ooxml_parser 0.3.0 → 0.6.0

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

@@ -1,12 +1,17 @@
-require 'filemagic'
+# 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
 
@@ -27,6 +32,11 @@ module OoxmlParser
       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
       attr_accessor :root_subfolder
       attr_accessor :theme
@@ -36,6 +46,10 @@ module OoxmlParser
       # @param path_to_file [String] file
       # @return [True, False] Check if file is protected by password on open
       def encrypted_file?(path_to_file)
+        if Gem.win_platform?
+          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`
@@ -56,6 +70,7 @@ module OoxmlParser
         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
@@ -64,6 +79,7 @@ module OoxmlParser
         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))
@@ -93,7 +109,8 @@ module OoxmlParser
       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.parse_rels(rels_path)
+
+        relationships = Relationships.new.parse_file(rels_path)
         relationships.target_by_id(id)
       end
     end

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

@@ -1,126 +1,117 @@
-# Module for helper methods for OOXMLDocumentObject
-module OoxmlDocumentObjectHelper
-  # Convert object to hash
-  # @return [Hash]
-  def to_hash
-    result_hash = {}
-    instance_variables.each do |current_attribute|
-      next if current_attribute == :@parent
-      attribute_value = instance_variable_get(current_attribute)
-      next unless attribute_value
-      if attribute_value.is_a?(Array)
-        attribute_value.each_with_index do |object_element, index|
-          result_hash["#{current_attribute}_#{index}".to_sym] = object_element.to_hash
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Module for helper methods for OOXMLDocumentObject
+  module OoxmlDocumentObjectHelper
+    # Convert object to hash
+    # @return [Hash]
+    def to_hash
+      result_hash = {}
+      instance_variables.each do |current_attribute|
+        next if current_attribute == :@parent
+
+        attribute_value = instance_variable_get(current_attribute)
+        next unless attribute_value
+
+        if attribute_value.is_a?(Array)
+          attribute_value.each_with_index do |object_element, index|
+            result_hash["#{current_attribute}_#{index}".to_sym] = object_element.to_hash
+          end
+        else
+          result_hash[current_attribute.to_sym] = if attribute_value.respond_to?(:to_hash)
+                                                    attribute_value.to_hash
+                                                  else
+                                                    attribute_value.to_s
+                                                  end
         end
-      else
-        result_hash[current_attribute.to_sym] = if attribute_value.respond_to?(:to_hash)
-                                                  attribute_value.to_hash
-                                                else
-                                                  attribute_value.to_s
-                                                end
       end
+      result_hash
     end
-    result_hash
-  end
-
-  private
 
-  VALUE_TO_SYMBOL_HASH = { l: :left,
-                           ctr: :center,
-                           r: :right,
-                           just: :justify,
-                           b: :bottom,
-                           t: :top,
-                           tr: :top_right,
-                           tl: :top_left,
-                           br: :bottom_right,
-                           bl: :bottom_left,
-                           contentLocked: :content_locked,
-                           dist: :distributed,
-                           tb: :horizontal,
-                           rl: :rotate_on_90,
-                           lr: :rotate_on_270,
-                           inset: :in,
-                           lu: :left_up,
-                           ru: :right_up,
-                           ld: :left_down,
-                           rd: :right_down,
-                           d: :down,
-                           u: :up,
-                           lg: :large,
-                           med: :medium,
-                           sm: :small,
-                           beneathText: :beneath_text,
-                           pageBottom: :page_bottom,
-                           sectEnd: :section_end,
-                           docEnd: :document_end,
-                           eachSect: :each_section,
-                           eachPage: :each_page,
-                           sngStrike: :single,
-                           dblStrike: :double,
-                           noStrike: :none,
-                           lgDash: :large_dash,
-                           dashDot: :dash_dot,
-                           dashDotDot: :dash_dot_dot,
-                           mediumDashed: :medium_dashed,
-                           mediumDashDot: :medium_dash_dot,
-                           mediumDashDotDot: :medium_dash_dot_dot,
-                           lgDashDot: :large_dash_dot,
-                           lgDashDotDot: :large_dash_dot_dot,
-                           sdtLocked: :sdt_locked,
-                           sdtContentLocked: :sdt_content_locked,
-                           sysDash: :system_dash,
-                           sysDot: :system_dot,
-                           sysDashDot: :system_dash_dot,
-                           sysDashDotDot: :system_dash_dot_dot,
-                           rightMargin: :right_margin,
-                           leftMargin: :left_margin,
-                           bottomMargin: :bottom_margin,
-                           topMargin: :top_margin,
-                           undOvr: :under_over,
-                           subSup: :subscript_superscript,
-                           lrTb: :left_to_right_top_to_bottom,
-                           tbRl: :top_to_bottom_right_to_left,
-                           btLr: :bottom_to_top_left_to_right,
-                           singlelevel: :single_level,
-                           maxMin: :max_min,
-                           middleDot: :middle_dot,
-                           minMax: :min_max,
-                           multilevel: :multi_level,
-                           nextTo: :next_to,
-                           hybridMultilevel: :hybrid_multi_level,
-                           rnd: :round,
-                           sq: :square }.freeze
+    private
 
