proiel 1.2.1 → 1.3.0

data/lib/proiel/proiel_xml/reader.rb

@@ -1,5 +1,5 @@
 #--
-# Copyright (c) 2015-2016 Marius L. Jøhndal
+# Copyright (c) 2015-2018 Marius L. Jøhndal
 #
 # See LICENSE in the top-level source directory for licensing terms.
 #++
@@ -7,6 +7,95 @@ module PROIEL
   module PROIELXML
     # @api private
     module Reader
+      class DictionarySource
+        include SAXMachine
+
+        attribute :idref, required: true
+        attribute :license, required: false
+        attribute :n, required: false
+      end
+
+      class DictionaryGloss
+        include SAXMachine
+
+        attribute :language, required: true
+        value :gloss
+      end
+
+      class DictionaryHomograph
+        include SAXMachine
+
+        attribute :lemma, required: true
+        attribute :'part-of-speech', as: :part_of_speech, required: true
+      end
+
+      class DictionarySlot2
+        include SAXMachine
+
+        attribute :form, required: true
+        attribute :n, required: true
+      end
+
+      class DictionarySlot1
+        include SAXMachine
+
+        elements :slot2, as: :slot2s, class: DictionarySlot2
+
+        attribute :morphology, required: true
+      end
+
+      class DictionaryArgument
+        include SAXMachine
+
+        attribute :relation, required: true
+        attribute :lemma, required: false
+        attribute :'part-of-speech', as: :part_of_speech, required: false
+        attribute :mood, required: false
+        attribute :case, required: false
+      end
+
+      class DictionaryToken
+        include SAXMachine
+
+        attribute :idref, required: true
+        attribute :flags, required: false
+      end
+
+      class DictionaryFrame
+        include SAXMachine
+
+        # We skip the intermediate grouping elements 'arguments' and 'tokens'
+        elements :argument, as: :arguments, class: DictionaryArgument
+        elements :token, as: :tokens, class: DictionaryToken
+      end
+
+      class DictionaryLemma
+        include SAXMachine
+
+        attribute :lemma, required: true
+        attribute :'part-of-speech', as: :part_of_speech, required: true
+        attribute :n, required: false
+
+        # We skip the intermediate grouping elements 'distribution', 'glosses', 'homographs', 'paradigm' and 'valency'
+        elements :source, as: :distribution, class: DictionarySource
+        elements :gloss, as: :glosses, class: DictionaryGloss
+        elements :homograph, as: :homographs, class: DictionaryHomograph
+        elements :slot1, as: :paradigm, class: DictionarySlot1
+        elements :frame, as: :valency, class: DictionaryFrame
+      end
+
+      # Parsing class for `dictionary` elements.
+      class Dictionary
+        include SAXMachine
+
+        attribute :language, required: true
+        attribute :dialect, required: false
+
+        # We skip the intermediate grouping elements 'sources' and 'lemmata'
+        elements :source, as: :sources, class: DictionarySource
+        elements :lemma, as: :lemmata, class: DictionaryLemma
+      end
+
       # Parsing class for `slash` elements.
       class Slash
         include SAXMachine
@@ -15,6 +104,22 @@ module PROIEL
         attribute :relation, required: true
       end
 
+      # Parsing class for `semantic-tag` elements.
+      class SemanticTag
+        include SAXMachine
+
+        attribute :attribute, required: true
+        attribute :value, required: true
+      end
+
+      # Parsing class for `note` elements.
+      class Note
+        include SAXMachine
+
+        attribute :originator, required: true
+        value :content
+      end
+
       # Parsing class for `token` elements.
       class Token
         include SAXMachine
@@ -37,6 +142,8 @@ module PROIEL
         attribute :'foreign-ids', as: :foreign_ids
 
         elements :slash, as: :slashes, class: Slash
+        elements :'semantic-tag', as: :semantic_tags, class: SemanticTag
+        elements :note, as: :notes, class: Note
       end
 
       # Parsing class for `sentence` elements.
@@ -54,6 +161,7 @@ module PROIEL
         attribute :'presentation-after', as: :presentation_after
 
         elements :token, as: :tokens, class: Token
+        elements :note, as: :notes, class: Note
       end
 
       # Parsing class for `div` elements.
@@ -67,6 +175,7 @@ module PROIEL
 
         element :title
         elements :sentence, as: :sentences, class: Sentence
+        elements :note, as: :notes, class: Note
       end
 
