spread2rdf 0.0.1pre.1 → 0.0.1

data/spread2rdf.gemspec

@@ -21,12 +21,14 @@ Gem::Specification.new do |gem|
   gem.required_ruby_version = '>= 1.9.3'
 
   gem.add_dependency('activesupport', '~> 3.2.3')
-  gem.add_dependency('awesome_print')
+  gem.add_dependency('awesome_print', '~> 1.1.0')
 
   gem.add_dependency('roo', '~> 1.12.2')
   gem.add_dependency('rubyzip', '~> 1.0.0') # for the roo-xlsm-fix
 
-  gem.add_dependency('linkeddata')
+  gem.add_dependency('linkeddata', '~> 1.0.0')
+
+  gem.add_dependency('ocra', '~> 1.3.1')
 
   gem.add_development_dependency('rake')
   gem.add_development_dependency('pry', '~> 0.9.12.2')

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: spread2rdf
 version: !ruby/object:Gem::Version
-  version: 0.0.1pre.1
+  version: 0.0.1
 platform: ruby
 authors:
 - Marcel Otto
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-10-16 00:00:00.000000000 Z
+date: 2013-10-23 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -28,16 +28,16 @@ dependencies:
   name: awesome_print
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - ! '>='
+    - - ~>
       - !ruby/object:Gem::Version
-        version: '0'
+        version: 1.1.0
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - ! '>='
+    - - ~>
       - !ruby/object:Gem::Version
-        version: '0'
+        version: 1.1.0
 - !ruby/object:Gem::Dependency
   name: roo
   requirement: !ruby/object:Gem::Requirement
@@ -70,16 +70,30 @@ dependencies:
   name: linkeddata
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - ! '>='
+    - - ~>
       - !ruby/object:Gem::Version
-        version: '0'
+        version: 1.0.0
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - ! '>='
+    - - ~>
       - !ruby/object:Gem::Version
-        version: '0'
+        version: 1.0.0
+- !ruby/object:Gem::Dependency
+  name: ocra
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ~>
+      - !ruby/object:Gem::Version
+        version: 1.3.1
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ~>
+      - !ruby/object:Gem::Version
+        version: 1.3.1
 - !ruby/object:Gem::Dependency
   name: rake
   requirement: !ruby/object:Gem::Requirement
@@ -133,6 +147,7 @@ extra_rdoc_files: []
 files:
 - .gitignore
 - Gemfile
+- Gemfile.ocra
 - LICENSE.txt
 - README.md
 - Rakefile
@@ -141,23 +156,32 @@ files:
 - lib/spread2rdf.rb
 - lib/spread2rdf/attributes.rb
 - lib/spread2rdf/cli.rb
+- lib/spread2rdf/coord.rb
 - lib/spread2rdf/extensions/roo_xlsm_fix.rb
-- lib/spread2rdf/helper.rb
+- lib/spread2rdf/mapping/cell.rb
+- lib/spread2rdf/mapping/column.rb
+- lib/spread2rdf/mapping/column_block.rb
+- lib/spread2rdf/mapping/default_cell_mappings.rb
+- lib/spread2rdf/mapping/element.rb
+- lib/spread2rdf/mapping/resource.rb
+- lib/spread2rdf/mapping/sheet.rb
+- lib/spread2rdf/mapping/spreadsheet.rb
+- lib/spread2rdf/mapping/statement.rb
+- lib/spread2rdf/mapping/worksheet.rb
 - lib/spread2rdf/namespace.rb
-- lib/spread2rdf/spreadsheet.rb
-- lib/spread2rdf/spreadsheet/column.rb
-- lib/spread2rdf/spreadsheet/column_mapping_context.rb
-- lib/spread2rdf/spreadsheet/coord.rb
-- lib/spread2rdf/spreadsheet/element.rb
-- lib/spread2rdf/spreadsheet/mapping_context.rb
-- lib/spread2rdf/spreadsheet/mapping_dsl.rb
-- lib/spread2rdf/spreadsheet/sheet.rb
-- lib/spread2rdf/spreadsheet/sheet_dsl.rb
-- lib/spread2rdf/spreadsheet/sheet_mapping_context.rb
-- lib/spread2rdf/spreadsheet/sub_sheet.rb
-- lib/spread2rdf/spreadsheet/sub_sheet_mapping_context.rb
-- lib/spread2rdf/spreadsheet/worksheet.rb
+- lib/spread2rdf/roo_helper.rb
+- lib/spread2rdf/schema/column.rb
+- lib/spread2rdf/schema/column_block.rb
+- lib/spread2rdf/schema/element.rb
+- lib/spread2rdf/schema/schema.rb
+- lib/spread2rdf/schema/sheet.rb
+- lib/spread2rdf/schema/sheet_dsl.rb
+- lib/spread2rdf/schema/spreadsheet.rb
+- lib/spread2rdf/schema/spreadsheet_dsl.rb
+- lib/spread2rdf/schema/statement_mapping_schema.rb
+- lib/spread2rdf/schema/worksheet.rb
 - lib/spread2rdf/version.rb
+- ontologies/unit-v1.1.ttl
 - spread2rdf.gemspec
 homepage: http://github.com/marcelotto/spread2rdf
 licenses:
@@ -174,9 +198,9 @@ required_ruby_version: !ruby/object:Gem::Requirement
       version: 1.9.3
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
-  - - ! '>'
+  - - ! '>='
     - !ruby/object:Gem::Version
-      version: 1.3.1
+      version: '0'
 requirements: []
 rubyforge_project: 
 rubygems_version: 2.1.9

