ooxml_parser 0.1.1 → 0.1.2

data/lib/ooxml_parser/common_parser/common_data/table/row/cell/merge.rb

@@ -8,10 +8,14 @@ module OoxmlParser
       @value = value
     end
 
-    def self.parse_merge_hash(hash)
+    # Parse Merge data
+    # @param [Nokogiri::XML:Element] node with Merge data
+    # @return [CellMerge] value of CellMerge
+    def self.parse(node)
       merge = CellMerge.new
-      merge.type = hash['type']
-      merge.count_of_merged_cells = hash['count_of_merged_cells']
+      merge.count_of_merged_cells = node.attribute('count_rows_in_span').nil? ? nil : node.attribute('count_rows_in_span').value
+      merge.value = node.attribute('val').nil? ? nil : node.attribute('val').value.to_sym
+      merge.type = node.attribute('type').nil? ? nil : node.attribute('type').value.to_sym
       merge
     end
   end

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

@@ -1,6 +1,6 @@
 module OoxmlParser
   # Class for describing Table Row Properties
-  class TableRowProperties
+  class TableRowProperties < OOXMLDocumentObject
     # @return [Float] Table Row Height
     attr_accessor :height
 

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

@@ -1,5 +1,5 @@
 module OoxmlParser
-  class TableGrid
+  class TableGrid < OOXMLDocumentObject
     attr_accessor :columns, :mass_of_width_cells
 
     def initialize(columns = [])

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

@@ -3,6 +3,7 @@ require_relative 'properties/table_cell_margin'
 require_relative 'properties/table_look'
 require_relative 'properties/table_style'
 require_relative 'properties/table_position'
+require_relative 'table_properties/table_borders'
 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,
@@ -16,7 +17,7 @@ module OoxmlParser
       @jc = :left
       @shd = nil
       @stretching = true
-      @table_borders = Borders.new
+      @table_borders = TableBorders.new
       @table_properties = nil
       @table_width = nil
       @grid_column = nil
@@ -30,7 +31,7 @@ module OoxmlParser
       table.table_width = @table_width
       table.shd = @shd
       table.stretching = @stretching
-      table.table_borders = @table_borders.copy
+      table.table_borders = @table_borders
       table.table_properties = @table_properties
       table.table_cell_margin = @table_cell_margin
       table.table_indent = @table_indent
@@ -46,6 +47,8 @@ module OoxmlParser
         case table_props_node_child.name
         when 'tableStyleId'
           table_properties.style = TableStyle.parse(style_id: table_props_node_child.text)
+        when 'tblBorders'
+          table_properties.table_borders = TableBorders.parse(table_props_node_child)
         when 'tblStyle'
           style = DocxParagraphRun.get_style_by_id(table_props_node_child.attribute('val').value)
           next if style.nil?

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

@@ -0,0 +1,33 @@
+module OoxmlParser
+  # Class for describing Table Border Propertie
+  class TableBorders < OOXMLDocumentObject
+    # @return [BordersProperties] left border propertie
+    attr_accessor :left
+    # @return [BordersProperties] right border propertie
+    attr_accessor :right
+    # @return [BordersProperties] top border propertie
+    attr_accessor :top
+    # @return [BordersProperties] bottom border propertie
+    attr_accessor :bottom
+
+    # Parse Table Borders data
+    # @param [Nokogiri::XML:Element] node with Table Borders data
+    # @return [TableBorders] value of Table Borders data
+    def self.parse(node)
+      borders = TableBorders.new
+      node.xpath('*').each do |cell_borders_node|
+        case cell_borders_node.name
+        when 'left'
+          borders.left = BordersProperties.parse(cell_borders_node)
+        when 'top'
+          borders.top = BordersProperties.parse(cell_borders_node)
+        when 'right'
+          borders.right = BordersProperties.parse(cell_borders_node)
+        when 'bottom'
+          borders.bottom = BordersProperties.parse(cell_borders_node)
+        end
+      end
+      borders
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/parser.rb

@@ -0,0 +1,17 @@
+module OoxmlParser
+  class Parser
+    # Base method to parse document of any type
+    # @param path_to_file [String] file
+    # @return [CommonDocumentStructure] structure of doc
+    def self.parse(path_to_file)
+      return nil if OOXMLDocumentObject.encrypted_file?(path_to_file)
+      path_to_zip_file = OOXMLDocumentObject.copy_file_and_rename_to_zip(path_to_file)
+      OOXMLDocumentObject.path_to_folder = path_to_zip_file.sub(File.basename(path_to_zip_file), '')
+      OOXMLDocumentObject.unzip_file(path_to_zip_file, OOXMLDocumentObject.path_to_folder)
+      model = yield
+      model.file_path = path_to_file
+      FileUtils.remove_dir(OOXMLDocumentObject.path_to_folder)
+      model
+    end
+  end
+end

