rsinger-enhanced_marc 0.1 → 0.1.2

data/README

@@ -1,7 +1,7 @@
-Enhanced MARC is a set of classes, modules and methods that sit on top of ruby-marc (http://rubyforge.org/projects/marc) to help parse MARC records more easily and conveniently.
+Enhanced MARC is a set of classes, modules and methods that sit on top of ruby-marc (http://rubyforge.org/projects/marc) to help parse the contents of MARC records more easily and conveniently.
 Installation:
   gem sources -a http://gems.github.com
-  sudo gem install rsinger-enhanced-marc
+  sudo gem install rsinger-enhanced_marc
   
 Usage:
 

data/Rakefile

@@ -21,3 +21,20 @@ Rake::RDocTask.new('doc') do |rd|
   rd.main = 'MARC::Record'
   rd.rdoc_dir = 'doc'
 end
+
+begin
+  require 'jeweler'
+  Jeweler::Tasks.new do |s|
+    
+    s.add_dependency('marc')
+    s.name = 'enhanced_marc'
+    s.author = 'Ross Singer'
+    s.email = 'rossfsinger@gmail.com'
+    s.homepage = 'http://github.com/rsinger/enhanced-marc/tree'
+    s.summary = 'A DSL for MARC data'
+    s.description = 'A set of enhancements to ruby-marc to make parsing MARC data easier'    
+  end
+rescue LoadError
+  puts "Jeweler not available. Install it with: sudo gem install technicalpickles-jeweler -s http://gems.github.com"
+end
+

data/VERSION

@@ -0,0 +1 @@
+0.1.2

data/lib/enhanced_marc/book_record.rb

@@ -0,0 +1,37 @@
+module MARC
+
+  # A class that represents an individual MARC record. Every record
+  # is made up of a collection of MARC::Field objects. 
+
+  class BookRecord < Record
+
+    def initialize
+      super
+      @leader[6] = 'a' if @leader[6,1] == ' '
+      @leader[7] = 'm' if @leader[7,1] == ' '    
+      @record_type = 'BKS'  
+      @bibliographic_level = @leader.get_blvl
+      self.extend BookType
+      self.inspect_fixed_fields
+    end
+    
+    def is_valid_type?
+      return false unless @leader[6,1].match(/[at]{1}/)
+      return true
+    end
+    def is_valid_blvl?
+      return false unless @leader[7,1].match(/[acdm]{1}/)    
+      return true
+    end
+    def self.new_from_record(record)
+      rec = BookRecord.new
+      record.instance_variables.each { | var |
+        rec.instance_variable_set(var, record.instance_variable_get(var))
+      }
+      return Exception.new("Incorrect type declaration in leader") unless rec.is_valid_type?
+      return Exception.new("Incorrect bibliographic declaration in leader") unless rec.is_valid_blvl?
+      return rec
+    end    
+    
+  end
+end

