471383147 (paper/cd-rom : alk. paper)  aDLCcDLCdDLC  apcc00aQA76.73.P22bM33 200000a005.13/32211 aMartinsson, Tobias,d1976-10aActivePerl with ASP and ADO /cTobias Martinsson.  aNew York :bJohn Wiley & Sons,c2000.  axxi, 289 p. :bill. ;c23 cm. +e1 computer  laser disc (4 3/4 in.)  a"Wiley Computer Publishing." 0aPerl (Computer program language)00aActive server pages.00aActiveX.

data/test/one.xml

@@ -0,0 +1,55 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<collection xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns="http://www.loc.gov/MARC21/slim" xsi:schemaLocation="http://www.loc.gov/MARC21/slim http://www.loc.gov/standards/marcxml/schema/MARC21slim.xsd">
+    <record>
+        <leader>     njm a22     uu 4500</leader>
+        <controlfield tag="001">afc99990058366</controlfield>
+        <controlfield tag="003">DLC</controlfield>
+        <controlfield tag="005">20071104155141.9</controlfield>
+        <controlfield tag="007">sd ummunniauub</controlfield>
+        <controlfield tag="008">071103s1939    xxufmnne||||||||| u eng||</controlfield>
+        <datafield tag="010" ind1=" " ind2=" ">
+            <subfield code="a">afc99990058366</subfield>
+        </datafield>
+        <datafield tag="040" ind1=" " ind2=" ">
+            <subfield code="a">DLC</subfield>
+            <subfield code="c">DLC</subfield>
+        </datafield>
+        <datafield tag="245" ind1="0" ind2="4">
+            <subfield code="a">The Texas ranger</subfield>
+            <subfield code="h">[sound recording] /</subfield>
+            <subfield code="c">Sung by Beale D. Taylor.</subfield>
+        </datafield>
+        <datafield tag="260" ind1=" " ind2=" ">
+            <subfield code="a">Medina, Texas,</subfield>
+            <subfield code="c">1939.</subfield>
+        </datafield>
+        <datafield tag="300" ind1=" " ind2=" ">
+            <subfield code="a">1 sound disc :</subfield>
+            <subfield code="b">analog, 33 1/3 rpm, mono. ;</subfield>
+            <subfield code="c">12 in.</subfield>
+        </datafield>
+        <datafield tag="651" ind1=" " ind2="0">
+            <subfield code="a">Medina</subfield>
+            <subfield code="z">Texas</subfield>
+            <subfield code="z">United States of America.</subfield>
+        </datafield>
+        <datafield tag="700" ind1="1" ind2=" ">
+            <subfield code="a">Lomax, John Avery, 1867-1948</subfield>
+            <subfield code="e">Recording engineer.</subfield>
+        </datafield>
+        <datafield tag="700" ind1="1" ind2=" ">
+            <subfield code="a">Lomax, Ruby T. (Ruby Terrill)</subfield>
+            <subfield code="e">Recording engineer.</subfield>
+        </datafield>
+        <datafield tag="700" ind1="1" ind2=" ">
+            <subfield code="a">Taylor, Beale D.</subfield>
+            <subfield code="e">Singer.</subfield>
+        </datafield>
+        <datafield tag="852" ind1=" " ind2=" ">
+            <subfield code="a">American Folklife Center, Library of Congress</subfield>
+        </datafield>
+        <datafield tag="852" ind1=" " ind2=" ">
+            <subfield code="a">DLC</subfield>
+        </datafield>
+    </record>
+</collection>