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

@@ -1,8 +1,10 @@
 # noinspection RubyInstanceMethodNamingConvention
 require_relative 'document_structure/comment'
 require_relative 'document_structure/docx_paragraph'
+require_relative 'document_structure/document_background'
 require_relative 'document_structure/document_properties'
 require_relative 'document_structure/document_style'
+require_relative 'document_structure/header_footer'
 require_relative 'document_structure/page_properties/page_properties'
 module OoxmlParser
   class DocumentStructure < CommonDocumentStructure
@@ -21,6 +23,15 @@ module OoxmlParser
       @document_styles = []
     end
 
+    def ==(other)
+      @elements == other.elements &&
+        @page_properties == other.page_properties &&
+        @notes == other.notes &&
+        @background == other.background &&
+        @document_properties == other.document_properties &&
+        @comments == other.comments
+    end
+
     def difference(other)
       Hash.object_to_hash(self).diff(Hash.object_to_hash(other))
     end
@@ -108,9 +119,8 @@ module OoxmlParser
       !document_style_by_name(name).nil?
     end
 
-    def self.parse(path)
+    def self.parse
       OOXMLDocumentObject.root_subfolder = 'word/'
-      OOXMLDocumentObject.path_to_folder = path
       OOXMLDocumentObject.xmls_stack = []
       OOXMLDocumentObject.namespace_prefix = 'w'
       @comments = []
@@ -138,28 +148,7 @@ module OoxmlParser
       doc = Nokogiri::XML(File.open(OOXMLDocumentObject.current_xml))
       doc.search('//w:document').each do |document|
         document.xpath('w:background').each do |background|
-          doc_structure.background = DocumentBackground.new
-          doc_structure.background.color1 = Color.from_int16(background.attribute('color').value)
-          background.xpath('v:background').each do |second_color|
-            unless second_color.attribute('targetscreensize').nil?
-              doc_structure.background.size = second_color.attribute('targetscreensize').value.sub(',', 'x')
-            end
-            second_color.xpath('v:fill').each do |fill|
-              if !fill.attribute('color2').nil?
-                doc_structure.background.color2 = Color.from_int16(fill.attribute('color2').value.split(' ').first.delete('#'))
-                doc_structure.background.type = fill.attribute('type').value
-              elsif !fill.attribute('id').nil?
-                path_to_image = path + 'word/' + ParserDocxHelper.get_link_from_rels(fill.attribute('id').value, path)
-                image_name = File.basename(path_to_image)
-                doc_structure.background.image = path + File.basename(path) + '/' + image_name
-                unless File.exist?(doc_structure.background.image.sub(File.basename(doc_structure.background.image), ''))
-                  FileUtils.mkdir(doc_structure.background.image.sub(File.basename(doc_structure.background.image), ''))
-                end
-                FileUtils.cp path_to_image, doc_structure.background.image
-                doc_structure.background.type = 'image'
-              end
-            end
-          end
+          doc_structure.background = DocumentBackground.parse(background)
         end
         document.xpath('w:body').each do |body|
           body.xpath('*').each do |element|
@@ -215,10 +204,10 @@ module OoxmlParser
           @default_table_properties = TableProperties.parse(table_pr_tag)
         end
         style.xpath('w:pPr').each do |table_paragraph_pr_tag|
-          DocxParagraph.parse_paragraph_style(table_paragraph_pr_tag, @default_table_paragraph, @default_table_character)
+          DocxParagraph.parse_paragraph_style(table_paragraph_pr_tag, DocumentStructure.default_table_paragraph_style, DocumentStructure.default_table_run_style)
         end
         style.xpath('w:rPr').each do |table_character_pr_tag|
-          DocxParagraphRun.parse(table_character_pr_tag, @default_table_character, @default_character)
+          DocxParagraphRun.parse(table_character_pr_tag, DocumentStructure.default_table_run_style, @default_character)
         end
       end
     end

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

@@ -1,4 +1,5 @@
 module OoxmlParser
+  # Class for describing Document Background `w:background`
   class DocumentBackground
     attr_accessor :color1, :size, :color2, :image, :type
 
