marcspec 1.1.1 → 1.5.0

data/CHANGES

@@ -1,3 +1,5 @@
+1.5.0 
+  * Added DSL; making it the preferred way to operate. Totally deprecating the aray-of-hashes stuff. 
 1.1.1
   * Fixed bug with :noMapKeyDefault => :passthrough for MVMap
 1.1.0

data/README.rdoc

@@ -4,7 +4,10 @@ The MARCSpec contains classes designed to make it (relatively) easy to specify
 a data field (my use case specifically is solr) in terms of sets of MARC fields and subfields.
 
 
-== Docs and examples in marc2solr
+== Docs and examples
+
+Docs are hosted at the [[wiki|http://github.com/billdueber/marcspec/wiki/]]
+
 
 Documented samples are available as part of the marc2solr project
 at http://github.com/billdueber/marc2solr -- look in the simple_sample area.

data/VERSION

@@ -1 +1 @@
-1.1.1
+1.5.0

data/lib/marcspec/controlfieldspec.rb

@@ -17,7 +17,7 @@ module MARCSpec
   # substrings are specified.
   
   class ControlFieldSpec
-    attr_accessor :tag, :range
+    attr_accessor :tag, :range, :rangehistory
     
     def initialize (tag, range=nil)
       unless MARC4J4R::ControlField.control_tag? tag
@@ -25,11 +25,12 @@ module MARCSpec
       end
       @tag = tag
       self.range = range
+      @rangehistory = []
     end
     
     def == other
       return ((self.tag == other.tag) and
-              (self.range = other.range))
+              (self.range == other.range))
     end
     
     
@@ -41,6 +42,7 @@ module MARCSpec
     # @return [MARCSpec::ControlFieldSpec] self
     
     def range= range
+      @rangehistory << @range if @range
       if range.nil?
         @range = nil
         return self

data/lib/marcspec/customspec.rb

@@ -50,24 +50,19 @@ module MARCSpec
     
     def initialize(opts)
       @solrField  = opts[:solrField]
-      @module = opts[:module]
-      @functionSymbol = opts[:functionSymbol]
+      @module = opts[:module] || nil
+      @functionSymbol = opts[:functionSymbol] || nil
 
-      unless @solrField and @module and @functionSymbol
-        raise ArgumentError, "Custom solr spec must have a field name in :solrField, module in :module, and the function name as a symbol in :functionSymbol"
-      end
-      
-      
       @functionArgs = opts[:functionArgs] || []
       
       @first = opts[:firstOnly] || false      
-      @default = opts[:default] || nil
+      @defaultValue = opts[:default] || nil
       @map = opts[:map] || nil
       @noMapKeyDefault = opts[:noMapKeyDefault] || nil
       
       if @solrField.is_a? Array
         @arity = @solrField.size
-        if @first or @default or @map or @noMapKeyDefault 
+        if @first or @defaultValue or @map or @noMapKeyDefault 
           raise ArgumentError, "Custom spec with multiple solrFields can't have :first, :map, :default, or :noMapKeyDefault set"
         end
       else
@@ -101,9 +96,9 @@ module MARCSpec
       PP.singleline_pp(@solrField, s)
       s.print(",\n ")
       s.print ":firstOnly => true,\n " if @first
-      if @default
+      if @defaultValue
         s.print(":default => ")
-        PP.singleline_pp(@default, s)
+        PP.singleline_pp(@defaultValue, s)
         s.print(",\n ")
       end
       if @map

