anvl 0.1.3 → 0.2.0

data/VERSION

@@ -1 +1 @@
-0.1.3
+0.2.0

data/anvl.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = %q{anvl}
-  s.version = "0.1.3"
+  s.version = "0.2.0"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Chris Beer"]
-  s.date = %q{2010-12-09}
+  s.date = %q{2010-12-12}
   s.email = %q{chris@cbeer.info}
   s.extra_rdoc_files = [
     "LICENSE.txt",
@@ -25,8 +25,13 @@ Gem::Specification.new do |s|
     "VERSION",
     "anvl.gemspec",
     "lib/anvl.rb",
+    "lib/anvl/document.rb",
+    "lib/anvl/element.rb",
+    "lib/anvl/erc.rb",
     "test/helper.rb",
-    "test/test_anvl.rb"
+    "test/test_anvl.rb",
+    "test/test_element.rb",
+    "test/test_erc.rb"
   ]
   s.homepage = %q{http://github.com/cbeer/anvl}
   s.licenses = ["MIT"]
@@ -35,7 +40,9 @@ Gem::Specification.new do |s|
   s.summary = %q{Ruby implementation of A Name-Value Language (ANVL)}
   s.test_files = [
     "test/helper.rb",
-    "test/test_anvl.rb"
+    "test/test_anvl.rb",
+    "test/test_element.rb",
+    "test/test_erc.rb"
   ]
 
   if s.respond_to? :specification_version then

data/lib/anvl.rb

@@ -1,115 +1,6 @@
+require 'anvl/element'
+require 'anvl/document'
 module ANVL
-  class Document # < Hash
-
-    def self.parse str
-      anvl = ANVL::Document.new  str
-      anvl
-    end
-
-    attr_reader :entries
-
-    def initialize obj = nil
-      @entries = Hash.new { |h,k| h[k] = [] }
-
-      case obj
-        when Hash
-          obj.each do |key, value|
-            @entries[key.to_sym] = value
-	  end
-	when String
-	  lines = obj.gsub(/\s?\n\s+/, ' ').split("\n")
-
-	  lines.each_with_index do |str, i|
-	    case str
-	      when /^#/
-          parse_comment str, i
-        when /:/  
-          parse_entry str, i
-	      else  # something is wrong..
-          nil
-	    end  
-	  end  
-
-      end if obj
-
-      add_entries_methods
-
-      gc!
-    end
-
-    def to_s
-      gc!
-      @entries.map do |key, value|
-        if value.is_a? Array
-          value.map do |v|
-            "#{key}: #{v.to_s.gsub(/\n/, "\n    ")}"
-          end
-        else
-          "#{key}: #{value.to_s.gsub(/\n/, "\n    ")}" 
-        end
-      end.join "\n"
-    end
-
-    def to_h
-      gc!
-      @entries
-    end
-
-    def [] key
-      return @entries[key]
-    end
-
-    def []= key, value
-      value = [value] unless value.is_a? Array
-      @entries[key] = value
-    end
-
-    def push hash
-      hash.each do |key, value|
-        @entries[key] = [@entries[key]] unless @entries[key].is_a? Array
-	if value.is_a? Array
-	  value.each do |v|      
-	    @entries[key] << v
-	  end
-        else
-          @entries[key] << value
-	end
-      end
-      gc!
-      @entries
-    end
-    alias_method :'<<', :push
-
-    private
-
-    def parse_comment str, line=0
-
-    end
-
-    def parse_entry str, line=0
-      key, value = str.split ":", 2
-      @entries[key.to_sym] << value.strip
-    end
-
-    def add_entries_methods
-      @entries.public_methods(false).reject { |x| self.respond_to? x }.each do |meth|
-        (class << self; self; end).class_eval do
-          define_method meth do |*args|
-            @entries.send meth, *args
-	  end 
-	end
-      end
-    end  
-
-    def gc!
-      @entries.delete_if { |key, value| value.nil? or (value.is_a? Array and value.empty?) }
-
-      @entries.each do |key, value|
-        @entries[key] = value.first if value.is_a? Array and value.length == 1
-      end
-    end
-  end
-
   def self.parse *args
     Document.parse *args
   end

