marc4j4r 0.1.6 → 0.2.2

data/lib/original_monolithic_file.rb

@@ -0,0 +1,518 @@
+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'
+
+
+# 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(hashify=true)
+    while self.hasNext
+      r = self.next
+      r.hashify if hashify
+      yield r
+    end
+  end  
+end
+
+
+module MARC4J4R
+  
+  # Do some simple substitutions to make things prettier. After this,
+  # we can use MARC4J4R::Record instead of Java::org.marc4j.marc.impl::RecordImpl
+  
+  Record =  Java::org.marc4j.marc.impl::RecordImpl
+  ControlField = Java::org.marc4j.marc.impl::ControlFieldImpl
+  DataField = Java::org.marc4j.marc.impl::DataFieldImpl
+  SubField = Java::org.marc4j.marc.impl::SubfieldImpl
+  
+
+  # 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
+    include Enumerable
+    alias_method :oldinit, :initialize
+    def initialize(fromwhere)
+      stream = nil
+      if fromwhere.is_a? Java::JavaIO::InputStream or fromwhere.is_a? Java::JavaIO::ByteArrayInputStream
+        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 then
+      return Java::org.marc4j.MarcStreamReader.new(input)
+    when :permissivemarc then
+      return Java::org.marc4j.MarcPermissiveStreamReader.new(input)
+    when :marcxml then
+      return Java::org.marc4j.MarcXmlReader.new(input)
+    when :alephsequential then
+      return MARC4J4R::AlephSequentialReader.new(input)
+    else
+      raise ArgumentError, "Reader type #{type} illegal:  must be :strictmarc, :permissivemarc, :marcxml, or :alephsequential"
+    end
+  end
+  module_function :reader
+  
+  
+  # Implement an AlephSequential reader
+  class AlephSequentialReader
+    include Enumerable
+    def initialize(fromwhere)
+      stream = nil
+      if fromwhere.is_a? Java::JavaIO::InputStream
+        stream = fromwhere.to_io
+      elsif fromwhere.is_a? IO
+        stream = fromwhere
+      else
+        stream = File.new(fromwhere)
+      end
+      
+      @handle = stream
+    end
+  
+    def each
+      record = nil
+      currentID = nil
+      
+      @handle.each_line do |l|
+        l.chomp!
+        next unless l =~ /\S/
+        vals = l.unpack('a9 a a3 c c a3 a*')
+        id, tag, ind1, ind2, data = vals[0], vals[2], vals[3], vals[4], vals[6]
+        # id, tag, ind1, ind2, junk, data = *(l.unpack('A10 a3 c c a3 A*'))
+        if id != currentID
+          if record
+            yield record
+          end
+          record = RecordImpl.new
+          currentID = id
+        end
+        if tag == 'LDR'
+          record.setLeader(Java::org.marc4j.marc.impl.LeaderImpl.new(data))
+        else
+          record << buildField(tag,ind1,ind2,data)
+        end
+      end
+      yield record
+    end
+    
+    
+    SUBREGEXP = /\$\$(.)/
+    def buildField (tag, ind1, ind2, data)
+      if Java::org.marc4j.marc.impl.Verifier.isControlField tag
+        return Java::org.marc4j.marc.impl.ControlFieldImpl.new(tag, data)
+      else
+        f = Java::org.marc4j.marc.impl.DataFieldImpl.new(tag, ind1, ind2)
+        data.split(SUBREGEXP)[1..-1].each_slice(2) do |code, value|
+          f.addSubfield Java::org.marc4j.marc.impl.SubfieldImpl.new(code[0].ord, value)
+        end
+        return f
+      end
+    end
+    
+  end # End of class AlephSequentialReader
+
+end
+
+
+# Open up RecordImpl to add some sugar, including Enumberable as well
+# @author Bill Dueber
+
+class Record
+  include Enumerable
+  
+  alias_method :<<, :addVariableField
+  alias_method :append, :addVariableField
+  alias_method :fields, :getVariableFields
+
+  # Export as a MARC-Hash, as described at 
+  # http://robotlibrarian.billdueber.com/marc-hash-the-saga-continues-now-with-even-less-structure/
+  # @return A marc-hash representation of the record, suitable for calling .to_json on or whatever
+  def to_marchash
+    h = {}
+    h['type'] = 'marc-hash'
+    h['version'] = [1,0]
+    h['leader'] = self.leader
+    
+    fields = []
+    
+    self.getVariableFields.each do |f|
+      if f.controlField?
