llt-review 0.0.1

data/lib/llt/review/treebank/difference/postag.rb

@@ -0,0 +1,37 @@
+module LLT
+  module Review::Treebank::Difference
+    class Postag < Generic
+      def initialize(tag, original, new)
+        super
+        compute_detailed_differences
+      end
+
+      def diff_id
+        @diff_id ||= "#{id}:#{map { |_, v| v.diff_id }.join('::')}"
+      end
+
+      private
+
+      # copied over right now from Postag until we figure out how to solve this
+      # more globally
+
+      POSTAG_SCHEMA = %i{
+        part_of_speech person number tense
+        mood voice gender case degree
+      }
+
+      def compute_detailed_differences
+        @original.each_char.with_index do |a, i|
+          b = @new[i]
+          add(Datapoint.new(POSTAG_SCHEMA[i], a, b)) unless a == b
+        end
+      end
+
+      def write_to_report(report, unique)
+        postags = report[:postags]
+        postags.add_wrong(unique)
+        postags[item.to_s].add_wrong(unique)
+      end
+    end
+  end
+end

data/lib/llt/review/treebank/difference/relation.rb

@@ -0,0 +1,11 @@
+module LLT
+  module Review::Treebank::Difference
+    class Relation < Attribute
+      def report_location
+        :relations
+      end
+    end
+  end
+end
+
+

data/lib/llt/review/treebank/difference/sentence.rb

@@ -0,0 +1,6 @@
+module LLT
+  module Review::Treebank::Difference
+    class Sentence < Review::Common::Difference::Sentence
+    end
+  end
+end

data/lib/llt/review/treebank/difference/word.rb

@@ -0,0 +1,8 @@
+module LLT
+  module Review::Treebank::Difference
+    class Word < Review::Common::Difference::Word
+      attr_reader :lemma, :head, :relation, :postag
+    end
+  end
+end
+

data/lib/llt/review/treebank/difference.rb

@@ -0,0 +1,15 @@
+module LLT
+  class Review::Treebank
+    module Difference
+      require 'llt/review/treebank/difference/sentence'
+      require 'llt/review/treebank/difference/word'
+      require 'llt/review/treebank/difference/generic'
+      require 'llt/review/treebank/difference/attribute'
+      require 'llt/review/treebank/difference/postag'
+      require 'llt/review/treebank/difference/datapoint'
+      require 'llt/review/treebank/difference/head'
+      require 'llt/review/treebank/difference/lemma'
+      require 'llt/review/treebank/difference/relation'
+    end
+  end
+end

data/lib/llt/review/treebank/gold.rb

@@ -0,0 +1,8 @@
+module LLT
+  class Review::Treebank
+    class Gold < Report
+      include Review::Common::Golden
+    end
+  end
+end
+

data/lib/llt/review/treebank/parser/helper.rb

@@ -0,0 +1,9 @@
+module LLT
+  class Review::Treebank::Parser
+    module Helper
+      def namespace
+        Review::Treebank
+      end
+    end
+  end
+end

data/lib/llt/review/treebank/parser/nokogiri_handler.rb

@@ -0,0 +1,30 @@
+require 'nokogiri'
+
+module LLT
+  class Review::Treebank::Parser
+    class NokogiriHandler < Nokogiri::XML::SAX::Document
+
+      include Review::Helpers::Parsing::Helper
+      include Review::Helpers::Parsing::Helper::ForNokogiri
+      include Helper
+
+      def parse(data)
+        Nokogiri::XML::SAX::Parser.new(self).parse(data)
+      end
+
+      def start_element(name, attrs = [])
+        case name
+        when 'word'     then register_word(attrs)
+        when 'sentence' then register_sentence(first_val(attrs))
+        end
+      end
+
+      private
+
+      def register_word(attrs)
+        super(attrs.shift.last) # need to shift, we don't want the id in the next step
+        attrs.each { |k, v| @word.send("#{k}=", v) }
+      end
+    end
+  end
+end

data/lib/llt/review/treebank/parser/ox_handler.rb

@@ -0,0 +1,42 @@
+require 'ox'
+
+module LLT
+  class Review::Treebank::Parser
+    class OxHandler < Ox::Sax
+
+      include Review::Helpers::Parsing::Helper
+      include Helper
+
+      def parse(data)
+        Ox.sax_parse(self, data)
+      end
+
+      def start_element(name)
+        case name
+        when :word     then @in_word = true
+        when :sentence then @in_sentence = true
+        end
+      end
+
+      def end_element(name)
+        case name
+        when :word     then @in_word = false
+        when :sentence then @in_sentence = false
+        end
+      end
+
+      def attr(name, value)
+        case
+        when @in_word
+          if name == :id
+            register_word(value)
+          else
+            @word.send("#{name}=", value)
+          end
+        when @in_sentence
+          register_sentence(value) if name == :id
+        end
+      end
+    end
+  end
+end

