roo 2.6.0 → 2.8.3

data/lib/roo/helpers/weak_instance_cache.rb

@@ -0,0 +1,41 @@
+# frozen_string_literal: true
+
+require "weakref"
+
+module Roo
+  module Helpers
+    module WeakInstanceCache
+      private
+
+      def instance_cache(key)
+        object = nil
+
+        if instance_variable_defined?(key) && (ref = instance_variable_get(key)) && ref.weakref_alive?
+          begin
+            object = ref.__getobj__
+          rescue => e
+            unless (defined?(::WeakRef::RefError) && e.is_a?(::WeakRef::RefError)) || (defined?(RefError) && e.is_a?(RefError))
+              raise e
+            end
+          end
+        end
+
+        unless object
+          object = yield
+          ObjectSpace.define_finalizer(object, instance_cache_finalizer(key))
+          instance_variable_set(key, WeakRef.new(object))
+        end
+
+        object
+      end
+
+      def instance_cache_finalizer(key)
+        proc do |object_id|
+          if instance_variable_defined?(key) && (ref = instance_variable_get(key)) && (!ref.weakref_alive? || ref.__getobj__.object_id == object_id)
+            remove_instance_variable(key)
+          end
+        end
+      end
+    end
+  end
+end

data/lib/roo/open_office.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'date'
 require 'nokogiri'
 require 'cgi'
@@ -5,14 +7,15 @@ require 'zip/filesystem'
 require 'roo/font'
 require 'roo/tempdir'
 require 'base64'
+require 'openssl'
 
 module Roo
   class OpenOffice < Roo::Base
     extend Roo::Tempdir
 
-    ERROR_MISSING_CONTENT_XML = 'file missing required content.xml'.freeze
-    XPATH_FIND_TABLE_STYLES   = "//*[local-name()='automatic-styles']".freeze
-    XPATH_LOCAL_NAME_TABLE    = "//*[local-name()='table']".freeze
+    ERROR_MISSING_CONTENT_XML = 'file missing required content.xml'
+    XPATH_FIND_TABLE_STYLES   = "//*[local-name()='automatic-styles']"
+    XPATH_LOCAL_NAME_TABLE    = "//*[local-name()='table']"
 
     # initialization and opening of a spreadsheet file
     # values for packed: :zip
@@ -22,7 +25,12 @@ module Roo
 
       @only_visible_sheets = options[:only_visible_sheets]
       file_type_check(filename, '.ods', 'an Roo::OpenOffice', file_warning, packed)
-      @tmpdir   = self.class.make_tempdir(self, find_basename(filename), options[:tmpdir_root])
+      # NOTE: Create temp directory and allow Ruby to cleanup the temp directory
+      #       when the object is garbage collected. Initially, the finalizer was
+      #       created in the Roo::Tempdir module, but that led to a segfault
+      #       when testing in Ruby 2.4.0.
+      @tmpdir = self.class.make_tempdir(self, find_basename(filename), options[:tmpdir_root])
+      ObjectSpace.define_finalizer(self, self.class.finalize(object_id))
       @filename = local_filename(filename, @tmpdir, packed)
       # TODO: @cells_read[:default] = false
       open_oo_file(options)
@@ -340,7 +348,7 @@ module Roo
     def find_cipher(*args)
       fail ArgumentError, 'Unknown algorithm ' + algorithm unless args[0] == 'http://www.w3.org/2001/04/xmlenc#aes256-cbc'
 
-      cipher = OpenSSL::Cipher.new('AES-256-CBC')
+      cipher = ::OpenSSL::Cipher.new('AES-256-CBC')
       cipher.decrypt
       cipher.padding = 0
       cipher.key     = find_cipher_key(cipher, *args[1..4])
@@ -353,7 +361,7 @@ module Roo
     def find_cipher_key(*args)
       fail ArgumentError, 'Unknown key derivation name ', args[1] unless args[1] == 'PBKDF2'
 
-      OpenSSL::PKCS5.pbkdf2_hmac_sha1(args[2], args[3], args[4], args[0].key_len)
+      ::OpenSSL::PKCS5.pbkdf2_hmac_sha1(args[2], args[3], args[4], args[0].key_len)
     end
 