data/lib/enhanced_marc/book_type.rb

@@ -0,0 +1,86 @@
+module BookType
+  include RecordType
+  public :is_govdoc?, :nature_of_contents, :is_conference?, :audience_level, :form, :has_index?
+  # Checks the leader and any 006 fields to 
+  # determine if the record is a manuscript.
+  # Returns a boolean.
+  def is_manuscript?
+    return true if @leader[6,1] == 't'
+    @fields.find_all {|f| ('006') === f.tag}.each { | fxd_fld |
+      return true if fxd_fld[0,1] == 't'
+    }       
+    return false
+  end
+  
+  def set_manuscript(value=false, field=nil)
+    if field
+      return Exception.new("Field is not an 006") unless field.tag == '006'
+      if value
+        field.value[0] = 't'
+      else
+        field.value[0] = 'a'
+      end
+    else  
+      if value
+        @leader[6] = 't'
+      else
+        @leader[6] = 'a'
+      end
+    end
+  end
+  
+  def literary_form(human_readable=false)
+    lit_map={'0'=>'Non-fiction', '1'=>'Fiction', 'c'=>'Comic',
+      'd'=>'Drama', 'e'=>'Essay', 'f'=>'Novel', 'h'=>'Humor/satire', 'i'=>'Letter', 'j'=>'Short story',
+      'm'=>'Mixed', 'p'=>'Poetry', 's'=>'Speech', 'u'=>'Unknown'}
+    human_readable = lit_map if human_readable
+    return self.field_parser({:match=>'BKS', :start=>33,:end=>1}, {:match=>/[at]{1}/, :start=>16,:end=>1}, human_readable)          
+  end    
+        
+  def is_biography?(human_readable=false)
+    biog_map = {'a'=>'Autobiography', 'b'=>'Individual biography',
+     'c'=>'Collective biography', 'd'=>'Contains biographical information'}
+    human_readable = lit_map if human_readable
+    return self.field_parser({:match=>'BKS', :start=>34,:end=>1}, {:match=>/[at]{1}/, :start=>17,:end=>1}, human_readable)               
+  end    
+  
+  def is_festschrift?
+    return true if self['008'].value[30,1] == "1" and @record_type == "BKS"
+    @fields.find_all {|f| ('006') === f.tag}.each { | fxd_fld |
+      next unless fxd_fld.value[0,1].match(/[at]{1}/) and fxd_fld.value[13,1] == "1"
+      return true
+    }        
+    return false
+  end    
+        
+  def is_illustrated?(human_readable=false)
+    ills_map = {'a'=>'Illustrations','b'=>'Maps','c'=>'Portraits','d'=>'Charts',
+      'e'=>'Plans', 'f'=>'Plates', 'g'=>'Music', 'h'=>'Facsimilies', 'i'=>'Coats of arms',
+      'j'=>'Genealogical tables', 'k'=>'Forms', 'j'=>'Genealogical tables', 'k'=>'Forms', 'l'=>'Samples',
+      'm'=>'Phonodisc', 'o'=>'Photographs', 'p'=>'Illuminations'}
+    contents = []
+    if self.record_type == 'BKS'
+      self['008'].value[18,4].split(//).each { | char | 
+        next if char == " "
+        if human_readable
+          contents << ills_map[char]
+        else
+          contents << char
+        end
+      }
+    end
+    @fields.find_all {|f| ('006') === f.tag}.each { | fxd_fld |
+      next unless fxd_fld.value[0,1].match(/[at]{1}/)     
+      fxd_fld.value[1,4].split(//).each { | char | 
+        next if char == " "
+        if human_readable
+          contents << ills_map[char]
+        else
+          contents << char
+        end
+      }
+    }       
+    return false if contents.empty?
+    return contents      
+  end        
+end

data/lib/enhanced_marc/computer_record.rb

@@ -0,0 +1,37 @@
+module MARC
+
+  # A class that represents an individual MARC record. Every record
+  # is made up of a collection of MARC::Field objects. 
+
+  class ComputerRecord < Record
+    
+    def initialize
+      super
+      @leader[6] = 'm' if @leader[6,1] == ' '
+      @leader[7] = 'm' if @leader[7,1] == ' '    
+      @record_type = 'COM'  
+      @bibliographic_level = @leader.get_blvl
+      self.extend ComputerType      
+      self.inspect_fixed_fields      
+    end
+    
+    def is_valid_type?
+      return false unless @leader[6,1].match('m')
+      return true
+    end
+    def is_valid_blvl?
+      return false unless @leader[7,1].match(/[abcdims]{1}/)
+      return true
+    end
+    def self.new_from_record(record)
+      rec = ComputerRecord.new
+      record.instance_variables.each { | var |
+        rec.instance_variable_set(var, record.instance_variable_get(var))
+      }
+      return Exception.new("Incorrect type declaration in leader") unless rec.is_valid_type?
+      return Exception.new("Incorrect bibliographic declaration in leader") unless rec.is_valid_blvl?
+      return rec
+    end    
+    
+  end
+end

