ooxml_parser 0.4.0 → 0.7.0

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
 
@@ -6,18 +9,19 @@ module OoxmlParser
       @elements = []
     end
 
+    # Parse note data
+    # @param params [Hash] data to parse
+    # @return [Note] result of parsing
     def self.parse(params)
       note = Note.new
       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 +32,19 @@ 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

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative 'document_grid'
 require_relative 'footnote_properties'
 require_relative 'page_size'
@@ -5,6 +7,7 @@ require_relative 'page_margins'
 require_relative 'columns'
 require_relative 'note'
 module OoxmlParser
+  # Class for data of PageProperties
   class PageProperties < OOXMLDocumentObject
     attr_accessor :type, :size, :margins, :document_grid, :num_type, :form_prot, :text_direction, :page_borders, :columns,
                   :notes

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

@@ -1,4 +1,7 @@
+# frozen_string_literal: true
+
 module OoxmlParser
+  # Class for data of PageSize
   class PageSize < OOXMLDocumentObject
     attr_accessor :height, :width, :orientation
 
@@ -8,6 +11,7 @@ module OoxmlParser
       @orientation = orientation
     end
 
+    # @return [String] convert to string
     def to_s
       'Height: ' + @height.to_s + ' Width: ' + @width.to_s + ' Orientation: ' + @orientation.to_s
     end
@@ -33,6 +37,7 @@ module OoxmlParser
       return 'ROC 16K' if same_dimensions?(PageSize.new(OoxmlSize.new(27.3, :centimeter), OoxmlSize.new(19.68, :centimeter)))
       return 'Envelope Choukei 3' if same_dimensions?(PageSize.new(OoxmlSize.new(23.49, :centimeter), OoxmlSize.new(11.99, :centimeter)))
       return 'Super B/A3' if same_dimensions?(PageSize.new(OoxmlSize.new(48.25, :centimeter), OoxmlSize.new(33.02, :centimeter)))
+
       "Unknown page size: Height #{@height} Width #{@width}"
     end
 

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

@@ -1,16 +1,29 @@
+# frozen_string_literal: true
+
 require_relative 'styles/document_defaults'
 module OoxmlParser
   # Class for parsing `styles.xml` file
   class Styles < OOXMLDocumentObject
     # @return [DocumentDefaults] defaults of document
     attr_accessor :document_defaults
+    # @return [Array<DocumentStyle>] array of document styles
+    attr_reader :styles
+
+    def initialize(parent: nil)
+      @styles = []
+      @parent = parent
+    end
 
+    # Parse styles data
+    # @return [Styles] result of parsing
     def parse
-      doc = Nokogiri::XML(File.open(OOXMLDocumentObject.path_to_folder + 'word/styles.xml'))
+      doc = parse_xml(OOXMLDocumentObject.path_to_folder + 'word/styles.xml')
       doc.xpath('w:styles/*').each do |node_child|
         case node_child.name
         when 'docDefaults'
           @document_defaults = DocumentDefaults.new(parent: self).parse(node_child)
+        when 'style'
+          @styles << DocumentStyle.new(parent: self).parse(node_child)
         end
       end
       self

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative 'document_defaults/paragraph_properties_default'
 require_relative 'document_defaults/run_properties_default'
 module OoxmlParser

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/styles/document_defaults/paragraph_properties_default.rb

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

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/styles/document_defaults/run_properties_default.rb

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

data/lib/ooxml_parser/docx_parser/docx_parser.rb

@@ -1,9 +1,15 @@
+# frozen_string_literal: true
+
 require_relative '../common_parser/common_data/ooxml_document_object'
 require_relative '../common_parser/common_data/valued_child'
 require_relative 'docx_data/document_structure'
 
 module OoxmlParser
+  # Basic class for DocxParser
   class DocxParser
+    # Parse docx file
+    # @param path_to_file [String] file path
+    # @return [DocumentStructure] result of parse
     def self.parse_docx(path_to_file)
       Parser.parse_format(path_to_file) do
         DocumentStructure.parse

data/lib/ooxml_parser/helpers/string_helper.rb

@@ -1,5 +1,7 @@
-# Class for working with strings
+# frozen_string_literal: true
+
 module OoxmlParser
+  # Class for working with strings
   class StringHelper
     class << self
       # This method check if string contains numeric

data/lib/ooxml_parser/name.rb

@@ -0,0 +1,9 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # This module holds name of product
+  module Name
+    # [String] name of the gem
+    STRING = 'ooxml_parser'
+  end
+end

