marc4j4r 0.1.3 → 0.1.4

data/doc/top-level-namespace.html

@@ -63,11 +63,11 @@
 <p class="children">
    
     
-      <strong class="modules">Modules:</strong> <a href="MARC4J4R.html" title="MARC4J4R (module)">MARC4J4R</a>
+      <strong class="modules">Modules:</strong> <a href="MARC.html" title="MARC (module)">MARC</a>, <a href="MARC4J4R.html" title="MARC4J4R (module)">MARC4J4R</a>
     
    
     
-      <strong class="classes">Classes:</strong> <a href="ControlFieldImpl.html" title="ControlFieldImpl (class)">ControlFieldImpl</a>, <a href="DataFieldImpl.html" title="DataFieldImpl (class)">DataFieldImpl</a>, <a href="RecordImpl.html" title="RecordImpl (class)">RecordImpl</a>, <a href="SubfieldImpl.html" title="SubfieldImpl (class)">SubfieldImpl</a>
+      <strong class="classes">Classes:</strong> <a href="ASReader.html" title="ASReader (class)">ASReader</a>, <a href="ControlFieldImpl.html" title="ControlFieldImpl (class)">ControlFieldImpl</a>, <a href="DataFieldImpl.html" title="DataFieldImpl (class)">DataFieldImpl</a>, <a href="RecordImpl.html" title="RecordImpl (class)">RecordImpl</a>, <a href="SubfieldImpl.html" title="SubfieldImpl (class)">SubfieldImpl</a>
     
   
 </p>
@@ -78,7 +78,7 @@
 </div>
     
     <div id="footer">
-  Generated on Fri Feb 12 16:38:07 2010 by 
+  Generated on Sat Apr  3 22:01:08 2010 by 
   <a href="http://yardoc.org" title="Yay! A Ruby Documentation Tool">yard</a>
   0.5.3 (ruby-1.8.7).
 </div>

data/lib/marc4j4r.rb

@@ -11,6 +11,28 @@ 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
 
   # Add some sugar to the MarcReader interface
@@ -31,10 +53,11 @@ module MARC4J4R
   # 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
+      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
@@ -54,6 +77,7 @@ module MARC4J4R
   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. 
@@ -75,43 +99,84 @@ module MARC4J4R
   
   def reader(input, type  = :strictmarc)
     case type
-    when :strictmarc
+    when :strictmarc then
       return Java::org.marc4j.MarcStreamReader.new(input)
-    when :permissivemarc
+    when :permissivemarc then
       return Java::org.marc4j.MarcPermissiveStreamReader.new(input)
-    when :marcxml
+    when :marcxml then
       return Java::org.marc4j.MarcXmlReader.new(input)
+    when :alephsequential then
+      return MARC4J4R::AlephSequentialReader.new(input)
     else
-      raise ArgumentError, "Reader type must be :strictmarc, :permissivemarc, or :marcxml"
+      raise ArgumentError, "Reader type #{type} illegal:  must be :strictmarc, :permissivemarc, :marcxml, or :alephsequential"
     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
+  
+  # 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
-  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
 
 
 
+
 include_class Java::org.marc4j.marc.impl::RecordImpl
 include_class Java::org.marc4j.marc.impl::ControlFieldImpl
 include_class Java::org.marc4j.marc.impl::DataFieldImpl
@@ -122,6 +187,38 @@ include_class Java::org.marc4j.marc.impl::SubfieldImpl
 
 class RecordImpl
   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
@@ -148,7 +245,8 @@ class RecordImpl
   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|
@@ -163,7 +261,11 @@ class RecordImpl
   
   def [] tag
     if defined? @hashedtags
-      return @hashedtags[tag][0]
+      if @hashedtags[tag]
+        return @hashedtags[tag][0]
+      else 
+        return nil
+      end
     else    
       return self.getVariableField(tag)
     end
@@ -209,7 +311,7 @@ class RecordImpl
   #  # 035    $a (OCoLC)ocm00001728
 
   def find_by_tag(tags, originalorder = false)
-    self.hashify unless @hashedtags
+    self.hashify unless @hashedtags and !originalorder
     if !tags.is_a? Array
       return @hashedtags[tags] || []
     end
@@ -221,18 +323,39 @@ class RecordImpl
     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
+    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 ControlFieldImpl
@@ -240,6 +363,14 @@ class ControlFieldImpl
     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
@@ -251,6 +382,20 @@ end
 
 class DataFieldImpl
   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
@@ -268,9 +413,13 @@ class DataFieldImpl
   # @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
+    # 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
 