data/lib/marcspec/dsl.rb

@@ -0,0 +1,139 @@
+module MARCSpec
+  # Here's where we put a simple DSL hook for SpecSet
+  
+  def self.build (&blk)
+    ss = SpecSet.new
+    ss.instance_eval(&blk)
+    return ss
+  end
+  
+  # Re-open SpecSet to add the necessary methods
+  
+  class SpecSet
+    
+    # create a normal field
+    def field(name, &blk)
+      $LOG.debug "Creating regular field #{name}"
+      sfs = SolrFieldSpec.new(:solrField=>name)
+      sfs.instance_eval(&blk)
+      self << sfs
+      return sfs
+    end
+    
+    # Create a constant field
+    def constant(name, &blk)
+      $LOG.debug "Creating constant field #{name}"
+      
+      constant = ConstantSolrSpec.new(:solrField=>name)
+      constant.instance_eval(&blk)
+      self << constant
+      return constant
+    end
+    
+    def custom(name, &blk)
+      $LOG.debug "Creating custom field #{name}"
+      custom = CustomSolrSpec.new(:solrField=>name)
+      custom.instance_eval(&blk)
+      
+      ##### Check to make sure it's all ok in here#####
+      self << custom
+      return custom
+    end
+    
+  end
+  
+  
+  class SolrFieldSpec
+    def spec(tag, &blk)
+      if tag.to_i == tag
+        tag = '%03d' % tag
+      end
+      
+      marcfieldspec = nil
+      if MARC4J4R::ControlField.control_tag? tag
+        marcfieldspec = MARCSpec::ControlFieldSpec.new(tag)
+      else
+        marcfieldspec = MARCSpec::VariableFieldSpec.new(tag)
+      end
+      
+      marcfieldspec.instance_eval(&blk) if block_given?
+      
+      # If we had multiple sub calls, get them from the codehistory
+      if marcfieldspec.is_a? MARCSpec::VariableFieldSpec
+        marcfieldspec.codehistory.uniq.compact.each do |c|
+          newmfs = marcfieldspec.clone
+          newmfs.codes = c
+          self << newmfs
+        end
+      end
+      
+      if marcfieldspec.is_a? MARCSpec::ControlFieldSpec
+        marcfieldspec.rangehistory.uniq.compact.each do |r|
+          newcfs = marcfieldspec.clone
+          newcfs.range = r
+          self << newcfs
+        end
+      end
+      
+      self << marcfieldspec
+      return marcfieldspec
+    end
+    
+    def firstOnly val=true
+      @first = val
+    end
+    
+    def default val
+      @defaultValue = val
+    end
+    
+    def mapname str
+      @_mapname = str
+    end
+    
+    def mapMissDefault str
+      @noMapKeyDefault = str
+    end
+    
+  end
+  
+  class ControlFieldSpec
+    def char c
+      self.range = c
+      return self
+    end
+    
+    alias_method :chars, :char
+  end
+  
+  class VariableFieldSpec
+    def sub c
+      self.codes = c
+      return self
+    end
+    
+    alias_method :subs, :sub
+  end
+    
+  
+  class CustomSolrSpec
+    def function(name, &blk)
+      self.functionSymbol = name.to_sym
+      self.instance_eval(&blk)
+    end
+    
+    def mod(constant)
+      self.module = constant
+    end
+    
+    def args(*arg_or_args)
+      self.functionArgs = arg_or_args
+    end
+  end
+  
+  class ConstantSolrSpec
+    def value(val)
+      self.constantValue = val
+    end
+  end
+end

data/lib/marcspec/map.rb

@@ -45,6 +45,13 @@ module MARCSpec
         raise e
       end
       
+      # Derive a name if there isn't one
+      unless rawmap[:mapname]
+        name = File.basename(filename)
+        name.gsub! /\..*$/, '' # remove the extension
+        rawmap[:mapname] = name
+      end
+      
       case rawmap[:maptype]
       when :kv
         return KVMap.new(rawmap[:mapname], rawmap[:map])

data/lib/marcspec/multivaluemap.rb

@@ -19,24 +19,56 @@ module MARCSpec
   # Again, note that if several keys are === to the passed argument, all the values will be returned. 
   
   class MultiValueMap   < Map
+        
+    # Override initialize and map= so we can do some optimization
+    
+    def initialize *args
+      super(*args)
+      self.optimize
+    end
+    
+    def map= map
+      @map = map
+      self.optimize
+    end
     
-    attr_accessor :mapname,:map
+    
+    def optimize
+      @super_regexp = Regexp.union @map.map{|pv| pv[0]}
+      inverted = {}
+      @map.each do |pv|
+        inverted[pv[1]] ||= []
+        inverted[pv[1]] << pv[0]
+      end
+      inverted.each_pair do |vals, patterns|
+        next unless patterns.size > 1
+        newpat = Regexp.union patterns
+        patterns.each do |p|
+          @map.delete_if{|pv| p == pv[0]}
+        end
+        @map << [newpat, vals]
+      end
+    end
     
