llt-review 0.0.1 → 0.0.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: c67647d0b4ba29bff3feb4066ed2d11eb2253a41
-  data.tar.gz: 6f8ba9c45e85319859765ce378ead8a654a2f12a
+  metadata.gz: 91ff18f1e72915936a2069917fa843b701ee7df8
+  data.tar.gz: f4398109dc7d16bdfed7980a6dba2359c7350c0a
 SHA512:
-  metadata.gz: 6bf66bd60ab8558ea51be1fa765c87540cbd419e993cdfe1ec4ce188284c19685c739f05f88115d4cb6b93bddcbdca91a71c19be378fa91c66653dbb6bc3427b
-  data.tar.gz: 9bba65d16cc077da638427d74aa31d84e258aa14f2456757903ea801cfd08e7b91999bfeee6526f503831d7c17e39342b65f2cf1c5bd8f9feac7389f4ed8082b
+  metadata.gz: 3e5087b17bef29c42939a587f667636f848773a6a5e0902936753413fafdeb51901327e222fd31eda844f9233555a69762adf236c82bf68c5e4658541569ccfa
+  data.tar.gz: 7eef4e5e71816b0fd7ef3d10e436c84f649c85e1d6ac23baedf9414c99b2b9f74640deac6807dca701e68bf246d65283f84cd25e72f5be4475de832ea98ae693

data/.gitignore

@@ -15,3 +15,4 @@ spec/reports
 test/tmp
 test/version_tmp
 tmp
+.sass-cache

data/config.ru

@@ -1,2 +1,8 @@
+require 'haml'
+require 'sass/plugin/rack'
 require 'llt/review/api'
+
+Sass::Plugin.options[:style] = :compressed
+use Sass::Plugin::Rack
+
 run Api

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

@@ -36,7 +36,7 @@ module LLT
       def set_languages(attrs)
         hsh = Hash[attrs]
         lang = hsh['xml:lang']
-        hsh['lnum'] == 'l1' ? @lang1 = lang : @lang2 = lang
+        hsh['lnum'] == 'L1' ? @lang1 = lang : @lang2 = lang
       end
     end
   end

data/lib/llt/review/api/helpers.rb

@@ -0,0 +1,47 @@
+module LLT
+  module Review::Api
+    module Helpers
+      def origin(el)
+        "Publication ##{extracted_id(el.id)} by #{el.sentences.annotators}"
+      end
+
+      def combined_extracted_id(comp)
+        ids = [comp.gold, comp.reviewable].map { |el| extracted_id(el.id)}.join('-')
+        "comp#{ids}"
+      end
+
+      def extracted_id(id)
+        last = id.split('/').last
+        /(.*?)(\.([\w]*?))?$/.match(last)[1]
+      end
+
+      def arethusa(rev, gold, lang, chunk = nil, word = nil)
+        path = "http://sosol.perseids.org/tools/arethusa/app/#/perseids_review_#{lang}_aldt"
+        path << "?doc=#{rev}&gold=#{gold}"
+        if chunk || word
+          path << "&chunk=#{chunk}" if chunk
+          path << "&w=#{word}" if word
+        end
+        path
+      end
+
+      def to_tooltip(cat, v)
+        %{#{cat}: <span class="success">#{v.original}</span> -> <span class="error">#{v.new}</span>}
+      end
+
+      def extract_heads(diff, s_id)
+        if heads = diff[:head]
+          [to_id(s_id, heads.original), to_id(s_id, heads.new)]
+        end
+      end
+
+      def to_id(s_id, w_id)
+        "#{s_id}-#{w_id}"
+      end
+
+      def to_percent(total, part)
+        ((part.to_f / total) * 100).round(2)
+      end
+    end
+  end
+end

data/lib/llt/review/api.rb

@@ -1,29 +1,34 @@
 require 'sinatra/base'
 require 'sinatra/respond_with'
 require 'llt/review'
+require 'llt/review/api/helpers'
 
 class Api < Sinatra::Base
   register Sinatra::RespondWith
 
-  get '/:type/diff' do
-    gold = Array(params[:gold])
-    rev  = Array(params[:reviewable])
+  set :root, File.expand_path('../../../..', __FILE__)
 
-    comp_param = params[:compare]
-    comparables = comp_param ? Array(comp_param).map(&:to_sym) : nil
 
