ooxml_parser 0.22.0 → 0.25.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f18fbe0fd1cdc17b4fa15edd44442ab0aab56a6620ba4539635b4e8c2d676b3c
-  data.tar.gz: de14faa95be23f0a4dc6fa9541adf184fed69bc769f837d30a2fd63321f7b765
+  metadata.gz: 409ce77d24b38512604d5215a2bd6a54f887685db71cd9942c81935ef678f3b0
+  data.tar.gz: 4f7edddf4bab2d5d6ab9485754b333a2943d2df96fdcc933d671e820cd9dc70d
 SHA512:
-  metadata.gz: 65800c4573bb1859c02d5795c0e45b6bd50b07349aaeba24e3c1565e09022f0a25c3ddff43780cb7642b02c76343ad929e3304e4bb4a75d555158c47ff47537e
-  data.tar.gz: b67e977b6af0047b52d2dce951fc57335bb80758bfd0a5188a9afa30a343ac408bcd227d41d14e0481e2ff4c3ae10dda283160dcd72962994f193aa69d8f59f4
+  metadata.gz: aad06b0a4dd775d5efe4411d43b87ebda68fe05a5a15924fb6773d77169141284baaf8047eee042634a2e6145649ac0fce2e9aa143d401587434598eeff5f73d
+  data.tar.gz: c2ab404a2ab3f559a312da5953da7a668b4ae1c359ff790bc4b4d577ccbd453e4314844125cc64e144c884f9e187389586c98f085c4d27ca5837d8a1b9e61f9d

data/lib/ooxml_parser/common_parser/common_data/alternate_content/drawing/graphic/shape/docx_shape/shape_style/{effect_reference.rb → style_matrix_reference.rb}

@@ -1,16 +1,16 @@
 # frozen_string_literal: true
 
 module OoxmlParser
-  # Class for parsing `a:effectRef` tags
-  class EffectReference < OOXMLDocumentObject
+  # Class for parsing `a:fillRef`, `a:effectRef`, `a:lnRef` tags
+  class StyleMatrixReference < OOXMLDocumentObject
     # @return [Integer] Style Matrix Index
     attr_reader :index
     # @return [Color] scheme color of EffectReference
     attr_reader :scheme_color
 
-    # Parse EffectReference object
+    # Parse StyleMatrixReference object
     # @param node [Nokogiri::XML:Element] node to parse
-    # @return [EffectReference] result of parsing
+    # @return [StyleMatrixReference] result of parsing
     def parse(node)
       node.attributes.each do |key, value|
         case key

data/lib/ooxml_parser/common_parser/common_data/alternate_content/drawing/graphic/shape/docx_shape/shape_style.rb

@@ -1,9 +1,7 @@
 # frozen_string_literal: true
 
-require_relative 'shape_style/effect_reference'
-require_relative 'shape_style/fill_reference'
+require_relative 'shape_style/style_matrix_reference'
 require_relative 'shape_style/font_reference'
-require_relative 'shape_style/line_reference'
 module OoxmlParser
   # Class for parsing `wps:style` tags
   class ShapeStyle < OOXMLDocumentObject
@@ -23,13 +21,13 @@ module OoxmlParser
       node.xpath('*').each do |node_child|
         case node_child.name
         when 'effectRef'
-          @effect_reference = EffectReference.new(parent: self).parse(node_child)
+          @effect_reference = StyleMatrixReference.new(parent: self).parse(node_child)
         when 'fillRef'
-          @fill_reference = FillReference.new(parent: self).parse(node_child)
+          @fill_reference = StyleMatrixReference.new(parent: self).parse(node_child)
         when 'fontRef'
           @font_reference = FontReference.new(parent: self).parse(node_child)
         when 'lnRef'
-          @line_reference = LineReference.new(parent: self).parse(node_child)
+          @line_reference = StyleMatrixReference.new(parent: self).parse(node_child)
         end
       end
       self