     # Given a passed_in_key (and optional default) return the set of values that match, as described
     # above.
     def [] key, default=nil
       rv = []
-      @map.each do |pv|
-        if pv[1].is_a? Proc
-          match = pv[0].match key
-          rv << pv[1].call(match) if match
-        else
-          rv << pv[1] if pv[0] === key
+      
+      if @super_regexp.match key # do *any* of them match?
+        @map.each do |pv|
+          if pv[1].is_a? Proc
+            match = pv[0].match key
+            rv << pv[1].call(match) if match
+          else
+            rv << pv[1] if pv[0] === key
+          end
         end
+        rv.flatten!
+        rv.compact!
+        rv.uniq!
       end
-      rv.flatten!
-      rv.compact!
-      rv.uniq!
+
       if rv.size > 0
         return rv
       else

data/lib/marcspec/solrfieldspec.rb

@@ -3,12 +3,13 @@ require 'marc4j4r/controlfield'
 
 module MARCSpec
   class SolrFieldSpec
-    attr_accessor :solrField, :first, :map, :noMapKeyDefault, :marcfieldspecs, :default, :arity
+    attr_accessor :solrField, :first, :map, :noMapKeyDefault, :marcfieldspecs, :defaultValue,  :_mapname
+    attr_reader :arity
 
     def initialize(opts)
       @solrField  = opts[:solrField]
       @first = opts[:firstOnly] || false      
-      @default = opts[:default] || nil
+      @defaultValue = opts[:default] || nil
       @map = opts[:map] || nil
       @noMapKeyDefault = opts[:noMapKeyDefault] || nil
       @arity = 1
@@ -42,10 +43,10 @@ module MARCSpec
       # If we got nothing, just return either nothing or the defualt,
       # if there is one. Don't screw around with mapping.
       if vals.size == 0
-        if @default.nil? # unless there's a default value, just return nothing
+        if @defaultValue.nil? # unless there's a default value, just return nothing
           return []
         else
-          return [@default]
+          return [@defaultValue]
         end
       end
       
@@ -68,6 +69,7 @@ module MARCSpec
       return ((other.solrField == self.solrField) and
              (other.first == self.first) and
              (other.map == self.map) and
+             (other.defaultValue == self.defaultValue) and
              (other.noMapKeyDefault == self.noMapKeyDefault) and
              (other.marcfieldspecs == self.marcfieldspecs))
     end
@@ -98,9 +100,9 @@ module MARCSpec
       PP.singleline_pp(@solrField, s)
       s.print(",\n ")
       s.print ":firstOnly => true,\n " if @first
-      if @default
+      if @defaultValue
         s.print(":default => ")
-        PP.singleline_pp(@default, s)
+        PP.singleline_pp(@defaultValue, s)
         s.print(",\n ")
       end
       if @map

data/lib/marcspec/specset.rb

@@ -24,6 +24,8 @@ module MARCSpec
     end
     
   end
+  
+  
 
   class SpecSet
     attr_accessor :tmaps, :solrfieldspecs, :benchmarks
@@ -54,6 +56,31 @@ module MARCSpec
     end
     
     
+    def buildSpecsFromDSLFile file
+      f = File.open(file)
+      $LOG.fatal("Can't open file #{file}") unless f
+      self.instance_eval(f.read)
+      self.check_and_fill_maps
+    end
+    
+    def check_and_fill_maps
+      @solrfieldspecs.each do |sfs|
+        if sfs._mapname
+          map = self.map(sfs._mapname)
+          if map
+            $LOG.debug "  Found map #{map.mapname} for solr field #{sfs.solrField}"
+            sfs.map = map
+          else
+            $LOG.error "  Cannot find map #{sfs._mapname} for solr field #{sfs.solrField}"
+            STDERR.puts "FATAL  Cannot find map #{sfs._mapname} for solr field #{sfs.solrField}"
+            Process.exit
+          end
+        end
+      end
+    end
+      
+    
+    
     def buildSpecsFromList speclist
       speclist.each do |spechash|
         if spechash[:module]
@@ -67,6 +94,7 @@ module MARCSpec
           map = self.map(spechash[:mapname])
           unless map
             $LOG.error "  Cannot find map #{spechash[:mapname]} for field #{spechash[:solrField]}"
+            Process.exit
           else
             $LOG.debug "  Found map #{spechash[:mapname]} for field #{spechash[:solrField]}"
             solrspec.map = map
@@ -80,7 +108,7 @@ module MARCSpec
     
     def add_spec solrfieldspec
       self.solrfieldspecs << solrfieldspec
-      @benchmarks[solrfieldspec.solrField] = Benchmark::Tms.new(0,0,0,0, 0, solrfieldspec.solrField)      
+      @benchmarks[solrfieldspec.solrField.to_s] = Benchmark::Tms.new(0,0,0,0, 0, solrfieldspec.solrField)      
     end
     
