ooxml_parser 0.1.1

data/lib/ooxml_parser/common_parser/common_data/table/table_properties.rb

@@ -0,0 +1,77 @@
+# Table Properties Data
+require_relative 'properties/table_cell_margin'
+require_relative 'properties/table_look'
+require_relative 'properties/table_style'
+require_relative 'properties/table_position'
+module OoxmlParser
+  class TableProperties < OOXMLDocumentObject
+    attr_accessor :jc, :table_width, :shd, :table_borders, :table_properties, :table_positon, :table_cell_margin, :table_indent, :stretching, :table_style, :row_banding_size,
+                  :column_banding_size, :table_look, :grid_column, :right_to_left, :style
+
+    alias table_properties table_positon
+
+    def initialize
+      @table_indent = nil
+      @table_cell_margin = nil
+      @jc = :left
+      @shd = nil
+      @stretching = true
+      @table_borders = Borders.new
+      @table_properties = nil
+      @table_width = nil
+      @grid_column = nil
+      @right_to_left = nil
+      @style = nil
+    end
+
+    def copy
+      table = TableProperties.new
+      table.jc = @jc
+      table.table_width = @table_width
+      table.shd = @shd
+      table.stretching = @stretching
+      table.table_borders = @table_borders.copy
+      table.table_properties = @table_properties
+      table.table_cell_margin = @table_cell_margin
+      table.table_indent = @table_indent
+      table.grid_column = @grid_column
+      table.right_to_left = @right_to_left
+      table.style = style
+      table
+    end
+
+    def self.parse(table_properties_node)
+      table_properties = TableProperties.new
+      table_properties_node.xpath('*').each do |table_props_node_child|
+        case table_props_node_child.name
+        when 'tableStyleId'
+          table_properties.style = TableStyle.parse(style_id: table_props_node_child.text)
+        when 'tblStyle'
+          style = DocxParagraphRun.get_style_by_id(table_props_node_child.attribute('val').value)
+          next if style.nil?
+          table_properties.table_style = TableStyle.parse(table_style_node: style)
+        when 'tblW'
+          table_properties.table_width = table_props_node_child.attribute('w').text.to_f / 567.0
+        when 'jc'
+          table_properties.jc = table_props_node_child.attribute('val').text.to_sym
+        when 'shd'
+          unless table_props_node_child.attribute('fill').nil?
+            background_color = Color.from_int16(table_props_node_child.attribute('fill').value)
+            table_properties.shd = background_color
+          end
+        when 'tblLook'
+          table_properties.table_look = TableLook.parse(table_props_node_child)
+        when 'tblInd'
+          table_properties.table_indent = table_props_node_child.attribute('w').text.to_f / 567.0
+        when 'tblpPr'
+          table_properties.table_positon = TablePosition.parse(table_props_node_child)
+        when 'tblCellMar'
+          table_properties.table_cell_margin = TableMargins.parse(table_props_node_child)
+        end
+      end
+      table_properties.table_look = TableLook.parse(table_properties_node) if table_properties.table_look.nil?
+      table_properties.right_to_left = OOXMLDocumentObject.option_enabled?(table_properties_node, 'rtl')
+      table_properties
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/table.rb

@@ -0,0 +1,41 @@
+require_relative 'table/row/row'
+require_relative 'table/table_properties'
+require_relative 'table/table_grid'
+require_relative 'table/margins/table_margins'
+require_relative 'table/margins/paragraph_margins'
+module OoxmlParser
+  class Table
+    attr_accessor :grid, :rows, :properties, :number
+
+    def initialize(rows = [])
+      @rows = rows
+    end
+
+    alias table_properties properties
+
+    def self.parse(table_node, number = 0, default_table_properties = TableProperties.new, default_paragraph = DocxParagraph.new, default_run = DocxParagraphRun.new)
+      table_properties = default_table_properties.copy
+      table_properties.jc = :left
+      table_paragraph = default_paragraph.copy
+      table_character = default_run.copy
+      table = Table.new
+      table_node.xpath('*').each do |table_node_child|
+        case table_node_child.name
+        when 'tblGrid'
+          table.grid = TableGrid.new
+          table_node_child.xpath('gridCol').each do |grid_col_node|
+            table.grid.columns << (grid_col_node.attribute('w').value.to_f / 360_000.0).round(2)
+          end
+        when 'tr'
+          table.rows << TableRow.parse(table_node_child)
+        when 'tblPr'
+          table.properties = TableProperties.parse(table_node_child)
+        when ''
+          DocxParagraph.parse_paragraph_style(table_node_child, table_paragraph, table_character)
+        end
+      end
+      table.number = number
+      table
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/underline.rb