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

@@ -12,23 +12,6 @@ module OoxmlParser
     end
 
     class << self
-      # @param [String] string to parse
-      # @return [Coordinates]
-      def parse_coordinates_from_string(string)
-        coordinates = Coordinates.new
-        begin
-          coordinates.list = string.match(/'\w+'/)[0].delete("''")
-        rescue StandardError
-          'Raise Exception'
-        end
-        string = string.split('!').last
-        if coordinates?(string)
-          coordinates.row = string.scan(/[0-9]/).join.to_i
-          coordinates.column = string.scan(/[A-Z]/).join
-        end
-        coordinates
-      end
-
       # Parse range of coordinates
       # @param arguments_string [String] data to parse
       # @return [Array] result
@@ -109,6 +92,18 @@ module OoxmlParser
       end
     end
 
+    # Parse string to coordinates
+    # @param [String] string to parse
+    # @return [Coordinates] result
+    def parse_string(string)
+      string = extract_list_from_string(string) if list_name_in_string?(string)
+      if Coordinates.coordinates?(string)
+        @row = string.scan(/\d/).join.to_i
+        @column = string.scan(/[A-Z]/).join
+      end
+      self
+    end
+
     # @return [Integer] number of column
     # This method takes @column string
     # and converts into integer
@@ -118,10 +113,6 @@ module OoxmlParser
       end
     end
 
-    alias get_column_number column_number
-    extend Gem::Deprecate
-    deprecate :get_column_number, 'column_number', 2020, 1
-
     # Compares rows of two cells
     # @param [Coordinates] other_cell other cell coordinates
     # @return [true, false] true, if row greater, than other row
@@ -151,5 +142,23 @@ module OoxmlParser
     def ==(other)
       other.is_a?(Coordinates) ? (@row == other.row && @column == other.column) : false
     end
+
+    private
+
+    # Check if there is list name in string
+    # @param [String] string to check
+    # @return [Boolean] result
+    def list_name_in_string?(string)
+      string.include?('!')
+    end
+
+    # Extract list name and leave coordinates
+    # @param [String] string to parse
+    # @return [String] coordinates string without list
+    def extract_list_from_string(string)
+      split = string.split('!')
+      @list = split[0...-1].join('!')
+      split[-1]
+    end
   end
 end

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

@@ -43,14 +43,14 @@ module OoxmlParser
       node.attributes.each do |key, value|
         case key
         when 'location'
-          @url = Coordinates.parse_coordinates_from_string(value.value)
+          @url = Coordinates.new.parse_string(value.value)
         when 'id'
           @id = value.value
           @url = OOXMLDocumentObject.get_link_from_rels(@id) unless @id.empty?
         when 'tooltip'
           @tooltip = value.value
         when 'ref'
-          @coordinates = Coordinates.parse_coordinates_from_string(value.value)
+          @coordinates = Coordinates.new.parse_string(value.value)
         when 'action'
           @action_link = value.value
         when 'highlightClick'

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

@@ -1,6 +1,5 @@
 # frozen_string_literal: true
 
-require 'filemagic' unless Gem.win_platform?
 require 'securerandom'
 require 'nokogiri'
 require 'zip'
@@ -59,28 +58,6 @@ module OoxmlParser
       # @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

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

@@ -22,8 +22,8 @@ module OoxmlParser
 
     # @return [String] text representation
     def to_s
-      "Value: `#{value}`, "\
-        "Color: `#{color}`, "\
+      "Value: `#{value}`, " \
+        "Color: `#{color}`, " \
         "Fill: `#{fill}`"
     end
 

data/lib/ooxml_parser/common_parser/common_data/paragraph/paragrpah_properties/paragraph_spacing.rb