     alias_method :<<, :add_spec
@@ -106,7 +134,7 @@ module MARCSpec
 
     def fill_hashlike_from_marc_benchmark r, hashlike
       @solrfieldspecs.each do |sfs|
-        @benchmarks[sfs.solrField] += Benchmark.measure do
+        @benchmarks[sfs.solrField.to_s] += Benchmark.measure do
           if sfs.arity == 1
             hashlike.add(sfs.solrField,sfs.marc_values(r, hashlike))
           else 

data/lib/marcspec/variablefieldspec.rb

@@ -15,12 +15,13 @@ module MARCSpec
   
   class VariableFieldSpec
 
-    attr_accessor :tag, :codes, :joiner, :ind1, :ind2
+    attr_accessor :tag, :codes, :joiner, :ind1, :ind2, :codehistory
 
     def initialize tag, codes=nil, joiner=' '
       @tag = tag
       @joiner = joiner || ' '
       self.codes = codes
+      @codehistory = []
     end
 
     def == other
@@ -30,6 +31,7 @@ module MARCSpec
     end
 
     def codes= c
+      @codehistory << @codes if @codes
       if c.nil?
         @codes = nil
         return nil

data/lib/marcspec.rb

@@ -11,7 +11,7 @@ require "marcspec/kvmap"
 require "marcspec/multivaluemap"
 require "marcspec/specset"
 require "marcspec/marcfieldspec"
-
+require "marcspec/dsl"
 
 # Build up a little module to include in MARC4J4R::Record that
 # gives us a way to cache computed values within the record itself

data/spec/data/simplemap.rb

@@ -0,0 +1,14 @@
+{
+ :maptype=>:kv,
+ :map => {"mdp"=>"University of Michigan",
+ "wu"=>"University of Wisconsin",
+ "gwla"=>"University of Michigan",
+ "miua"=>"University of Michigan",
+ "miun"=>"University of Michigan",
+ "inu"=>"Indiana University",
+ "uc1"=>"University of California",
+ "uc2"=>"University of California",
+ "pst"=>"Penn State University",
+ "umn"=>"University of Minnesota"}
+
+}