data/lib/enhanced_marc/computer_type.rb

@@ -0,0 +1,14 @@
+module ComputerType
+  include RecordType
+  public :form, :audience_level, :is_govdoc?
+  def file_type(human_readable=false)
+    return false unless self.contains_type?("COM")
+    file_map = {'a'=>'Numeric data', 'b'=>'Computer program', 'c'=>'Representational',
+      'd'=>'Document', 'e'=>'Bibliographic data', 'f'=>'Font', 'g'=>'Game',
+      'h'=>'Sounds', 'i'=>'Interactive multimedia', 'j'=>'Online', 'm'=>'Combination',
+      'u'=>'Unknown', 'z'=>'Other'}
+    human_readable = file_map if human_readable
+    return self.field_parser({:match=>'COM', :start=>26,:end=>1}, {:match=>'m', :start=>9,:end=>1}, human_readable)        
+  end    
+
+end

data/lib/enhanced_marc/leader.rb

@@ -0,0 +1,132 @@
+module MARC
+  
+  # A class for accessing the MARC Leader
+  class Leader < String
+    attr_reader :leader, :record_type, :bibliographic_level, :encoding_level, :fixed_fields
+    def initialize(leader)
+      super  
+      # leader defaults:
+      # http://www.loc.gov/marc/bibliographic/ecbdldrd.html
+      self[10..11] = '22'
+      self[20..23] = '4500'           
+    end
+    
+    def parse_leader
+      self.get_type
+      self.get_blvl
+      self.get_elvl
+    end
+    
+    def get_type_code
+      return self[6,1]    
+    end
+    
+    def get_blvl_code
+      return self[7,1]
+    end
+    
+    def get_type
+      return @record_type = self.type_translator(self.get_type_code, self.get_blvl_code)
+    end
+    
+    def is_archival?
+      return true if self[8,1] == 'a'
+      return false
+    end
+
+    def get_blvl
+      blvls = {
+        'a'=>'Monographic component part',
+        'b'=>'Serial component part',
+        'c'=>'Collection',
+        'd'=>'Subunit',
+        'i'=>'Integrating resource',
+        'm'=>'Monograph/Item',
+        's'=>'Serial'
+      }
+      return @bibliographic_level = blvls[self.get_blvl_code]
+    end
+
+    def set_type(type)
+      if type.length == 1
+        translated_type = self.types(type)
+        raise ArgumentError, "Invalid Type" if translated_type.nil?
+      elsif type.length > 1
+        translated_type = type
+        type = self.types(type)
+        raise ArgumentError, "Invalid Type" if type.nil?
+      else
+        raise ArgumentError, "Invalid Type"
+      end      
+      self[6] = type 
+    end
+    
+    def type_translator(type_code, blvl_code)
+      valid_types = ['a','t','g','k','r','o','p','e','f','c','d','i','j','m']
+      unless valid_types.index(type_code)    
+        raise ArgumentError, "Invalid Type!" 
+        return
+      end
+      rec_types = {
+        'BKS' => { :type => /[at]{1}/,	:blvl => /[acdm]{1}/ },
+	      'SER' => { :type => /[a]{1}/,	:blvl => /[bs]{1}/ },
+        'VIS' => { :type => /[gkro]{1}/,	:blvl => /[abcdms]{1}/ },
+        'MIX' => { :type => /[p]{1}/,	:blvl => /[cd]{1}/ },
+        'MAP' => { :type => /[ef]{1}/,	:blvl => /[abcdms]{1}/ },
+        'SCO' => { :type => /[cd]{1}/,	:blvl => /[abcdms]{1}/ },
+        'REC' => { :type => /[ij]{1}/,	:blvl => /[abcdms]{1}/ },
+        'COM' => { :type => /[m]{1}/,	:blvl => /[abcdms]{1}/ }
+      } 
+      
+      rec_types.each_key { | type |
+        return type if type_code.match(rec_types[type][:type]) and blvl_code.match(rec_types[type][:blvl])
+      } 
+      raise ArgumentError, "Invalid BLvl!"
+      return nil
+    end
+    
+    def get_elvl_code
+      return self[17,1]
+    end
+    
+    def get_elvl
+      elvls = {
+        ' '=>'Full',
+        '1'=>'Full, not examined',
+        '2'=>'Less-than-full',
+        '3'=>'Abbreviated',
+        '4'=>'Core',
+        '5'=>'Partial',
+        '7'=>'Minimal',
+        '8'=>'Prepublication',
+        'I'=>'Full-level input by OCLC participants',
+        'K'=>'Less-than-full input by OCLC participants',
+        'L'=>'Full-level input added from a batch process',
+        'M'=>'Less-than-full added from a batch process',
+        'E'=>'System-identified MARC error in batchloaded record',
+        'J'=>'Deleted record'
+      }
+      return @encoding_level = elvls[self.get_elvl_code]
+    end
+    
+    def ELvl
+      self.get_elvl unless @encoding_level
+      return @encoding_level
+    end
+    
+    def get_desc_code
+      return self[18,1]
+    end
+    
+    def get_desc
+      codes = {' '=>'Non-ISBD', 'a'=>'AACR2', 'i'=>'ISBD', 'u'=>'Unknown'}
+      return @descriptive_cataloging_form = codes[self.get_desc_code]      
+    end
+    
+    def Desc
+      self.get_desc unless @descriptive_cataloging_form
+      return @descriptive_cataloging_form
+    end
+    
+  end
+end