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

@@ -1,67 +1,62 @@
-require_relative 'presentation/presentation_comment'
+# frozen_string_literal: true
+
+require_relative 'presentation/comment_authors'
+require_relative 'presentation/presentation_comments'
 require_relative 'presentation/presentation_helpers'
 require_relative 'presentation/presentation_theme'
 require_relative 'presentation/slide'
 require_relative 'presentation/slide_size'
+require_relative 'presentation/table_styles'
 module OoxmlParser
+  # Basic class for all parsed pptx data
   class Presentation < CommonDocumentStructure
     include PresentationHelpers
-    attr_accessor :slides, :theme, :slide_size, :comments
+    attr_accessor :slides, :theme, :slide_size
     # @return [Relationships] relationships of presentation
     attr_accessor :relationships
+    # @return [TableStyles] table styles data
+    attr_accessor :table_styles
+    # @return [CommentAuthors] authors of presentation
+    attr_reader :comment_authors
+    # @return [PresentationComments] comments of presentation
+    attr_reader :comments
 
-    def initialize(slides = [], theme = nil, comments = [])
-      @slides = slides
-      @theme = theme
-      @comments = comments
+    def initialize(params = {})
+      @slides = []
+      @comments = []
+      super
     end
 
-    def self.parse
+    # Parse data of presentation
+    # @return [Presentation] parsed presentation
+    def parse
+      @content_types = ContentTypes.new(parent: self).parse
       OOXMLDocumentObject.root_subfolder = 'ppt/'
       OOXMLDocumentObject.xmls_stack = []
       OOXMLDocumentObject.add_to_xmls_stack('ppt/presentation.xml')
-      doc = Nokogiri::XML(File.open(OOXMLDocumentObject.current_xml))
-      presentation = Presentation.new
-      presentation.theme = PresentationTheme.parse('ppt/theme/theme1.xml')
+      doc = parse_xml(OOXMLDocumentObject.current_xml)
+      @theme = PresentationTheme.parse('ppt/theme/theme1.xml')
+      @table_styles = TableStyles.new(parent: self).parse
+      @comment_authors = CommentAuthors.new(parent: self).parse
+      @comments = PresentationComments.new(parent: self).parse
       presentation_node = doc.search('//p:presentation').first
       presentation_node.xpath('*').each do |presentation_node_child|
         case presentation_node_child.name
         when 'sldSz'
-          presentation.slide_size = SlideSize.new(parent: presentation).parse(presentation_node_child)
+          @slide_size = SlideSize.new(parent: self).parse(presentation_node_child)
         when 'sldIdLst'
           presentation_node_child.xpath('p:sldId').each do |silde_id_node|
             id = nil
             silde_id_node.attribute_nodes.select { |node| id = node.to_s if node.namespace && node.namespace.prefix == 'r' }
-            presentation.slides << Slide.new(parent: presentation,
-                                             xml_path: "#{OOXMLDocumentObject.root_subfolder}/#{OOXMLDocumentObject.get_link_from_rels(id)}")
-                                        .parse
+            @slides << Slide.new(parent: self,
+                                 xml_path: "#{OOXMLDocumentObject.root_subfolder}/#{OOXMLDocumentObject.get_link_from_rels(id)}")
+                            .parse
           end
         end
       end
-      presentation.comments = PresentationComment.parse_list
       OOXMLDocumentObject.xmls_stack.pop
-      presentation.relationships = Relationships.parse_rels("#{OOXMLDocumentObject.path_to_folder}/ppt/_rels/presentation.xml.rels")
-      presentation
-    end
-
-    class << self
-      # @return [FontStyle] current font_style
-      attr_writer :current_font_style
-
-      def current_font_style
-        @current_font_style = FontStyle.new if @current_font_style.nil?
-        @current_font_style
-      end
-
-      # @return [String] default_font_typeface
-      def default_font_typeface
-        'Arial'
-      end
-
-      # @return [FixNum] default_font_typeface
-      def default_font_size
-        18
-      end
+      @relationships = Relationships.new(parent: self).parse_file("#{OOXMLDocumentObject.path_to_folder}/ppt/_rels/presentation.xml.rels")
+      self
     end
   end
 end