@@ -6,5 +7,30 @@ module OoxmlParser
       @color1 = color1
       @type = type
     end
+
+    # Parse DocumentBackground object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [DocumentBackground] result of parsing
+    def self.parse(node)
+      background = DocumentBackground.new
+      background.color1 = Color.from_int16(node.attribute('color').value)
+      node.xpath('v:background').each do |second_color|
+        unless second_color.attribute('targetscreensize').nil?
+          background.size = second_color.attribute('targetscreensize').value.sub(',', 'x')
+        end
+        second_color.xpath('v:fill').each do |fill|
+          if !fill.attribute('color2').nil?
+            background.color2 = Color.from_int16(fill.attribute('color2').value.split(' ').first.delete('#'))
+            background.type = fill.attribute('type').value
+          elsif !fill.attribute('id').nil?
+            path_to_media_file = OOXMLDocumentObject.get_link_from_rels(fill.attribute('id').value)
+            path_to_image = OOXMLDocumentObject.copy_media_file("#{OOXMLDocumentObject.root_subfolder}/#{path_to_media_file.gsub('..', '')}")
+            background.image = path_to_image
+            background.type = 'image'
+          end
+        end
+      end
+      background
+    end
   end
 end

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

@@ -12,7 +12,12 @@ module OoxmlParser
     # @return [DocumentProperties]
     def self.parse
       properties = DocumentProperties.new
-      doc_props = XmlSimple.xml_in(File.open("#{OOXMLDocumentObject.path_to_folder}docProps/app.xml"))
+      properties_file = "#{OOXMLDocumentObject.path_to_folder}docProps/app.xml"
+      unless File.exist?(properties_file)
+        warn "There is no 'docProps/app.xml' in docx. It may be some problem with it"
+        return properties
+      end
+      doc_props = XmlSimple.xml_in(File.open(properties_file))
       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

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

@@ -332,21 +332,21 @@ module OoxmlParser
           end
         when 'footnoteReference'
           character_style.footnote = HeaderFooter.new(:header)
-          character_style.footnote.id = r_node_child.attribute('id').value
+          character_style.footnote.id = r_node_child.attribute('id').value.to_i
           character_style.footnote.parse('footnote', character_style.footnote.id, @default_paragraph, @default_character)
         when 'endnoteReference'
           character_style.endnote = HeaderFooter.new(:footer)
-          character_style.endnote.id = r_node_child.attribute('id').value
+          character_style.endnote.id = r_node_child.attribute('id').value.to_i
           character_style.endnote.parse('endnote', character_style.endnote.id, @default_paragraph, @default_character)
         when 'pict'
           r_node_child.xpath('*').each do |pict_node_child|
             case pict_node_child.name
             when 'shape'
-              character_style.shape = Shape.parse(pict_node_child, :shape, OOXMLDocumentObject.path_to_folder)
+              character_style.shape = Shape.parse(pict_node_child, :shape)
             when 'rect'
-              character_style.shape = Shape.parse(pict_node_child, :rectangle, OOXMLDocumentObject.path_to_folder)
+              character_style.shape = Shape.parse(pict_node_child, :rectangle)
             when 'oval'
-              character_style.shape = Shape.parse(pict_node_child, :oval, OOXMLDocumentObject.path_to_folder)
+              character_style.shape = Shape.parse(pict_node_child, :oval)
             when 'shapetype'
             end
           end

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

@@ -9,31 +9,30 @@ module OoxmlParser
       @elements = elements
     end
 
-    def self.parse(shape_node, type, path)
+    def self.parse(shape_node, type)
       shape = Shape.new
       shape.type = type
       shape_node.attribute('style').value.to_s.split(';').each do |property|
-        case
-        when property.include?('margin-top')
+        if property.include?('margin-top')
           shape.properties.margins.top = property.split(':').last
-        when property.include?('margin-left')
+        elsif property.include?('margin-left')
           shape.properties.margins.left = property.split(':').last
-        when property.include?('margin-right')
+        elsif property.include?('margin-right')
           shape.properties.margins.right = property.split(':').last
-        when property.include?('width')
+        elsif property.include?('width')
           shape.properties.size.width = property.split(':').last
-        when property.include?('height')
+        elsif property.include?('height')
           shape.properties.size.height = property.split(':').last
-        when property.include?('z-index')
+        elsif property.include?('z-index')
           shape.properties.z_index = property.split(':').last
-        when property.include?('position')
+        elsif property.include?('position')
           shape.properties.position = property.split(':').last
         end
       end
       shape.properties.fill_color = Color.from_int16(shape_node.attribute('fillcolor').value.to_s.sub('#', '').split(' ').first) unless shape_node.attribute('fillcolor').nil?
       shape.properties.stroke.weight = shape_node.attribute('strokeweight').value unless shape_node.attribute('strokeweight').nil?
       shape.properties.stroke.color = Color.from_int16(shape_node.attribute('strokecolor').value.to_s.sub('#', '').split(' ').first) unless shape_node.attribute('strokecolor').nil?
-      shape.elements = TextBox.parse_list(shape_node.xpath('v:textbox').first, path) unless shape_node.xpath('v:textbox').first.nil?
+      shape.elements = TextBox.parse_list(shape_node.xpath('v:textbox').first) unless shape_node.xpath('v:textbox').first.nil?
       shape
     end
   end