@@ -0,0 +1,40 @@
+# Class for working Underline style parameter
+module OoxmlParser
+  class Underline
+    attr_accessor :style, :color
+
+    def initialize(style = :none, color = nil)
+      @style = (style == 'single') ? :single : style
+      @color = color
+    end
+
+    def ==(other)
+      if other.is_a? Underline
+        @style.to_sym == other.style.to_sym && @color == other.color
+      elsif other.is_a? Symbol
+        @style.to_sym == other
+      else
+        false
+      end
+    end
+
+    def to_s
+      if @color.nil?
+        @style.to_s
+      else
+        "#{@style} #{@color}"
+      end
+    end
+
+    def self.parse(attribute_value)
+      underline = Underline.new
+      case attribute_value
+      when 'sng'
+        underline.style = :single
+      when 'none'
+        underline.style = :none
+      end
+      underline
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_document_structure.rb

@@ -0,0 +1,7 @@
+module OoxmlParser
+  # Common document structure for DOCX, XLSX, PPTX file
+  class CommonDocumentStructure < OOXMLDocumentObject
+    # @return [String] path to original file
+    attr_accessor :file_path
+  end
+end

data/lib/ooxml_parser/configuration.rb

@@ -0,0 +1,20 @@
+module OoxmlParser
+  class Configuration
+    attr_accessor :units
+
+    def initialize
+      @units = :centimeters
+    end
+
+    # @return [Float] by which divide units values
+    def units_delimiter
+      return (20 * 635.0) if @units == :points
+      return (566.929 * 635.0) if @units == :centimeters
+      return 1 if @units == :dxa
+      return 1440 if @units == :inches
+      return (1.0 / 635.0) if @units == :emu
+      warn "Cannot recognize #{@units} unit. Will use dxa by default"
+      1
+    end
+  end
+end

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

@@ -0,0 +1,28 @@
+# Comment Data
+module OoxmlParser
+  class Comment < OOXMLDocumentObject
+    attr_accessor :id, :paragraphs
+
+    def initialize(id = nil, paragraphs = [])
+      @id = id
+      @paragraphs = paragraphs
+    end
+
+    def self.parse_list
+      comments = []
+      comments_filename = "#{OOXMLDocumentObject.path_to_folder}/#{OOXMLDocumentObject.root_subfolder}/comments.xml"
+      return [] unless File.exist?(comments_filename)
+      doc = Nokogiri::XML(File.open(comments_filename))
+      doc.search('//w:comments').each do |document|
+        document.xpath('w:comment').each do |comment_tag|
+          comment = Comment.new(comment_tag.attribute('id').value)
+          comment_tag.xpath('w:p').each_with_index do |p, index|
+            comment.paragraphs << DocxParagraph.parse(p, index)
+          end
+          comments << comment.dup
+        end
+      end
+      comments
+    end
+  end
+end

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

@@ -0,0 +1,10 @@
+module OoxmlParser
+  class DocumentBackground
+    attr_accessor :color1, :size, :color2, :image, :type
+
+    def initialize(color1 = nil, type = 'simple')
+      @color1 = color1
+      @type = type
+    end
+  end
+end

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

@@ -0,0 +1,21 @@
+# Document Properties
+module OoxmlParser
+  class DocumentProperties < OOXMLDocumentObject
+    attr_accessor :pages, :words
+
+    def initialize(pages = nil, words = nil)
+      @pages = pages
+      @words = words
+    end
+
+    # Parse Document properties
+    # @return [DocumentProperties]
+    def self.parse
+      properties = DocumentProperties.new
+      doc_props = XmlSimple.xml_in(File.open("#{OOXMLDocumentObject.path_to_folder}docProps/app.xml"))
+      properties.pages = doc_props['Pages'].first.to_i unless doc_props['Pages'].nil?
+      properties.words = doc_props['Words'].first.to_i unless doc_props['Words'].nil?
+      properties
+    end
+  end
+end

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

@@ -0,0 +1,70 @@
+module OoxmlParser
+  # Class for describing styles containing in +styles.xml+
+  class DocumentStyle < OOXMLDocumentObject
+    # @return [Symbol] Type of style (+:paragraph+ or +:table+)
+    attr_accessor :type
+    # @return [FixNum] number of style
+    attr_accessor :style_id
+    # @return [String] name of style
+    attr_accessor :name
+    # @return [FixNum] id of style on which this style is based
+    attr_accessor :based_on
+    # @return [FixNum] id of next style
+    attr_accessor :next_style
+    # @return [DocxParagraphRun] run properties
+    attr_accessor :run_properties
+    # @return [DocxParagraph] run properties
+    attr_accessor :paragraph_properties
+    # @return [True, False] Latent Style Primary Style Setting
+    # Used to determine if current style is visible in style list in editors
+    # According to http://www.wordarticles.com/Articles/WordStyles/LatentStyles.php
+    attr_accessor :q_format
+    alias visible? q_format
+
+    def initialize
+      @q_format = false
+    end
+
+    # Parse single document style
+    # @return [DocumentStyle]
+    def self.parse(node)
+      document_style = DocumentStyle.new
+      node.attributes.each do |key, value|
+        case key
+        when 'type'
+          document_style.type = value.value.to_sym
+        when 'styleId'
+          document_style.style_id = value.value.to_i
+        end
+      end
+      node.xpath('*').each do |subnode|
+        case subnode.name
+        when 'name'
+          document_style.name = subnode.attribute('val').value
+        when 'basedOn'
+          document_style.based_on = subnode.attribute('val').value.to_i
+        when 'next'
+          document_style.next_style = subnode.attribute('val').value.to_i
+        when 'rPr'
+          document_style.run_properties = DocxParagraphRun.parse(subnode)
+        when 'pPr'
+          document_style.paragraph_properties = DocxParagraph.parse_paragraph_style(subnode)
+        when 'qFormat'
+          document_style.q_format = true
+        end
+      end
+      document_style
+    end
+
+    # Parse all document style list
+    # @return [Array, DocumentStyle]
+    def self.parse_list
+      styles_array = []
+      doc = Nokogiri::XML(File.open(OOXMLDocumentObject.path_to_folder + 'word/styles.xml'))
+      doc.search('//w:style').each do |style|
+        styles_array << DocumentStyle.parse(style)
+      end
+      styles_array
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/accent.rb

