ooxml_parser 0.14.1-mingw32

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

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Stuff for help working with paragraph
+  module DocxParagraphHelper
+    # @return [Nil, CommentExtended] extended data for this comment
+    def comment_extend_data
+      return if @paragraph_id.nil?
+
+      root_object.comments_extended.by_id(@paragraph_id)
+    end
+
+    # Temp method to return background color
+    # Need to be compatible with older versions
+    # @return [OoxmlParser::Color]
+    def background_color
+      paragraph_properties.shade.to_background_color
+    end
+  end
+end

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

@@ -0,0 +1,111 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Module for some helpers for ParagraphRun
+  module DocxParagraphRunHelpers
+    # Parse other properties
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [DocxParagraphRun] result of parse
+    def parse_properties(node)
+      self.font_style = FontStyle.new
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'rFonts'
+          node_child.attributes.each do |font_attribute, value|
+            case font_attribute
+            when 'asciiTheme'
+              theme = node_child.attribute('asciiTheme').value
+              next unless root_object.theme
+
+              self.font = root_object.theme.font_scheme.major_font.latin.typeface if theme.include?('major')
+              self.font = root_object.theme.font_scheme.minor_font.latin.typeface if theme.include?('minor')
+            when 'ascii'
+              self.font = value.value
+              break
+            end
+          end
+        when 'sz'
+          self.size = node_child.attribute('val').value.to_i / 2.0
+        when 'highlight'
+          self.highlight = node_child.attribute('val').value
+        when 'vertAlign'
+          self.vertical_align = node_child.attribute('val').value.to_sym
+        when 'effect'
+          self.effect = node_child.attribute('val').value
+        when 'position'
+          self.position = (node_child.attribute('val').value.to_f / (28.0 + 1.0 / 3.0) / 2.0).round(1)
+        when 'em'
+          self.em = node_child.attribute('val').value
+        when 'spacing'
+          self.spacing = (node_child.attribute('val').value.to_f / 566.9).round(1)
+        when 'textFill'
+          self.text_fill = TextFill.new(parent: self).parse(node_child)
+        when 'textOutline'
+          self.text_outline = TextOutline.new(parent: self).parse(node_child)
+        when 'bCs', 'b'
+          font_style.bold = option_enabled?(node_child)
+        when 'iCs', 'i'
+          font_style.italic = option_enabled?(node_child)
+        when 'caps'
+          self.caps = :caps
+        when 'smallCaps'
+          self.caps = :small_caps if option_enabled?(node_child)
+        when 'color'
+          parse_color_tag(node_child)
+        when 'shd'
+          @shade = Shade.new(parent: self).parse(node_child)
+        when 'u', 'uCs'
+          parse_underline(node_child)
+        when 'strike'
+          font_style.strike = :single if option_enabled?(node_child)
+        when 'dstrike'
+          font_style.strike = :double if option_enabled?(node_child)
+        end
+      end
+      self
+    end
+
+    # Temp method to return background color
+    # Need to be compatible with older versions
+    # @return [OoxmlParser::Color]
+    def background_color
+      shade.to_background_color
+    end
+
+    private
+
+    # Parse `color` tag
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [nil]
+    def parse_color_tag(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'val'
+          self.font_color = Color.new(parent: self).parse_hex_string(value.value)
+        when 'themeColor'
+          if root_object.theme && root_object.theme.color_scheme[value.value.to_sym]
+            break if value.value == 'text2' || value.value == 'background2' || value.value.include?('accent') # Don't know why. Just works
+
+            self.font_color = root_object.theme.color_scheme[value.value.to_sym].color.dup
+          end
+        when 'themeShade'
+          font_color.calculate_with_shade!(value.value.hex.to_f / 255.0)
+        when 'themeTint'
+          font_color.calculate_with_tint!(1.0 - (value.value.hex.to_f / 255.0))
+        end
+      end
+    end
+
+    # Parse `u` or `uCs` tag
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [nil]
+    def parse_underline(node)
+      font_style.underlined = Underline.new
+
+      return unless option_enabled?(node)
+
+      font_style.underlined.style = node.attribute('val').value
+      font_style.underlined.color = Color.new(parent: font_style.underlined).parse_hex_string(node.attribute('color').value) unless node.attribute('color').nil?
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_paragraph_run/object/ole_object.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `o:OLEObject` tags
+  class OleObject < OOXMLDocumentObject
+    # @return [String] id of ole_object
+    attr_accessor :id
+    # @return [FileReference] data referenced in object
+    attr_accessor :file_reference
+
+    # Parse OleObject object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [OleObject] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'id'
+          @id = value.value
+        end
+      end
+      @file_reference = FileReference.new(parent: self).parse(node)
+      self
+    end
+  end
+end