     # Block decrypt raw bytes from the zip file based on the cipher
@@ -555,7 +563,7 @@ module Roo
     end
 
     def read_labels
-      @label ||= Hash[doc.xpath('//table:named-range').map do |ne|
+      @label ||= doc.xpath('//table:named-range').each_with_object({}) do |ne, hash|
         #-
         # $Sheet1.$C$5
         #+
@@ -563,8 +571,8 @@ module Roo
         sheetname, coords = attribute(ne, 'cell-range-address').to_s.split('.$')
         col, row          = coords.split('$')
         sheetname         = sheetname[1..-1] if sheetname[0, 1] == '$'
-        [name, [sheetname, row, col]]
-      end]
+        hash[name] = [sheetname, row, col]
+      end
     end
 
     def read_styles(style_elements)

data/lib/roo/spreadsheet.rb

@@ -24,7 +24,7 @@ module Roo
           options[:file_warning] = :ignore
           extension.tr('.', '').downcase.to_sym
         else
-          res = ::File.extname((path =~ /\A#{::URI.regexp}\z/) ? ::URI.parse(::URI.encode(path)).path : path)
+          res = ::File.extname((path =~ /\A#{::URI::DEFAULT_PARSER.make_regexp}\z/) ? ::URI.parse(::URI.encode(path)).path : path)
           res.tr('.', '').downcase.to_sym
         end
       end

data/lib/roo/tempdir.rb

@@ -2,7 +2,7 @@ module Roo
   module Tempdir
     def finalize_tempdirs(object_id)
       if @tempdirs && (dirs_to_remove = @tempdirs[object_id])
-        @tempdirs[object_id] = nil
+        @tempdirs.delete(object_id)
         dirs_to_remove.each do |dir|
           ::FileUtils.remove_entry(dir)
         end
@@ -10,16 +10,11 @@ module Roo
     end
 
     def make_tempdir(object, prefix, root)
-      root ||= ENV['ROO_TMP']
-      # folder is cleaned up in .finalize_tempdirs
+      root ||= ENV["ROO_TMP"]
+      # NOTE: This folder is cleaned up by finalize_tempdirs.
       ::Dir.mktmpdir("#{Roo::TEMP_PREFIX}#{prefix}", root).tap do |tmpdir|
-        @tempdirs ||= {}
-        if @tempdirs[object.object_id]
-          @tempdirs[object.object_id] << tmpdir
-        else
-          @tempdirs[object.object_id] = [tmpdir]
-          ObjectSpace.define_finalizer(object, method(:finalize_tempdirs))
-        end
+        @tempdirs ||= Hash.new { |h, k| h[k] = [] }
+        @tempdirs[object.object_id] << tmpdir
       end
     end
   end

data/lib/roo/utils.rb

@@ -1,35 +1,49 @@
+# frozen_string_literal: true
+
 module Roo
   module Utils
     extend self
 
     LETTERS = ('A'..'Z').to_a
 
-    def split_coordinate(str)
-      @split_coordinate ||= {}
+    def extract_coordinate(s)
+      num = letter_num = 0
+      num_only = false
 
-      @split_coordinate[str] ||= begin
-        letter, number = split_coord(str)
-        x = letter_to_number(letter)
-        y = number
-        [y, x]
+      s.each_byte do |b|
+        if !num_only && (index = char_index(b))
+          letter_num *= 26
+          letter_num += index
+        elsif index = num_index(b)
+          num_only = true
+          num *= 10
+          num += index
+        else
+          fail ArgumentError
+        end
       end
+      fail ArgumentError if letter_num == 0 || !num_only
+
+      Excelx::Coordinate.new(num, letter_num)
     end
 
-    alias_method :ref_to_key, :split_coordinate
+    alias_method :ref_to_key, :extract_coordinate
 
-    def split_coord(s)
-      if s =~ /([a-zA-Z]+)([0-9]+)/
-        letter = Regexp.last_match[1]
-        number = Regexp.last_match[2].to_i
-      else
-        fail ArgumentError
-      end
-      [letter, number]
+    def split_coordinate(str)
+      warn "[DEPRECATION] `Roo::Utils.split_coordinate` is deprecated.  Please use `Roo::Utils.extract_coordinate` instead."
+      extract_coordinate(str)
+    end
+
+
+
+    def split_coord(str)
+      coord = extract_coordinate(str)
+      [number_to_letter(coord.column), coord.row]
     end
 
     # convert a number to something like 'AB' (1 => 'A', 2 => 'B', ...)
     def number_to_letter(num)
-      result = ""
+      result = +""
 
       until num.zero?
         num, index = (num - 1).divmod(26)
@@ -56,11 +70,30 @@ module Roo
       cells = str.split(':')
       return 1 if cells.count == 1
       raise ArgumentError.new("invalid range string: #{str}. Supported range format 'A1:B2'") if cells.count != 2
-      x1, y1 = split_coordinate(cells[0])
-      x2, y2 = split_coordinate(cells[1])
+      x1, y1 = extract_coordinate(cells[0])
+      x2, y2 = extract_coordinate(cells[1])
       (x2 - (x1 - 1)) * (y2 - (y1 - 1))
     end
 