@@ -0,0 +1,11 @@
+# Accent Data
+module OoxmlParser
+  class Accent
+    attr_accessor :symbol, :element
+
+    def initialize(symbol = nil, element = nil)
+      @symbol = symbol
+      @element = element
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/bar.rb

@@ -0,0 +1,10 @@
+module OoxmlParser
+  class Bar
+    attr_accessor :position, :element
+
+    def initialize(position = 'bottom', element = nil)
+      @position = position
+      @element = element
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/box.rb

@@ -0,0 +1,11 @@
+# Box data
+module OoxmlParser
+  class Box
+    attr_accessor :borders, :element
+
+    def initialize(borders = false, element = nil)
+      @borders = borders
+      @element = element
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/delimeter.rb

@@ -0,0 +1,12 @@
+# Delimiter Data
+module OoxmlParser
+  class Delimiter
+    attr_accessor :begin_character, :value, :end_character
+
+    def initialize(begin_character = '(', value = nil, end_character = ')')
+      @begin_character = begin_character
+      @value = value
+      @end_character = end_character
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/fraction.rb

@@ -0,0 +1,11 @@
+# Fraction Data
+module OoxmlParser
+  class Fraction
+    attr_accessor :numerator, :denominator
+
+    def initialize(numerator = nil, denominator = nil)
+      @numerator = numerator
+      @denominator = denominator
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/function.rb

@@ -0,0 +1,11 @@
+# Function Data
+module OoxmlParser
+  class Function
+    attr_accessor :name, :argument
+
+    def initialize(name = nil, argument = nil)
+      @name = name
+      @argument = argument
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/group_char.rb

@@ -0,0 +1,13 @@
+# Group Char Data
+module OoxmlParser
+  class GroupChar
+    attr_accessor :symbol, :position, :vertical_align, :element
+
+    def initialize(symbol = nil, position = nil, vertical_align = nil, element = nil)
+      @symbol = symbol
+      @position = position
+      @vertical_align = vertical_align
+      @element = element
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/index.rb

@@ -0,0 +1,12 @@
+# Index data
+module OoxmlParser
+  class Index
+    attr_accessor :value, :top_index, :bottom_index
+
+    def initialize(value = nil, bottom_index = nil, top_index = nil)
+      @value = value
+      @bottom_index = bottom_index
+      @top_index = top_index
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/limit.rb

@@ -0,0 +1,12 @@
+# Limit Data
+module OoxmlParser
+  class Limit
+    attr_accessor :type, :element, :limit
+
+    def initialize(type = nil, element = nil, limit = nil)
+      @type = type
+      @element = element
+      @limit = limit
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/matrix/matrix_row.rb

@@ -0,0 +1,10 @@
+# Matrix Row data
+module OoxmlParser
+  class MatrixRow
+    attr_accessor :columns
+
+    def initialize(columns_count = 1)
+      @columns = Array.new(columns_count)
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/matrix.rb

@@ -0,0 +1,10 @@
+require_relative 'matrix/matrix_row'
+module OoxmlParser
+  class Matrix
+    attr_accessor :rows
+
+    def initialize(rows = [])
+      @rows = rows
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/operator.rb

@@ -0,0 +1,13 @@
+# Operator Data
+module OoxmlParser
+  class Operator
+    attr_accessor :operator, :bottom_value, :top_value, :value
+
+    def initialize(operator = nil, bottom_value = nil, top_value = nil, value = nil)
+      @operator = operator
+      @bottom_value = bottom_value
+      @top_value = top_value
+      @value = value
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/radical.rb

@@ -0,0 +1,11 @@
+# Radical Data
+module OoxmlParser
+  class Radical
+    attr_accessor :degree, :value
+
+    def initialize(degree = 2, value = nil)
+      @degree = degree
+      @value = value
+    end
+  end
+end