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

@@ -0,0 +1,23 @@
+# frozen_string_literal: true
+
+require_relative 'object/ole_object'
+module OoxmlParser
+  # Class for parsing `w:object`
+  class RunObject < OOXMLDocumentObject
+    # @return [OleObject] ole object of object
+    attr_accessor :ole_object
+
+    # Parse RunObject object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [RunObject] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'OLEObject'
+          @ole_object = OleObject.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for working with Shape Margins
+  class ShapeMargins
+    attr_accessor :left, :right, :top
+
+    def initialize(left = nil, right = nil, top = nil)
+      @left = left
+      @right = right
+      @top = top
+    end
+  end
+end

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

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for working with Shape Size
+  class ShapeSize
+    attr_accessor :width, :height
+
+    def initialize(width = nil, height = nil)
+      @width = width
+      @height = height
+    end
+  end
+end

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

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for working with Stroke
+  class Stroke
+    attr_accessor :weight, :color
+
+    def initialize(weight = nil, color = nil)
+      @weight = weight
+      @color = color
+    end
+  end
+end

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

@@ -0,0 +1,19 @@
+# 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
+
+    def initialize(position = nil, margins = ShapeMargins.new, size = ShapeSize.new, z_index = nil, stroke = Stroke.new)
+      @position = position
+      @margins = margins
+      @size = size
+      @z_index = z_index
+      @stroke = stroke
+    end
+  end
+end

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

@@ -0,0 +1,77 @@
+# frozen_string_literal: true
+
+require_relative 'shape/shape_properties'
+module OoxmlParser
+  # Class for parsing `shape`, `rect`, `oval` tags
+  class Shape < OOXMLDocumentObject
+    attr_accessor :type, :properties, :elements
+
+    def initialize(type = nil,
+                   properties = ShapeProperties.new,
+                   elements = [],
+                   parent: nil)
+      @type = type
+      @properties = properties
+      @elements = elements
+      super(parent: parent)
+    end
+
+    # Parse Shape object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Shape] result of parsing
+    def parse(node, type)
+      @type = type
+      node.attributes.each do |key, value|
+        case key
+        when 'fillcolor'
+          @properties.fill_color = Color.new(parent: self)
+                                        .parse_hex_string(value_to_hex(value))
+        when 'strokecolor'
+          @properties.stroke.color = Color.new(parent: self)
+                                          .parse_hex_string(value_to_hex(value))
+        when 'strokeweight'
+          @properties.stroke.weight = value.value
+        when 'style'
+          parse_style(value.value)
+        end
+      end
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'textbox'
+          @elements = TextBox.parse_list(node_child, parent: self)
+        end
+      end
+      self
+    end
+
+    private
+
+    # @param style [String] value to parse
+    def parse_style(style)
+      style.split(';').each do |property|
+        if property.include?('margin-top')
+          @properties.margins.top = property.split(':').last
+        elsif property.include?('margin-left')
+          @properties.margins.left = property.split(':').last
+        elsif property.include?('width')
+          @properties.size.width = property.split(':').last
+        elsif property.include?('height')
+          @properties.size.height = property.split(':').last
+        elsif property.include?('z-index')
+          @properties.z_index = property.split(':').last
+        elsif property.include?('position')
+          @properties.position = property.split(':').last
+        end
+      end
+    end
+
+    # @param value [String] value to extract hex string
+    # @return [String] hex string value
+    def value_to_hex(value)
+      value.to_s
+           .sub('#', '')
+           .split
+           .first
+    end
+  end
+end

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

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Data about `textFill` object
+  class TextFill < OOXMLDocumentObject
+    attr_accessor :color_scheme
+
+    # Parse TextFill object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [TextFill] result of parsing
+    def parse(node)
+      @color_scheme = DocxColorScheme.new(parent: self).parse(node)
+      self
+    end
+  end
+end

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

@@ -0,0 +1,29 @@
+# frozen_string_literal: true
+
+# Class for parsing text outline and its properties
+module OoxmlParser
+  # Class for parsing `w:textOutline`
+  class TextOutline < OOXMLDocumentObject
+    attr_accessor :width, :color_scheme
+
+    def initialize(parent: nil)
+      @width = OoxmlSize.new(0)
+      @color_scheme = :none
+      super
+    end
+
+    # Parse TextOutline object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [TextOutline] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'w'
+          @width = OoxmlSize.new(value.value.to_f, :emu)
+        end
+      end
+      @color_scheme = DocxColorScheme.new(parent: self).parse(node)
+      self
+    end
+  end
+end