@@ -38,8 +38,8 @@ module OoxmlParser
     private
 
     # This is dirty workaround for situations
-    # Then @line_rule parsed after @line so getting
-    # @line value is totally screwed up
+    # Then `@line_rule` parsed after `@line` so getting
+    # `@line` value is totally screwed up
     # @param [Nokogiri::XML:Node] node with ParagraphSpacing
     # @return [Hash] hash with sorted values
     # TODO: Totally redone parsing of spacing to remove this workaround

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

@@ -8,11 +8,11 @@ module OoxmlParser
 
     # @return [String] result of convert of object to string
     def to_s
-      "Table position left: #{left}, "\
-        "right: #{right}, "\
-        "top: #{top}, "\
-        "bottom #{bottom}, "\
-        "position x: #{position_x}, "\
+      "Table position left: #{left}, " \
+        "right: #{right}, " \
+        "top: #{top}, " \
+        "bottom #{bottom}, " \
+        "position x: #{position_x}, " \
         "position y: #{position_y}"
     end
 

data/lib/ooxml_parser/common_parser/parser/encryption_checker.rb

@@ -0,0 +1,43 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Check if file is encrypted
+  class EncryptionChecker
+    def initialize(file_path, ignore_system: false)
+      @file_path = file_path
+      @ignore_system = ignore_system
+    end
+
+    # @return [Boolean] is file encrypted
+    def encrypted?
+      return false unless compatible_system?
+
+      # Support of Encrypted 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/0eb7c1b83341cc954620b45d2e2d65ee7df1a4e7/ChangeLog#L623
+      if mime_type.include?('encrypted') ||
+         mime_type.include?('Composite Document File V2 Document, No summary info') ||
+         mime_type.include?('application/CDFV2-corrupt')
+        warn("File #{@file_path} is encrypted. Can't parse it")
+        return true
+      end
+      false
+    end
+
+    private
+
+    # Check if system is compatible with `file` utility
+    # @return [Boolean] Result of check
+    def compatible_system?
+      return true if !Gem.win_platform? && !@ignore_system
+
+      warn 'Checking file for encryption is not supported on Windows'
+      false
+    end
+
+    # @return [String] mime type of file
+    def mime_type
+      @mime_type ||= `file -b --mime "#{@file_path}"`
+    end
+  end
+end

data/lib/ooxml_parser/common_parser/parser.rb

@@ -1,5 +1,7 @@
 # frozen_string_literal: true
 
+require_relative 'parser/encryption_checker'
+
 module OoxmlParser
   # Basic class for OoxmlParser
   class Parser
@@ -7,7 +9,7 @@ module OoxmlParser
     # @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)
+      return nil if EncryptionChecker.new(path_to_file).encrypted?
 
       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), '')

data/lib/ooxml_parser/common_parser.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+require_relative 'common_parser/common_data/ooxml_document_object'
+require_relative 'common_parser/common_data/color'
+require_relative 'common_parser/common_data/coordinates'
+require_relative 'common_parser/common_data/font_style'
+require_relative 'common_parser/common_data/borders_properties'
+require_relative 'common_parser/common_data/hyperlink'
+require_relative 'common_parser/common_data/paragraph'
+require_relative 'common_parser/common_data/alternate_content/alternate_content'
+require_relative 'common_parser/common_data/colors/presentation_fill'
+require_relative 'common_parser/common_data/table'
+require_relative 'common_parser/common_data/relationships'
+require_relative 'common_parser/parser'
+require_relative 'common_parser/common_document_structure'

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

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Class for parsing `w:instrText` object
+  class InstructionText < TextValue
+    # @return [Boolean] is current object for hyperlink
+    def hyperlink?
+      value.include?('HYPERLINK')
+    end
+
+    # @return [Boolean] is current object for page number
+    def page_number?
+      value.match?(/PAGE\s+\\\*/)
+    end
+
+    # @return [Hyperlink] convert InstructionText to Hyperlink
+    def to_hyperlink
+      Hyperlink.new(value.sub('HYPERLINK ', '').split(' \\o ').first,
+                    value.sub('HYPERLINK', '').split(' \\o ').last)
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/{docx_data/document_structure → document_structure}/docx_paragraph/docx_paragraph_run.rb

