ooxml_parser 0.1.2 → 0.2.0

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

@@ -0,0 +1,20 @@
+module OoxmlParser
+  # Class for parsing `settings.xml` file
+  class DocumentSettings < OOXMLDocumentObject
+    # @return [OoxmlSize] size of default tab
+    attr_accessor :default_tab_stop
+
+    # Parse Settings object
+    # @return [DocumentSettings] result of parsing
+    def parse
+      doc = Nokogiri::XML(File.open(OOXMLDocumentObject.path_to_folder + 'word/settings.xml'))
+      doc.xpath('w:settings/*').each do |node_child|
+        case node_child.name
+        when 'defaultTabStop'
+          @default_tab_stop = OoxmlSize.new.parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -0,0 +1,18 @@
+module OoxmlParser
+  # method to help to work with DocumentStructure
+  module DocumentStructureHelpers
+    # @return [True, false] if structure contain any user data
+    def with_data?
+      return true if @elements.length > 1
+      @elements.each do |current_element|
+        return true if current_element.with_data?
+      end
+      @notes.each do |current_note|
+        current_note.elements.each do |note_element|
+          return true if note_element.with_data?
+        end
+      end
+      false
+    end
+  end
+end

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

@@ -0,0 +1,37 @@
+module OoxmlParser
+  # Helper methods for working with Style List
+  module DocumentStyleHelper
+    # Return document style by its name
+    # @param name [String] name of style
+    # @return [DocumentStyle, nil]
+    def document_style_by_name(name)
+      root_object.document_styles.each do |style|
+        return style if style.name == name
+      end
+      nil
+    end
+
+    # Return document style by its id
+    # @param id [String] id of style
+    # @return [DocumentStyle, nil]
+    def document_style_by_id(id)
+      root_object.document_styles.each do |style|
+        return style if style.style_id == id
+      end
+      nil
+    end
+
+    # Return document style which is based on
+    # @return [DocumentStyle]
+    def based_on_style
+      document_style_by_id(@based_on)
+    end
+
+    # Check if style exists in current document
+    # @param name [String] name of style
+    # @return [True, False]
+    def style_exist?(name)
+      !document_style_by_name(name).nil?
+    end
+  end
+end

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

@@ -1,6 +1,9 @@
+require_relative 'document_style/document_style_helper'
 module OoxmlParser
   # Class for describing styles containing in +styles.xml+
   class DocumentStyle < OOXMLDocumentObject
+    include TableStylePropertiesHelper
+    include DocumentStyleHelper
     # @return [Symbol] Type of style (+:paragraph+ or +:table+)
     attr_accessor :type
     # @return [FixNum] number of style
@@ -15,54 +18,80 @@ module OoxmlParser
     attr_accessor :run_properties
     # @return [DocxParagraph] run properties
     attr_accessor :paragraph_properties
+    # @return [TableProperties] properties of table
+    attr_accessor :table_properties
+    # @return [Array, TableStyleProperties] list of table style properties
+    attr_accessor :table_style_properties_list
+    # @return [TableRowProperties] properties of table row
+    attr_accessor :table_row_properties
+    # @return [CellProperties] properties of table cell
+    attr_accessor :table_cell_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
+    def initialize(parent: nil)
       @q_format = false
+      @table_style_properties_list = []
+      @parent = parent
+    end
+
+    def to_s
+      "Table style properties list: #{@table_style_properties_list.join(',')}"
+    end
+
+    def inspect
+      to_s
     end
 
