ooxml_parser 0.14.1-mingw32

data/lib/ooxml_parser/common_parser/common_data/ooxml_document_object/ooxml_object_attribute_helper.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Module for stroing some attribute helpers for objects
+  module OoxmlObjectAttributeHelper
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @param attribute_name [String] name of attribute
+    # @return [True, False] is option enabled
+    def option_enabled?(node, attribute_name = 'val')
+      return true if node.attributes.empty?
+      return true if node.to_s == '1'
+      return false if node.to_s == '0'
+      return false if node.attribute(attribute_name).nil?
+
+      status = node.attribute(attribute_name).value
+      !%w[false off 0].include?(status)
+    end
+
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @param attribute_name [String] name of attribute
+    # @return [True, False] is attribute enabled
+    def attribute_enabled?(node, attribute_name = 'val')
+      return true if node.to_s == '1'
+      return false if node.to_s == '0'
+      return false if node.attribute(attribute_name).nil?
+
+      status = node.attribute(attribute_name).value
+      %w[true on 1].include?(status)
+    end
+  end
+end

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

@@ -0,0 +1,144 @@
+# frozen_string_literal: true
+
+require 'filemagic' unless Gem.win_platform?
+require 'securerandom'
+require 'nokogiri'
+require 'zip'
+require_relative 'ooxml_document_object/ooxml_document_object_helper'
+require_relative 'ooxml_document_object/ooxml_object_attribute_helper'
+
+module OoxmlParser
+  # Basic class for any OOXML Document Object
+  class OOXMLDocumentObject
+    include OoxmlDocumentObjectHelper
+    include OoxmlObjectAttributeHelper
+    # @return [OOXMLDocumentObject] object which hold current object
+    attr_accessor :parent
+
+    def initialize(parent: nil)
+      @parent = parent
+    end
+
+    # Compare this object to other
+    # @param other [Object] any other object
+    # @return [True, False] result of comparision
+    def ==(other)
+      return false if self.class != other.class
+
+      instance_variables.each do |current_attribute|
+        next if current_attribute == :@parent
+        return false unless instance_variable_get(current_attribute) == other.instance_variable_get(current_attribute)
+      end
+      true
+    end
+
+    # @return [True, false] if structure contain any user data
+    def with_data?
+      true
+    end
+
+    # @return [Nokogiri::XML::Document] result of parsing xml via nokogiri
+    def parse_xml(xml_path)
+      Nokogiri::XML(File.open(xml_path), &:strict)
+    end
+
+    class << self
+      # @return [String] path to root subfolder
+      attr_accessor :root_subfolder
+      # @return [PresentationTheme] list of themes
+      attr_accessor :theme
+      # @return [Array<String>] stack of xmls
+      attr_accessor :xmls_stack
+      # @return [String] path to root folder
+      attr_accessor :path_to_folder
+
+      # @param path_to_file [String] file
+      # @param ignore_system [True, False] should host system be ignored, since
+      #   this method is OS-dependent
+      # @return [True, False] Check if file is protected by password on open
+      def encrypted_file?(path_to_file, ignore_system: false)
+        if Gem.win_platform? || ignore_system
+          warn 'FileMagic and checking file for encryption is not supported on Windows'
+          return false
+        end
+        file_result = FileMagic.new(:mime).file(path_to_file)
+        # Support of Encrtypted status in `file` util was introduced in file v5.20
+        # but LTS version of ubuntu before 16.04 uses older `file` and it return `Composite Document`
+        # https://github.com/file/file/blob/master/ChangeLog#L217
+        if file_result.include?('encrypted') ||
+           file_result.include?('Composite Document File V2 Document, No summary info') ||
+           file_result.include?('application/CDFV2-corrupt')
+          warn("File #{path_to_file} is encrypted. Can't parse it")
+          return true
+        end
+        false
+      end
+
+      # Copy this file and rename to zip
+      # @param path [String] path to file
+      # @return [String] path to result zip
+      def copy_file_and_rename_to_zip(path)
+        file_name = File.basename(path)
+        tmp_folder = Dir.mktmpdir('ruby-ooxml-parser')
+        file_path = "#{tmp_folder}/#{file_name}"
+        FileUtils.rm_rf(tmp_folder) if File.directory?(tmp_folder)
+        FileUtils.mkdir_p(tmp_folder)
+        raise "Cannot find file by path #{path}" unless File.exist?(path)
+
+        FileUtils.cp path, tmp_folder
+        file_path
+      end
+
+      # Unzip specified file
+      # @param path_to_file [String] path to zip file
+      # @param destination [String] folder to extract
+      # @return [void]
+      def unzip_file(path_to_file, destination)
+        Zip.warn_invalid_date = false
+        Zip::File.open(path_to_file) do |zip_file|
+          raise LoadError, "There is no files in zip #{path_to_file}" if zip_file.entries.empty?
+
+          zip_file.each do |file|
+            file_path = File.join(destination, file.name)
+            FileUtils.mkdir_p(File.dirname(file_path))
+            zip_file.extract(file, file_path) unless File.exist?(file_path)
+          end
+        end
+      end
+
+      # @return [String] dir to base of file
+      def dir
+        "#{OOXMLDocumentObject.path_to_folder}#{File.dirname(OOXMLDocumentObject.xmls_stack.last)}/"
+      end
+
+      # @return [String] path to current xml file
+      def current_xml
+        OOXMLDocumentObject.path_to_folder + OOXMLDocumentObject.xmls_stack.last
+      end
+
+      # Add file to parsing stack
+      # @param path [String] path of file to add to stack
+      # @return [void]
+      def add_to_xmls_stack(path)
+        OOXMLDocumentObject.xmls_stack << if path.include?('..')
+                                            "#{File.dirname(OOXMLDocumentObject.xmls_stack.last)}/#{path}"
+                                          elsif path.start_with?(OOXMLDocumentObject.root_subfolder)
+                                            path
+                                          else
+                                            OOXMLDocumentObject.root_subfolder + path
+                                          end
+      end
+
+      # Get link to file from rels file
+      # @param id [String] file to get
+      # @return [String] result
+      def get_link_from_rels(id)
+        rels_path = dir + "_rels/#{File.basename(OOXMLDocumentObject.xmls_stack.last)}.rels"
+        raise LoadError, "Cannot find .rels file by path: #{rels_path}" unless File.exist?(rels_path)
+
+        relationships = Relationships.new.parse_file(rels_path)
+        relationships.target_by_id(id)
+      end
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragraph_properties.rb