-    klass = params[:type]
-    # return an error if klass is neither treebank nor Alignment
+  helpers LLT::Review::Api::Helpers
 
-    diff = LLT::Review.const_get(klass.capitalize).new
-    puts comparables
-    diff.diff(gold, rev, comparables)
+  get '/:type/diff' do
+    process_params(params)
+    diff = process_diff(params)
 
     respond_to do |f|
       f.xml { diff.to_xml }
     end
   end
 
+  get '/:type/diff/:view' do
+    process_params(params)
+    diff = process_diff(params)
+
+    if params[:view] == 'html'
+      haml :index, locals: { diff: diff }
+    end
+  end
+
   get '/:type/report' do
     uris = Array(params[:uri])
     klass = params[:type]
@@ -34,4 +39,38 @@ class Api < Sinatra::Base
       f.xml { reporter.to_xml(:report)}
     end
   end
+
+  def process_params(params)
+    if backend = params[:backend]
+      p = case backend
+          when 'perseids'     then 'sosol.perseids.org/sosol'
+          when 'perseids-dev' then 'dev.alpheios.net:3000'
+          end
+
+      expand_perseids_urls(params, :gold, p)
+      expand_perseids_urls(params, :reviewable, p)
+    end
+  end
+
+  def process_diff(params)
+    gold = Array(params[:gold])
+    rev  = Array(params[:reviewable])
+
+    comp_param = params[:compare]
+    comparables = comp_param ? Array(comp_param).map(&:to_sym) : nil
+
+    klass = params[:type]
+    # return an error if klass is neither treebank nor Alignment
+
+    diff = LLT::Review.const_get(klass.capitalize).new
+    diff.diff(gold, rev, comparables)
+    diff
+  end
+
+  def expand_perseids_urls(params, key, sosol_path)
+    t = Array(params[key])
+    params[key] = t.map do |publication_id|
+      "http://#{sosol_path}/dmm_api/item/TreebankCite/#{publication_id}"
+    end
+  end
 end

data/lib/llt/review/common/golden.rb

@@ -1,12 +1,18 @@
 module LLT
   class Review::Common
     module Golden
-      attr_reader :sentences
+      attr_accessor :sentences
 
       # Check Comparison#report to learn more
       def clone
         cloned = super
+        # need to go to some lengths here, as Parsing::Result objects,
+        # which hide behind @sentences, have important inst vars of
+        # their own now and are not simply HashContainables
+        parse_result = @sentences.clone;
         cloned.replace_with_clone(:sentences, :report)
+        parse_result.container.merge!(cloned.sentences)
+        cloned.sentences = parse_result
         cloned
       end
     end

data/lib/llt/review/helpers/parsing/helper/annotator.rb

@@ -0,0 +1,15 @@
+module LLT
+  module Review::Helpers::Parsing::Helper
+    class Annotator
+      attr_accessor :short, :name, :address, :uri
+
+      def complete?
+        @short && @name && @address && @uri
+      end
+
+      def name_and_short_to_s
+        "#{@name} (#{@short})"
+      end
+    end
+  end
+end

data/lib/llt/review/helpers/parsing/helper/annotators.rb

@@ -0,0 +1,25 @@
+module LLT
+  module Review::Helpers::Parsing::Helper
+    class Annotators
+      def initialize
+        @annotators = []
+      end
+
+      def add(annotator)
+        @annotators << annotator
+      end
+
+      def to_s
+        meth = :name_and_short_to_s
+        case @annotators.length
+        when 0 then "unknown annotators"
+        when 1 then @annotators.first.send(meth)
+        when 2 then @annotators.map { |a| a.send(meth) }.join(' and ')
+        else
+          with_punct = @annotators[0..-2].map { |a| a.send(meth) }.join(', ')
+          "#{with_punct} and #{@annotators.last.send(meth)}"
+        end
+      end
+    end
+  end
+end

data/lib/llt/review/helpers/parsing/helper.rb

@@ -1,6 +1,8 @@
 module LLT
   module Review::Helpers::Parsing::Helper
     require 'llt/review/helpers/parsing/helper/for_nokogiri'
+    require 'llt/review/helpers/parsing/helper/annotators'
+    require 'llt/review/helpers/parsing/helper/annotator'
 
     def initialize
       @result = Review::Helpers::Parsing::Result.new
@@ -22,9 +24,42 @@ module LLT
       @sentence.add(@word)
     end
 