data/lib/spread2rdf/helper.rb

@@ -1,14 +0,0 @@
-module Spread2RDF
-  module Helper
-
-  module_function
-
-    # TODO: include this in the MappingContext(s)
-    def resource_name(string)
-      string
-        .gsub(', ', '-')
-        .gsub(' ', '-')
-    end
-
-  end
-end

data/lib/spread2rdf/spreadsheet/column.rb

@@ -1,48 +0,0 @@
-module Spread2RDF
-  class Spreadsheet
-    class Sheet
-      class Column < Element
-
-        attr_reader :coord # this is set by Worksheet#index_columns!
-
-        self.attributes = {
-            predicate:  nil,
-            object:     nil,
-            statement:  nil
-        }
-
-        def initialize(sheet, options = {}, &block)
-          super
-        end
-
-        alias sheet parent
-
-        def worksheet
-          parent = self.parent
-          parent = parent.parent until parent.is_a? Worksheet or parent.nil?
-          parent
-        end
-
-        def map(range, context)
-          #puts "mapping #{self} in #{range} ..."
-          case range
-            when Integer
-              coord = Coord[row: range, column: self.coord]
-              worksheet.cell_mapping[coord.to_sym] ||= mapping =
-                  create_context(context, row: range,
-                                 subject: context.subject, predicate: predicate)
-              mapping.object
-            when Range
-              range.map { |row| self.map(row, context) }
-            else raise ArgumentError
-          end
-        end
-
-        def to_s
-          "#{super} of #{sheet}"
-        end
-
-      end
-    end
-  end
-end