data/lib/llt/review/treebank/parser.rb

@@ -0,0 +1,13 @@
+module LLT
+  class Review::Treebank
+    class Parser
+      require 'llt/review/treebank/parser/helper'
+      autoload :NokogiriHandler, 'llt/review/treebank/parser/nokogiri_handler'
+      autoload :OxHandler,       'llt/review/treebank/parser/ox_handler'
+
+      include Review::Helpers::Parsing
+    end
+  end
+end
+
+

data/lib/llt/review/treebank/postag.rb

@@ -0,0 +1,65 @@
+module LLT
+  class Review::Treebank
+    class Postag
+      attr_reader :id
+
+      def initialize(postag)
+        @id = :postag
+        @postag = postag
+      end
+
+      def to_s
+        @postag
+      end
+
+      # A little violating... move this to Datapoints
+      def report
+        @report ||= begin
+          # Questionable what the total numbers of datapoints should be.
+          # Count empty points as well?
+          data = Report::Datapoints.new(@postag.size)
+          add_datapoints_container(data)
+          data.each_with_index do |(_, container), i|
+            rtr = container.reports_to_request
+            val = @postag[i]
+            next unless val && val != '-'
+            container.add(Report::Postag::Datapoint.new(rtr, val))
+            container.increment
+          end
+          data
+        end
+      end
+
+
+      # All these constants cannot stay. Hardcoding the meaning of every postag
+      # datapoint is a no-go.
+
+      POSTAG_SCHEMA = %i{
+        part_of_speech person number tense
+        mood voice gender case degree
+      }
+
+      def analysis
+        @analysis ||= begin
+          Hash[POSTAG_SCHEMA.zip(@postag.each_char)]
+        end
+      end
+
+      def datapoints
+        POSTAG_SCHEMA.size
+      end
+
+      PLURALIZED_POSTAG_SCHEMA = %i{
+        parts_of_speech persons numbers tenses
+        moods voices genders cases degrees
+      }
+
+      CONTAINER_TABLE = PLURALIZED_POSTAG_SCHEMA.zip(POSTAG_SCHEMA)
+      def add_datapoints_container(data)
+        CONTAINER_TABLE.each do |pl, sg|
+          data.add(Report::Generic.new(pl, 0, sg))
+        end
+      end
+    end
+  end
+end

data/lib/llt/review/treebank/report/datapoints.rb

@@ -0,0 +1,12 @@
+module LLT
+  class Review::Treebank::Report
+    class Datapoints < Generic
+      include Review::Helpers::Reportable
+
+      def initialize(total = 1)
+        super(:datapoints, total)
+      end
+    end
+  end
+end
+

data/lib/llt/review/treebank/report/generic.rb

@@ -0,0 +1,10 @@
+module LLT
+  class Review::Treebank::Report
+    class Generic < Review::Common::Report::Generic
+      def collect_reports(words)
+        return unless @reports_to_request
+        words.each { |_, word| add(word[@reports_to_request].report) }
+      end
+    end
+  end
+end

data/lib/llt/review/treebank/report/lemma.rb

@@ -0,0 +1,8 @@
+module LLT
+  class Review::Treebank::Report
+    class Lemma
+      include Review::Helpers::Reportable
+    end
+  end
+end
+

data/lib/llt/review/treebank/report/postag/datapoint.rb

@@ -0,0 +1,16 @@
+module LLT
+  class Review::Treebank::Report::Postag
+    class Datapoint
+      include Review::Helpers::Reportable
+
+      def initialize(tag, id, total = 1)
+        super(id, total)
+        @tag = tag
+      end
+
+      def xml_tag
+        @tag
+      end
+    end
+  end
+end

data/lib/llt/review/treebank/report/postag.rb

@@ -0,0 +1,7 @@
+module LLT
+  class Review::Treebank::Report
+    class Postag
+      include Review::Helpers::Reportable
+    end
+  end
+end