data/lib/enhanced_marc/leftovers.rb

@@ -0,0 +1,122 @@
+    def get_fixed_field_position
+      ff_pos = {
+        'Ctry' => {
+          '008' => {
+            'BKS' => {:start => 15, :len => 3, :def => ' ' },
+            'SER' => {:start => 15, :len => 3, :def => ' ' },
+            'VIS' => {:start => 15, :len => 3, :def => ' ' },
+            'MIX' => {:start => 15, :len => 3, :def => ' ' },
+            'MAP' => {:start => 15, :len => 3, :def => ' ' },
+            'SCO' => {:start => 15, :len => 3, :def => ' ' },
+            'REC' => {:start => 15, :len => 3, :def => ' ' },
+            'COM' => {:start => 15, :len => 3, :def => ' ' }
+          }
+        },
+        'Lang' => {
+          '008' => {
+            'BKS' => {:start => 35, :len => 3, :def => ' ' },
+            'SER' => {:start => 35, :len => 3, :def => ' ' },
+            'VIS' => {:start => 35, :len => 3, :def => ' ' },
+            'MIX' => {:start => 35, :len => 3, :def => ' ' },
+            'MAP' => {:start => 35, :len => 3, :def => ' ' },
+            'SCO' => {:start => 35, :len => 3, :def => ' ' },
+            'REC' => {:start => 35, :len => 3, :def => ' ' },
+            'COM' => {:start => 35, :len => 3, :def => ' ' }
+          }
+        },
+        'MRec' => {
+          '008' => {
+            'BKS' => {:start => 38, :len => 3, :def => ' ' },
+            'SER' => {:start => 38, :len => 3, :def => ' ' },
+            'VIS' => {:start => 38, :len => 3, :def => ' ' },
+            'MIX' => {:start => 38, :len => 3, :def => ' ' },
+            'MAP' => {:start => 38, :len => 3, :def => ' ' },
+            'SCO' => {:start => 38, :len => 3, :def => ' ' },
+            'REC' => {:start => 38, :len => 3, :def => ' ' },
+            'COM' => {:start => 38, :len => 3, :def => ' ' }
+          }
+        },
+        'DtSt' => {
+          '008' => {
+            'BKS' => {:start => 6, :len => 1, :def => ' ' },
+            'SER' => {:start => 6, :len => 1, :def => 'c' },
+            'VIS' => {:start => 6, :len => 1, :def => ' ' },
+            'MIX' => {:start => 6, :len => 1, :def => ' ' },
+            'MAP' => {:start => 6, :len => 1, :def => ' ' },
+            'SCO' => {:start => 6, :len => 1, :def => ' ' },
+            'REC' => {:start => 6, :len => 1, :def => ' ' },
+            'COM' => {:start => 6, :len => 1, :def => ' ' },
+          }
+        },
+        'Type' => {
+          'ldr' => {
+            'BKS' => {:start => 6, :len => 1, :def => 'a' },
+            'SER' => {:start => 6, :len => 1, :def => 'a' },
+            'VIS' => {:start => 6, :len => 1, :def => 'g' },
+            'MIX' => {:start => 6, :len => 1, :def => 'p' },
+            'MAP' => {:start => 6, :len => 1, :def => 'e' },