+    def coordinates_in_range(str)
+      return to_enum(:coordinates_in_range, str) unless block_given?
+      coordinates = str.split(":", 2).map! { |s| extract_coordinate s }
+
+      case coordinates.size
+      when 1
+        yield coordinates[0]
+      when 2
+        tl, br = coordinates
+        rows = tl.row..br.row
+        cols = tl.column..br.column
+        rows.each do |row|
+          cols.each do |column|
+            yield Excelx::Coordinate.new(row, column)
+          end
+        end
+      end
+    end
+
     def load_xml(path)
       ::File.open(path, 'rb') do |file|
         ::Nokogiri::XML(file)
@@ -69,10 +102,27 @@ module Roo
 
     # Yield each element of a given type ('row', 'c', etc.) to caller
     def each_element(path, elements)
+      elements = Array(elements)
       Nokogiri::XML::Reader(::File.open(path, 'rb'), nil, nil, Nokogiri::XML::ParseOptions::NOBLANKS).each do |node|
-        next unless node.node_type == Nokogiri::XML::Reader::TYPE_ELEMENT && Array(elements).include?(node.name)
+        next unless node.node_type == Nokogiri::XML::Reader::TYPE_ELEMENT && elements.include?(node.name)
         yield Nokogiri::XML(node.outer_xml).root if block_given?
       end
     end
+
+    private
+
+    def char_index(byte)
+      if byte >= 65 && byte <= 90
+        byte - 64
+      elsif byte >= 97 && byte <= 122
+        byte - 96
+      end
+    end
+
+    def num_index(byte)
+      if byte >= 48 && byte <= 57
+        byte - 48
+      end
+    end
   end
 end

data/lib/roo/version.rb

@@ -1,3 +1,3 @@
 module Roo
-  VERSION = "2.6.0"
+  VERSION = "2.8.3"
 end

data/lib/roo.rb

@@ -1,3 +1,6 @@
+# frozen_string_literal: true
+
+require 'roo/version'
 require 'roo/constants'
 require 'roo/errors'
 require 'roo/spreadsheet'
@@ -9,7 +12,7 @@ module Roo
   autoload :Excelx,       'roo/excelx'
   autoload :CSV,          'roo/csv'
 
-  TEMP_PREFIX = 'roo_'.freeze
+  TEMP_PREFIX = 'roo_'
 
   CLASS_FOR_EXTENSION = {
     ods: Roo::OpenOffice,

data/roo.gemspec

@@ -4,22 +4,25 @@ $LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
 require 'roo/version'
 
 Gem::Specification.new do |spec|
-  spec.name          = 'roo'
-  spec.version       = Roo::VERSION
-  spec.authors       = ['Thomas Preymesser', 'Hugh McGowan', 'Ben Woosley', 'Oleksandr Simonov', 'Steven Daniels']
-  spec.email         = ['ruby.ruby.ruby.roo@gmail.com', 'oleksandr@simonov.me']
-  spec.summary       = 'Roo can access the contents of various spreadsheet files.'
-  spec.description   = "Roo can access the contents of various spreadsheet files. It can handle\n* OpenOffice\n* Excelx\n* LibreOffice\n* CSV"
-  spec.homepage      = 'http://github.com/roo-rb/roo'
-  spec.license       = 'MIT'
+  spec.name                   = 'roo'
+  spec.version                = Roo::VERSION
+  spec.authors                = ['Thomas Preymesser', 'Hugh McGowan', 'Ben Woosley', 'Oleksandr Simonov', 'Steven Daniels', 'Anmol Chopra']
+  spec.email                  = ['ruby.ruby.ruby.roo@gmail.com', 'oleksandr@simonov.me']
+  spec.summary                = 'Roo can access the contents of various spreadsheet files.'
+  spec.description            = "Roo can access the contents of various spreadsheet files. It can handle\n* OpenOffice\n* Excelx\n* LibreOffice\n* CSV"
+  spec.homepage               = 'https://github.com/roo-rb/roo'
+  spec.license                = 'MIT'
 
-  spec.files         = `git ls-files -z`.split("\x0")
+  spec.files                  = `git ls-files -z`.split("\x0")
   spec.files.reject! { |fn| fn.include?('test/files') }
-  spec.require_paths = ['lib']
+  spec.require_paths          = ['lib']
+
+  spec.required_ruby_version  = ">= 2.3.0"
 
   spec.add_dependency 'nokogiri', '~> 1'
-  spec.add_dependency 'rubyzip', '~> 1.1', '< 2.0.0'
+  spec.add_dependency 'rubyzip', '>= 1.3.0', '< 3.0.0'
 
   spec.add_development_dependency 'rake', '~> 10.1'
   spec.add_development_dependency 'minitest', '~> 5.4', '>= 5.4.3'
+  spec.add_development_dependency 'rack', '~> 1.6', '< 2.0.0'
 end