@@ -0,0 +1,109 @@
+# frozen_string_literal: true
+
+require_relative 'paragrpah_properties/numbering_properties'
+require_relative 'paragrpah_properties/paragraph_borders'
+require_relative 'paragrpah_properties/paragraph_spacing'
+require_relative 'paragrpah_properties/paragraph_stlye_ref'
+require_relative 'paragrpah_properties/spacing'
+require_relative 'paragrpah_properties/tabs'
+module OoxmlParser
+  # Class for data for ParagraphProperties
+  class ParagraphProperties < OOXMLDocumentObject
+    attr_accessor :align, :numbering, :level, :spacing, :spacing_before, :spacing_after, :indent, :margin_left, :margin_right
+    # @return [FrameProperties] frame properties
+    attr_reader :frame_properties
+    # @return [Tabs] list of tabs
+    attr_accessor :tabs
+    # @return [RunProperties] properties of run
+    attr_accessor :run_properties
+    # @return [Borders] borders of paragraph
+    attr_accessor :paragraph_borders
+    # @return [ParagraphStyleRef] Referenced Paragraph Style
+    attr_accessor :paragraph_style_ref
+    # @return [True, False] Specifies that the paragraph
+    # (or at least part of it) should be rendered on
+    # the same page as the next paragraph when possible
+    attr_accessor :keep_next
+    # @return [PageProperties] properties of section
+    attr_accessor :section_properties
+    # @return [True, False] This element specifies that any space specified
+    # before or after this paragraph, specified using the spacing element (17.3.1.33),
+    # should not be applied when the preceding and following paragraphs are of
+    # the same paragraph style, affecting the top and bottom spacing respectively
+    attr_accessor :contextual_spacing
+    # @return [Symbol] The alignment or justification to be applied to a paragraph
+    attr_accessor :justification
+    # @return [Shade] Shade property
+    attr_accessor :shade
+
+    def initialize(numbering = NumberingProperties.new, parent: nil)
+      @numbering = numbering
+      @spacing = Spacing.new(OoxmlSize.new(0),
+                             OoxmlSize.new(0),
+                             OoxmlSize.new(1, :centimeter),
+                             :multiple)
+      @keep_next = false
+      @tabs = []
+      super(parent: parent)
+    end
+
+    # Parse ParagraphProperties object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [ParagraphProperties] result of parsing
+    def parse(node)
+      @spacing.parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'algn'
+          @align = value_to_symbol(value)
+        when 'lvl'
+          @level = value.value.to_i
+        when 'indent'
+          @indent = OoxmlSize.new(value.value.to_f, :emu)
+        when 'marR'
+          @margin_right = OoxmlSize.new(value.value.to_f, :emu)
+        when 'marL'
+          @margin_left = OoxmlSize.new(value.value.to_f, :emu)
+        end
+      end
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'buSzPct'
+          @numbering.size = node_child.attribute('val').value
+        when 'buFont'
+          @numbering.font = node_child.attribute('typeface').value
+        when 'buChar'
+          @numbering.symbol = node_child.attribute('char').value
+        when 'buAutoNum'
+          @numbering.type = node_child.attribute('type').value.to_sym
+          @numbering.start_at = node_child.attribute('startAt').value if node_child.attribute('startAt')
+        when 'framePr'
+          @frame_properties = FrameProperties.new(parent: self).parse(node_child)
+        when 'tabs', 'tabLst'
+          @tabs = Tabs.new(parent: self).parse(node_child)
+        when 'ind'
+          @indent = Indents.new(parent: self).parse(node_child)
+        when 'rPr'
+          @run_properties = RunProperties.new(parent: self).parse(node_child)
+        when 'pBdr'
+          @paragraph_borders = ParagraphBorders.new(parent: self).parse(node_child)
+        when 'pStyle'
+          @paragraph_style_ref = ParagraphStyleRef.new(parent: self).parse(node_child)
+        when 'keepNext'
+          @keep_next = true
+        when 'sectPr'
+          @section_properties = PageProperties.new(parent: self).parse(node_child, @parent, DocxParagraphRun.new)
+        when 'shd'
+          @shade = Shade.new(parent: self).parse(node_child)
+        when 'spacing'
+          @spacing = ParagraphSpacing.new(parent: self).parse(node_child)
+        when 'jc'
+          @justification = value_to_symbol(node_child.attribute('val'))
+        when 'contextualSpacing'
+          @contextual_spacing = option_enabled?(node_child)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragraph_run/run_properties/outline.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+# Class for parsing Run Outline properties
+module OoxmlParser
+  # Class for parsing `w:ln` tags
+  class Outline < OOXMLDocumentObject
+    # @return [OoxmlSize] width of outline
+    attr_reader :width
+    # @return [DocxColorScheme] color of outline
+    attr_reader :color_scheme
+
+    def initialize(parent: nil)
+      @width = OoxmlSize.new(0)
+      super
+    end
+
+    # Parse Outline object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Outline] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'w'
+          @width = OoxmlSize.new(value.value.to_f, :emu)
+        end
+      end
+      @color_scheme = DocxColorScheme.new(parent: self).parse(node)
+      self
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragraph_run/run_properties/position.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `w:position` object
+  class Position < OOXMLDocumentObject
+    # @return [String] value of position
+    attr_accessor :value
+
+    # Parse Position
+    # @param [Nokogiri::XML:Node] node with Position
+    # @return [Position] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'val'
+          @value = OoxmlSize.new(value.value.to_f, :half_point)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragraph_run/run_properties/run_spacing.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `w:sz` object
+  class RunSpacing < OOXMLDocumentObject
+    # @return [Integer] value of size
+    attr_accessor :value
+
+    # Parse RunSpacing
+    # @param [Nokogiri::XML:Node] node with RunSpacing
+    # @return [RunSpacing] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'val'
+          @value = OoxmlSize.new(value.value.to_f)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragraph_run/run_properties/run_style.rb