+    def register_format(format)
+      @result.format = format
+    end
+
+    def register_language(language)
+      @result.lang = language
+    end
+
     def namespace
       # has should be implemented by classes that use this module
       self.class
     end
+
+    def annotator
+      @annotator ||= Annotator.new
+    end
+
+    def add_complete_annotator
+      if annotator.complete?
+        @result.annotators.add(annotator)
+        @annotator = nil
+      end
+    end
+
+    def set_annotator_variable(attr, val)
+      instance_variable_set("@in_#{attr}", val);
+    end
+
+    def parse_annotator_values(str)
+      params = [:short, :name, :address, :uri]
+      params.each do |param|
+        if instance_variable_get("@in_#{param}")
+          annotator.send("#{param}=", str)
+          add_complete_annotator
+        end
+      end
+    end
   end
 end

data/lib/llt/review/helpers/parsing/result.rb

@@ -1,6 +1,12 @@
 module LLT
   module Review::Helpers::Parsing
     class Result
+      attr_accessor :format, :lang
+
+      def annotators
+        @annotators ||= Helper::Annotators.new
+      end
+
       include Core::Structures::HashContainable
     end
   end

data/lib/llt/review/helpers/reportable.rb

@@ -16,6 +16,10 @@ module LLT
         each { |_, el| el.init_diff }
       end
 
+      def percentage(category = :right)
+        ((send(category).to_f / @total) * 100).round(2)
+      end
+
       def add(element)
         if el = @container[element.id]
           el.add_total(element)

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

@@ -16,6 +16,28 @@ module LLT
         case name
         when 'word'     then register_word(attrs)
         when 'sentence' then register_sentence(first_val(attrs))
+        when 'treebank' then register_treebank_attrs(attrs)
+        when 'annotator' then @in_annotator = true
+        end
+
+        if @in_annotator
+          set_annotator_variable(name, true)
+        end
+      end
+
+      def end_element(name, attrs = [])
+        if name == 'annotator'
+          @in_annotator = false
+        end
+
+        if @in_annotator
+          set_annotator_variable(name, false)
+        end
+      end
+
+      def characters(string)
+        if @in_annotator
+          parse_annotator_values(string)
         end
       end
 
@@ -25,6 +47,12 @@ module LLT
         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
+
+      def register_treebank_attrs(attrs)
+        hsh = Hash[attrs]
+        register_language(hsh['xml:lang'])
+        register_format(hsh['format'])
+      end
     end
   end
 end

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

@@ -15,6 +15,12 @@ module LLT
         case name
         when :word     then @in_word = true
         when :sentence then @in_sentence = true
+        when :annotator then @in_annotator = true
+        when :treebank then @in_treebank = true
+        end
+
+        if @in_annotator
+          set_annotator_variable(name, true)
         end
       end
 
@@ -22,6 +28,11 @@ module LLT
         case name
         when :word     then @in_word = false
         when :sentence then @in_sentence = false
+        when :annotator then @in_annotator = false
+        end
+
+        if @in_annotator
+          set_annotator_variable(name, false)
         end
       end
 
@@ -35,6 +46,15 @@ module LLT
           end
         when @in_sentence
           register_sentence(value) if name == :id
+        when @in_treebank
+          register_language(value) if name == :"xml:lang"
+          register_format(value)   if name == :format
+        end
+      end
+
+      def text(value)
+        if @in_annotator
+          parse_annotator_values(value)
         end
       end
     end

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

@@ -16,13 +16,11 @@ module LLT
       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/lib/llt/review/version.rb

@@ -1,5 +1,5 @@
 module LLT
   class Review
-    VERSION = "0.0.1"
+    VERSION = "0.0.2"
   end
 end

data/llt-review.gemspec

@@ -20,6 +20,9 @@ Gem::Specification.new do |spec|
 
   spec.add_development_dependency "bundler", "~> 1.5"
   spec.add_development_dependency "rake"
-  spec.add_development_dependency "rspec"
+  spec.add_development_dependency "rspec", "2.14"
   spec.add_development_dependency "simplecov", "~> 0.7"
+
+  spec.add_dependency 'sass'
+  spec.add_dependency 'haml'
 end