data/spec/dsl_spec.rb

@@ -0,0 +1,370 @@
+require 'spec_helper'
+
+# The contents of @one
+#
+#
+# LEADER 00734njm a2200217uu 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
+
+
+# Create a helper for the custom functions
+
+module SPECHelper
+  def self.test doc, r
+    return 'Hello'
+  end
+  
+  def self.single doc, r, arg
+    return "Hello " + arg.to_s
+  end
+  
+  def self.double doc, r, one, two
+    return ["Hello", one, two].join(' ')
+  end
+  
+  def self.any doc, r, *args
+    return 'Hello ' + args.map{|s| s.to_s}.join(' ')
+  end
+
+  def self.multihead doc, r
+    return ['one', 'two']
+  end
+
+end
+
+describe "DSL" do
+  before do 
+    @one = MARC4J4R::Reader.new("#{DIR}/data/one.dat").first
+  end
+  
+  describe "constant DSL" do
+  
+    it "can add a constant solrfieldspec" do
+      ss = MARCSpec.build do
+        constant('id') do
+          value "Bill"
+        end
+      end
+      ss.solrfieldspecs.size.should.equal 1
+      ss.hash_from_marc(@one)['id'].should.equal ['Bill']
+    end
+  end
+
+  describe "custom DSL" do
+  
+    # before do 
+    #   @one = MARC4J4R::Reader.new("#{DIR}/data/one.dat").first
+    # end
+  
+    it "builds a bare-bones custom" do
+      ss = MARCSpec.build do
+        custom('hello') do
+          function(:test) {
+            mod SPECHelper
+          }
+        end
+      end
+    
+      ss.hash_from_marc(@one)['hello'].should.equal ['Hello']
+    end
+  
+    it "builds a custom that can take a single argument" do
+      ss = MARCSpec.build do
+        custom('hello') do
+          function(:single) {
+            mod SPECHelper
+            args 'Bill'
+          }
+        end
+      end
+
+      ss.hash_from_marc(@one)['hello'].should.equal ['Hello Bill']
+    end
+  
+    it "builds a custom that can take two arguments" do
+      ss = MARCSpec.build do
+        custom('hello') do
+          function(:double) {
+            mod SPECHelper
+            args 'Bill', 'Dueber'
+          }
+        end
+      end
+
+      ss.hash_from_marc(@one)['hello'].should.equal ['Hello Bill Dueber']
+    end
+ 
+    it "builds a custom that can take two arguments" do
+      ss = MARCSpec.build do
+        custom('hello') do
+          function(:any) {
+            mod SPECHelper
+            args 1,2,3,4,5
+          }
+        end
+      end
+
+      ss.hash_from_marc(@one)['hello'].should.equal ['Hello 1 2 3 4 5']
+    end
+    
+    it "works with multihead" do
+      ss = MARCSpec.build do
+        custom(['a', 'b']) do
+          function(:multihead) {
+            mod SPECHelper
+          }
+        end
+      end
+
+      ss.hash_from_marc(@one)['a'].should.equal ['one']
+      ss.hash_from_marc(@one)['b'].should.equal ['two']
+    end
+  end
+
+
+  describe "control fields DSL" do
+    # before do 
+    #   @one = MARC4J4R::Reader.new("#{DIR}/data/one.dat").first
+    # end
+  
+    it "should get a standard with a single whole control spec" do
+      ss = MARCSpec.build do
+        field("id") do
+          spec('001') 
+        end
+      end
+      ss.hash_from_marc(@one)['id'].should.equal ['afc99990058366']
+    end
+
+    it "should allow integer tags" do
+      ss = MARCSpec.build do
+        field("id") do
+          spec(001) 
+        end
+      end
+      ss.hash_from_marc(@one)['id'].should.equal ['afc99990058366']
+    end
+
+    it "can get a single char" do
+      ss = MARCSpec.build do
+        field("tst") do
+          spec(001) {
+            char 2
+          }
+        end
+      end
+      ss.hash_from_marc(@one)['tst'].should.equal ['c']
+    end
+
+    it "can get a range" do
+      ss = MARCSpec.build do
+        field("tst") do
+          spec(001) {
+            chars 2..6
+          }
+        end
+      end
+      ss.hash_from_marc(@one)['tst'].should.equal ['c9999']
+    end
+    
+    it "allows multiple char/chars per control field" do
+      ss = MARCSpec.build do
+        field("tst") do
+          spec(001) {
+            char  2
+            chars 2..6
+          }
+        end
+      end
+      ss.hash_from_marc(@one)['tst'].should.equal ['c', 'c9999']
+    end
+      
+
+  end
+
+
+  describe "variable fields DSL" do
+    # before do 
+    #   @one = MARC4J4R::Reader.new("#{DIR}/data/one.dat").first
+    # end
+  
+    it "can get a whole variable field" do
+      ss = MARCSpec.build do
+        field("tst") do
+          spec(260) 
+        end
+      end
+      ss.hash_from_marc(@one)['tst'].should.equal ['Medina, Texas, 1939.']
+    end
+  
+    it "can get a single subfield" do
+      ss = MARCSpec.build do
+        field("tst") do
+          spec(260) {
+            sub 'a'
+          }
+        end
+      end
+      ss.hash_from_marc(@one)['tst'].should.equal ['Medina, Texas,']
+    end
+
+
+    it "can get multiple subfields" do
+      ss = MARCSpec.build do
+        field("tst") do
+          spec(245) {
+            sub 'ac'
+          }
+        end
+      end
+      ss.hash_from_marc(@one)['tst'].should.equal ['The Texas ranger Sung by Beale D. Taylor.']
+    end
+
+    it "can get multiple subfields as array" do
+      ss = MARCSpec.build do
+        field("tst") do
+          spec(245) {
+            subs ['a', 'c']
+          }
+        end
+      end
+      ss.hash_from_marc(@one)['tst'].should.equal ['The Texas ranger Sung by Beale D. Taylor.']
+    end 
+  
+    it "can get multiple different subfields from the same field" do
+      ss = MARCSpec.build do
+        field("tst") do
+          spec(245) {
+            sub 'a'
+            sub 'c'
+          }
+        end
+      end
+      ss.hash_from_marc(@one)['tst'].should.equal ['The Texas ranger', 'Sung by Beale D. Taylor.']
+    end  
+  
+    it "can handle multiple specs" do
+      ss = MARCSpec.build do
+        field('tst') do
+          spec(245) {
+            sub 'a'
+          }
+          spec(245) {
+            sub 'c'
+          }
+        end
+      end
+      ss.hash_from_marc(@one)['tst'].should.equal ['The Texas ranger', 'Sung by Beale D. Taylor.']
+    end
+  end
+
+  describe "SolrFieldSpec modifiers DSL" do
+    it "works with firstOnly" do
+      ss = MARCSpec.build do
+        field('tst') do
+          firstOnly
+          
+          spec(700) {
+            sub 'a'
+          }
+        end
+      end
+      
+      ss.hash_from_marc(@one)['tst'].should.equal ['Lomax, John Avery, 1867-1948']
+    end
+    
+    it "works with default" do
+      ss = MARCSpec.build do
+        field('tst') do
+          default 'Default value'
+
+          spec(777) {
+            sub 'a'
+          }
+        end
+      end
+
+      ss.hash_from_marc(@one)['tst'].should.equal ['Default value']
+    end
+  end
+  
+  describe "use as config file" do
+    it "works in practice" do
+      string = %q|
+        field('tst') do
+          default 'Default'
+          spec(999)
+        end
+        
+        field('id') do
+          spec(001) {
+            chars 2..6
+          }
+        end
+      |
+
+      ss = MARCSpec::SpecSet.new
+      ss.instance_eval(string)
+      ss.hash_from_marc(@one)['tst'].should.equal ['Default']
+      ss.hash_from_marc(@one)['id'].should.equal ['c9999']      
+    end
+    
+    it "works in compact form" do
+      string = %q|
+        field('tst') do
+          default 'Default'
+          spec(999)
+        end
+        
+        field('id') do
+          spec(001) {chars 2..6}
+        end
+      |
+
+      ss = MARCSpec::SpecSet.new
+      ss.instance_eval(string)
+      ss.hash_from_marc(@one)['tst'].should.equal ['Default']
+      ss.hash_from_marc(@one)['id'].should.equal ['c9999']      
+    end
+    
+    it "bails on a missing map" do
+      string = %q|
+        field('tst') do
+          default 'Default'
+          mapname 'nosuchmap'
+          spec(999)
+        end
+        
+        field('id') do
+          spec(001) {chars 2..6}
+        end
+      |
+      f = Tempfile.new('ss')
+      f.puts string
+      path = f.path
+      f.close
+      ss = MARCSpec::SpecSet.new
+      lambda{ss.buildSpecsFromDSLFile(path)}.should.raise SystemExit
+      f.unlink
+      
+    end
+    
+  end
+
+
+end