data/lib/spread2rdf/spreadsheet/column_mapping_context.rb

@@ -1,156 +0,0 @@
-module Spread2RDF
-  class Spreadsheet
-    class Sheet
-      class Column
-        class MappingContext < Spreadsheet::MappingContext
-
-          self.attributes = {
-              subject:    nil,
-              predicate:  nil,
-              row:        nil
-          }
-
-          alias column element
-          alias property predicate
-
-          attr_reader :value
-
-          def initialize(sheet, parent_context = nil, attr = {})
-            super
-            @value = cell_value(row: row, column: column.coord)
-            return if @value.blank?
-            statements_to_object
-            worksheet.graph << self.graph
-          end
-
-          def cell_coord
-            Coord[row: row, column: column.coord]
-          end
-
-          def subject
-            @subject or parent_context.try(:subject)
-          end
-
-          def object
-            @object ||= @value && map_to_object(value)
-          end
-
-          def value_of_column(name)
-            other_column = sheet.column[name]
-            raise "couldn't find column #{name} when mapping #{column}" if
-                other_column.nil?
-            cell_value(row: row, column: other_column.coord)
-          end
-
-          def object_of_column(name)
-            other_column = sheet.column[name]
-            raise "couldn't find column #{name} when mapping #{column}" if
-                other_column.nil?
-            cell(row: row, column: other_column.coord).object
-          end
-
-          ######################################################################
-          # Value-to-object mapping
-
-          private
-
-          def map_to_object(value)
-            case object_mapping_mode
-              when :to_string
-                value
-              when :resource_ref
-                resolve_resource_ref
-              when :new_resource
-                create_resource_object
-              when :custom
-                # TODO execute a mapping block in the context of Column::MappingContext
-              else
-                raise 'internal error: unknown column mapping type'
-            end
-          end
-
-
-          def object_mapping_mode
-            case
-              when column.object.nil?         then :to_string
-              when column.object.is_a?(Proc)  then :custom
-              when !column.object[:uri].nil?  then :new_resource
-              when !column.object[:from].nil? then :resource_ref
-              else
-                raise "mapping specification error: don't know how to map #{column}"
-            end
-          end
-
-          def resolve_resource_ref
-            source = column.object[:from]
-            source = { worksheet: source } if source.is_a? Symbol
-            raise ArgumentError, "expecting a Hash as source, but got #{source}" unless source.is_a? Hash
-            source_worksheet = source[:worksheet]
-            source_worksheet = spreadsheet.worksheet[source_worksheet]
-            raise "#{column}: couldn't find source worksheet #{source[:worksheet]}" if source_worksheet.nil?
-            source_predicate = source[:predicate] || RDF::RDFS.label
-            result = source_worksheet.graph.query([nil, source_predicate, value])
-            raise "#{column}: couldn't find a resource for #{value} in #{source_worksheet}" if result.empty?
-            raise "#{column}: found multiple resources for #{value} in #{source_worksheet}: #{result.map(&:subject)}" if result.count > 1
-            result.first.subject
-          end
-
-          # TODO: Should we reuse/share mapping logic with Sheet::MappingContext (#subject etc.)?
-          def create_resource_object
-            case
-              when (column.object.try(:fetch, :uri, nil) || object) == :bnode
-                RDF::Node.new
-              else
-                raise NotImplementedError
-            end
-          end
-
-
-          ######################################################################
-          # Statement mapping
-
-          def statement_mapping_mode
-            case
-              when column.statement == :none then :ignore
-              when column.statement == :none then :ignore
-              when column.predicate.nil?     then :ignore
-              when restriction_mode          then :restriction
-              else                                :default
-            end
-          end
-
-          def restriction_mode
-            restriction_mode = column.statement
-            case restriction_mode
-              when :restriction then RDF::OWL.hasValue
-              when Hash         then restriction_mode[:restriction]
-              else nil
-            end
-          end
-
-          def statements_to_object
-            case statement_mapping_mode
-              when :default
-                statement(subject, predicate, object)
-              when :restriction
-                restriction_class = RDF::Node.new
-                statements(
-                    [ subject, RDF::RDFS.subClassOf, restriction_class ],
-                    [ restriction_class, RDF.type, RDF::OWL.Restriction ],
-                    [ restriction_class, RDF::OWL.onProperty, predicate ],
-                    [ restriction_class, restriction_mode, object ]
-                )
-            end
-            exec(value, &column.block) if column.block
-          end
-
-          def exec(value, &block)
-            #puts "executing block of #{@___column___} in row #{row}"
-            self.instance_exec(value, &block)
-          end
-
-        end
-      end
-    end
-  end
-end