data/public/javascripts/script.js

@@ -0,0 +1,91 @@
+$(document).ready(function() {
+
+  function hsv2rgb(h, s, v) {
+  }
+
+  function percentToColor(percent) {
+    var h = Math.floor((100 - percent) * 120/ 100);
+    var s = 1, v = 1;
+
+    var rgb, i, data = [];
+    h = h / 60;
+    i = Math.floor(h);
+    data = [v*(1-s), v*(1-s*(h-i)), v*(1-s*(1-(h-i)))];
+    switch(i) {
+      case 0:
+        rgb = [v, data[2], data[0]]; break;
+      case 1:
+        rgb = [data[1], v, data[0]]; break;
+      case 2:
+        rgb = [data[0], v, data[2]]; break;
+      case 3:
+        rgb = [data[0], data[1], v]; break;
+      case 4:
+        rgb = [data[2], data[0], v]; break;
+      default:
+        rgb = [v, data[0], data[1]]; break;
+    }
+
+    return '#' + rgb.map(function(x){
+      return ("0" + Math.round(x*255).toString(16)).slice(-2);
+    }).join('');
+  }
+
+
+  function addTokenTooltips() {
+    $('.token.tooltip').tooltipsy({
+      content: function($el) { return $el.attr('tooltip'); },
+      delay: 0
+    });
+
+    $('.token.head-error').each(function() {
+      var el = $(this);
+      el.hover(function() {
+        $('#' + el.attr('hr')).toggleClass('success-bg');
+        $('#' + el.attr('hw')).toggleClass('error-bg');
+      });
+    });
+  }
+
+  function addWordsDiffBindings() {
+    var sel = '.word-diff-table';
+    $(sel).hide();
+    $('.word-diff-container').click(function() {
+      $(this).find(sel).toggle(300);
+    });
+  }
+
+  function addPercentageColors() {
+    $('.percentage').each(function() {
+      var el = $(this);
+      var percent = el.attr('percentage');
+      var color = percentToColor(100 - percent);
+      el.css('color', color);
+    });
+  }
+
+  function addCategoryToggler() {
+    $('.toggler').each(function() {
+      var el = $(this);
+      var target = el.siblings('#' + el.attr('toggler'));
+      el.click(function() { target.toggle(300); });
+      target.hide();
+    });
+  }
+
+  function addReportToggler() {
+    $('.comparison_toggle').each(function() {
+      var el = $(this);
+      var target = $('#' + el.attr('c_id'));
+      el.click(function() { target.toggle(300); });
+      el.addClass('clickable');
+      target.hide();
+    });
+  }
+
+  addWordsDiffBindings();
+  addTokenTooltips();
+  addPercentageColors();
+  addCategoryToggler();
+  addReportToggler();
+});