data/lib/llt/review/treebank/report/postags.rb

@@ -0,0 +1,27 @@
+module LLT
+  class Review::Treebank::Report
+    class Postags < Generic
+      def initialize(total = 1)
+        super(:postags, total)
+      end
+
+      def collect_reports(words)
+        words.each do |_, word|
+          postag = word[:postag]
+          add(Postag.new(postag.to_s))
+          add(postag.report)
+        end
+      end
+
+      # Sorting is all good, but we want the postags to show up
+      # in front of the datapoints
+      def sort
+        shifter = :datapoints
+        sorted = super
+        sorted[shifter] = sorted.delete(shifter)
+        sorted
+      end
+    end
+  end
+end
+

data/lib/llt/review/treebank/report/relation.rb

@@ -0,0 +1,7 @@
+module LLT
+  class Review::Treebank::Report
+    class Relation
+      include Review::Helpers::Reportable
+    end
+  end
+end

data/lib/llt/review/treebank/report.rb

@@ -0,0 +1,19 @@
+module LLT
+  class Review::Treebank
+    class Report < Review::Common::Report
+      require 'llt/review/treebank/report/generic'
+      require 'llt/review/treebank/report/datapoints'
+      require 'llt/review/treebank/report/postags'
+      require 'llt/review/treebank/report/lemma'
+      require 'llt/review/treebank/report/postag'
+      require 'llt/review/treebank/report/postag/datapoint'
+      require 'llt/review/treebank/report/relation'
+
+      private
+
+      def namespace
+        Report
+      end
+    end
+  end
+end

data/lib/llt/review/treebank/reviewable.rb

@@ -0,0 +1,6 @@
+module LLT
+  class Review::Treebank
+    class Reviewable < Review::Common::Reviewable
+    end
+  end
+end

data/lib/llt/review/treebank/sentence.rb

@@ -0,0 +1,25 @@
+module LLT
+  class Review::Treebank
+    class Sentence < Review::Common::Sentence
+      private
+
+      def report_container
+        reports = {
+          words: nil,
+          heads: nil,
+          relations: :relation,
+          lemmata: :lemma,
+        }
+
+        reports.each_with_object({}) do |(tag, requested), hsh|
+          hsh[tag] = Report::Generic.new(tag, size, requested)
+        end.merge(postags: Report::Postags.new(size))
+      end
+
+      def diff_namespace
+        Review::Treebank::Difference
+      end
+    end
+  end
+end
+

data/lib/llt/review/treebank/word.rb

@@ -0,0 +1,53 @@
+module LLT
+  class Review::Treebank
+    class Word
+      include Core::Structures::HashContainable
+
+      attr_accessor :form, :lemma, :head, :relation
+      attr_reader :postag
+
+      Attr = Struct.new(:id, :attribute) do
+        def to_s
+          attribute
+        end
+
+        def report
+          @report ||= Report.const_get(id.capitalize).new(attribute)
+        end
+      end
+
+      %i{ lemma head relation }.each do |type|
+        define_method("#{type}=") { |val| add(Attr.new(type, val)) }
+      end
+
+      def postag=(tag)
+        add(Postag.new(tag))
+      end
+
+      COMPARABLE_ELEMENTS = %i{ lemma postag head relation }
+
+      def compare(other, diff_container, comparables)
+        comparables = COMPARABLE_ELEMENTS unless comparables # default value
+        comparables.each do |comparator|
+          a, b = [self, other].map { |w| w[comparator].to_s }
+          if a != b
+            d = diff_container[id] ||= Difference::Word.new(self)
+            d.add(new_difference(comparator, a, b))
+          end
+        end
+      end
+
+      private
+
+      def new_difference(comparator, original, new)
+        klass = Difference.const_get(comparator.capitalize)
+        klass.new(self[comparator], original, new)
+      end
+
+      # used when parsers try to check in attributes we are not interested in
+      def method_missing(meth, *args, &blk)
+        super unless meth =~ /=$/
+      end
+    end
+  end
+end

data/lib/llt/review/treebank.rb

@@ -0,0 +1,22 @@
+module LLT
+  class Review::Treebank < Review
+    require 'llt/review/treebank/parser'
+
+    require 'llt/review/treebank/report'
+    require 'llt/review/treebank/gold'
+    require 'llt/review/treebank/reviewable'
+    require 'llt/review/treebank/comparison'
+
+    require 'llt/review/treebank/sentence'
+    require 'llt/review/treebank/word'
+    require 'llt/review/treebank/postag'
+
+    require 'llt/review/treebank/difference'
+
+    private
+
+    def root_identifier
+      'treebank'
+    end
+  end
+end