data/lib/spread2rdf/spreadsheet/coord.rb

@@ -1,51 +0,0 @@
-module Spread2RDF
-  class Spreadsheet
-    class Coord < Struct.new(:column, :row)
-      def initialize(*args)
-        case args.length
-          when 2 then super
-          when 1
-            case args = args.first
-              when Hash
-                super(args[:column], args[:row])
-              when Symbol, String
-                coord = args.to_s
-                raise "Invalid cell coordinates #{coord}" unless coord =~ /(\w+)(\d+)/
-                super(Regexp.last_match[1], Regexp.last_match[2].to_i)
-              else raise ArgumentError, "can't handle argument #{args}"
-            end
-          else raise ArgumentError, "too many arguments: #{args}"
-        end
-      end
-
-      def column_as_number
-        Roo::Base.letter_to_number(column)
-      end
-
-      def column_as_index
-        column_as_number - 1
-      end
-
-      def increment_column(count = 1)
-        self.class.increment_column(self.column, count)
-      end
-
-      def to_s
-        "#{column}#{row}"
-      end
-
-      def to_sym
-        to_s.to_sym
-      end
-
-      class << self
-        alias [] new
-
-        def increment_column(column, count=1)
-          Roo::Base.number_to_letter(Roo::Base.letter_to_number(column) + count)
-        end
-      end
-
-    end
-  end
-end

data/lib/spread2rdf/spreadsheet/mapping_context.rb

@@ -1,67 +0,0 @@
-module Spread2RDF
-  class Spreadsheet
-    class MappingContext
-      include Attributes
-
-      self.attributes = {
-      }
-
-      attr_reader :element
-      attr_reader :parent_context
-
-      attr_reader :graph
-
-      def initialize(element, parent_context, attr = {})
-        @element = element
-        @parent_context = parent_context
-        @graph = RDF::Repository.new
-        init_attributes(attr)
-      end
-
-      def sheet
-        @element.sheet
-      end
-
-      def worksheet
-        @element.worksheet
-      end
-
-      def spreadsheet
-        @element.spreadsheet
-      end
-
-      def cell_value(coord)
-        worksheet.cell(coord)
-      end
-
-      def cell(coord)
-        coord = Coord[coord] unless coord.is_a? Coord
-        worksheet.cell_mapping[coord.to_sym]
-      end
-
-      def to_s
-        "#{self.class.name.split('::')[-2..-1].join('::')} of #{element}"
-      end
-
-      ##########################################################################
-      # statement generators
-
-      private
-
-      def add_statement(*args)
-        args = args.first if args.count == 1 and args.first.is_a? Array
-        #puts "adding statement: #{args.inspect}"
-        raise "internal error: trying to add a bad triple with nil value: #{args}" if args.count != 3 or args.one? { |arg| arg.nil? }
-        @graph << RDF::Statement.new(*args)
-      end
-      alias statement add_statement
-
-      def add_statements(*args)
-        args = args.first if args.count == 1 and args.first.is_a? Array
-        args.each { |arg| statement(arg) }
-      end
-      alias statements add_statements
-
-    end
-  end
-end

data/lib/spread2rdf/spreadsheet/mapping_dsl.rb

@@ -1,23 +0,0 @@
-module Spread2RDF
-  class Spreadsheet
-    class MappingDSL
-
-      def initialize(schema)
-        @schema = schema
-      end
-
-      def namespaces(namespaces)
-        namespaces.each { |name, namespace| Namespace[name] = namespace }
-      end
-
-      def worksheet(name, options={}, &block)
-        source_name = options[:source_name] = name
-        name = ( options.delete(:name) || source_name ).to_sym
-        worksheet = @schema.worksheet[name] ||= Worksheet.new(@schema)
-        worksheet.update_attributes options.merge(name: name, source_name: source_name)
-        Sheet::DSL.new(worksheet, &block)
-      end
-
-    end
-  end
-end