data/spec/maps_spec.rb

@@ -1,21 +1,12 @@
 require 'spec_helper'
 
-describe "Maps" do
+describe "KV Maps" do
   before do
     @kvmap = MARCSpec::KVMap.new('kvmap', {'one' => '1', 'two' => ['2', 'zwei']})
-    @mvmap = MARCSpec::MultiValueMap.new('mvmap', [
-                                                  [/bi/, 'Bill'], 
-                                                  [/mo/i, 'Molly'], 
-                                                  [/ll/, 'Bill'], 
-                                                  [/lly/i, ['One', 'Two']], 
-                                                  [/^.*?\s+(.*)$/, Proc.new{|m| m[1]}]
-                                                  ]
-                                          )
   end
   
   it "knows its name" do
     @kvmap.mapname.should.equal 'kvmap'
-    @mvmap.mapname.should.equal 'mvmap'
   end
   
   it "gets simple value from a kv map" do
@@ -30,21 +21,90 @@ describe "Maps" do
     @kvmap['ddd'].should.equal nil
   end
 
-  it "gets nothing on nonmatches for mvmap" do
-    @mvmap['ddd'].should.equal nil
+  it "gets default if set for nonmatches with KVMap" do
+    @kvmap['ddd', 'default'].should.equal 'default'
+  end
+
+  it "gets key if default is :passthrough for nonmatches with KVMap" do
+    @kvmap['ddd', :passthrough].should.equal 'ddd'
   end
+
   
-  it "gets default if set for nonmatches with KVMap" do
+  it "correctly uses default value" do
     @kvmap['ddd', 'default'].should.equal 'default'
+    @kvmap['one', 'default'].should.equal '1'
+  end
+  
+  it "should round-trip a kvmap" do
+    s = @kvmap.asPPString
+    newkvmap = MARCSpec::KVMap.fromPPString s
+    newkvmap.should.equal @kvmap
+  end
+  
+  
+  it "should read a kv solrmarc file" do
+    map = MARCSpec::KVMap.from_solrmarc_file "#{DIR}/data/umich/translation_maps/country_map.properties"
+    map.mapname.should.equal 'country_map'
+    map["nl"].should.equal "New Caledonia"
+  end
+  
+  it "should correctly deal with solrmarc files with escaped chars (via \\)" do
+    map = MARCSpec::KVMap.from_solrmarc_file "#{DIR}/data/umich/translation_maps/location_map.properties"
+    map['AAEL'].should.equal 'AAEL'
+    map['AAEL MICE'].should.equal 'AAEL MICE'
+    map['BUHR AAEL'].should.equal 'BUHR'
+  end
+  
+  
+  it "can dump/load a kv map via generic map interface" do
+    map = MARCSpec::KVMap.from_solrmarc_file "#{DIR}/data/umich/translation_maps/country_map.properties"
+    f = Tempfile.new('kvmap')
+    f.puts map.asPPString
+    path = f.path
+    f.close
+    map2 = MARCSpec::Map.fromFile(path)
+    f.unlink
+    map.class.should.equal MARCSpec::KVMap    
+    map.should.equal map2
+  end
+  
+  it "can name a map based on the filename when using fromFile(path)" do
+    map = MARCSpec::Map.fromFile("#{DIR}/data/simplemap.rb")
+    map.mapname.should.equal 'simplemap'
   end
 