     # Parse single document style
     # @return [DocumentStyle]
-    def self.parse(node)
-      document_style = DocumentStyle.new
+    def parse(node)
       node.attributes.each do |key, value|
         case key
         when 'type'
-          document_style.type = value.value.to_sym
+          @type = value.value.to_sym
         when 'styleId'
-          document_style.style_id = value.value.to_i
+          @style_id = value.value
         end
       end
       node.xpath('*').each do |subnode|
         case subnode.name
         when 'name'
-          document_style.name = subnode.attribute('val').value
+          @name = subnode.attribute('val').value
         when 'basedOn'
-          document_style.based_on = subnode.attribute('val').value.to_i
+          @based_on = subnode.attribute('val').value
         when 'next'
-          document_style.next_style = subnode.attribute('val').value.to_i
+          @next_style = subnode.attribute('val').value
         when 'rPr'
-          document_style.run_properties = DocxParagraphRun.parse(subnode)
+          @run_properties = DocxParagraphRun.new.parse_properties(subnode)
         when 'pPr'
-          document_style.paragraph_properties = DocxParagraph.parse_paragraph_style(subnode)
+          @paragraph_properties = DocxParagraph.new(parent: self).parse_paragraph_style(subnode)
+        when 'tblPr'
+          @table_properties = TableProperties.new(parent: self).parse(subnode)
+        when 'trPr'
+          @table_row_properties = TableRowProperties.new(parent: self).parse(subnode)
+        when 'tcPr'
+          @table_cell_properties = CellProperties.new(parent: self).parse(subnode)
+        when 'tblStylePr'
+          @table_style_properties_list << TableStyleProperties.new(parent: self).parse(subnode)
         when 'qFormat'
-          document_style.q_format = true
+          @q_format = true
         end
       end
-      document_style
+      fill_empty_table_styles
+      self
     end
 
     # Parse all document style list
     # @return [Array, DocumentStyle]
-    def self.parse_list
+    def self.parse_list(parent)
       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)
+        styles_array << DocumentStyle.new(parent: parent).parse(style)
       end
       styles_array
     end

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

@@ -0,0 +1,21 @@
+module OoxmlParser
+  # Class for parsing `w:bookmarkStart`, `w:bookmarkEnd` tags
+  class Bookmark < OOXMLDocumentObject
+    attr_accessor :id, :name
+
+    # Parse Bookmark object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Bookmark] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'id'
+          @id = value.value
+        when 'name'
+          @name = value.value
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -1,11 +1,25 @@
-# Accent Data
 module OoxmlParser
-  class Accent
+  # Class for `acc` data
+  class Accent < OOXMLDocumentObject
     attr_accessor :symbol, :element
 
-    def initialize(symbol = nil, element = nil)
-      @symbol = symbol
-      @element = element
+    # Parse Accent object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Accent] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'accPr'
+          node_child.xpath('*').each do |node_child_child|
+            case node_child_child.name
+            when 'chr'
+              @symbol = node_child_child.attribute('val').value
+            end
+          end
+        end
+      end
+      @element = DocxFormula.new(parent: self).parse(node)
+      self
     end
   end
 end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/argument_properties/argument_size.rb

@@ -0,0 +1,20 @@
+module OoxmlParser
+  # Class for parsing `m:argSz` object
+  class ArgumentSize < OOXMLDocumentObject
+    # @return [String] value of size
+    attr_accessor :value
+
+    # Parse ArgumentSize
+    # @param [Nokogiri::XML:Node] node with ArgumentSize
+    # @return [ArgumentSize] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'val'
+          @value = value.value.to_f
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -0,0 +1,21 @@
+require_relative 'argument_properties/argument_size'
+module OoxmlParser
+  # Class for parsing `m:argPr` object
+  class ArgumentProperties < OOXMLDocumentObject
+    # @return [ArgumentSize] size of argument
+    attr_accessor :argument_size
+
+    # Parse ArgumentProperties
+    # @param [Nokogiri::XML:Node] node with ArgumentProperties
+    # @return [ArgumentProperties] result of parsing
+    def parse(node)
+      node.xpath('*').each do |argument_child|
+        case argument_child.name
+        when 'argSz'
+          @argument_size = ArgumentSize.new(parent: self).parse(argument_child)
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -1,10 +1,30 @@
 module OoxmlParser
-  class Bar
+  # Class for `bar` data
+  class Bar < OOXMLDocumentObject
     attr_accessor :position, :element
 
-    def initialize(position = 'bottom', element = nil)
-      @position = position
-      @element = element
+    def initialize(parent: nil)
+      @position = :bottom
+      @parent = parent
+    end
+
+    # Parse Bar object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Bar] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'barPr'
+          node_child.xpath('*').each do |node_child_child|
+            case node_child_child.name
+            when 'pos'
+              @position = node_child_child.attribute('val').value.to_sym
+            end
+          end
+        end
+      end
+      @element = DocxFormula.new(parent: self).parse(node)
+      self
     end
   end
 end

