ooxml_parser 0.3.0 → 0.6.0

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_paragraph_run/shape/shape_properties.rb

@@ -1,7 +1,10 @@
+# frozen_string_literal: true
+
 require_relative 'shape_properties/shape_margins'
 require_relative 'shape_properties/shape_size'
 require_relative 'shape_properties/stroke'
 module OoxmlParser
+  # Class for working with Shape Properties
   class ShapeProperties
     attr_accessor :fill_color, :anchor_x, :anchor_y, :shadow, :margins, :size, :opacity, :position, :z_index, :stroke
 

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_paragraph_run/shape/shape_properties/shape_margins.rb

@@ -1,4 +1,7 @@
+# frozen_string_literal: true
+
 module OoxmlParser
+  # Class for working with Shape Margins
   class ShapeMargins
     attr_accessor :left, :right, :top
 

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_paragraph_run/shape/shape_properties/shape_size.rb

@@ -1,4 +1,7 @@
+# frozen_string_literal: true
+
 module OoxmlParser
+  # Class for working with Shape Size
   class ShapeSize
     attr_accessor :width, :height
 

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_paragraph_run/shape/shape_properties/stroke.rb

@@ -1,4 +1,7 @@
+# frozen_string_literal: true
+
 module OoxmlParser
+  # Class for working with Stroke
   class Stroke
     attr_accessor :weight, :color
 

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_paragraph_run/text_fill.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Data about `textFill` object
   class TextFill < OOXMLDocumentObject

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_paragraph_run/text_outline.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 # Class for parsing text outline and its properties
 module OoxmlParser
   # Class for parsing `w:textOutline`

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/field_simple.rb

@@ -0,0 +1,41 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `w:fldSimple` tags
+  class FieldSimple < OOXMLDocumentObject
+    # @return [String] instruction value
+    attr_reader :instruction
+    # @return [Array<ParagraphRun>] instruction value
+    attr_reader :runs
+
+    def initialize(parent: nil)
+      @runs = []
+      @parent = parent
+    end
+
+    # Parse FieldSimple object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [FieldSimple] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'instr'
+          @instruction = value.value.to_s
+        end
+      end
+
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'r'
+          @runs << ParagraphRun.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+
+    # @return [True, False] is field simple page numbering
+    def page_numbering?
+      instruction.include?('PAGE')
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/frame_properties.rb

@@ -1,4 +1,7 @@
+# frozen_string_literal: true
+
 module OoxmlParser
+  # Class for data with FrameProperties
   class FrameProperties < OOXMLDocumentObject
     attr_accessor :drop_cap, :lines, :wrap, :vertical_anchor, :horizontal_anchor, :width, :height, :height_rule,
                   :horizontal_align, :vertical_align, :anchor_lock, :vertical_space, :horizontal_space,

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/indents.rb

@@ -1,4 +1,7 @@
+# frozen_string_literal: true
+
 module OoxmlParser
+  # Class for working with Indents data
   class Indents < OOXMLDocumentObject
     attr_accessor :first_line_indent, :left_indent, :right_indent, :hanging_indent
 

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/inserted.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'date'
 
 module OoxmlParser

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/sdt/sdt_content.rb

@@ -1,14 +1,13 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for parsing `w:w:sdtContent` tags
   class SDTContent < OOXMLDocumentObject
-    # @return [Array, ParagraphRun] runs of sdt
-    attr_reader :runs
-    # @return [Array, ParagraphRun] paragraphs of sdt
-    attr_reader :paragraphs
+    # @return [Array <ParagraphRun, Table, ParagraphRun>] list of all elements in SDT
+    attr_reader :elements
 
     def initialize(parent: nil)
-      @runs = []
-      @paragraphs = []
+      @elements = []
       @parent = parent
     end
 
@@ -19,12 +18,29 @@ module OoxmlParser
       node.xpath('*').each do |node_child|
         case node_child.name
         when 'p'
-          @paragraphs << DocxParagraph.new(parent: self).parse(node_child)
+          @elements << DocxParagraph.new(parent: self).parse(node_child)
         when 'r'
-          @runs << ParagraphRun.new(parent: self).parse(node_child)
+          @elements << ParagraphRun.new(parent: self).parse(node_child)
+        when 'tbl'
+          @elements << Table.new(parent: self).parse(node_child)
         end
       end
       self
     end
+
+    # @return [Array<DocxParagraphs>] list of paragraphs
+    def paragraphs
+      @elements.select { |obj| obj.is_a?(OoxmlParser::DocxParagraph) }
+    end
+
+    # @return [Array<ParagraphRun>] list of runs
+    def runs
+      @elements.select { |obj| obj.is_a?(OoxmlParser::ParagraphRun) }
+    end
+
+    # @return [Array<Table>] list of tables
+    def tables
+      @elements.select { |obj| obj.is_a?(OoxmlParser::Table) }
+    end
   end
 end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/sdt/sdt_properties.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for parsing `w:sdtPr` tags
   class SDTProperties < OOXMLDocumentObject