       # Parsing class for `source` elements.
@@ -74,10 +183,12 @@ module PROIEL
         include SAXMachine
 
         attribute :id, required: true
-        attribute :'alignment-id', as: :alignment_id, required: false
+        attribute :'alignment-id', as: :alignment_id, class: String, required: false
         attribute :language, required: true
+        attribute :dialect, required: false
 
         element :title
+        element :alternative_title
         element :author
         element :citation_part
         element :principal
@@ -107,7 +218,11 @@ module PROIEL
         element :printed_text_publisher
         element :printed_text_place
         element :printed_text_date
+        element :chronology_composition
+        element :chronology_manuscript
+
         elements :div, as: :divs, class: Div
+        elements :note, as: :notes, class: Note
       end
 
       # Parsing class for `relations/value` elements.
@@ -174,6 +289,25 @@ module PROIEL
         attribute :summary, required: true
       end
 
+      # Parsing class for `lemma` elements.
+      class Lemma
+        include SAXMachine
+
+        attribute :form, required: true
+        attribute :'part-of-speech', as: :part_of_speech, required: true
+        attribute :gloss, required: false
+
+        elements :'semantic-tag', as: :semantic_tags, class: SemanticTag
+        elements :note, as: :notes, class: Note
+      end
+
+      # Parsing class for `dictionary` elements.
+      class Dictionary
+        include SAXMachine
+
+        elements :lemma, as: :lemmas, class: Lemma
+      end
+
       # Parsing class for `information_statuses` elements.
       class InformationStatuses
         include SAXMachine
@@ -189,6 +323,7 @@ module PROIEL
         element :parts_of_speech, as: :parts_of_speech, class: PartsOfSpeech
         element :morphology, class: Morphology
         element :information_statuses, as: :information_statuses, class: InformationStatuses
+        element :dictionary, as: :dictionary, class: Dictionary
       end
 
       # Parsing class for `proiel` elements.
@@ -199,6 +334,7 @@ module PROIEL
         attribute :'schema-version', as: :schema_version, required: true
 
         elements :source, as: :sources, class: Source
+        elements :dictionary, as: :dictionaries, class: Dictionary
         element :annotation, class: Annotation
       end
 

data/lib/proiel/proiel_xml/schema.rb

@@ -16,7 +16,7 @@ module PROIEL
       # @return [String] schema version number
       #
       def self.current_proiel_xml_schema_version
-        '2.1'
+        '3.0'
       end
 
       # Invalid PROIEL XML schema version error.
@@ -41,6 +41,8 @@ module PROIEL
             '2.0'
           when '2.1'
             '2.1'
+          when '3.0'
+            '3.0'
           when NilClass
             '1.0'
           else
@@ -70,7 +72,7 @@ module PROIEL
       # @raise ArgumentError
       #
       def self.proiel_xml_schema_filename(schema_version)
-        if schema_version == '1.0' or schema_version == '2.0' or schema_version == '2.1'
+        if schema_version == '1.0' or schema_version == '2.0' or schema_version == '2.1' or schema_version == '3.0'
           File.join(File.dirname(__FILE__),
                     "proiel-#{schema_version}",
                     "proiel-#{schema_version}.xsd")

data/lib/proiel/sentence.rb

@@ -113,10 +113,13 @@ module PROIEL
     # Returns the printable form of the sentence with all token forms and any
     # presentation data.
     #
+    # @param custom_token_formatter [Lambda] formatting function for tokens
+    # which is passed the token as its sole argument
+    #
     # @return [String] the printable form of the sentence
-    def printable_form(options = {})
+    def printable_form(custom_token_formatter: nil)
       [presentation_before,
-       @children.reject(&:is_empty?).map { |t| t.printable_form(options) },
+       @children.reject(&:is_empty?).map { |t| t.printable_form(custom_token_formatter: custom_token_formatter) },
        presentation_after].compact.join
     end
 

data/lib/proiel/source.rb

@@ -15,6 +15,9 @@ module PROIEL
     # @return [String] language of the source as an ISO 639-3 language tag
     attr_reader :language
 
+    # @return [String] dialect of the source
+    attr_reader :dialect
+
     # @return [DateTime] export time for the source
     attr_reader :export_time
 
@@ -26,7 +29,7 @@ module PROIEL
     attr_reader :alignment_id
 
     # Creates a new source object.
-    def initialize(parent, id, export_time, language, metadata, alignment_id, &block)
+    def initialize(parent, id, export_time, language, dialect, metadata, alignment_id, &block)
       @treebank = parent
       @id = id.freeze
 
@@ -34,6 +37,7 @@ module PROIEL
       @export_time = export_time.nil? ? nil : DateTime.parse(export_time).freeze
 