data/public/javascripts/tooltipsy.min.js

@@ -0,0 +1,20 @@
+/* tooltipsy by Brian Cray
+ * Lincensed under GPL2 - http://www.gnu.org/licenses/gpl-2.0.html
+ * Option quick reference:
+ * - alignTo: "element" or "cursor" (Defaults to "element")
+ * - offset: Tooltipsy distance from element or mouse cursor, dependent on alignTo setting. Set as array [x, y] (Defaults to [0, -1])
+ * - content: HTML or text content of tooltip. Defaults to "" (empty string), which pulls content from target element's title attribute
+ * - show: function(event, tooltip) to show the tooltip. Defaults to a show(100) effect
+ * - hide: function(event, tooltip) to hide the tooltip. Defaults to a fadeOut(100) effect
+ * - delay: A delay in milliseconds before showing a tooltip. Set to 0 for no delay. Defaults to 200
+ * - css: object containing CSS properties and values. Defaults to {} to use stylesheet for styles
+ * - className: DOM class for styling tooltips with CSS. Defaults to "tooltipsy"
+ * - showEvent: Set a custom event to bind the show function. Defaults to mouseenter
+ * - hideEvent: Set a custom event to bind the show function. Defaults to mouseleave
+ * Method quick reference:
+ * - $('element').data('tooltipsy').show(): Force the tooltip to show
+ * - $('element').data('tooltipsy').hide(): Force the tooltip to hide
+ * - $('element').data('tooltipsy').destroy(): Remove tooltip from DOM
+ * More information visit http://tooltipsy.com/
+ */
+;(function(a){a.tooltipsy=function(c,b){this.options=b;this.$el=a(c);this.title=this.$el.attr("title")||"";this.$el.attr("title","");this.random=parseInt(Math.random()*10000);this.ready=false;this.shown=false;this.width=0;this.height=0;this.delaytimer=null;this.$el.data("tooltipsy",this);this.init()};a.tooltipsy.prototype={init:function(){var e=this,d,b=e.$el,c=b[0];e.settings=d=a.extend({},e.defaults,e.options);d.delay=+d.delay;if(typeof d.content==="function"){e.readify()}if(d.showEvent===d.hideEvent&&d.showEvent==="click"){b.toggle(function(f){if(d.showEvent==="click"&&c.tagName=="A"){f.preventDefault()}if(d.delay>0){e.delaytimer=window.setTimeout(function(){e.show(f)},d.delay)}else{e.show(f)}},function(f){if(d.showEvent==="click"&&c.tagName=="A"){f.preventDefault()}window.clearTimeout(e.delaytimer);e.delaytimer=null;e.hide(f)})}else{b.bind(d.showEvent,function(f){if(d.showEvent==="click"&&c.tagName=="A"){f.preventDefault()}e.delaytimer=window.setTimeout(function(){e.show(f)},d.delay||0)}).bind(d.hideEvent,function(f){if(d.showEvent==="click"&&c.tagName=="A"){f.preventDefault()}window.clearTimeout(e.delaytimer);e.delaytimer=null;e.hide(f)})}},show:function(i){if(this.ready===false){this.readify()}var b=this,f=b.settings,h=b.$tipsy,k=b.$el,d=k[0],g=b.offset(d);if(b.shown===false){if((function(m){var l=0,e;for(e in m){if(m.hasOwnProperty(e)){l++}}return l})(f.css)>0){b.$tip.css(f.css)}b.width=h.outerWidth();b.height=h.outerHeight()}if(f.alignTo==="cursor"&&i){var j=[i.clientX+f.offset[0],i.clientY+f.offset[1]];if(j[0]+b.width>a(window).width()){var c={top:j[1]+"px",right:j[0]+"px",left:"auto"}}else{var c={top:j[1]+"px",left:j[0]+"px",right:"auto"}}}else{var j=[(function(){if(f.offset[0]<0){return g.left-Math.abs(f.offset[0])-b.width}else{if(f.offset[0]===0){return g.left-((b.width-k.outerWidth())/2)}else{return g.left+k.outerWidth()+f.offset[0]}}})(),(function(){if(f.offset[1]<0){return g.top-Math.abs(f.offset[1])-b.height}else{if(f.offset[1]===0){return g.top-((b.height-b.$el.outerHeight())/2)}else{return g.top+b.$el.outerHeight()+f.offset[1]}}})()]}h.css({top:j[1]+"px",left:j[0]+"px"});b.settings.show(i,h.stop(true,true))},hide:function(c){var b=this;if(b.ready===false){return}if(c&&c.relatedTarget===b.$tip[0]){b.$tip.bind("mouseleave",function(d){if(d.relatedTarget===b.$el[0]){return}b.settings.hide(d,b.$tipsy.stop(true,true))});return}b.settings.hide(c,b.$tipsy.stop(true,true))},readify:function(){this.ready=true;this.$tipsy=a('<div id="tooltipsy'+this.random+'" style="position:fixed;z-index:2147483647;display:none">').appendTo("body");this.$tip=a('<div class="'+this.settings.className+'">').appendTo(this.$tipsy);this.$tip.data("rootel",this.$el);var c=this.$el;var b=this.$tip;this.$tip.html(this.settings.content!=""?(typeof this.settings.content=="string"?this.settings.content:this.settings.content(c,b)):this.title)},offset:function(b){return this.$el[0].getBoundingClientRect()},destroy:function(){if(this.$tipsy){this.$tipsy.remove();a.removeData(this.$el,"tooltipsy")}},defaults:{alignTo:"element",offset:[0,-1],content:"",show:function(c,b){b.fadeIn(100)},hide:function(c,b){b.fadeOut(100)},css:{},className:"tooltipsy",delay:200,showEvent:"mouseenter",hideEvent:"mouseleave"}};a.fn.tooltipsy=function(b){return this.each(function(){new a.tooltipsy(this,b)})}})(jQuery);