+            'SCO' => {:start => 6, :len => 1, :def => 'c' },
+            'REC' => {:start => 6, :len => 1, :def => 'i' },
+            'COM' => {:start => 6, :len => 1, :def => 'm' },
+          }
+        },
+        'Ctrl' => {
+          'ldr' => {
+            'BKS' => {:start => 8, :len => 1, :def => ' ' },
+            'SER' => {:start => 8, :len => 1, :def => ' ' },
+            'VIS' => {:start => 8, :len => 1, :def => ' ' },
+            'MIX' => {:start => 8, :len => 1, :def => ' ' },
+            'MAP' => {:start => 8, :len => 1, :def => ' ' },
+            'SCO' => {:start => 8, :len => 1, :def => ' ' },
+            'REC' => {:start => 8, :len => 1, :def => ' ' },
+            'COM' => {:start => 8, :len => 1, :def => ' ' },
+          }
+        },
+        'BLvl' => {
+        	'ldr'=> {
+        		'BKS' => {:start => 7, :len => 1, :def => 'm' },
+        		'SER' => {:start => 7, :len => 1, :def => 's' },
+        		'VIS' => {:start => 7, :len => 1, :def => 'm' },
+        		'MIX' => {:start => 7, :len => 1, :def => 'c' },
+        		'MAP' => {:start => 7, :len => 1, :def => 'm' },
+        		'SCO' => {:start => 7, :len => 1, :def => 'm' },
+        		'REC' => {:start => 7, :len => 1, :def => 'm' },
+        		'COM' => {:start => 7, :len => 1, :def => 'm' },
+        	}
+        },
+        'Desc' => {
+        	'ldr' => {
+            'BKS' => {:start => 18, :len => 1, :def => ' ' },
+            'SER' => {:start => 18, :len => 1, :def => ' ' },
+            'VIS' => {:start => 18, :len => 1, :def => ' ' },
+            'MIX' => {:start => 18, :len => 1, :def => ' ' },
+            'MAP' => {:start => 18, :len => 1, :def => ' ' },
+            'SCO' => {:start => 18, :len => 1, :def => ' ' },
+            'REC' => {:start => 18, :len => 1, :def => ' ' },
+            'COM' => {:start => 18, :len => 1, :def => ' ' },
+        	}
+        },
+        'ELvl' => {
+        	'ldr' => {
+            'BKS' => {:start => 17, :len => 1, :def => ' ' },
+            'SER' => {:start => 17, :len => 1, :def => ' ' },
+            'VIS' => {:start => 17, :len => 1, :def => ' ' },
+            'MIX' => {:start => 17, :len => 1, :def => ' ' },
+            'MAP' => {:start => 17, :len => 1, :def => ' ' },
+            'SCO' => {:start => 17, :len => 1, :def => ' ' },
+            'REC' => {:start => 17, :len => 1, :def => ' ' },
+            'COM' => {:start => 17, :len => 1, :def => ' ' },
+        	}
+        },
+        'TMat' => {
+        	'008' => {
+        		'VIS' => {:start => 33, :len => 1, :def => ' ' },        		
+        	},
+        	'006' => {
+        		'VIS' => {:start => 16, :len => 1, :def => ' ' },
+        	}
+        },
+       }
+
+    
+    end