marc4j4r 0.1.3 → 0.1.4

data/test/bench.rb

@@ -0,0 +1,63 @@
+require 'rubygems'
+require '../lib/marc4j4r'
+require 'benchmark'
+
+filename = 'test.mrc'
+tags = %w(
+  245
+  035
+  100
+  010
+  001
+  050
+  300
+  600
+  856
+  260
+)
+
+Benchmark.bmbm do |x|
+  
+  x.report("0 tags; nohash") do
+    reader = MARC4J4R.reader(filename)
+    reader.each(false) do |r|
+      tags.each do |t|
+#        fields = r.find_by_tag(t, true)
+      end
+    end
+  end
+  x.report("0 tags; hash") do
+    reader = MARC4J4R.reader(filename)
+    reader.each do |r|
+      tags.each do |t|
+#        fields = r.find_by_tag(t, true)
+      end
+    end
+  end
+  
+    
+  (1..10).each do |i|
+    x.report("#{tags.size * i} tags; nohash") do
+      reader = MARC4J4R.reader(filename)
+      reader.each(false) do |r|
+        i.times do
+          tags.each do |t|
+            fields = r.find_by_tag(t, true)
+          end
+        end
+      end
+    end
+    x.report("#{tags.size * i} tags; hash") do
+      reader = MARC4J4R.reader(filename)
+      reader.each do |r|
+        i.times do 
+          tags.each do |t|
+            fields = r.find_by_tag(t)
+          end
+        end
+      end
+    end
+  end
+end
+    
+  

data/test/one.seq

@@ -0,0 +1,30 @@
+006988435 LDR   L 00000nam^a2200277Ia^4500
+006988435 001   L 006988435
+006988435 006   L m^^^^^^^^d^^^^^^^^
+006988435 007   L cr^cn|||||||||
+006988435 008   L 070103s2007^^^^nyu^^^^^sb^^^^001^0^eng^d
+006988435 010   L $$z2007060391
+006988435 020   L $$z9781403977564
+006988435 035   L $$a(CaPaEBR)ebrtrial10194097
+006988435 040   L $$aCaPaEBR$$cCaPaEBR$$dMiU
+006988435 05014 L $$aPL207$$b.C39 2007eb
+006988435 1001  L $$aÇayır, Kenan.
+006988435 24510 L $$aIslamic literature in contemporary Turkey$$h[electronic resource]$$bfrom epic to novel /$$cKenan Çayır.
+006988435 250   L $$a1st ed.
+006988435 260   L $$aNew York :$$bPalgrave Macmillan,$$c2007.
+006988435 504   L $$aIncludes bibliographical references (p. 189-198) and index.
+006988435 650 0 L $$aIslamic literature, Turkish$$xHistory and criticism.
+006988435 650 0 L $$aTurkish literature$$xHistory and criticism.
+006988435 7102  L $$aebrary, Inc.
+006988435 85640 L $$3Ann Arbor campus:$$zAccess to the ebrary online version restricted; authentication may be required:$$uhttp://site.ebrary.com/lib/umich/Doc?id=10194097
+006988435 85640 L $$3Flint campus:$$zAccess to the ebrary online version restricted; authentication may be required:$$uhttp://libproxy.umflint.edu:2048/login?url=http://site.ebrary.com/lib/umich/Doc?id=10194097
+006988435 8526  L $$aMiU$$bELEC$$hSee URL for access
+006988435 8526  L $$aMiFliC$$bFLINT$$cELEC$$hSee URL for access
+006988435 970   L $$aBK$$bBook
+006988435 970   L $$aCE$$bElectronic Resource
+006988435 971   L $$aMiU
+006988435 971   L $$aMiFliC
+006988435 972   L $$c20091022
+006988435 972   L $$c20091022
+006988435 973   L $$aAO$$bavail_online
+006988435 998   L $$cebrary-dld ERLoad20091022

data/test/readtest.rb

@@ -0,0 +1,120 @@
+require 'benchmark'
+require '../lib/marc4j4r.rb'
+curdir = File.dirname(__FILE__)
+
+Dir.glob("#{curdir}/../../jruby_marc_to_solr/jars/*.jar") do |x|
+  require x
+end
+
+include_class Java::org.solrmarc.marc.MarcAlephSequentialReader
+
+
+Benchmark.bmbm do |x|
+
+  x.report("Java AS") do
+    count = 0
+    3.times do
+      reader = MarcAlephSequentialReader.new(java.io.FileInputStream.new('test.seq'.to_java_string))
+      reader.each do |r|
+        count += 1
+      end
+    end
+    # puts "AS read #{count} records"
+  end
+  
+  x.report("strict") do
+    count = 0
+    3.times do
+      reader = MARC4J4R.reader('test.mrc', :strictmarc)
+      reader.each do |r|
+        count += 1
+      end
+    end
+    # puts "Strict binary read #{count} records"
+  end
+
+    x.report("xml") do
+      count = 0
+      3.times do
+        reader = MARC4J4R.reader('test.xml', :marcxml)
+        reader.each do |r|
+          count += 1
+        end
+      end
+      # puts "XML read #{count} records"
+    end
+
+
+  x.report("alephsequential") do
+    count = 0
+    3.times do
+      reader = MARC4J4R.reader('test.seq', :alephsequential)
+      reader.each do |r|
+        count += 1
+      end
+    end
+    # puts "AS read #{count} records"
+  end
+
+
+    
+
+end
+
+__END__
+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

data/test/t.rb

@@ -0,0 +1,6 @@
+File.open('test.seq') do |f|
+  f.each_line do |x|
+    next unless x[10..12] == 'LDR'
+    puts x if x.size == 43
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: marc4j4r
 version: !ruby/object:Gem::Version 
-  version: 0.1.3
+  version: 0.1.4
 platform: ruby
 authors: 
 - BillDueber
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-02-15 00:00:00 -05:00
+date: 2010-04-07 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -69,12 +69,15 @@ files:
 - doc/js/jquery.js
 - doc/method_list.html
 - doc/top-level-namespace.html
+- jars/MarcImporter.jar
 - jars/marc4j.jar
 - lib/marc4j4r.rb
+- test/batch.seq
+- test/bench.rb
 - test/helper.rb
 - test/one.dat
+- test/one.seq
 - test/one.xml
-- test/test_marc4j4r.rb
 has_rdoc: true
 homepage: http://github.com/billdueber/marc4j4r
 licenses: []
@@ -104,5 +107,7 @@ signing_key:
 specification_version: 3
 summary: Use marc4j java library in JRuby in a more ruby-ish way
 test_files: 
+- test/bench.rb
 - test/helper.rb
-- test/test_marc4j4r.rb
+- test/readtest.rb
+- test/t.rb

data/test/test_marc4j4r.rb

@@ -1,76 +0,0 @@
-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