       @language = language.freeze
+      @dialect = dialect ? dialect.freeze : nil
       @metadata = metadata.freeze
 
       raise ArgumentError, 'string or nil expected' unless alignment_id.nil? or alignment_id.is_a?(String)
@@ -50,9 +54,12 @@ module PROIEL
     # Returns the printable form of the source with all token forms and any
     # presentation data.
     #
+    # @param custom_token_formatter [Lambda] formatting function for tokens
+    # which is passed the token as its sole argument
+    #
     # @return [String] the printable form of the source
-    def printable_form(options = {})
-      @children.map { |d| d.printable_form(options) }.compact.join
+    def printable_form(custom_token_formatter: nil)
+      @children.map { |d| d.printable_form(custom_token_formatter: custom_token_formatter) }.compact.join
     end
 
     # Accesses metadata fields.

data/lib/proiel/treebank.rb

@@ -1,5 +1,5 @@
 #--
-# Copyright (c) 2015 Marius L. Jøhndal
+# Copyright (c) 2015-2018 Marius L. Jøhndal
 #
 # See LICENSE in the top-level source directory for licensing terms.
 #++
@@ -23,9 +23,13 @@ module PROIEL
     # @return [Array<Source>] sources in the treebank
     attr_reader :sources
 
+    # @return [Array<Dictionary>] dictionaries in the treebank
+    attr_reader :dictionaries
+
     # Available metadata elements for sources.
     METADATA_ELEMENTS = %i(
       title
+      alternative_title
       author
       citation_part
       principal
@@ -55,6 +59,8 @@ module PROIEL
       printed_text_publisher
       printed_text_place
       printed_text_date
+      chronology_composition
+      chronology_manuscript
     )
 
     # Creates a new treebank object.
@@ -62,6 +68,7 @@ module PROIEL
       @annotation_schema = nil
       @schema_version = nil
       @sources = []
+      @dictionaries = []
 
       @source_index = {}
       @div_index = {}
@@ -85,12 +92,18 @@ module PROIEL
         tf = PROIELXML::Reader.parse_io(f)
 
         tf.proiel.sources.each do |s|
-          @sources << Source.new(self, s.id, tf.proiel.export_time, s.language,
+          @sources << Source.new(self, s.id, tf.proiel.export_time, s.language, s.dialect,
                                  bundle_metadata(s), s.alignment_id) do |source|
             build_divs(s, source)
           end
 
-          index_objects!(@sources.last)
+          index_source_objects!(@sources.last)
+        end
+
+        tf.proiel.dictionaries.each do |s|
+          @dictionaries << Dictionary.new(self, tf.proiel.export_time, s.language, s.dialect, s)
+
+          index_dictionary_objects!(@dictionaries.last)
         end
 
         annotation_schema = AnnotationSchema.new(tf.proiel.annotation)
@@ -198,7 +211,7 @@ module PROIEL
       end
     end
 
-    def index_objects!(source)
+    def index_source_objects!(source)
       @source_index[source.id] = source
 
       source.divs.each do |div|
@@ -213,5 +226,9 @@ module PROIEL
         end
       end
     end
+
+    def index_dictionary_objects!(dictionary)
+      # TODO
+    end
   end
 end

data/lib/proiel/version.rb

@@ -5,5 +5,5 @@
 #++
 module PROIEL
   # Gem version
-  VERSION = '1.2.1'
+  VERSION = '1.3.0'
 end

data/lib/proiel/visualization/graphviz.rb

@@ -2,7 +2,7 @@ module PROIEL
   module Visualization
     module Graphviz
       DEFAULT_GRAPHVIZ_BINARY = 'dot'.freeze
-      DEFAULT_TEMPLATES = %i(classic linearized packed)
+      DEFAULT_TEMPLATES = %i(classic linearized packed modern aligned-modern)
       SUPPORTED_OUTPUT_FORMATS = %i(png svg)
 
       class GraphvizError < Exception
@@ -21,7 +21,7 @@ module PROIEL
       def self.generate(template, graph, output_format, options = {})
         raise ArgumentError, 'string or symbol expected' unless template.is_a?(String) or template.is_a?(Symbol)
 
-        dot_code = generate_dot(template, graph)
+        dot_code = generate_dot(template, graph, options)
 
         if output_format.to_sym == :dot
           dot_code
@@ -58,7 +58,9 @@ module PROIEL
         result
       end
 
-      def self.generate_dot(template, graph)
+      def self.generate_dot(template, graph, options)
+        raise ArgumentError, 'invalid direction' unless options[:direction].nil? or %(TD LR).include?(options[:direction])
+
         filename = template_filename(template)
 
         content = File.read(filename)
@@ -66,12 +68,14 @@ module PROIEL
         template = ERB.new(content, nil, '-')
         template.filename = filename
 