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/structured_document_tag.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative 'sdt/sdt_content'
 require_relative 'sdt/sdt_properties'
 module OoxmlParser

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/style_parametres.rb

@@ -1,5 +1,7 @@
-# Style Parameter Data
+# frozen_string_literal: true
+
 module OoxmlParser
+  # Style Parameter Data
   class StyleParametres < OOXMLDocumentObject
     attr_accessor :q_format, :hidden, :name
 

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/header_footer.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class Header Footer classes
   class HeaderFooter < OOXMLDocumentObject
@@ -41,9 +43,10 @@ module OoxmlParser
     def parse(node)
       @id = node.attribute('id').value.to_i
       parse_type(node)
-      doc = Nokogiri::XML(File.open(OOXMLDocumentObject.path_to_folder + xml_path))
+      doc = parse_xml(OOXMLDocumentObject.path_to_folder + xml_path)
       doc.search(xpath_for_search).each do |footnote|
         next unless footnote.attribute('id').value.to_i == @id
+
         paragraph_number = 0
         footnote.xpath('w:p').each do |paragraph|
           @elements << DocumentStructure.default_paragraph_style.dup.parse(paragraph, paragraph_number, DocumentStructure.default_run_style, parent: self)

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/numbering.rb

@@ -1,7 +1,10 @@
+# frozen_string_literal: true
+
 require_relative 'numbering/abstract_numbering'
 require_relative 'numbering/numbering_definition'
 
 module OoxmlParser
+  # Class for data of Numbering
   class Numbering < OOXMLDocumentObject
     # @return [Array, AbstractNumbering] abstract numbering list
     attr_accessor :abstract_numbering_list
@@ -18,10 +21,12 @@ module OoxmlParser
       abstract_num_id = nil
       @numbering_definition_list.each do |num_def|
         next unless num_id == num_def.id
+
         abstract_num_id = num_def.abstract_numbering_id.value
         break
       end
       return nil if abstract_num_id.nil?
+
       @abstract_numbering_list.each do |abstract_num_item|
         return abstract_num_item if abstract_num_id == abstract_num_item.id
       end
@@ -30,7 +35,8 @@ module OoxmlParser
     def parse
       numbering_xml = OOXMLDocumentObject.path_to_folder + 'word/numbering.xml'
       return nil unless File.exist?(numbering_xml)
-      node = Nokogiri::XML(File.open(numbering_xml), 'r:UTF-8')
+
+      node = parse_xml(File.open(numbering_xml))
       node.xpath('w:numbering/*').each do |numbering_child_node|
         case numbering_child_node.name
         when 'abstractNum'

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/numbering/abstract_numbering.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative 'abstract_numbering/multilevel_type'
 require_relative 'abstract_numbering/numbering_level'
 module OoxmlParser

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/numbering/abstract_numbering/multilevel_type.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # This element specifies the type of numbering defined by a given abstract numbering type
   class MultilevelType < OOXMLDocumentObject

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/numbering/abstract_numbering/numbering_level.rb

@@ -1,7 +1,8 @@
+# frozen_string_literal: true
+
 require_relative 'numbering_level/level_justification'
 require_relative 'numbering_level/level_text'
 require_relative 'numbering_level/numbering_format'
-require_relative 'numbering_level/start'
 require_relative 'numbering_level/suffix'
 module OoxmlParser
   # This element specifies the appearance and behavior of a numbering level
@@ -43,7 +44,7 @@ module OoxmlParser
       node.xpath('*').each do |num_level_child|
         case num_level_child.name
         when 'start'
-          @start = Start.new(parent: self).parse(num_level_child)
+          @start = ValuedChild.new(:integer, parent: self).parse(num_level_child)
         when 'numFmt'
           @numbering_format = NumberingFormat.new(parent: self).parse(num_level_child)
         when 'lvlText'

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/numbering/abstract_numbering/numbering_level/level_justification.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for storing Level Justification, `lvlJc` tag
   class LevelJustification < OOXMLDocumentObject

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/numbering/abstract_numbering/numbering_level/level_text.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for storing Level Text, `lvlText` tag
   class LevelText < OOXMLDocumentObject

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/numbering/abstract_numbering/numbering_level/numbering_format.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for storing Numbering Format data, `numFmt` tag
   class NumberingFormat < OOXMLDocumentObject

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/numbering/abstract_numbering/numbering_level/suffix.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for storing Suffix `w:suff`
   # This element specifies the content which

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/numbering/numbering_definition.rb

@@ -1,11 +1,12 @@
-require_relative 'numbering_definition/abstract_numbering_id'
+# frozen_string_literal: true
+
 module OoxmlParser
   # This element specifies a unique instance of numbering information that can be referenced by zero or more
   # paragraphs within the parent WordprocessingML document.
   class NumberingDefinition < OOXMLDocumentObject
     # @return [Integer] num id
     attr_accessor :id
-    # @return [AbstractNumberingId] abstract numbering id
+    # @return [ValuedChild] abstract numbering id
     attr_accessor :abstract_numbering_id
 