data/lib/spread2rdf/spreadsheet/sheet.rb

@@ -1,128 +0,0 @@
-module Spread2RDF
-  class Spreadsheet
-    class Sheet < Element
-
-      self.attributes = {
-          subject:          nil,
-          start:            :A2,
-          row_count_per_resource: nil
-      }
-
-      def initialize(parent, options={}, &block)
-        super(parent, options, &block)
-        @column = {}
-      end
-
-      def worksheet
-        return self if self.is_a? Worksheet
-        parent = self.parent
-        parent = parent.parent until parent.is_a? Worksheet or parent.nil?
-        parent
-      end
-
-      def column(name = nil)
-        return @column if name.nil?
-        name = name.to_sym
-        @column[name] or ( parent.is_a?(Sheet) and parent.column(name) ) or nil
-      end
-
-      def columns
-        @column.values
-      end
-
-      def cell(coord)
-        coord = Coord[coord]
-        spreadsheet.roo.cell(coord.column, coord.row, worksheet.source_name)
-      end
-
-      def start_coord
-        Coord[start]
-      end
-
-      def row_range
-        raise NotImplementedError, 'subclasses of Sheet must implement this method'
-      end
-
-      def column_range
-        first = columns.first.coord
-        first = first.begin if first.is_a? Range
-        last = columns.last.coord
-        last = last.end if last.is_a? Range
-        first .. last
-      end
-
-      def each_column(&block)
-        columns.each do |column|
-          if column.is_a? SubSheet
-            column.each_column(&block)
-          else
-            yield column
-          end
-        end
-      end
-
-      def subject_column
-        #return nil unless subject_mapping_type == :from_column
-        column_name = self.subject.try(:fetch, :column, nil) || :uri
-        @column[column_name]
-      end
-
-      def fix_row_count_per_resource
-        row_count_per_resource or ( !subject_column && 1 ) or nil
-      end
-
-      def add_triple(*args)
-        raise "internal error: trying to add a bad triple with nil value in #{self}: #{args}" if args.count != 3 or args.one? { |arg| arg.nil? }
-        worksheet.graph << RDF::Statement.new(*args)
-      end
-
-      def map(row_range = self.row_range, context = nil)
-        #puts "processing #{self} ..."
-        return [] if row_range.nil?
-        subjects = rows_per_resource(row_range).map do |resource_range|
-          mapping = create_context(context, row_range: resource_range)
-          spreadsheet.worksheet_mapping[worksheet.name] = mapping unless self.is_a? SubSheet
-          mapping.subject
-        end
-        subjects
-      end
-
-    private
-
-      def rows_per_resource(row_range)
-        return [] if row_range.nil?
-        rows = if fix_row_count = fix_row_count_per_resource
-            row_range.find_all do |row|
-              (row - row_range.begin) % fix_row_count == 0
-            end
-          else
-            subject_column_coord = self.subject_column.try(:coord)
-            raise "no subject column for #{self}" if subject_column_coord.blank?
-            row_range.find_all do |row|
-              not cell(row: row, column: subject_column_coord).blank?
-            end
-          end
-        rows_per_resource = []
-        rows.each_with_index do |first_row, i|
-          last_row = (i+1 == rows.count ? row_range.end : rows[i+1]-1)
-          rows_per_resource << Range.new(first_row, last_row)
-        end
-        rows_per_resource
-      end
-
-      def roo_select
-        spreadsheet.roo.default_sheet = worksheet.source_name
-      end
-
-      def roo
-        last_default_sheet = spreadsheet.roo.default_sheet
-        roo_select
-        return nil unless block_given?
-        result = yield
-        spreadsheet.roo.default_sheet = last_default_sheet
-        result
-      end
-
-    end
-  end
-end

data/lib/spread2rdf/spreadsheet/sheet_dsl.rb

@@ -1,34 +0,0 @@
-module Spread2RDF
-  class Spreadsheet
-    class Sheet
-      class DSL
-        def initialize(sheet, &block)
-          @sheet = sheet
-          instance_exec(&block) if block_given?
-        end
-
-        def column(name, options={}, &block)
-          name = name.to_sym
-          column = @sheet.column[name] ||= Column.new(@sheet, &block)
-          column.update_attributes options.merge(name: name)
-          column # TODO: chaining logic ...?
-        end
-
-        def sub_sheet(name, options={}, &block)
-          name = name.to_sym
-          sub_sheet = @sheet.column[name] ||= SubSheet.new(@sheet)
-          sub_sheet.update_attributes options.merge(name: name)
-          Sheet::DSL.new(sub_sheet, &block)
-        end
-        alias column_block sub_sheet
-
-        def cell(coord, options = {}, &block)
-          content = @sheet.cell(coord)
-          content = block.call(content) if block_given?
-          content
-        end
-
-      end
-    end
-  end
-end