data/lib/anvl/document.rb

@@ -0,0 +1,130 @@
+module ANVL
+  class Document # < Hash
+    def self.parse str
+      anvl = self.new  str
+      anvl
+    end
+
+    attr_reader :entries
+
+    def initialize obj = nil
+      @entries = Array.new
+
+      case obj
+        when Hash
+          self.push obj
+
+	when String
+	  lines = obj.gsub(/\s?\n\s+/, ' ').split("\n")
+
+	  lines.each_with_index do |str, i|
+	    case str
+	      when /^#/
+                parse_comment str, i
+              when /:/  
+                parse_entry str, i
+	      else  # something is wrong..
+                nil
+	    end  
+	  end  
+
+      end if obj
+
+      add_entries_methods
+
+      gc!
+    end
+
+    def to_s
+      gc!
+      @entries.map do |obj|
+        obj.to_anvl
+      end.join "\n"
+    end
+
+    def to_h
+      gc!
+      h = {}
+
+      @entries.map do |obj|
+        h[(obj.display_label || obj.label).to_sym] ||= []
+        h[(obj.display_label || obj.label).to_sym] << obj.to_s
+      end  
+
+      h.each do |label, value|
+        h[label] = value.first if value.length == 1
+      end
+
+      h
+    end
+
+    def [] display_label, args = {}
+      v = @entries.select { |x| x =~ display_label }
+      v &&= v.map { |x| x.to_s } unless args[:raw]
+      v &&= v.first if v.length == 1
+      v
+    end
+    alias_method :fetch, :[]
+
+    def []= display_label, value, append = false
+      label = convert_label display_label
+      value = [value] unless value.is_a? Array
+      @entries.delete_if { |x| x =~ label } unless append
+      value.each do |v|
+        case v
+          when Hash
+            @entries << element_class.new({ :document => self, :label => label, :value => v }.merge(v))
+          else
+            @entries << element_class.new({ :document => self, :label => label, :value => v })
+        end
+      end
+    end
+
+    alias_method :store, :[]=
+
+    def push hash
+      hash.each do |label, value|
+        self.store label, value, true
+      end  
+      gc!
+      @entries
+    end
+    alias_method :'<<', :push
+
+    protected
+    def element_class
+      ANVL::Element
+    end
+
+    def parse_comment str, line=0
+
+    end
+
+    def parse_entry str, line=0
+      label, value = str.split ":", 2
+      self.store label, value, true
+    end
+
+    def add_entries_methods
+      @entries.reject { |x| self.respond_to? x[:label] }.each do |obj|
+        (class << self; self; end).class_eval do
+          define_method obj.label do |*args|
+            @entries[obj.label]
+	  end 
+	end
+      end
+    end  
+
+    def gc!
+    end
+
+    def format_value str = ''
+      str = str.to_s
+      str &&= str.gsub(/\n/, "\n    ")
+    end
+
+    def convert_label label
+      label.to_s
+    end
+  end
+end

data/lib/anvl/element.rb

@@ -0,0 +1,61 @@
+module ANVL
+  class Element 
+    attr_reader :document, :label, :display_label, :value
+    def initialize args
+      @document = args[:document]
+      @display_label = args[:display_label] || convert_label(args[:label])
+      @label = convert_label(args[:label])
+      @value = args[:value].to_s.strip
+    end
+
+    def [] key
+      self.send key.to_sym
+    end
+
+    def to_s
+      str = @value
+      str &&= str.gsub(/\s*\n\s+/, ' ')
+
+      str
+    end
+
+    def to_anvl
+      "#{@display_label || @label}:#{format_value_for_anvl}"
+    end
+
+    def <=> obj
+      @value <=> obj
+    end
+
+    def == obj
+      case obj
+        when String
+          return @value == obj if obj.is_a? String
+        when Element  
+          @label == obj.label && @value = obj.display_label
+        else
+          false
+      end
+    end
+
+    def =~ str
+      str = str.to_s
+      str == @label or str == @display_label or convert_label(str) == @label
+    end
+
+    def push value
+      @document.store @display_label,  value, true
+    end
+    alias_method :'<<', :push
+
+    private
+    def format_value_for_anvl
+      " " + @value.gsub(/\n/, "\n    ")
+    end
+
+    def convert_label label
+      label
+    end
+
+  end
+end