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

@@ -1,11 +1,20 @@
-# Box data
 module OoxmlParser
-  class Box
+  # Class for `box`, `borderBox` data
+  class Box < OOXMLDocumentObject
     attr_accessor :borders, :element
 
-    def initialize(borders = false, element = nil)
-      @borders = borders
-      @element = element
+    def initialize(parent: nil)
+      @borders = false
+      @parent = parent
+    end
+
+    # Parse Box object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Box] result of parsing
+    def parse(node)
+      @borders = true if node.name == 'borderBox'
+      @element = MathRun.new(parent: self).parse(node)
+      self
     end
   end
 end

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

@@ -1,12 +1,33 @@
-# Delimiter Data
 module OoxmlParser
-  class Delimiter
+  # Class for `d` data
+  class Delimiter < OOXMLDocumentObject
     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
+    def initialize(parent: nil)
+      @begin_character = '('
+      @end_character = ')'
+      @parent = parent
+    end
+
+    # Parse Delimiter object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Delimiter] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'dPr'
+          node_child.xpath('*').each do |node_child_child|
+            case node_child_child.name
+            when 'begChr'
+              @begin_character = node_child_child.attribute('val').value
+            when 'endChr'
+              @end_character = node_child_child.attribute('val').value
+            end
+          end
+        end
+      end
+      @value = DocxFormula.new(parent: self).parse(node)
+      self
     end
   end
 end

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

@@ -1,11 +1,21 @@
-# Fraction Data
 module OoxmlParser
-  class Fraction
+  # Class for 'f' data
+  class Fraction < OOXMLDocumentObject
     attr_accessor :numerator, :denominator
 
-    def initialize(numerator = nil, denominator = nil)
-      @numerator = numerator
-      @denominator = denominator
+    # Parse Fraction object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Fraction] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'num'
+          @numerator = DocxFormula.new(parent: self).parse(node_child)
+        when 'den'
+          @denominator = DocxFormula.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_formula/function.rb

@@ -1,11 +1,20 @@
-# Function Data
 module OoxmlParser
-  class Function
+  # Class for `func` data
+  class Function < OOXMLDocumentObject
     attr_accessor :name, :argument
 
-    def initialize(name = nil, argument = nil)
-      @name = name
-      @argument = argument
+    # Parse Function object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Function] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'fName'
+          @name = DocxFormula.new(parent: self).parse(node_child)
+        end
+      end
+      @argument = DocxFormula.new(parent: self).parse(node)
+      self
     end
   end
 end

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

@@ -1,13 +1,29 @@
-# Group Char Data
 module OoxmlParser
-  class GroupChar
+  # Class for `groupChr` data
+  class GroupChar < OOXMLDocumentObject
     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
+    # Parse GroupChar object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [GroupChar] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'groupChrPr'
+          node_child.xpath('*').each do |node_child_child|
+            case node_child_child.name
+            when 'chr'
+              @symbol = node_child_child.attribute('val').value
+            when 'pos'
+              @position = node_child_child.attribute('val').value
+            when 'vertJc'
+              @vertical_align = node_child_child.attribute('val').value
+            end
+          end
+        end
+      end
+      @element = DocxFormula.new(parent: self).parse(node)
+      self
     end
   end
 end

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

@@ -1,12 +1,22 @@
-# Index data
 module OoxmlParser
-  class Index
+  # Class for 'sSubSup', 'sSup', 'sSub' data
+  class Index < OOXMLDocumentObject
     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
+    # Parse Index object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Index] result of parsing
+    def parse(node)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'sup'
+          @top_index = DocxFormula.new(parent: self).parse(node_child)
+        when 'sub'
+          @bottom_index = DocxFormula.new(parent: self).parse(node_child)
+        end
+      end
+      @value = DocxFormula.new(parent: self).parse(node)
+      self
     end
   end
 end

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

@@ -1,12 +1,26 @@
-# Limit Data
 module OoxmlParser
-  class Limit
+  # Class for `limUpp`, `limLow` data
+  class Limit < OOXMLDocumentObject
     attr_accessor :type, :element, :limit
 