data/lib/llt/review/version.rb

@@ -0,0 +1,5 @@
+module LLT
+  class Review
+    VERSION = "0.0.1"
+  end
+end

data/lib/llt/review.rb

@@ -0,0 +1,118 @@
+require 'llt/core/api/helpers'
+require 'llt/core/structures/hash_containable'
+require 'llt/core/structures/hash_containable/generic'
+require "llt/review/version"
+
+module LLT
+  # This is pretty much the only messy class this whole gem contains.
+  # Do something about it!
+  class Review
+    require 'llt/review/helpers'
+
+    require 'llt/review/common'
+    require 'llt/review/treebank'
+    require 'llt/review/alignment'
+
+    include Core::Api::Helpers
+
+    def diff(gold, reviewables, comparables = nil)
+      parses = parse_files(Gold: gold, Reviewable: reviewables)
+
+      @gold, @reviewables = parses.partition do |parse_data|
+        parse_data.instance_of?(self.class.const_get(:Gold))
+      end
+
+      compare(comparables)
+      diff_report
+      all_diffs
+    end
+
+    def report(*uris)
+      @reports = parse_files(Report: uris)
+      @reports.each(&:report)
+      @reports
+    end
+
+    def to_xml(type = :diff)
+      root_name = "#{root_identifier}-#{type}"
+      XML_DECLARATION + wrap_with_tag(root_name, header + send("#{type}_to_xml"))
+    end
+
+    def all_diffs
+      @all_diffs ||= @reviewables.map do |reviewable|
+        reviewable.diff.values
+      end.flatten
+    end
+    alias_method :diffs, :all_diffs
+    alias_method :comparisons, :all_diffs
+
+    private
+
+    def diff_report
+      if @reviewables.one?
+        all_diffs.each(&:report)
+      else
+        diff_report_with_cloned_reports
+      end
+    end
+
+    # Check the comment at Comparison#report for more info
+    def diff_report_with_cloned_reports
+      used_golds = []
+      all_diffs.each do |d|
+        d.report(to_clone_or_not_to_clone?(used_golds, d.gold.id))
+      end
+    end
+
+    def to_clone_or_not_to_clone?(used, id)
+      used.include?(id) ? true : (used << id; false)
+    end
+
+    def compare(comparables = nil)
+      @gold.each do |gold|
+        @reviewables.each { |reviewable| reviewable.compare(gold, comparables) }
+      end
+    end
+
+    def parse_files(files)
+      to_parse = files.flat_map { |klass, uris| uris.map { |uri| [klass, uri] } }
+      parse_threaded(to_parse)
+    end
+
+    def parse_threaded(uris_with_classes)
+      threads = uris_with_classes.map do |klass, uri|
+        Thread.new do
+          data = get_from_uri(uri)
+          self.class.const_get(klass).new(uri, parse(data))
+        end
+      end
+      threads.map { |t| t.join; t.value }
+    end
+
+    def header
+      wrap_with_tag('files', header_files.map(&:xml_heading).join)
+    end
+
+    def header_files
+      [@gold, @reviewables, @reports].flatten.compact
+    end
+
+    def wrap_with_tag(tag, content)
+      "<#{tag}>" +
+        content +
+      "</#{tag}>"
+    end
+
+    def diff_to_xml
+      wrap_with_tag(:comparisons, @reviewables.map(&:to_xml).join)
+    end
+
+    def report_to_xml
+      wrap_with_tag(:reports, @reports.map(&:to_xml).join)
+    end
+
+    def parse(data)
+      self.class.const_get(:Parser).new.parse(data)
+    end
+  end
+end

data/llt-review.gemspec

@@ -0,0 +1,25 @@
+# coding: utf-8
+lib = File.expand_path('../lib', __FILE__)
+$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
+require 'llt/review/version'
+
+Gem::Specification.new do |spec|
+  spec.name          = "llt-review"
+  spec.version       = LLT::Review::VERSION
+  spec.authors       = ["LFDM"]
+  spec.email         = ["1986gh@gmail.com"]
+  spec.summary       = %q{Reviews annotations}
+  spec.description   = spec.summary
+  spec.homepage      = ""
+  spec.license       = "MIT"
+
+  spec.files         = `git ls-files`.split($/)
+  spec.executables   = spec.files.grep(%r{^bin/}) { |f| File.basename(f) }
+  spec.test_files    = spec.files.grep(%r{^(test|spec|features)/})
+  spec.require_paths = ["lib"]
+
+  spec.add_development_dependency "bundler", "~> 1.5"
+  spec.add_development_dependency "rake"
+  spec.add_development_dependency "rspec"
+  spec.add_development_dependency "simplecov", "~> 0.7"
+end