@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `w:rStyle` object
+  class RunStyle < OOXMLDocumentObject
+    # @return [Integer] value of size
+    attr_accessor :value
+
+    # Parse RunStyle
+    # @param [Nokogiri::XML:Node] node with RunStyle
+    # @return [RunStyle] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'val'
+          @value = value.value.to_s
+        end
+      end
+      self
+    end
+
+    # @return [DocumentStyle] style which was referenced in RunStyle
+    def referenced
+      root_object.document_style_by_id(@value)
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragraph_run/run_properties/shade.rb

@@ -0,0 +1,60 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `w:shd` object
+  class Shade < OOXMLDocumentObject
+    # @return [Symbol] value of shade
+    attr_accessor :value
+    # @return [Symbol] color of shade
+    attr_accessor :color
+    # @return [Color] fill of shade
+    attr_accessor :fill
+
+    def initialize(value: nil,
+                   color: nil,
+                   fill: nil,
+                   parent: nil)
+      @value = value
+      @color = color
+      @fill = fill
+      super(parent: parent)
+    end
+
+    # @return [String] text representation
+    def to_s
+      "Value: `#{value}`, "\
+        "Color: `#{color}`, "\
+        "Fill: `#{fill}`"
+    end
+
+    # Parse Shade
+    # @param [Nokogiri::XML:Node] node with Shade
+    # @return [Shade] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'val'
+          @value = value.value.to_sym
+        when 'color'
+          @color = value.value.to_sym
+        when 'fill'
+          @fill = Color.new(parent: self).parse_hex_string(value.value.to_s)
+        end
+      end
+      self
+    end
+
+    # Helper method to get background color
+    # @return [OoxmlParser::Color]
+    def to_background_color
+      return nil unless fill
+
+      background_color = fill
+      background_color.set_style(value) if value
+      background_color
+    end
+
+    extend Gem::Deprecate
+    deprecate :to_background_color, 'use shade direct methods', 2077, 1
+  end
+end