+end
+
+describe "MVMaps" do
+  before do
+    @mvmap = MARCSpec::MultiValueMap.new('mvmap', [
+                                                  [/bi/, 'Bill'], 
+                                                  [/mo/i, 'Molly'], 
+                                                  [/ll/, 'Bill'], 
+                                                  [/lly/i, ['One', 'Two']], 
+                                                  [/^.*?\s+(.*)$/, Proc.new{|m| m[1]}]
+                                                  ]
+                                          )
+    @mvmapCollapse = MARCSpec::MultiValueMap.new('mvmap', [
+        [/^bill/i, 'William'],
+        [/^will.*/i, 'William'],
+        [/dueber/i, 'Dueber'],
+        [/duebs/i, 'Dueber']
+      ])                                          
+  end
+  
+  it "knows its name" do
+    @mvmap.mapname.should.equal 'mvmap'
+  end
+  
+  it "gets nothing on nonmatches for mvmap" do
+    @mvmap['ddd'].should.equal nil
+  end
+  
   it "gets default if set for nonmatches with MVMap" do
     @mvmap['ddd', 'default'].should.equal 'default'
   end
   
-  it "gets key if default is :passthrough for nonmatches with KVMap" do
-    @kvmap['ddd', :passthrough].should.equal 'ddd'
-  end
 
   it "gets key if default is :passthrough for nonmatches with KVMap" do
     @mvmap['ddd', :passthrough].should.equal 'ddd'
@@ -57,18 +117,10 @@ describe "Maps" do
     @mvmap['mobi'].sort.should.equal ['Bill', 'Molly'].sort
     @mvmap['Molly'].sort.should.equal ['Molly', 'Bill', 'One', 'Two'].sort
   end
-  
+
   it "correctly uses default value" do
     @mvmap['bi', 'default'].should.equal ['Bill']
     @mvmap['ddd', 'default'].should.equal 'default'
-    @kvmap['ddd', 'default'].should.equal 'default'
-    @kvmap['one', 'default'].should.equal '1'
-  end
-  
-  it "should round-trip a kvmap" do
-    s = @kvmap.asPPString
-    newkvmap = MARCSpec::KVMap.fromPPString s
-    newkvmap.should.equal @kvmap
   end
   
   it "should round trip a multivaluemap without a Proc" do 
@@ -98,38 +150,13 @@ describe "Maps" do
      @mvmap['one'].should.equal ['one']
      @mvmap['two'].should.equal ['two']
   end
-  
-  it "should read a kv solrmarc file" do
-    map = MARCSpec::KVMap.from_solrmarc_file "#{DIR}/data/umich/translation_maps/country_map.properties"
-    map.mapname.should.equal 'country_map'
-    map["nl"].should.equal "New Caledonia"
-  end
-  
-  it "should correctly deal with solrmarc files with escaped chars (via \\)" do
-    map = MARCSpec::KVMap.from_solrmarc_file "#{DIR}/data/umich/translation_maps/location_map.properties"
-    map['AAEL'].should.equal 'AAEL'
-    map['AAEL MICE'].should.equal 'AAEL MICE'
-    map['BUHR AAEL'].should.equal 'BUHR'
-  end
-  
+
   it "should read a pattern solrmarc file" do
     map = MARCSpec::MultiValueMap.from_solrmarc_file "#{DIR}/data/umich/translation_maps/library_map.properties"
     map.mapname.should.equal 'library_map'
     map['UMTRI Stuff'].should.equal ['Transportation Research Institute Library (UMTRI)']
     map['HATCH DOCS'].should.equal ['Hatcher Graduate', 'Hatcher Graduate Documents Center']
   end
-  
-  it "can dump/load a kv map via generic map interface" do
-    map = MARCSpec::KVMap.from_solrmarc_file "#{DIR}/data/umich/translation_maps/country_map.properties"
-    f = Tempfile.new('kvmap')
-    f.puts map.asPPString
-    path = f.path
-    f.close
-    map2 = MARCSpec::Map.fromFile(path)
-    f.unlink
-    map.class.should.equal MARCSpec::KVMap    
-    map.should.equal map2
-  end
 
   it "can dump/load a multivalue map via generic map interface" do
     map = MARCSpec::MultiValueMap.from_solrmarc_file "#{DIR}/data/umich/translation_maps/library_map.properties"
@@ -143,5 +170,13 @@ describe "Maps" do
     map.should.equal map2
   end
 