data/lib/anvl/erc.rb

@@ -0,0 +1,109 @@
+require 'anvl'
+module ANVL
+  class Erc < Document
+    STORY_TO_OFFSET = { 'erc' => 0, 'about-erc' => 10, 'support-erc' => 20, 'meta-erc' => 30 }
+
+    LABEL_TO_SYN = { 
+      'who' => 'h1', 'what' => 'h2', 'when' => 'h3', 'where' => 'h4', 'how' => 'h5',
+      'about-who' => 'h11', 'about-what' => 'h12', 'about-when' => 'h13', 'about-where' => 'h14', 'about-how' => 'h15',
+      'support-who' => 'h21', 'support-what' => 'h22', 'support-when' => 'h23', 'support-where' => 'h24',
+      'meta-who' => 'h31', 'meta-what' => 'h32', 'meta-when' => 'h33', 'meta-where' => 'h34'
+    }
+
+    SYN_TO_KERNEL = LABEL_TO_SYN.invert
+
+
+    class Element < ANVL::Element
+      def to_s
+        str = super
+        str &&= format_initial_comma_to_recover_word_order str
+        str &&= decode_element_value_encoding str
+        str
+      end
+
+      def =~ str
+        bool = super
+
+        label = convert_label str
+        bool ||= (label =~ Regexp.new("\\(#{Regexp.escape(@label)}\\)"))
+      end
+
+      private
+      def format_value_for_anvl
+        super.sub(/^\s+([,;\|])/, '\1')
+      end
+
+      def format_initial_comma_to_recover_word_order str, sort_point = ','
+        return str unless str[0, 1] == sort_point
+        arr = str.split(sort_point)
+        arr.shift
+
+       insig = nil
+        if str[-1,1] == sort_point
+          insig = arr.pop.strip
+        end
+
+        word = arr.pop
+        arr = [arr.join ","]
+        arr.unshift word
+        arr.unshift insig
+
+        arr.compact.map(&:strip).reject { |x|  x.empty?}.join " "
+      end
+
+      def decode_element_value_encoding str
+        h = { '%sp' => ' ', '%ex' => '!', '%dq' => '"', '%ns' => '#', '%do' => '$', '%pe' => '%', '%am' => '&', '%sq' => '\'', '%op' => '(', '%cp' => ')', '%as' => '*', '%pl' => '+', '%co' => ',', '%sl' => '/', '%cn' => ':', '%sc' => ';', '%lt' => '<', '%eq' => '=', '%gt' => '>', '%qu' => '?', '%at' => '@', '%ox' => '[', '%ls' => '\\', '%cx' => ']', '%vb' => '|', '%nu' => "\0", '%%' => '%' }
+        s = str.dup
+        h.each do |key, value|
+         s.gsub! key, value
+        end
+        s
+      end
+
+      def convert_label label
+        str = label.to_s.downcase.gsub(' ', '_')
+
+        str = ANVL::Erc::LABEL_TO_SYN[str] if ANVL::Erc::LABEL_TO_SYN[str]
+
+        str
+      end
+
+    end
+
+    def []= display_label, value, append = false
+      if value[0,1] != '|' and STORY_TO_OFFSET[display_label] and !value.strip.empty?
+        process_abbr_form display_label, value 
+      elsif value[0,1] != ';' and value =~ /;/
+        value.split(';').each do |v|
+          self.store display_label, v, true
+        end
+      else
+        super
+      end
+    end
+    alias_method :store, :[]=
+
+    def complete?
+      !(self.fetch(:h1).empty? or self.fetch(:h2).empty? or self.fetch(:h3).empty? or self.fetch(:h4).empty?) ||
+      !(self.fetch(:h11).empty? or self.fetch(:h12).empty? or self.fetch(:h13).empty? or self.fetch(:h14).empty?)  ||
+      !(self.fetch(:h21).empty? or self.fetch(:h22).empty? or self.fetch(:h23).empty? or self.fetch(:h24).empty?) ||
+      !(self.fetch(:h31).empty? or self.fetch(:h32).empty? or self.fetch(:h33).empty? or self.fetch(:h34).empty?)
+    end
+
+    protected
+
+    def element_class
+      ANVL::Erc::Element
+    end
+
+    def process_abbr_form display_label, value
+      offset = STORY_TO_OFFSET[display_label]
+      elements = value.split('|')
+      elements.each_with_index do |e, i|
+        next if e.strip.empty?
+        self.store "h#{offset + i + 1}", e.strip
+      end
+    end
+
+  end
+end