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

@@ -0,0 +1,39 @@
+# frozen_string_literal: true
+
+require_relative 'comment_authors/comment_author'
+module OoxmlParser
+  # Class for parsing `commentAuthors.xml` file
+  class CommentAuthors < OOXMLDocumentObject
+    # @return [Array<CommentAuthor>] list of comment authors
+    attr_reader :list
+
+    def initialize(parent: nil)
+      @list = []
+      @parent = parent
+    end
+
+    # Parse CommentAuthors object
+    # @param file [Nokogiri::XML:Element] node to parse
+    # @return [CommentAuthors] result of parsing
+    def parse(file = "#{OOXMLDocumentObject.path_to_folder}/#{OOXMLDocumentObject.root_subfolder}/commentAuthors.xml")
+      return nil unless File.exist?(file)
+
+      document = parse_xml(File.open(file))
+      node = document.xpath('*').first
+
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'cmAuthor'
+          @list << CommentAuthor.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+
+    # @param id [Integer] id of author
+    # @return [CommentAuthor] author by id
+    def author_by_id(id)
+      list.detect { |author| author.id == id }
+    end
+  end
+end

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

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `cmAuthor` tag
+  class CommentAuthor < OOXMLDocumentObject
+    # @return [Integer] Comment Author ID
+    attr_reader :id
+    # @return [String] Author Name
+    attr_reader :name
+    # @return [String] Author Initials
+    attr_reader :initials
+    # @return [Integer] Index of Comment Author's last comment
+    attr_reader :last_index
+    # @return [Integer] Comment Author Color Index
+    attr_reader :color_index
+
+    # Parse CommentAuthor object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [CommentAuthor] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'id'
+          @id = value.value.to_i
+        when 'name'
+          @name = value.value.to_s
+        when 'initials'
+          @initials = value.value.to_s
+        when 'lastIdx'
+          @last_index = value.value.to_i
+        when 'clrIdx'
+          @color_index = value.value.to_i
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -0,0 +1,33 @@
+# frozen_string_literal: true
+
+require_relative 'presentation_comments/presentation_comment'
+module OoxmlParser
+  # Class for parsing `comment1.xml` file
+  class PresentationComments < OOXMLDocumentObject
+    # @return [Array<PresentationComment>] list of comments
+    attr_reader :list
+
+    def initialize(parent: nil)
+      @list = []
+      @parent = parent
+    end
+
+    # Parse PresentationComments object
+    # @param file [Nokogiri::XML:Element] node to parse
+    # @return [PresentationComments] result of parsing
+    def parse(file = "#{OOXMLDocumentObject.path_to_folder}/#{OOXMLDocumentObject.root_subfolder}/comments/comment1.xml")
+      return nil unless File.exist?(file)
+
+      document = parse_xml(File.open(file))
+      node = document.xpath('*').first
+
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'cm'
+          @list << PresentationComment.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -0,0 +1,39 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `p:cm` tag
+  class PresentationComment < OOXMLDocumentObject
+    # @return [Integer] id of author
+    attr_reader :author_id
+    # @return [OOXMLCoordinates] position of comment
+    attr_reader :position
+    # @return [String] text of comment
+    attr_reader :text
+
+    # @return [CommentAuthor] author of comment
+    def author
+      root_object.comment_authors.author_by_id(@author_id)
+    end
+
+    # Parse PresentationComment object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [PresentationComment] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'authorId'
+          @author_id = value.value.to_i
+        end
+      end
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'pos'
+          @position = OOXMLCoordinates.parse(node_child, x_attr: 'x', y_attr: 'y')
+        when 'text'
+          @text = node_child.text.to_s
+        end
+      end
+      self
+    end
+  end
+end