+  it "collapses ok" do
+    @mvmapCollapse['bill'].should.equal ['William']
+    @mvmapCollapse['william'].should.equal ['William']
+    @mvmapCollapse['bill dueber'].sort.should.equal ['William', 'Dueber'].sort
+    @mvmapCollapse['Will "duebes" Dueber'].sort.should.equal ['William', 'Dueber'].sort
+    @mvmapCollapse['notinthere'].should.equal nil
+  end
+    
   
-end
+end

data/spec/solrfieldspec_spec.rb

@@ -189,17 +189,6 @@ describe "CustomSolrSpec" do
     @map = MARCSpec::KVMap.new('nameOfTheMap', {@titleACValue.upcase => @mapValue, @default=>@mapValueForDefault})
   end
   
-  it "requires solrfield, module, and function" do
-    lambda{
-      css = MARCSpec::CustomSolrSpec.new(:solrField=>'solrField')
-    }.should.raise ArgumentError
-    lambda{
-      css = MARCSpec::CustomSolrSpec.new(:solrField=>'solrField', :module=>A::B)
-    }.should.raise ArgumentError
-    lambda{
-      css = MARCSpec::CustomSolrSpec.new(:module=>A::B, :functionSymbol=>:titleUp)
-    }.should.raise ArgumentError
-  end
   
   it "works with no args or map" do
     css = MARCSpec::CustomSolrSpec.new(:solrField=>'solrField', :module=>A::B, :functionSymbol=>:titleUp)

data/spec/specset_spec.rb

@@ -118,6 +118,12 @@ describe "SpecSet Basics" do
     h['two'].should.equal [2]
     h['letters'].should.equal ['a', 'b']
   end
+  
+  it "bails if it can't find a map" do
+    @speclist << {:solrField => 'tst', :mapname=>'nosuch', :specs => [['245']]}
+    lambda{@ss.buildSpecsFromList(@speclist)}.should.raise SystemExit 
+  end
+  
 end
 
 

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: marcspec
 version: !ruby/object:Gem::Version 
-  hash: 17
+  hash: 3
   prerelease: false
   segments: 
   - 1
-  - 1
-  - 1
-  version: 1.1.1
+  - 5
+  - 0
+  version: 1.5.0
 platform: ruby
 authors: 
 - BillDueber
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-09-15 00:00:00 -04:00
+date: 2010-09-20 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -99,6 +99,7 @@ files:
 - lib/marcspec/constantspec.rb
 - lib/marcspec/controlfieldspec.rb
 - lib/marcspec/customspec.rb
+- lib/marcspec/dsl.rb
 - lib/marcspec/kvmap.rb
 - lib/marcspec/leaderspec.rb
 - lib/marcspec/map.rb
@@ -111,6 +112,7 @@ files:
 - spec/controlfieldspec_spec.rb
 - spec/data/batch.dat
 - spec/data/one.dat
+- spec/data/simplemap.rb
 - spec/data/umich/translation_maps/area_map.properties
 - spec/data/umich/translation_maps/availability_map_ht.properties
 - spec/data/umich/translation_maps/availability_map_umich.properties
@@ -125,6 +127,7 @@ files:
 - spec/data/umich/translation_maps/library_map.properties
 - spec/data/umich/translation_maps/location_map.properties
 - spec/data/umich/umich_index.properties
+- spec/dsl_spec.rb
 - spec/leaderspec_spec.rb
 - spec/maps_spec.rb
 - spec/marcfieldspecs_spec.rb
@@ -169,6 +172,8 @@ summary: Extract data from MARC records and send to Solr
 test_files: 
 - spec/cachespot_spec.rb
 - spec/controlfieldspec_spec.rb
+- spec/data/simplemap.rb
+- spec/dsl_spec.rb
 - spec/leaderspec_spec.rb
 - spec/maps_spec.rb
 - spec/marcfieldspecs_spec.rb