-  # Convert value to human readable symbol
-  # @param [String] value to convert
-  # @return [Symbol]
-  def value_to_symbol(value)
-    symbol = VALUE_TO_SYMBOL_HASH[value.value.to_sym]
-    return value.value.to_sym if symbol.nil?
-    symbol
-  end
+    VALUE_TO_SYMBOL_HASH = { l: :left,
+                             ctr: :center,
+                             r: :right,
+                             just: :justify,
+                             b: :bottom,
+                             t: :top,
+                             tr: :top_right,
+                             tl: :top_left,
+                             br: :bottom_right,
+                             bl: :bottom_left,
+                             contentLocked: :content_locked,
+                             dist: :distributed,
+                             tb: :horizontal,
+                             rl: :rotate_on_90,
+                             lr: :rotate_on_270,
+                             inset: :in,
+                             lu: :left_up,
+                             ru: :right_up,
+                             ld: :left_down,
+                             rd: :right_down,
+                             d: :down,
+                             u: :up,
+                             lg: :large,
+                             med: :medium,
+                             sm: :small,
+                             beneathText: :beneath_text,
+                             greaterThan: :greater_than,
+                             pageBottom: :page_bottom,
+                             sectEnd: :section_end,
+                             docEnd: :document_end,
+                             eachSect: :each_section,
+                             eachPage: :each_page,
+                             sngStrike: :single,
+                             dblStrike: :double,
+                             noStrike: :none,
+                             lgDash: :large_dash,
+                             dashDot: :dash_dot,
+                             dashDotDot: :dash_dot_dot,
+                             mediumDashed: :medium_dashed,
+                             mediumDashDot: :medium_dash_dot,
+                             mediumDashDotDot: :medium_dash_dot_dot,
+                             lgDashDot: :large_dash_dot,
+                             lgDashDotDot: :large_dash_dot_dot,
+                             sdtLocked: :sdt_locked,
+                             sdtContentLocked: :sdt_content_locked,
+                             sysDash: :system_dash,
+                             sysDot: :system_dot,
+                             sysDashDot: :system_dash_dot,
+                             sysDashDotDot: :system_dash_dot_dot,
+                             rightMargin: :right_margin,
+                             leftMargin: :left_margin,
+                             bottomMargin: :bottom_margin,
+                             topMargin: :top_margin,
+                             undOvr: :under_over,
+                             subSup: :subscript_superscript,
+                             lrTb: :left_to_right_top_to_bottom,
+                             tbRl: :top_to_bottom_right_to_left,
+                             btLr: :bottom_to_top_left_to_right,
+                             singlelevel: :single_level,
+                             maxMin: :max_min,
+                             middleDot: :middle_dot,
+                             minMax: :min_max,
+                             multilevel: :multi_level,
+                             nextTo: :next_to,
+                             hybridMultilevel: :hybrid_multi_level,
+                             rnd: :round,
+                             sq: :square }.freeze
 
-  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
-    !(status == 'false' || status == 'off' || status == '0')
-  end
+    # Convert value to human readable symbol
+    # @param [String] value to convert
+    # @return [Symbol]
+    def value_to_symbol(value)
+      symbol = VALUE_TO_SYMBOL_HASH[value.value.to_sym]
+      return value.value.to_sym if symbol.nil?
 