-    def initialize(type = nil, element = nil, limit = nil)
-      @type = type
-      @element = element
-      @limit = limit
+    def initialize(parent: nil)
+      @type = :upper
+      @parent = parent
+    end
+
+    # Parse Accent object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Accent] result of parsing
+    def parse(node)
+      @type = :lower if node.name == 'limLow'
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'lim'
+          @limit = DocxFormula.new(parent: self).parse(node_child)
+        end
+      end
+      @element = DocxFormula.new(parent: self).parse(node)
+      self
     end
   end
 end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/math_run/math_run_properties.rb

@@ -0,0 +1,25 @@
+module OoxmlParser
+  # Class for parsing `m:rPr` object
+  class MathRunProperties < OOXMLDocumentObject
+    # @return [True, False] is run with break
+    attr_accessor :break
+
+    def initialize(parent: nil)
+      @break = false
+      @parent = parent
+    end
+
+    # Parse MathRunProperties
+    # @param [Nokogiri::XML:Node] node with MathRunProperties
+    # @return [MathRunProperties] result of parsing
+    def parse(node)
+      node.xpath('*').each do |math_run_child|
+        case math_run_child.name
+        when 'brk'
+          @break = true
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -0,0 +1,25 @@
+require_relative 'math_run/math_run_properties'
+module OoxmlParser
+  # Class for parsing `m:r` object
+  class MathRun < OOXMLDocumentObject
+    # @return [MathRunProperties] properties of run
+    attr_accessor :properties
+    # @return [String] text of formula
+    attr_accessor :text
+
+    # Parse MathRun
+    # @param [Nokogiri::XML:Node] node with MathRun
+    # @return [MathRun] result of parsing
+    def parse(node)
+      node.xpath('*').each do |math_run_child|
+        case math_run_child.name
+        when 'rPr'
+          @properties = MathRunProperties.new(parent: self).parse(math_run_child)
+        when 't'
+          @text = math_run_child.text
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -1,10 +1,11 @@
-# Matrix Row data
 module OoxmlParser
-  class MatrixRow
+  # Class for 'm:mr' data
+  class MatrixRow < OOXMLDocumentObject
     attr_accessor :columns
 
-    def initialize(columns_count = 1)
+    def initialize(columns_count = 1, parent: nil)
       @columns = Array.new(columns_count)
+      @parent = parent
     end
   end
 end

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

@@ -1,10 +1,43 @@
 require_relative 'matrix/matrix_row'
 module OoxmlParser
-  class Matrix
+  # Class for 'm' data
+  class Matrix < OOXMLDocumentObject
     attr_accessor :rows
 
-    def initialize(rows = [])
-      @rows = rows
+    def initialize(parent: nil)
+      @rows = []
+      @parent = parent
+    end
+
+    # Parse Matrix object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Matrix] result of parsing
+    def parse(node)
+      columns_count = 1
+      j = 0
+      node.xpath('m:mPr').each do |m_pr|
+        m_pr.xpath('m:mcs').each do |mcs|
+          mcs.xpath('m:mc').each do |mc|
+            mc.xpath('m:mcPr').each do |mc_pr|
+              mc_pr.xpath('m:count').each do |count|
+                columns_count = count.attribute('val').value.to_i
+              end
+            end
+          end
+        end
+      end
+
+      node.xpath('m:mr').each do |mr|
+        i = 0
+        @rows << MatrixRow.new(columns_count, parent: self)
+        mr.xpath('m:e').each do |e|
+          @rows[j].columns[i] = DocxFormula.new(parent: self).parse(e)
+          i += 1
+        end
+        j += 1
+      end
+
+      self
     end
   end
 end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph/docx_formula/nary/nary_properties/nary_grow.rb

@@ -0,0 +1,20 @@
+module OoxmlParser
+  # Class for parsing `m:grow` object
+  class NaryGrow < OOXMLDocumentObject
+    # @return [String] value of grow
+    attr_accessor :value
+
+    # Parse NaryGrow
+    # @param [Nokogiri::XML:Node] node with NaryGrow
+    # @return [NaryGrow] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'val'
+          @value = value.value
+        end
+      end
+      self
+    end
+  end
+end