-        TemplateContext.new(graph).generate(template)
+        TemplateContext.new(graph, options[:direction] || 'TD').generate(template)
       end
 
       class TemplateContext
-        def initialize(graph)
+        def initialize(graph, direction, title = '')
           @graph = graph
+          @direction = direction
+          @title = title
         end
 
         def generate(template)

data/lib/proiel/visualization/graphviz/aligned-modern.dot.erb

@@ -0,0 +1,83 @@
+digraph "<%= @title -%>" {
+  charset="UTF-8";
+  graph [truecolor=true,bgcolor=transparent];
+  rankdir="<%= @direction -%>";
+  nodesep=0.1;
+  ranksep=0.25;
+
+  <%- @graph.left.each_with_index do |tokens, i| -%>
+    <%= "rootL#{i}" -%> [label="",shape=point];
+
+    <%- tokens.select { |t| t.empty_token_sort != 'P' }.each do |token| -%>
+      <%- if token.empty_token_sort -%>
+        <%= node token.id, token.relation.to_s.upcase, shape: :none, fontcolor: :gray -%>
+      <%- else -%>
+        <%= node token.id, token.relation.to_s.upcase, shape: :none -%>
+      <%- end -%>
+
+      <%- if token.relation -%>
+        <%= edge (token.head ? token.head.id : "rootL#{i}"), token.id, '', weight: 1.0, color: :orange, arrowhead: :none -%>
+      <%- end -%>
+
+      <%- token.slashes.each do |(relation, target)| -%>
+        <%= edge token.id, target, relation.to_s.upcase, weight: 0.0, fontcolor: :blue, color: :blue, style: :dashed %>
+      <%- end -%>
+    <%- end -%>
+
+    <%- tokens.reject(&:empty_token_sort).each do |token| -%>
+      <%= edge token.id, "T#{token.id}", nil, weight: 10, arrowhead: :none -%>
+    <%- end -%>
+  <%- end -%>
+
+  {
+    rank="same";
+
+    <%- @graph.left.each do |tokens| -%>
+      <%- tokens.reject(&:empty_token_sort).each do |token| -%>
+        <%= node "T#{token.id}", token.form, shape: :none, fontcolor: :blue, tooltip: [token.lemma, token.part_of_speech, token.morphology].join("\n") -%>
+      <%- end -%>
+
+      <%= tokens.reject(&:empty_token_sort).map { |token| "T#{token.id}" }.join('->') -%> [style="invis"];
+    <%- end -%>
+  }
+
+  <%- @graph.right.each_with_index do |tokens, i| -%>
+    <%= "rootR#{i}" -%> [label="",shape=point];
+
+    <%- tokens.select { |t| t.empty_token_sort != 'P' }.each do |token| -%>
+      <%- if token.empty_token_sort -%>
+        <%= node token.id, token.relation.to_s.upcase, shape: :none, fontcolor: :gray -%>
+      <%- else -%>
+        <%= node token.id, token.relation.to_s.upcase, shape: :none -%>
+      <%- end -%>
+
+      <%- if token.relation -%>
+        <%= edge token.id, (token.head ? token.head.id : "rootR#{i}"), '', weight: 1.0, color: :orange, arrowhead: :none -%>
+      <%- end -%>
+
+      <%- token.slashes.each do |(relation, target)| -%>
+        <%= edge token.id, target, relation.to_s.upcase, weight: 0.0, fontcolor: :blue, color: :blue, style: :dashed %>
+      <%- end -%>
+    <%- end -%>
+
+    <%- tokens.reject(&:empty_token_sort).each do |token| -%>
+      <%= edge "T#{token.id}", token.id, nil, weight: 10, arrowhead: :none -%>
+    <%- end -%>
+  <%- end -%>
+
+  {
+    rank="same";
+
+    <%- @graph.right.each do |tokens| -%>
+      <%- tokens.reject(&:empty_token_sort).each do |token| -%>
+        <%= node "T#{token.id}", token.form, shape: :none, fontcolor: :blue, tooltip: [token.lemma, token.part_of_speech, token.morphology].join("\n") -%>
+      <%- end -%>
+
+      <%= tokens.reject(&:empty_token_sort).map { |token| "T#{token.id}" }.join('->') -%> [style="invis"];
+    <%- end -%>
+  }
+
+  <%- @graph.alignments.each do |x, y| -%>
+    <%= "T#{x}" -%> -> <%= "T#{y}" -%> [color=blue,dir=none];
+  <%- end -%>
+}