+        fields << [f.tag, f.value]
+      else 
+        farray = [f.tag, f.indicator1 || ' ', f.indicator2 || ' ']
+        subs = []
+        f.each do |subfield|
+          subs << [subfield.code, subfield.value]
+        end
+        farray.push subs
+        fields << farray
+      end
+    end
+    h['fields'] = fields
+    return h
+  end
+
+  # 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
+      if @hashedtags[tag]
+        return @hashedtags[tag][0]
+      else 
+        return nil
+      end
+    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 Just get all fields for a single tag
+  #  ohThirtyFives = r.find_by_tag('035')
+  #
+  # @example Get a bunch of standard identifiers
+  #   standardIDs = r.find_by_tag(['022', '020', '010'])
+  #
+  # @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    $a (OCoLC)ocm00001728
+
+  def find_by_tag(tags, originalorder = false)
+    self.hashify unless @hashedtags and !originalorder
+    if !tags.is_a? Array
+      return @hashedtags[tags] || []
+    end
+    if originalorder
+      return self.find_all {|f| tags.include? f.tag}
+    else
+      # puts "Tags is #{tags}: got #{@hashedtags.values_at(*tags)}"
+      return @hashedtags.values_at(*tags).flatten.compact
+    end
+  end
+  
+  
+
+  # Return the record as valid MARC-XML
+  # @return String A MARC-XML representation of the record, including the XML header
+  def to_xml
+    return @xml if @xml
+    begin
+      @xml = java.io.StringWriter.new
+      res = javax.xml.transform.stream.StreamResult.new(@xml)
+      writer =  org.marc4j.MarcXmlWriter.new(res)
+      writer.write(self)
+      writer.writeEndDocument
+      return @xml.toString
+    rescue
+      "Woops! to_xml failed for record #{self['001'].data}: #{$!}"
+    end
+	end
+	
+	def to_marc
+    begin
+      s = Java::java.io.ByteArrayOutputStream.new
+      writer =  org.marc4j.MarcStreamWriter.new(s)
+      writer.write(self)
+      @marcbinary = s.to_string
+      puts @marcbinary
+      return @marcbinary
+    rescue
+      # "Woops! to_marc failed for record #{self['001'].data}: #{$!}"
+      "Whoops! Failed: #{$!}"
+    end
+  end
+    
+	
+end
+
+class ControlField
+  def value
+    return self.data
+  end
+  
+  def controlField?
+    return true
+  end 
+  
+  def self.control_tag? tag
+    return Java::org.marc4j.marc.impl.Verifier.isControlField tag
+  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
+
+  def == other
+    self.tag == other.tag && self.value == other.value
+  end
+
+end
+
+class DataField
+  include Enumerable
+  
+  alias_method :<<, :addSubfield
+  
+  
+  def controlField?