@@ -2,6 +2,7 @@
 
 # noinspection RubyTooManyInstanceVariablesInspection
 require_relative 'docx_paragraph_run/docx_paragraph_run_helpers'
+require_relative 'docx_paragraph_run/instruction_text'
 require_relative 'docx_paragraph_run/object'
 require_relative 'docx_paragraph_run/text_outline'
 require_relative 'docx_paragraph_run/text_fill'
@@ -15,9 +16,9 @@ module OoxmlParser
                   :link, :highlight, :effect, :caps, :w,
                   :position, :em, :spacing, :break, :touch, :shape, :footnote, :endnote, :fld_char, :style,
                   :comments, :alternate_content, :page_number, :text_outline, :text_fill
-    # @return [String] type of instruction used for upper level of run
-    # http://officeopenxml.com/WPfieldInstructions.php
-    attr_accessor :instruction
+    # @return [InstructionText] text of instruction
+    #   See ECMA-376, 17.16.23 instrText (Field Code)
+    attr_reader :instruction_text
     # @return [RunProperties] properties of run
     attr_accessor :run_properties
     # @return [RunObject] object of run
@@ -96,10 +97,10 @@ module OoxmlParser
           parse_properties(node_child)
           @run_properties = RunProperties.new(parent: self).parse(node_child)
         when 'instrText'
-          if node_child.text.include?('HYPERLINK')
-            hyperlink = Hyperlink.new(node_child.text.sub('HYPERLINK ', '').split(' \\o ').first, node_child.text.sub('HYPERLINK', '').split(' \\o ').last)
-            @link = hyperlink
-          elsif node_child.text[/PAGE\s+\\\*/]
+          @instruction_text = InstructionText.new(parent: self).parse(node_child)
+          if @instruction_text.hyperlink?
+            @link = @instruction_text.to_hyperlink
+          elsif @instruction_text.page_number?
             @text = '*PAGE NUMBER*'
           end
         when 'fldChar'

data/lib/ooxml_parser/docx_parser/{docx_data/document_structure → document_structure}/docx_paragraph/indents.rb

@@ -25,7 +25,7 @@ module OoxmlParser
     # Convert to string
     # @return [String] result of conversion
     def to_s
-      "first line indent: #{@first_line_indent}, left indent: #{@left_indent}, "\
+      "first line indent: #{@first_line_indent}, left indent: #{@left_indent}, " \
         "right indent: #{@right_indent}, hanging indent: #{@hanging_indent}"
     end
 

data/lib/ooxml_parser/docx_parser/{docx_data/document_structure → document_structure}/docx_paragraph.rb

@@ -18,7 +18,7 @@ module OoxmlParser
   # Class for data of DocxParagraph
   class DocxParagraph < OOXMLDocumentObject
     include DocxParagraphHelper
-    attr_accessor :number, :bookmark_start, :bookmark_end, :align, :spacing, :ind, :numbering,
+    attr_accessor :number, :align, :spacing, :ind, :numbering,
                   :character_style_array, :page_break, :borders, :keep_lines,
                   :contextual_spacing, :sector_properties, :page_numbering, :section_break, :style, :keep_next,
                   :orphan_control
@@ -39,8 +39,6 @@ module OoxmlParser
 
     def initialize(parent: nil)
       @number = 0
-      @bookmark_start = []
-      @bookmark_end = []
       @align = :left
       @spacing = Spacing.new
       @ind = Indents.new
@@ -62,8 +60,6 @@ module OoxmlParser
     # @return [void]
     def initialize_copy(source)
       super
-      @bookmark_start = source.bookmark_start.clone
-      @bookmark_end = source.bookmark_end.clone
       @character_style_array = source.character_style_array.clone
       @spacing = source.spacing.clone
     end