-  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
-    status == 'true' || status == 'on' || status == '1'
-  end
+      symbol
+    end
 
-  def root_object
-    tree_object = self
-    tree_object = tree_object.parent until tree_object.parent.nil?
-    tree_object
+    def root_object
+      tree_object = self
+      tree_object = tree_object.parent until tree_object.parent.nil?
+      tree_object
+    end
   end
 end

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/paragraph.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative 'paragraph/paragraph_properties'
 require_relative 'paragraph/paragraph_run'
 require_relative 'paragraph/text_field'

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

@@ -1,9 +1,12 @@
+# 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/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

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative 'paragraph_run/run_properties'
 require_relative 'paragraph_run/text'
 module OoxmlParser
@@ -32,5 +34,27 @@ module OoxmlParser
       end
       self
     end
+
+    # @return [True, False] is current run empty
+    def empty?
+      text.empty?
+    end
+
+    def instruction
+      parent.instruction
+    end
+
+    def page_number
+      parent.page_numbering?
+    end
+
+    def link
+      parent.parent.hyperlink
+    end
+
+    extend Gem::Deprecate
+    deprecate :instruction, 'parent.instruction', 2020, 1
+    deprecate :page_number, 'parent.page_numbering?', 2020, 1
+    deprecate :link, 'parent.parent.hyperlink', 2020, 1
   end
 end

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

@@ -1,4 +1,5 @@
-require_relative 'run_properties/language'
+# frozen_string_literal: true
+
 require_relative 'run_properties/outline'
 require_relative 'run_properties/position'
 require_relative 'run_properties/size'
@@ -20,8 +21,8 @@ module OoxmlParser
     attr_accessor :spacing
     # @return [RunSpacing] get color
     attr_accessor :color
-    # @return [Language] language property
-    attr_accessor :language
+    # @return [ValuedChild] language property
+    attr_reader :language
     # @return [Position] position property
     attr_accessor :position
     # @return [Shade] shade property
@@ -44,7 +45,7 @@ module OoxmlParser
     # @param node [Nokogiri::XML:Element] node to parse
     # @return [RunProperties] result of parsing
     def parse(node)
-      @font_style = Presentation.current_font_style.dup
+      @font_style = root_object.default_font_style.dup
       node.attributes.each do |key, value|
         case key
         when 'sz'
@@ -91,7 +92,7 @@ module OoxmlParser
         when 'color'
           @color = OoxmlColor.new(parent: self).parse(node_child)
         when 'solidFill'
-          @font_color = Color.parse_color(node_child.xpath('*').first)
+          @font_color = Color.new(parent: self).parse_color(node_child.xpath('*').first)
         when 'latin'
           @font_name = node_child.attribute('typeface').value
         when 'b'
@@ -113,7 +114,7 @@ module OoxmlParser
         when 'ln'
           @outline = Outline.new(parent: self).parse(node_child)
         when 'lang'
-          @language = Language.new(parent: self).parse(node_child)
+          @language = ValuedChild.new(:string, parent: self).parse(node_child)
         when 'position'
           @position = Position.new(parent: self).parse(node_child)
         when 'shd'
@@ -123,8 +124,8 @@ module OoxmlParser
         end
       end
       @font_color = DocxColorScheme.new(parent: self).parse(node)
-      @font_name = Presentation.default_font_typeface if @font_name.empty?
-      @font_size = Presentation.default_font_size if @font_size.nil?
+      @font_name = root_object.default_font_typeface if @font_name.empty?
+      @font_size ||= root_object.default_font_size
       self
     end
   end

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 # Class for parsing Run Outline properties
 module OoxmlParser
   # Class for parsing `w:ln` tags

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

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for parsing `w:position` object
   class Position < OOXMLDocumentObject