ooxml_parser 0.14.1-mingw32

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

@@ -0,0 +1,52 @@
+# frozen_string_literal: true
+
+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 for Table data
+  class Table < OOXMLDocumentObject
+    attr_accessor :grid, :rows, :properties, :number
+
+    def initialize(rows = [], parent: nil)
+      @rows = rows
+      super(parent: parent)
+    end
+
+    alias table_properties properties
+
+    # @return [String] result of convert of object to string
+    def to_s
+      "Rows: #{@rows.join(',')}"
+    end
+
+    # @return [String] inspect of object for debug means
+    def inspect
+      to_s
+    end
+
+    # Parse Table object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Table] result of parsing
+    def parse(node,
+              number = 0,
+              default_table_properties = TableProperties.new)
+      table_properties = default_table_properties.dup
+      table_properties.jc = :left
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'tblGrid'
+          @grid = TableGrid.new(parent: self).parse(node_child)
+        when 'tr'
+          @rows << TableRow.new(parent: self).parse(node_child)
+        when 'tblPr'
+          @properties = TableProperties.new(parent: self).parse(node_child)
+        end
+      end
+      @number = number
+      self
+    end
+  end
+end

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

@@ -0,0 +1,50 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `u` tags
+  class Underline < OOXMLDocumentObject
+    attr_accessor :style, :color
+
+    def initialize(style = :none, color = nil, parent: nil)
+      @style = style == 'single' ? :single : style
+      @color = color
+      super(parent: parent)
+    end
+
+    # Compare this object to other
+    # @param other [Object] any other object
+    # @return [True, False] result of comparision
+    def ==(other)
+      case other
+      when Underline
+        @style.to_sym == other.style.to_sym && @color == other.color
+      when Symbol
+        @style.to_sym == other
+      else
+        false
+      end
+    end
+
+    # @return [String] result of convert of object to string
+    def to_s
+      if @color.nil?
+        @style.to_s
+      else
+        "#{@style} #{@color}"
+      end
+    end
+
+    # Parse Underline object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Underline] result of parsing
+    def parse(node)
+      case node
+      when 'sng'
+        @style = :single
+      when 'none'
+        @style = :none
+      end
+      self
+    end
+  end
+end

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

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for working with any tag contained only value
+  class ValuedChild < OOXMLDocumentObject
+    # @return [String] value of tag
+    attr_accessor :value
+    # @return [String] type of value
+    attr_reader :type
+
+    def initialize(type = :string, parent: nil)
+      @type = type
+      super(parent: parent)
+    end
+
+    # Parse ValuedChild object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [ValuedChild] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'val'
+          @value = value.value.to_s if type == :string
+          @value = value_to_symbol(value) if type == :symbol
+          @value = value.value.to_i if type == :integer
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_document_structure.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+require_relative 'common_data/content_types'
+module OoxmlParser
+  # Common document structure for DOCX, XLSX, PPTX file
+  class CommonDocumentStructure < OOXMLDocumentObject
+    # @return [String] path to original file
+    attr_accessor :file_path
+    # @return [Integer] default font size
+    attr_reader :default_font_size
+    # @return [Integer] default font typeface
+    attr_reader :default_font_typeface
+    # @return [FontStyle] Default font style of presentation
+    attr_accessor :default_font_style
+    # @return [ContentTypes] data about content types
+    attr_accessor :content_types
+
+    def initialize(params = {})
+      @default_font_size = params.fetch(:default_font_size, 18)
+      @default_font_typeface = params.fetch(:default_font_typeface, 'Arial')
+      @default_font_style = FontStyle.new
+      super(parent: nil)
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/parser.rb

@@ -0,0 +1,49 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Basic class for OoxmlParser
+  class Parser
+    # Base method to yield parse document of any type
+    # @param path_to_file [String] file
+    # @return [CommonDocumentStructure] structure of doc
+    def self.parse_format(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 if model
+      FileUtils.rm_rf(OOXMLDocumentObject.path_to_folder)
+      model
+    end
+
+    # 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)
+      Parser.parse_format(path_to_file) do
+        format = Parser.recognize_folder_format
+        case format
+        when :docx
+          DocumentStructure.parse
+        when :xlsx
+          XLSXWorkbook.new.parse
+        when :pptx
+          Presentation.new.parse
+        else
+          warn "#{path_to_file} is a simple zip file without OOXML content"
+        end
+      end
+    end
+
+    # Recognize folder format
+    # @param directory [String] path to dirctory
+    # @return [Symbol] type of document
+    def self.recognize_folder_format(directory = OOXMLDocumentObject.path_to_folder)
+      return :docx if Dir.exist?("#{directory}/word")
+      return :xlsx if Dir.exist?("#{directory}/xl")
+      return :pptx if Dir.exist?("#{directory}/ppt")
+    end
+  end
+end

data/lib/ooxml_parser/configuration.rb

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for storing configuration
+  class Configuration
+    # @return [Integer] accuracy of digits in fraction part
+    attr_accessor :accuracy
+
+    def initialize
+      @accuracy = 2
+    end
+  end
+end