 
@@ -345,6 +494,10 @@ end
 
 class SubfieldImpl
   
+  def == other
+    return self.code == other.code and self.data == other.data
+  end
+  
   def value
     return self.data
   end

data/test/batch.seq

@@ -0,0 +1,118 @@
+000004262 LDR   L ^^^^^nam^a22003011^^4500
+000004262 001   L 000004262
+000004262 005   L 19891107000000.0
+000004262 006   L m^^^^^^^^d^^^^^^^^
+000004262 007   L cr^bn^---auaua
+000004262 008   L 880715r19721939nyua^^^^^b^^^^00100^eng^^
+000004262 010   L $$a72001885/MN
+000004262 020   L $$a0306705133
+000004262 035   L $$a(RLIN)MIUG0297757-B
+000004262 035   L $$a(CaOTULAS)159822964
+000004262 035   L $$a(OCoLC)ocm00297757
+000004262 035   L $$asdr-inu1221428
+000004262 040   L $$aDLC$$cDLC$$dMiU
+000004262 043   L $$an------
+000004262 0500  L $$aML3557$$b.D3645 1972
+000004262 082   L $$a784.7/51
+000004262 1001  L $$aDensmore, Frances,$$d1867-1957.
+000004262 24510 L $$aNootka and Quileute music.
+000004262 260   L $$aNew York,$$bDa Capo Press,$$c1972.
+000004262 300   L $$axxvi, 358 p.$$billus.$$c23 cm.
+000004262 4900  L $$aDa Capo Press music reprint series
+000004262 500   L $$aReprint of the 1939 ed., which was issued as Bulletin 124 of Smithsonian Institution. Bureau of American Ethnology.
+000004262 504   L $$aBibliography: p. 349.
+000004262 538   L $$aMode of access: Internet.
+000004262 650 0 L $$aNootka Indians$$xMusic.
+000004262 650 0 L $$aQuileute Indians$$xMusic.
+000004262 8520  L $$aMiU$$bMUSIC$$hML3557 .D4255 1972
+000004262 852   L $$ainu$$bSDR$$cINU
+000004262 970   L $$aBK$$bBook
+000004262 970   L $$aCE$$bElectronic Resource
+000004262 971   L $$aMiU
+000004262 972   L $$c20040625
+000004262 973   L $$aHT$$bavail_ht
+000004262 973   L $$aAC$$bavail_circ
+000004262 974   L $$uinu.30000053901769$$ric$$d20100226
+000004262 998   L $$s9665
+000005951 LDR   L ^^^^^nam^a22002771^^4500
+000005951 001   L 000005951
+000005951 005   L 19970922000000.0
+000005951 006   L m^^^^^^^^d^^^^^^^^
+000005951 007   L cr^bn^---auaua
+000005951 008   L 880715s1968^^^^dcu^^^^^^b^^^|00010^eng^c
+000005951 010   L $$ahew68000006
+000005951 035   L $$a(RLIN)MIUG0425743-B
+000005951 035   L $$a(CaOTULAS)159824865
+000005951 035   L $$a(OCoLC)ocm00425743
+000005951 035   L $$asdr-ucsc.b16458175
+000005951 040   L $$a*U.S. Dept. of Health Education, and Welfare. Li$$cDLC$$dMiU
+000005951 0500  L $$aHD7123$$b.A39 no. 22
+000005951 082   L $$a301.5/5
+000005951 1001  L $$aKreps, Juanita Morris.
+000005951 24510 L $$aLifetime allocation of work and leisure,$$cby Juanita M. Kreps.
+000005951 260   L $$a[Washington,$$bFor sale by Supt. of Docs., U.S. Govt. Print. Off.,$$c1968]
+000005951 300   L $$aix, 44 p.$$c24 cm.
+000005951 4901  L $$aUnited States. Social Security Administration. Office of Research and Statistics. Research report$$vno. 22
+000005951 504   L $$aBibliographical footnotes.
+000005951 538   L $$aMode of access: Internet.