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

@@ -0,0 +1,147 @@
+# frozen_string_literal: true
+
+# noinspection RubyTooManyInstanceVariablesInspection
+require_relative 'docx_paragraph_run/docx_paragraph_run_helpers'
+require_relative 'docx_paragraph_run/object'
+require_relative 'docx_paragraph_run/text_outline'
+require_relative 'docx_paragraph_run/text_fill'
+require_relative 'docx_paragraph_run/shape'
+
+module OoxmlParser
+  # Class for working with DocxParagraphRun
+  class DocxParagraphRun < OOXMLDocumentObject
+    include DocxParagraphRunHelpers
+    attr_accessor :number, :font, :vertical_align, :size, :font_color, :font_style, :text, :drawings,
+                  :link, :highlight, :effect, :caps, :w,
+                  :position, :em, :spacing, :break, :touch, :shape, :footnote, :endnote, :fld_char, :style,
+                  :comments, :alternate_content, :page_number, :text_outline, :text_fill
+    # @return [String] type of instruction used for upper level of run
+    # http://officeopenxml.com/WPfieldInstructions.php
+    attr_accessor :instruction
+    # @return [RunProperties] properties of run
+    attr_accessor :run_properties
+    # @return [RunObject] object of run
+    attr_accessor :object
+    # @return [Shade] shade properties
+    attr_accessor :shade
+
+    def initialize(parent: nil)
+      @number = 0
+      @font = 'Arial'
+      @vertical_align = :baseline
+      @size = 11
+      @font_color = Color.new
+      @font_style = FontStyle.new
+      @text = ''
+      @drawings = []
+      @w = false
+      @position = 0.0
+      @spacing = 0.0
+      @break = false
+      @comments = []
+      @page_number = false
+      super
+    end
+
+    # Constructor for copy of object
+    # @param source [DocxParagraphRun] original object
+    # @return [void]
+    def initialize_copy(source)
+      super
+      @drawings = source.drawings.clone
+      @comments = source.comments.clone
+    end
+
+    # @return [Array, nil] Drawings of Run
+    def drawing
+      # TODO: Rewrite all tests without this methos
+      @drawings.empty? ? nil : drawings.first
+    end
+
+    # @return [True, False] is current run empty
+    def empty?
+      text.empty? &&
+        !alternate_content &&
+        !drawing &&
+        !object &&
+        !shape &&
+        !footnote &&
+        !endnote &&
+        !@break
+    end
+
+    # Compare this object to other
+    # @param other [Object] any other object
+    # @return [True, False] result of comparision
+    def ==(other)
+      ignored_attributes = %i[@number @parent]
+      all_instance_variables = instance_variables
+      significan_attribues = all_instance_variables - ignored_attributes
+      significan_attribues.each do |current_attributes|
+        return false unless instance_variable_get(current_attributes) == other.instance_variable_get(current_attributes)
+      end
+      true
+    end
+
+    # Parse object
+    # @param r_tag [Nokogiri::XML:Node] node with DocxParagraphRun
+    # @param char_number [Integer] number of run
+    # @param parent [OOXMLDocumentObject] parent of run
+    # @return [void]
+    def parse(r_tag, char_number, parent: nil)
+      @parent = parent
+      r_tag.xpath('*').each do |node_child|
+        case node_child.name
+        when 'rPr'
+          parse_properties(node_child)
+          @run_properties = RunProperties.new(parent: self).parse(node_child)
+        when 'instrText'
+          if node_child.text.include?('HYPERLINK')
+            hyperlink = Hyperlink.new(node_child.text.sub('HYPERLINK ', '').split(' \\o ').first, node_child.text.sub('HYPERLINK', '').split(' \\o ').last)
+            @link = hyperlink
+          elsif node_child.text[/PAGE\s+\\\*/]
+            @text = '*PAGE NUMBER*'
+          end
+        when 'fldChar'
+          @fld_char = node_child.attribute('fldCharType').value.to_sym
+        when 't'
+          @text += node_child.text
+        when 'noBreakHyphen'
+          @text += '–'
+        when 'tab'
+          @text += "\t"
+        when 'drawing'
+          @drawings << DocxDrawing.new(parent: self).parse(node_child)
+        when 'AlternateContent'
+          @alternate_content = AlternateContent.new(parent: self).parse(node_child)
+        when 'br'
+          if node_child.attribute('type').nil?
+            @break = :line
+            @text += "\r"
+          else
+            case node_child.attribute('type').value
+            when 'page', 'column'
+              @break = node_child.attribute('type').value.to_sym
+            end
+          end
+        when 'footnoteReference'
+          @footnote = HeaderFooter.new(parent: self).parse(node_child)
+        when 'endnoteReference'
+          @endnote = HeaderFooter.new(parent: self).parse(node_child)
+        when 'pict'
+          node_child.xpath('*').each do |pict_node_child|
+            case pict_node_child.name
+            when 'shape'
+              @shape = Shape.new(parent: self).parse(pict_node_child, :shape)
+            when 'rect'
+              @shape = Shape.new(parent: self).parse(pict_node_child, :rectangle)
+            end
+          end
+        when 'object'
+          @object = RunObject.new(parent: self).parse(node_child)
+        end
+      end
+      @number = char_number
+    end
+  end
+end