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

@@ -29,7 +29,7 @@ module OoxmlParser
         when 'yAlign'
           frame_properties.vertical_align = value.value.to_sym
         when 'anchorLock'
-          frame_properties.anchor_lock = OOXMLDocumentObject.option_enabled?(frame_pr_node, anchorLock)
+          frame_properties.anchor_lock = OOXMLDocumentObject.option_enabled?(frame_pr_node, 'anchorLock')
         when 'vSpace'
           frame_properties.vertical_space = (value.value.to_f / 566.9).round(2)
         when 'hSpace'

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

@@ -1,5 +1,5 @@
 module OoxmlParser
-  class DocumentGrid
+  class DocumentGrid < OOXMLDocumentObject
     attr_accessor :type, :line_pitch, :char_space
 
     def initialize(type = nil, line_pitch = nil, char_space = nil)

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

@@ -28,10 +28,10 @@ module OoxmlParser
       sect_pr.xpath('w:pgBorders').each do |pg_borders|
         page_borders = Borders.new
         unless pg_borders.attribute('display').nil?
-          page_borders.display = pg_borders.attribute('display').value
+          page_borders.display = pg_borders.attribute('display').value.to_sym
         end
         unless pg_borders.attribute('offsetFrom').nil?
-          page_borders.offset_from = pg_borders.attribute('offsetFrom').value
+          page_borders.offset_from = pg_borders.attribute('offsetFrom').value.to_sym
         end
         pg_borders.xpath('w:bottom').each do |bottom|
           page_borders.bottom = BordersProperties.parse(bottom)
@@ -45,7 +45,7 @@ module OoxmlParser
         pg_borders.xpath('w:right').each do |right|
           page_borders.bottom = BordersProperties.parse(right)
         end
-        page_properties.page_border = page_borders
+        page_properties.page_borders = page_borders
       end
       sect_pr.xpath('w:type').each do |type|
         page_properties.type = type.attribute('val').value
@@ -68,17 +68,10 @@ module OoxmlParser
       sect_pr.xpath('w:cols').each do |columns_grid|
         page_properties.columns = Columns.parse(columns_grid)
       end
-      sect_pr.xpath('w:headerReference').each do |header_reference|
-        target = OOXMLDocumentObject.get_link_from_rels(header_reference.attribute('id').value)
+      sect_pr.xpath('w:headerReference', 'w:footerReference').each do |doc_edge_reference|
+        target = OOXMLDocumentObject.get_link_from_rels(doc_edge_reference.attribute('id').value)
         OOXMLDocumentObject.add_to_xmls_stack("word/#{target}")
-        note = Note.parse(default_paragraph, default_character, target, header_reference.attribute('type').value, File.basename(target).sub('.xml', ''))
-        page_properties.notes << note
-        OOXMLDocumentObject.xmls_stack.pop
-      end
-      sect_pr.xpath('w:footerReference').each do |footer_reference|
-        target = OOXMLDocumentObject.get_link_from_rels(footer_reference.attribute('id').value)
-        OOXMLDocumentObject.add_to_xmls_stack("word/#{target}")
-        note = Note.parse(default_paragraph, default_character, target, footer_reference.attribute('type').value, File.basename(target).sub('.xml', ''))
+        note = Note.parse(default_paragraph, default_character, target, doc_edge_reference.attribute('type').value, File.basename(target).sub('.xml', ''))
         page_properties.notes << note
         OOXMLDocumentObject.xmls_stack.pop
       end

data/lib/ooxml_parser/docx_parser/docx_parser.rb

@@ -4,13 +4,9 @@ require_relative 'docx_data/document_structure'
 module OoxmlParser
   class DocxParser
     def self.parse_docx(path_to_file)
-      path_to_zip_file = OOXMLDocumentObject.copy_file_and_rename_to_zip(path_to_file)
-      OOXMLDocumentObject.unzip_file(path_to_zip_file, path_to_zip_file.sub(File.basename(path_to_zip_file), ''))
-      OOXMLDocumentObject.path_to_folder = path_to_zip_file.sub(File.basename(path_to_zip_file), '')
-      docx_model = DocumentStructure.parse(OOXMLDocumentObject.path_to_folder)
-      docx_model.file_path = path_to_file
-      FileUtils.remove_dir(OOXMLDocumentObject.path_to_folder)
-      docx_model
+      Parser.parse(path_to_file) do
+        DocumentStructure.parse
+      end
     end
   end
 end