+    return false
+  end
+  
+  # Broken. Need to check subs as well
+  def == other
+    self.tag == other.tag and
+    self.indicator1 == other.indicator1 and
+    self.indicator2 == other.indicator2
+  end
+
+  # 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
+    # need to send a char value that the underlying java can deal with
+    sub = self.getSubfield(code[0].ord)
+    if (sub)
+      return sub.getData
+    else 
+      return nil
+    end
+  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. $c 1969
+  # rec['260'].sub_values('a') #=> ["New York,"]
+  # rec['260'].sub_values(['a', 'c']) #=> ["New York,", "1969"]
+  # rec['260'].sub_values(['c', 'a']) #=> ["New York,", "1969"]
+  # rec['260'].sub_values(['c', 'a'], true) #=> ["1969", "New York"]
+  
+  def sub_values(code, myorder = false)
+    
+    # Do a little razzle-dazzle for the common case when a single code is given
+    if not [Set, Array].include? code.class
+      c = code
+    elsif code.size == 1
+      c = code.first
+    end
+    if c
+      return self.find_all { |s| c == s.code}.map {|s| s.data}
+    end
+    
+    # unless [Set, Array].include? code.class
+    #   code = [code] 
+    #   # puts "Arrayified for code #{code} / #{code.class}"
+    # end
+    if myorder
+      subs = []
+      code.each do |c|
+        subs << self.find_all {|s| c == s.code}
+      end
+      return subs.flatten.map {|s| s.data}
+    else
+      return self.find_all{|s| code.include? s.code}.map {|s| s.data}
+    end
+  end
+
+  # Get first indicator as a one-character string
+  def indicator1
+    return self.getIndicator1.chr
+  end
+
+  # Get second indicator as a one-character string
+  def indicator2
+    return self.getIndicator2.chr
+  end
+
+  # Iterate over the subfields
+  def each
+    self.getSubfields.each do |s|
+      yield s
+    end
+  end
+
+  # Get the concatentated values of the subfields in order the appear in the field
+  # @param [String] joiner The string used to join the subfield values
+  def value joiner=' '
+    data = self.getSubfields.map {|s| s.data}
+    return data.join(joiner)
+  end  
+end
+
+class SubField
+  
+  def == other
+    return ((self.code == other.code) and (self.data == other.data))
+  end
+  
+  def value
+    return self.data
+  end
+  
+  def code
+    return self.getCode.chr
+  end
+
+  def to_s
+    return '$' + self.code + " " + self.data
+  end
+end

data/spec/batch.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.00647pam  2200241 a 4500001001300000003000600013005001700019008004100036010001700077020001500094040001800109042000800127050002600135082001500161100002600176245006700202260003800269263000900307300001100316650003700327650002500364700001600389fol05754809 IMchF20000601115601.0000203s2000    mau           001 0 eng    a   00022023   a1565926994  aDLCcDLCdDLC  apcc00aQA76.73.P22bD47 200000a005.742211 aDescartes, Alligator.10aProgramming the Perl DBI /cAlligator Descartes and Tim Bunce.  aCmabridge, MA :bO'Reilly,c2000.  a1111  ap. cm. 0aPerl (Computer program language) 0aDatabase management.1 aBunce, Tim.00605cam  22002054a 4500001001300000003000600013005001700019008004100036010001700077040001800094042000800112050002700120082001700147100002100164245005500185260004500240300002600285504005100311650003700362fol05843555 IMchF20000525142739.0000318s1999    cau      b    001 0 eng    a   00501349   aDLCcDLCdDLC  apcc00aQA76.73.P22bB763 199900a005.13/32211 aBrown, Martin C.10aPerl :bprogrammer's reference /cMartin C. Brown.  aBerkeley :bOsborne/McGraw-Hill,cc1999.  axix, 380 p. ;c22 cm.  aIncludes bibliographical references and index. 0aPerl (Computer program language)00579cam  22002054a 4500001001300000003000600013005001700019008004100036010001700077020001500094040001800109042000800127050002700135082001700162100002100179245005500200260004500255300003600300650003700336fol05843579 IMchF20000525142716.0000318s1999    caua          001 0 eng    a   00502116   a0072120002  aDLCcDLCdDLC  apcc00aQA76.73.P22bB762 199900a005.13/32211 aBrown, Martin C.10aPerl :bthe complete reference /cMartin C. Brown.  