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

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Comment Data
+  class Comment < OOXMLDocumentObject
+    # @return [Integer] id of comment
+    attr_reader :id
+    # @return [Array<DocxParagraph>] array of paragraphs
+    attr_reader :paragraphs
+
+    def initialize(id = nil, paragraphs = [], parent: nil)
+      @id = id
+      @paragraphs = paragraphs
+      super(parent: parent)
+    end
+
+    # Parse Comment object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Comment] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'id'
+          @id = value.value.to_i
+        end
+      end
+
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'p'
+          @paragraphs << DocxParagraph.new.parse(node_child, 0, parent: self)
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -0,0 +1,36 @@
+# frozen_string_literal: true
+
+require_relative 'comments/comment'
+module OoxmlParser
+  # Class for parsing `comments.xml` file
+  class Comments < OOXMLDocumentObject
+    # @return [Array<Comment>] list of comments
+    attr_reader :comments_array
+
+    def initialize(params = {})
+      @comments_array = []
+      @file = params.fetch(:file, "#{OOXMLDocumentObject.path_to_folder}word/comments.xml")
+      super(parent: params[:parent])
+    end
+
+    # @return [Comment] accessor
+    def [](key)
+      @comments_array[key]
+    end
+
+    # Parse CommentsExtended object
+    # @return [Comments] result of parsing
+    def parse
+      return nil unless File.file?(@file)
+
+      doc = parse_xml(@file)
+      doc.xpath('w:comments/*').each do |node_child|
+        case node_child.name
+        when 'comment'
+          @comments_array << Comment.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `w15:commentEx` tag
+  class CommentExtended < OOXMLDocumentObject
+    # @return [Integer] id of paragraph
+    attr_accessor :paragraph_id
+    # @return [True, False] is done?
+    attr_accessor :done
+
+    # Parse CommentExtended object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [CommentExtended] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'paraId'
+          @paragraph_id = value.value.to_i
+        when 'done'
+          @done = attribute_enabled?(value.value)
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -0,0 +1,42 @@
+# frozen_string_literal: true
+
+require_relative 'comments_extended/comment_extended'
+module OoxmlParser
+  # Class for parsing `commentsExtended.xml` file
+  class CommentsExtended < OOXMLDocumentObject
+    def initialize(parent: nil)
+      @comments_extended_array = []
+      super
+    end
+
+    # @return [Array, CommentsExtended] accessor
+    def [](key)
+      @comments_extended_array[key]
+    end
+
+    # Parse CommentsExtended object
+    # @return [CommentsExtended] result of parsing
+    def parse
+      file_to_parse = "#{OOXMLDocumentObject.path_to_folder}word/commentsExtended.xml"
+      return nil unless File.exist?(file_to_parse)
+
+      doc = parse_xml(file_to_parse)
+      doc.xpath('w15:commentsEx/*').each do |node_child|
+        case node_child.name
+        when 'commentEx'
+          @comments_extended_array << CommentExtended.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+
+    # @param id [Integer] id of comment
+    # @return [CommentExtended] comment by id
+    def by_id(id)
+      @comments_extended_array.each do |cur_comment|
+        return cur_comment if cur_comment.paragraph_id == id
+      end
+      nil
+    end
+  end
+end

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

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for describing Document Background `w:background`
+  class DocumentBackground < OOXMLDocumentObject
+    attr_reader :color1, :size, :color2, :type
+    # @return [FileReference] image structure
+    attr_reader :file_reference
+    # @return [Fill] fill data
+    attr_reader :fill
+
+    def initialize(parent: nil)
+      @color1 = nil
+      @type = 'simple'
+      super
+    end
+
+    # Parse DocumentBackground object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [DocumentBackground] result of parsing
+    def parse(node)
+      @color1 = Color.new(parent: self).parse_hex_string(node.attribute('color').value)
+      node.xpath('v:background').each do |second_color|
+        @size = second_color.attribute('targetscreensize').value.sub(',', 'x') unless second_color.attribute('targetscreensize').nil?
+        second_color.xpath('*').each do |node_child|
+          case node_child.name
+          when 'fill'
+            @fill = Fill.new(parent: self).parse(node_child)
+          end
+        end
+      end
+      self
+    end
+  end
+end

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

@@ -0,0 +1,33 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Document Properties
+  class DocumentProperties < OOXMLDocumentObject
+    attr_accessor :pages, :words
+
+    # Parse Document properties
+    # @return [DocumentProperties]
+    def parse
+      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 self
+      end
+      node = parse_xml(properties_file)
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'Properties'
+          node_child.xpath('*').each do |node_child_child|
+            case node_child_child.name
+            when 'Pages'
+              @pages = node_child_child.text.to_i
+            when 'Words'
+              @words = node_child_child.text.to_i
+            end
+          end
+        end
+      end
+      self
+    end
+  end
+end