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation.rb

@@ -12,8 +12,7 @@ module OoxmlParser
       @comments = comments
     end
 
-    def self.parse(path_to_folder)
-      OOXMLDocumentObject.path_to_folder = path_to_folder
+    def self.parse
       OOXMLDocumentObject.root_subfolder = 'ppt/'
       OOXMLDocumentObject.xmls_stack = []
       OOXMLDocumentObject.namespace_prefix = 'a'

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/presentation_comment/presentation_comment_author.rb

@@ -17,7 +17,8 @@ module OoxmlParser
       end
       doc.xpath('p:cmAuthorLst/p:cmAuthor').each do |author_node|
         comment_authors[author_node.attribute('id').value] = PresentationCommentAuthor.new(
-          author_node.attribute('id').value, author_node.attribute('name').value, author_node.attribute('initials').value)
+          author_node.attribute('id').value, author_node.attribute('name').value, author_node.attribute('initials').value
+        )
       end
       comment_authors
     end

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/presentation_theme.rb

@@ -17,7 +17,7 @@ module OoxmlParser
       end
       doc = Nokogiri::XML(File.open(OOXMLDocumentObject.current_xml))
       doc.xpath('a:theme').each do |theme_node|
-        OOXMLDocumentObject.theme.name = theme_node.attribute('name').value
+        OOXMLDocumentObject.theme.name = theme_node.attribute('name').value if theme_node.attribute('name')
         theme_node.xpath('a:themeElements/*').each do |theme_element_node|
           case theme_element_node.name
           when 'clrScheme'

data/lib/ooxml_parser/pptx_parser/pptx_data/presentation/slide.rb

@@ -20,20 +20,26 @@ module OoxmlParser
       @elements.select { |cur_element| cur_element.is_a?(GraphicFrame) }
     end
 
-    def content_horizontal_align(object, slide_size)
-      transform = nil
+    # Get transform property of object, by object type
+    # @param object [Symbol] type of object: :image, :chart, :table, :shape
+    # @return [OOXMLDocumentObject] needed object
+    def transform_of_object(object)
       case object
       when :image
-        transform = elements.find { |e| e.is_a? Picture }.properties.transform
+        return elements.find { |e| e.is_a? Picture }.properties.transform
       when :chart
-        transform = elements.find { |e| e.is_a? GraphicFrame }.transform
+        return elements.find { |e| e.is_a? GraphicFrame }.transform
       when :table
-        transform = elements.find { |e| e.is_a? GraphicFrame }.transform
+        return elements.find { |e| e.is_a? GraphicFrame }.transform
       when :shape
-        transform = elements.find { |e| !e.shape_properties.preset.nil? }.shape_properties.transform
+        return elements.find { |e| !e.shape_properties.preset.nil? }.shape_properties.transform
       else
         raise "Dont know this type object - #{object}"
       end
+    end
+
+    def content_horizontal_align(object, slide_size)
+      transform = transform_of_object(object)
       return :left if transform.offset.x == 0
       return :center if ((slide_size.width / 2) - (transform.extents.x / 2)).round(1) == transform.offset.x.round(1)
       return :right if (slide_size.width - transform.extents.x).round(1) == transform.offset.x.round(1)
@@ -41,19 +47,7 @@ module OoxmlParser
     end
 
     def content_vertical_align(object, slide_size)
-      transform = nil
-      case object
-      when :image
-        transform = elements.find { |e| e.is_a? Picture }.properties.transform
-      when :chart
-        transform = elements.find { |e| e.is_a? GraphicFrame }.transform
-      when :table
-        transform = elements.find { |e| e.is_a? GraphicFrame }.transform
-      when :shape
-        transform = elements.find { |e| !e.shape_properties.preset.nil? }.shape_properties.transform
-      else
-        raise "Dont know this type object - #{object}"
-      end
+      transform = transform_of_object(object)
       return :top if transform.offset.y == 0
       return :middle if ((slide_size.height / 2) - (transform.extents.y / 2)).round(1) == transform.offset.y.round(1)
       return :bottom if (slide_size.height - transform.extents.y).round(1) == transform.offset.y.round(1)