     # Parse NumberingDefinition data
@@ -22,7 +23,7 @@ module OoxmlParser
       node.xpath('*').each do |numbering_child_node|
         case numbering_child_node.name
         when 'abstractNumId'
-          @abstract_numbering_id = AbstractNumberingId.new(parent: self).parse(numbering_child_node)
+          @abstract_numbering_id = ValuedChild.new(:integer, parent: self).parse(numbering_child_node)
         end
       end
       self

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/page_properties/columns.rb

@@ -1,11 +1,16 @@
+# frozen_string_literal: true
+
 require_relative 'columns/column'
 module OoxmlParser
+  # Class for data of Columns
   class Columns < OOXMLDocumentObject
-    attr_accessor :separator
     attr_accessor :count
     attr_accessor :equal_width
+
     alias equal_width? equal_width
     attr_accessor :column_array
+    # @return [Boolean] Draw Line Between Columns
+    attr_reader :separator
     # @return [OoxmlSize] space between columns
     attr_accessor :space
 
@@ -28,7 +33,7 @@ module OoxmlParser
         when 'num'
           @count = value.value.to_i
         when 'sep'
-          @separator = value.value
+          @separator = option_enabled?(node, 'sep')
         when 'equalWidth'
           @equal_width = option_enabled?(node, 'equalWidth')
         when 'space'

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/page_properties/columns/column.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for parsing `w:col` object
   class Column < OOXMLDocumentObject

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/page_properties/document_grid.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for parsing `docGrid` tags
   class DocumentGrid < OOXMLDocumentObject

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/page_properties/footnote_properties.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for parsing `footnotePr` tags
   class FootnoteProperties < OOXMLDocumentObject

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/page_properties/note.rb

@@ -1,4 +1,7 @@
+# frozen_string_literal: true
+
 module OoxmlParser
+  # Class with data of Note
   class Note < OOXMLDocumentObject
     attr_accessor :type, :elements, :assigned_to
 
@@ -11,13 +14,11 @@ module OoxmlParser
       note.type = params[:type]
       note.assigned_to = params[:assigned_to]
       note.parent = params[:parent]
-      doc = Nokogiri::XML(File.open(OOXMLDocumentObject.path_to_folder + "word/#{params[:target]}"))
+      doc = note.parse_xml(note.file_path(params[:target]))
       if note.type.include?('footer')
         xpath_note = '//w:ftr'
       elsif note.type.include?('header')
         xpath_note = '//w:hdr'
-      else
-        raise "Cannot parse unknown Note type: #{type}"
       end
       doc.search(xpath_note).each do |ftr|
         number = 0
@@ -28,21 +29,19 @@ module OoxmlParser
           elsif sub_element.name == 'tbl'
             note.elements << Table.new(parent: note).parse(sub_element, number)
             number += 1
-          elsif sub_element.name == 'std'
-            sub_element.xpath('w:p').each do |p|
-              note.elements << params[:default_paragraph].copy.parse(p, number, params[:default_character])
-              number += 1
-            end
-            sub_element.xpath('w:sdtContent').each do |sdt_content|
-              sdt_content.xpath('w:p').each do |p|
-                note.elements << params[:default_paragraph].copy.parse(p, number, params[:default_character])
-                number += 1
-              end
-            end
           end
         end
       end
       note
     end
+
+    # @param target [String] name of target
+    # @return [String] path to note xml file
+    def file_path(target)
+      file = "#{OOXMLDocumentObject.path_to_folder}word/#{target}"
+      return file if File.exist?(file)
+
+      "#{OOXMLDocumentObject.path_to_folder}#{target}" unless File.exist?(file)
+    end
   end
 end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/page_properties/page_margins.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for parsing `pgMar` tags
   class PageMargins < OOXMLDocumentObject
@@ -17,23 +19,23 @@ module OoxmlParser
     # Parse BordersProperties
     # @param [Nokogiri::XML:Element] node with PageMargins
     # @return [PageMargins] value of PageMargins
-    def parse(node)
+    def parse(node, unit = :dxa)
       node.attributes.each do |key, value|
         case key
         when 'top'
-          @top = OoxmlSize.new(value.value.to_f)
+          @top = OoxmlSize.new(value.value.to_f, unit)
         when 'left'
-          @left = OoxmlSize.new(value.value.to_f)
+          @left = OoxmlSize.new(value.value.to_f, unit)
         when 'right'
-          @right = OoxmlSize.new(value.value.to_f)
+          @right = OoxmlSize.new(value.value.to_f, unit)
         when 'bottom'
-          @bottom = OoxmlSize.new(value.value.to_f)
+          @bottom = OoxmlSize.new(value.value.to_f, unit)
         when 'header'
-          @header = OoxmlSize.new(value.value.to_f)
+          @header = OoxmlSize.new(value.value.to_f, unit)
         when 'footer'
-          @footer = OoxmlSize.new(value.value.to_f)
+          @footer = OoxmlSize.new(value.value.to_f, unit)
         when 'gutter'
-          @gutter = OoxmlSize.new(value.value.to_f)
+          @gutter = OoxmlSize.new(value.value.to_f, unit)
         end
       end
       self