data/test/test_marc4j4r.rb

@@ -0,0 +1,76 @@
+require 'helper'
+
+# one.xml
+# LEADER 00000njm a2200000uu 4500
+# 001    afc99990058366
+# 003    DLC
+# 005    20071104155141.9
+# 007    sd ummunniauub
+# 008    071103s1939    xxufmnne||||||||| u eng||
+# 010    $a afc99990058366
+# 040    $a DLC $c DLC
+# 245 04 $a The Texas ranger $h [sound recording] / $c Sung by Beale D. Taylor.
+# 260    $a Medina, Texas, $c 1939.
+# 300    $a 1 sound disc : $b analog, 33 1/3 rpm, mono. ; $c 12 in.
+# 651  0 $a Medina $z Texas $z United States of America.
+# 700 1  $a Lomax, John Avery, 1867-1948 $e Recording engineer.
+# 700 1  $a Lomax, Ruby T. (Ruby Terrill) $e Recording engineer.
+# 700 1  $a Taylor, Beale D. $e Singer.
+# 852    $a American Folklife Center, Library of Congress
+# 852    $a DLC
+
+class TestMarc4j4r < Test::Unit::TestCase
+  
+  def setup
+    reader = MARC4J4R.reader(File.dirname(__FILE__) + '/one.xml', :marcxml)
+    @r = reader.next
+  end
+
+  should "get the leader as a string" do 
+    assert_equal '00000njm a2200000uu 4500', @r.leader
+  end
+  
+  should "get all fields with the given tag" do 
+    assert_equal 3, @r.find_by_tag('700').size
+  end
+  
+  should "get all fields with any of the given tags" do
+    assert_equal 6, @r.find_by_tag(['010','700', '852']).size
+  end
+  
+  should "get an empty array trying to find a non-existent tag" do
+    assert_equal [], @r.find_by_tag('002')
+  end
+  
+  should "not return anything for a non-existent tag" do
+    assert_equal 1, @r.find_by_tag(['010', '002']).size
+  end
+  
+  should "get the value of a control tag" do
+    assert_equal 'DLC', @r['003'].value
+  end
+  
+  should "get a subfield value via field[]" do
+    assert_equal 'Sung by Beale D. Taylor.', @r['245']['c']
+  end
+  
+  should "joing all values of a field with a space" do
+    assert_equal "DLC DLC", @r['040'].value
+  end
+    
+  should "Get the first field with a given tag via []" do
+    assert_equal '700 1  $a Lomax, John Avery, 1867-1948 $e Recording engineer.', @r['700'].to_s
+  end
+  
+  should "get the subfield values in order of the original record" do
+    assert_equal ['Medina, Texas,', '1939.'], @r['260'].sub_values(['a', 'c'])
+    assert_equal ['Medina, Texas,', '1939.'], @r['260'].sub_values(['c', 'a'])
+  end
+  
+  should "get the subfield values in order of the codes I pass" do
+    assert_equal [ '1939.', 'Medina, Texas,'], @r['260'].sub_values(['c', 'a'], true)
+  end
+    
+  
+  
+end

metadata

@@ -0,0 +1,108 @@
+--- !ruby/object:Gem::Specification 
+name: marc4j4r
+version: !ruby/object:Gem::Version 
+  version: 0.1.0
+platform: ruby
+authors: 
+  - BillDueber
+autorequire: 
+bindir: bin
+cert_chain: []
+
+date: 2010-02-14 00:00:00 -05:00
+default_executable: 
+dependencies: 
+  - !ruby/object:Gem::Dependency 
+    name: thoughtbot-shoulda
+    type: :development
+    version_requirement: 
+    version_requirements: !ruby/object:Gem::Requirement 
+      requirements: 
+        - - ">="
+          - !ruby/object:Gem::Version 
+            version: "0"
+      version: 
+  - !ruby/object:Gem::Dependency 
+    name: yard
+    type: :development
+    version_requirement: 
+    version_requirements: !ruby/object:Gem::Requirement 
+      requirements: 
+        - - ">="
+          - !ruby/object:Gem::Version 
+            version: "0"
+      version: 
+description: Syntactic sugar and some extra methods to deal with classes in the marc4j java package
+email: bill@dueber.com
+executables: []
+
+extensions: []
+
+extra_rdoc_files: 
+  - LICENSE
+  - README.markdown
+files: 
+  - .document
+  - .gitignore
+  - LICENSE
+  - README.markdown
+  - Rakefile
+  - VERSION
+  - doc/ControlFieldImpl.html
+  - doc/DataFieldImpl.html
+  - doc/Java/OrgMarc4j/MarcReader.html
+  - doc/MARC4J4R.html
+  - doc/MARC4J4R/Reader.html
+  - doc/RecordImpl.html
+  - doc/SubfieldImpl.html
+  - doc/_index.html
+  - doc/class_list.html
+  - doc/css/common.css
+  - doc/css/full_list.css
+  - doc/css/style.css
+  - doc/file.README.html
+  - doc/file_list.html
+  - doc/frames.html
+  - doc/index.html
+  - doc/js/app.js
+  - doc/js/full_list.js
+  - doc/js/jquery.js
+  - doc/method_list.html
+  - doc/top-level-namespace.html
+  - jars/marc4j.jar
+  - lib/marc4j4r.rb
+  - test/helper.rb
+  - test/one.dat
+  - test/one.xml
+  - test/test_marc4j4r.rb
+has_rdoc: true
+homepage: http://github.com/billdueber/marc4j4r
+licenses: []
+
+post_install_message: 
+rdoc_options: 
+  - --charset=UTF-8
+require_paths: 
+  - lib
+required_ruby_version: !ruby/object:Gem::Requirement 
+  requirements: 
+    - - ">="
+      - !ruby/object:Gem::Version 
+        version: "0"
+  version: 
+required_rubygems_version: !ruby/object:Gem::Requirement 
+  requirements: 
+    - - ">="
+      - !ruby/object:Gem::Version 
+        version: "0"
+  version: 
+requirements: []
+
+rubyforge_project: 
+rubygems_version: 1.3.5
+signing_key: 
+specification_version: 3
+summary: Use marc4j java library in JRuby in a more ruby-ish way
+test_files: 
+  - test/helper.rb
+  - test/test_marc4j4r.rb