data/test/test_anvl.rb

@@ -38,7 +38,8 @@ a: 2'
     assert_equal({:a => 'a' }, h.to_h)
     h[:a] = ['a', 'b']
     assert_equal({:a => ['a', 'b'] }, h.to_h)
-    h[:a] << 'c'
+    h[:a].inspect
+    h.store :a, 'c', true
     assert_equal({:a => ['a', 'b', 'c'] }, h.to_h)
     assert_equal(['a', 'b', 'c'], h[:a])
 
@@ -49,10 +50,10 @@ a: 2'
     assert_equal({:a => ['a', 'b', 'c', 'd'] }, h.to_h)
 
     h << { :c => 1 }
-    assert_equal(1, h[:c])
+    assert_equal("1", h[:c])
 
     h << { :c => 2 }
-    assert_equal([1, 2], h[:c])
+    assert_equal(["1", "2"], h[:c])
 
     str = h.to_s
     assert_match(/^a: a$/, str)
@@ -76,9 +77,33 @@ a: 2'
   
   def test_fmt_first_draft
     str = ANVL.to_anvl({:entry => [""], :who => ['Gilbert, W.S. | Sullivan, Arthur'], :what => ["The Yeomen of the Guard"], :"when/created" => [1888]})
+    h = ANVL.parse({:entry => [""], :who => ['Gilbert, W.S. | Sullivan, Arthur'], :what => ["The Yeomen of the Guard"], :"when/created" => [1888]})
     assert_match(/entry:/, str)
     assert_match(/who: Gilbert, W.S. | Sullivan, Arthur/, str)
     assert_match(/what: The Yeomen of the Guard/, str)
     assert_match(/when\/created: 1888/, str)
   end
+
+  def test_display_label
+    h = ANVL::Document.new
+    h['a'] = {:display_label => 'A', :value => '123' }
+    assert_equal("123", h['a'])
+
+    assert_match(/A:/, h.to_s)
+  end
+  
+  def test_str_vs_sym
+    str = 'erc:
+who:    Lederberg, Joshua
+what:   Studies of Human Families for Genetic Linkage
+when:   1974
+where:  http://profiles.nlm.nih.gov/BB/AA/TT/tt.pdf
+note:   This is an arbitrary note inside a
+        small descriptive record.'
+    h = ANVL::Document.parse str
+
+    assert_equal("Lederberg, Joshua", h['who'])
+    assert_equal("Lederberg, Joshua", h[:who])
+  end
+
 end

data/test/test_element.rb

@@ -0,0 +1,23 @@
+require 'helper'
+
+class TestErcElement < Test::Unit::TestCase
+  def test_basic
+    h = ANVL::Erc::Element.new :label => 'abc', :value => '123'
+    assert_equal('123', h.to_s)
+    assert_equal('abc: 123', h.to_anvl)
+  end
+
+  def test_initial_comma_to_recover_natural_word_order
+    h = ANVL::Erc::Element.new :label => 'abc', :value => ',  van Gogh, Vincent'
+    assert_equal('Vincent van Gogh', h.to_s)
+    assert_equal('abc:,  van Gogh, Vincent', h.to_anvl)
+
+    h = ANVL::Erc::Element.new :label => 'abc', :value => ',  Howell, III, PhD, 1922-1987, Thurston'
+    assert_equal('Thurston Howell, III, PhD, 1922-1987', h.to_s)
+    assert_equal('abc:,  Howell, III, PhD, 1922-1987, Thurston', h.to_anvl)
+
+    h = ANVL::Erc::Element.new :label => 'abc', :value => ',  McCartney, Paul, Sir,'
+    assert_equal('Sir Paul McCartney', h.to_s)
+    assert_equal('abc:,  McCartney, Paul, Sir,', h.to_anvl)
+  end
+end