aBerkeley :bOsborne/McGraw-Hill,cc1999.  axxxv, 1179 p. :bill. ;c24 cm. 0aPerl (Computer program language)00801nam  22002778a 4500001001300000003000600013005001700019008004100036010001700077020001500094040001300109042000800122050002600130082001800156100002000174245008800194250003200282260004100314263000900355300001100364650003700375650003600412650002600448700002500474700002400499fol05848297 IMchF20000524125727.0000518s2000    mau           001 0 eng    a   00041664   a1565924193  aDLCcDLC  apcc00aQA76.73.P22bG84 200000a005.2/7622211 aGuelich, Scott.10aCGI programming with Perl /cScott Guelich, Shishir Gundavaram & Gunther Birznieks.  a2nd ed., expanded & updated  aCambridge, Mass. :bO'Reilly,c2000.  a0006  ap. cm. 0aPerl (Computer program language) 0aCGI (Computer network protocol) 0aInternet programming.1 aGundavaram, Shishir.1 aBirznieks, Gunther.00665nam  22002298a 4500001001300000003000600013005001700019008004100036010001700077020001500094040001300109042000800122050002700130082001700157111005200174245008600226250001200312260004100324263000900365300001100374650005000385fol05865950 IMchF20000615103017.0000612s2000    mau           100 0 eng    a   00055759   a0596000138  aDLCcDLC  apcc00aQA76.73.P22bP475 200000a005.13/32212 aPerl Conference 4.0d(2000 :cMonterey, Calif.)10aProceedings of the Perl Conference 4.0 :bJuly 17-20, 2000, Monterey, California.  a1st ed.  aCambridge, Mass. :bO'Reilly,c2000.  a0006  ap. cm. 0aPerl (Computer program language)vCongresses.00579nam  22002178a 4500001001300000003000600013005001700019008004100036010001700077020001500094040001300109042000800122050002600130082001700156100002800173245006200201260004100263263000900304300001100313650003700324fol05865956 IMchF20000615102948.0000612s2000    mau           000 0 eng    a   00055770   a1565926099  aDLCcDLC  apcc00aQA76.73.P22bB43 200000a005.13/32211 aBlank-Edelman, David N.10aPerl for system administration /cDavid N. Blank-Edelman.  aCambridge, Mass. :bO'Reilly,c2000.  a0006  ap. cm. 0aPerl (Computer program language)00661nam  22002538a 4500001001300000003000600013005001700019008004100036010001700077020001500094040001300109042000800122050002600130082001700156100001700173245006700190250001200257260004100269263000900310300001100319650003700330700002300367700001700390fol05865967 IMchF20000615102611.0000614s2000    mau           000 0 eng    a   00055799   a0596000278  aDLCcDLC  apcc00aQA76.73.P22bW35 200000a005.13/32211 aWall, Larry.10aProgramming Perl /cLarry Wall, Tom Christiansen & Jon Orwant.  a3rd ed.  aCambridge, Mass. :bO'Reilly,c2000.  a0007  ap. cm. 0aPerl (Computer program language)1 aChristiansen, Tom.1 aOrwant, Jon.00603cam  22002054a 4500001001300000003000600013005001700019008004100036010001700077020001500094040001800109042000800127050002600135082001700161100003200178245006000210260005700270300003300327650003700360fol05872355 IMchF20000706095105.0000315s1999    njua          001 0 eng    a   00500678   a013020868X  aDLCcDLCdDLC  apcc00aQA76.73.P22bL69 199900a005.13/32211 aLowe, Vincentq(Vincent D.)10aPerl programmer's interactive workbook /cVincent Lowe.  aUpper Saddle River, NJ :bPrentice Hall PTP,cc1999.  axx, 633 p. :bill. ;c23 cm. 0aPerl (Computer program language)00696nam  22002538a 4500001001300000003000600013005001700019008004100036010001700077020002800094040001300122042000800135050002600143082001700169100002600186245004400212260005100256263000900307300001100316500002000327650003700347650001700384650004100401fol05882032 IMchF20000707091904.0000630s2000    cau           001 0 eng    a   00058174   a0764547291 (alk. paper)  aDLCcDLC  apcc00aQA76.73.P22bF64 200000a005.13/32212 aFoster-Johnson, Eric.10aCross-platform Perl /cEric F. Johnson.  aFoster City, CA :bIDG Books Worldwide,c2000.  a0009  ap. cm.  aIncludes index. 0aPerl (Computer program language) 0aWeb servers. 0aCross-platform software development.