ooxml_parser 0.21.0 → 0.24.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 6d49ee5c48167b1fec3d73c1f146b1f5039ad13591d91cd01076c2026a4da9d5
-  data.tar.gz: c903db5d14db5c9603c1a42b8c3a96666ea421ceb6264c3de142a23e6b4f4871
+  metadata.gz: 7709109a298642b27297b3646d9b94c2dd73dd5ba6d52468dfe7c90c221d4809
+  data.tar.gz: '09db15c81f7f0c31d2a0bd7f5afcb7606c8455baa7140b1d0a340234ec34e74f'
 SHA512:
-  metadata.gz: 7e1a32945ce66a7b668a734526a5cdb3f936a4ef78f0e88dfb3955e10505a21e7fae326cc3f515977bbee0b04e7cdd0d584963243b2c756de41f25c9be93514f
-  data.tar.gz: 1f01fbad38dd7c8ad7271d01fdc93ebb693f71ec0af1483bec680f3afbee8a6f32ded803cd01e889aba7c767de6bab3b6ce732a07ed5d771dff9faf95c023ffd
+  metadata.gz: 515a28a6d616cb0bd272645ab0ec4d8c5365fcbb96f58141552a70eec3a34dd3e22ff207f997f1baa4624216b3452a351e803f5b93c6edbb73f12ad344b07448
+  data.tar.gz: e81fc4ad01b4fc326eea90a3a8758b95b9f7fe56cb4c6c46d53c2b50d0ba472007ee6c7412d0510d9550632c1ff6ff52202b6519ea8f92adb66c443137bef52c

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/alternate_content/drawing/graphic/shape/docx_shape.rb

@@ -10,9 +10,16 @@ module OoxmlParser
   # Class for parsing `sp`, `wsp` tags
   class DocxShape < OOXMLDocumentObject
     attr_accessor :non_visual_properties, :properties, :style, :body_properties, :text_body
+    # @return [True, False] Specifies if text in shape is locked when sheet is protected
+    attr_reader :locks_text
 
     alias shape_properties properties
 
+    def initialize(parent: nil)
+      @locks_text = true
+      super
+    end
+
     # @return [True, false] if structure contain any user data
     def with_data?
       return true if @text_body.nil?
@@ -27,6 +34,13 @@ module OoxmlParser
     # @param node [Nokogiri::XML:Element] node to parse
     # @return [DocxShape] result of parsing
     def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'fLocksText'
+          @locks_text = attribute_enabled?(value)
+        end
+      end
+
       node.xpath('*').each do |node_child|
         case node_child.name
         when 'nvSpPr'

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

@@ -1,9 +1,9 @@
 # frozen_string_literal: true
 
-require 'filemagic' unless Gem.win_platform?
 require 'securerandom'
 require 'nokogiri'
 require 'zip'
+require 'ooxml_decrypt'
 require_relative 'ooxml_document_object/nokogiri_parsing_exception'
 require_relative 'ooxml_document_object/ooxml_document_object_helper'
 require_relative 'ooxml_document_object/ooxml_object_attribute_helper'
@@ -58,28 +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
@@ -95,6 +73,20 @@ module OoxmlParser
         file_path
       end
 
+      # Decrypt file protected with password
+      # @param path [String] path to file
+      # @param password [String] password to file
+      # @return [String] path to decrypted file
+      def decrypt_file(path, password)
+        file_name = File.basename(path)
+        tmp_folder = Dir.mktmpdir('ruby-ooxml-parser')
+        decrypted_path = "#{tmp_folder}/#{file_name}"
+        binary_password = password.encode('utf-16le').bytes.pack('c*').encode('binary')
+        OoxmlDecrypt::EncryptedFile.decrypt_to_file(path, binary_password, decrypted_path)
+
+        decrypted_path
+      end
+
       # Unzip specified file
       # @param path_to_file [String] path to zip file
       # @param destination [String] folder to extract

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), '')
@@ -21,7 +23,8 @@ module OoxmlParser
     # 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)
+    def self.parse(path_to_file, password: nil)
+      path_to_file = OOXMLDocumentObject.decrypt_file(path_to_file, password) if password
       Parser.parse_format(path_to_file) do
         format = Parser.recognize_folder_format
         case format

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/docx_parser.rb → docx_parser.rb}

@@ -1,8 +1,7 @@
 # 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'
+require_relative 'common_parser/common_data/valued_child'
+require_relative 'docx_parser/document_structure'
 
 module OoxmlParser
   # Basic class for DocxParser