data/test/test_erc.rb

@@ -0,0 +1,134 @@
+require 'helper'
+require 'anvl/erc'
+
+class TestErc < Test::Unit::TestCase
+  def test_parse_empty
+    h = ANVL::Erc.parse ''
+    assert_equal(0, h.entries.length)
+  end
+
+  def test_basic
+    str = 'erc:
+who:    Gibbon, Edward
+what:   The Decline and Fall of the Roman Empire
+when:   1781
+where:  http://www.ccel.org/g/gibbon/decline/'
+
+    h = ANVL::Erc.parse str
+    assert_equal("", h[:erc])
+    assert_equal("Gibbon, Edward", h[:who])
+    assert_equal("The Decline and Fall of the Roman Empire", h[:what])
+    assert_equal("1781", h[:when])
+    assert_equal("http://www.ccel.org/g/gibbon/decline/", h[:where])
+  end  
+
+  def test_complete
+    str = 'erc:
+what:    The Digital Dilemma
+where:  http://books.nap.edu/html/digital%5Fdilemma
+    '
+
+    h = ANVL::Erc.parse str
+    assert_equal(false, h.complete?)
+
+    h[:who] = '---'
+    h[:when] = '---'
+    assert(h.complete?)
+  end
+
+  def test_complete_meta
+    str = 'meta-erc:  NLM | pm9546494 | 19980418
+               | http://ark.nlm.nih.gov/12025/pm9546494??'
+    h = ANVL::Erc.parse str
+    assert(h.complete?)
+  end
+
+  def test_missing_about
+    str = 'about-erc:   | Bispectrum ; Nonlinearity ; Epilepsy
+                   ; Cooperativity ; Subdural ; Hippocampus'
+    h = ANVL::Erc.parse str
+    assert_contains(h['about-what'], 'Bispectrum')
+    assert_contains(h['about-what'], 'Nonlinearity')
+    assert_contains(h['about-what'], 'Epilepsy')
+    assert_contains(h['about-what'], 'Cooperativity')
+    assert_contains(h['about-what'], 'Subdural')
+    assert_contains(h['about-what'], 'Hippocampus')
+  end
+
+  def test_abbr
+    str = 'erc: Gibbon, Edward | The Decline and Fall of the Roman Empire
+         | 1781 | http://www.ccel.org/g/gibbon/decline/'
+
+    h = ANVL::Erc.parse str
+    assert_equal("Gibbon, Edward", h[:who])
+    assert_equal("The Decline and Fall of the Roman Empire", h[:what])
+    assert_equal("1781", h[:when])
+    assert_equal("http://www.ccel.org/g/gibbon/decline/", h[:where])
+  end
+
+  def test_multiple_values
+    str = 'erc:
+who:  Smith, J; Wong, D; Khan, H'
+    h = ANVL::Erc.parse str
+
+    assert_contains(h[:who], 'Smith, J')
+    assert_contains(h[:who], 'Wong, D')
+    assert_contains(h[:who], 'Khan, H')
+  end
+
+  def test_intl
+    str = 'erc:
+who:    Lederberg, Joshua
+h2:   Studies of Human Families for Genetic Linkage'
+    h = ANVL::Erc.parse str
+    assert_equal("Lederberg, Joshua", h[:who])
+    assert_equal("Lederberg, Joshua", h['who'])
+    assert_equal("Lederberg, Joshua", h[:h1])
+    assert_equal("Lederberg, Joshua", h['wer(h1)'])
+    assert_equal("Studies of Human Families for Genetic Linkage", h[:h2])
+    assert_equal("Studies of Human Families for Genetic Linkage", h[:what])
+    assert_equal("Studies of Human Families for Genetic Linkage", h['was(h2)'])
+  end
+
+  def test_label_structure
+    str = 'erc:'
+    h = ANVL::Erc.parse str
+
+    h['marc_856'] = 'abc'
+
+    assert_equal("abc", h['marc_856'])
+    h['MARC 856'] = '123'
+
+    assert_equal("123", h['marc_856'])
+  end
+
+  def test_initial_comma
+    str = 'erc:
+who:,  van Gogh, Vincent