+000005951 650 0 L $$aAge and employment
+000005951 650 0 L $$aRetirement
+000005951 830 0 L $$aResearch report (United States. Social Security Administration. Office of Research and Statistics) ;$$vno. 22
+000005951 8520  L $$aMiU$$bTAUB$$hHD6279 .K925
+000005951 8520  L $$aMiU$$bTAUB$$hHD 7123 .A28 no.22
+000005951 8520  L $$aMiU$$bBUHR$$cGRAD$$hHD 7123 .A28 no.22
+000005951 852   L $$auc1$$bSDR$$cUCSC
+000005951 970   L $$aBK$$bBook
+000005951 970   L $$aCE$$bElectronic Resource
+000005951 971   L $$aMiU
+000005951 972   L $$c20040625
+000005951 972   L $$c20040625
+000005951 972   L $$c20040625
+000005951 973   L $$aHT$$bavail_ht
+000005951 973   L $$aAC$$bavail_circ
+000005951 974   L $$umdp.39015038814847$$ric
+000005951 974   L $$umdp.39015072105847$$ric$$d20091001
+000005951 974   L $$zno.22$$uuc1.32106000924198$$ric$$d20091202
+000005951 998   L $$s9665
+000009811 LDR   L ^^^^^nam^a22003251^^4500
+000009811 001   L 000009811
+000009811 005   L 19970922000000.0
+000009811 006   L m^^^^^^^^d^^^^^^^^
+000009811 007   L cr^bn^---auaua
+000009811 008   L 880715s1971^^^^dcua^^^^^bs^^^00000^eng^^
+000009811 010   L $$a72179703
+000009811 035   L $$a(RLIN)MIUG0695316-B
+000009811 035   L $$a(CaOTULAS)159829127
+000009811 035   L $$a(OCoLC)ocm00695316
+000009811 035   L $$asdr-ucsc.b13985097
+000009811 040   L $$aDLC$$cDLC$$dMiU$$dCStRLIN
+000009811 043   L $$an-us---
+000009811 0500  L $$aHD7123$$b.A39 no. 37$$aHD7102.U4
+000009811 082   L $$a368.4/00973 s$$a368.4/2/00973
+000009811 086   L $$aII0 aHE 3.49:37
+000009811 1101  L $$aUnited States.$$bSocial Security Administration.$$bOffice of Research and Statistics.
+000009811 24510 L $$aFinancing mental health care under medicare and medicaid.
+000009811 260   L $$a[Washington;$$bFor sale by the Supt. of Docs., U.S. Govt. Print. Off.,$$c1971]
+000009811 300   L $$avii, 52 p.$$billus.$$c27 cm.
+000009811 4901  L $$aIts Research report$$vno. 37
+000009811 500   L $$aSupt. of Docs. no.: HE 3.49:37
+000009811 504   L $$aIncludes bibliographical references.
+000009811 538   L $$aMode of access: Internet.
+000009811 650 0 L $$aMental health insurance$$zUnited States.
+000009811 650 0 L $$aMedicare
+000009811 650 0 L $$aMedicaid
+000009811 830 0 L $$aResearch report (United States. Social Security Administration. Office of Research and Statistics) ;$$vno. 37
+000009811 8520  L $$aMiU$$bBUHR$$cGRAD$$hHD 7123 .A28 no.37
+000009811 8520  L $$aMiU$$bTAUB$$hHD 7123 .A28 no.37
+000009811 852   L $$auc1$$bSDR$$cUCSC
+000009811 970   L $$aBK$$bBook
+000009811 970   L $$aCE$$bElectronic Resource
+000009811 970   L $$aXS$$bStatistics
+000009811 971   L $$aMiU
+000009811 972   L $$c20040625
+000009811 972   L $$c20040625
+000009811 973   L $$aHT$$bavail_ht
+000009811 973   L $$aAC$$bavail_circ
+000009811 974   L $$zno.37$$umdp.39015004063635$$ric
+000009811 974   L $$zno.37$$uuc1.32106000924313$$ric$$d20091202
+000009811 998   L $$s9665