+who:,  Howell, III, PhD, 1922-1987, Thurston
+who:,  Acme Rocket Factory, Inc., The
+who:,  Mao Tse Tung
+who:,  McCartney, Pat, Ms,
+who:,  McCartney, Paul, Sir,
+who:,  McCartney, Petra, Dr,
+what:, Health and Human Services, United States Government
+    Department of, The,'
+
+    h = ANVL::Erc.parse str
+    assert_contains(h[:who], "Vincent van Gogh")
+    assert_contains(h[:who], "Thurston Howell, III, PhD, 1922-1987")
+    assert_contains(h[:who], "The Acme Rocket Factory, Inc.")
+    assert_contains(h[:who], "Mao Tse Tung")
+    assert_contains(h[:who], "Ms Pat McCartney")
+    assert_contains(h[:who], "Sir Paul McCartney")
+    assert_contains(h[:who], "Dr Petra McCartney")
+    assert_equal(h[:what], "The United States Government Department of Health and Human Services")
+  end
+
+  def test_value_encoding
+    str = 'erc:
+who: %sp%ex%dq'
+    h = ANVL::Erc.parse str
+    assert_contains(" !\"", h[:who])
+  end
+end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: anvl
 version: !ruby/object:Gem::Version 
-  hash: 29
+  hash: 23
   prerelease: false
   segments: 
   - 0
-  - 1
-  - 3
-  version: 0.1.3
+  - 2
+  - 0
+  version: 0.2.0
 platform: ruby
 authors: 
 - Chris Beer
@@ -15,10 +15,13 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-12-09 00:00:00 -05:00
+date: 2010-12-12 00:00:00 -05:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
+  type: :development
+  prerelease: false
+  name: shoulda
   version_requirements: &id001 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
@@ -28,11 +31,11 @@ dependencies:
         segments: 
         - 0
         version: "0"
-  name: shoulda
   requirement: *id001
+- !ruby/object:Gem::Dependency 
   type: :development
   prerelease: false
-- !ruby/object:Gem::Dependency 
+  name: bundler
   version_requirements: &id002 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
@@ -44,11 +47,11 @@ dependencies:
         - 0
         - 0
         version: 1.0.0
-  name: bundler
   requirement: *id002
+- !ruby/object:Gem::Dependency 
   type: :development
   prerelease: false
-- !ruby/object:Gem::Dependency 
+  name: jeweler
   version_requirements: &id003 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
@@ -60,11 +63,11 @@ dependencies:
         - 5
         - 1
         version: 1.5.1
-  name: jeweler
   requirement: *id003
+- !ruby/object:Gem::Dependency 
   type: :development
   prerelease: false
-- !ruby/object:Gem::Dependency 
+  name: rcov
   version_requirements: &id004 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
@@ -74,10 +77,7 @@ dependencies:
         segments: 
         - 0
         version: "0"
-  name: rcov
   requirement: *id004
-  type: :development
-  prerelease: false
 description: 
 email: chris@cbeer.info
 executables: []
@@ -97,8 +97,13 @@ files:
 - VERSION
 - anvl.gemspec
 - lib/anvl.rb
+- lib/anvl/document.rb
+- lib/anvl/element.rb
+- lib/anvl/erc.rb
 - test/helper.rb
 - test/test_anvl.rb
+- test/test_element.rb
+- test/test_erc.rb
 has_rdoc: true
 homepage: http://github.com/cbeer/anvl
 licenses: 
@@ -136,3 +141,5 @@ summary: Ruby implementation of A Name-Value Language (ANVL)
 test_files: 
 - test/helper.rb
 - test/test_anvl.rb
+- test/test_element.rb
+- test/test_erc.rb