data/spec/lib/roo/base_spec.rb

@@ -127,10 +127,22 @@ describe Roo::Base do
     end
   end
 
-  describe '#row' do
-    it 'should return the specified row' do
+  describe "#row" do
+    it "should return the specified row" do
       expect(spreadsheet.row(12)).to eq([41.0, 42.0, 43.0, 44.0, 45.0, nil, nil])
-      expect(spreadsheet.row(16)).to eq([nil, '"Hello world!"', 'forty-three', 'forty-four', 'forty-five', nil, nil])
+      expect(spreadsheet.row(16)).to eq([nil, '"Hello world!"', "forty-three", "forty-four", "forty-five", nil, nil])
+    end
+
+    it "should return the specified row if default_sheet is set by a string" do
+      spreadsheet.default_sheet = "my_sheet"
+      expect(spreadsheet.row(12)).to eq([41.0, 42.0, 43.0, 44.0, 45.0, nil, nil])
+      expect(spreadsheet.row(16)).to eq([nil, '"Hello world!"', "forty-three", "forty-four", "forty-five", nil, nil])
+    end
+
+    it "should return the specified row if default_sheet is set by an integer" do
+      spreadsheet.default_sheet = 0
+      expect(spreadsheet.row(12)).to eq([41.0, 42.0, 43.0, 44.0, 45.0, nil, nil])
+      expect(spreadsheet.row(16)).to eq([nil, '"Hello world!"', "forty-three", "forty-four", "forty-five", nil, nil])
     end
   end
 
@@ -146,6 +158,11 @@ describe Roo::Base do
         expect { spreadsheet.row_with([/Missing Header/]) }.to \
           raise_error(Roo::HeaderRowNotFoundError)
       end
+
+      it 'returns missing headers' do
+        expect { spreadsheet.row_with([/Header/, /Missing Header 1/, /Missing Header 2/]) }.to \
+          raise_error(Roo::HeaderRowNotFoundError, '[/Missing Header 1/, /Missing Header 2/]')
+      end
     end
   end
 
@@ -173,6 +190,31 @@ describe Roo::Base do
     end
   end
 
+  describe "#default_sheet=" do
+    it "should correctly set the default sheet if passed a string" do
+      spreadsheet.default_sheet = "my_sheet"
+      expect(spreadsheet.default_sheet).to eq("my_sheet")
+    end
+
+    it "should correctly set the default sheet if passed an integer" do
+      spreadsheet.default_sheet = 0
+      expect(spreadsheet.default_sheet).to eq("my_sheet")
+    end
+
+    it "should correctly set the default sheet if passed an integer for the second sheet" do
+      spreadsheet.default_sheet = 1
+      expect(spreadsheet.default_sheet).to eq("blank sheet")
+    end
+
+    it "should raise an error if passed a sheet that does not exist as an integer" do
+      expect { spreadsheet.default_sheet = 10 }.to raise_error RangeError
+    end
+
+    it "should raise an error if passed a sheet that does not exist as a string" do
+      expect { spreadsheet.default_sheet = "does_not_exist" }.to raise_error RangeError
+    end
+  end
+
   describe '#to_yaml' do
     it 'should convert the spreadsheet to yaml' do
       expect(spreadsheet.to_yaml({}, 5, 1, 5, 1)).to eq("--- \n" + yaml_entry(5, 1, 'date', '1961-11-21'))

data/spec/lib/roo/excelx/relationships_spec.rb

@@ -0,0 +1,43 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+
+describe Roo::Excelx::Relationships do
+  subject(:relationships) { Roo::Excelx::Relationships.new Roo::Excelx.new(path).rels_files[0] }
+
+  describe "#include_type?" do
+    [
+      ["with hyperlink type", "test/files/link.xlsx", true, false],
+      ["with nil path", "test/files/Bibelbund.xlsx", false, false],
+      ["with comments type", "test/files/comments-google.xlsx", false, true],
+    ].each do |context_desc, file_path, hyperlink_value, comments_value|
+      context context_desc do
+        let(:path) { file_path }
+
+        it "should return #{hyperlink_value} for hyperlink" do
+          expect(subject.include_type?("hyperlink")).to be hyperlink_value
+        end
+
+        it "should return #{hyperlink_value} for link" do
+          expect(subject.include_type?("link")).to be hyperlink_value
+        end
+
+        it "should return false for hypelink" do
+          expect(subject.include_type?("hypelink")).to be false
+        end
+
+        it "should return false for coment" do
+          expect(subject.include_type?("coment")).to be false
+        end
+
+        it "should return #{comments_value} for comments" do
+          expect(subject.include_type?("comments")).to be comments_value
+        end
+
+        it "should return #{comments_value} for comment" do
+          expect(subject.include_type?("comment")).to be comments_value
+        end
+      end
+    end
+  end
+end

data/spec/lib/roo/excelx/sheet_doc_spec.rb

@@ -0,0 +1,11 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+
+describe Roo::Excelx::SheetDoc do
+  subject(:blank_children) { Roo::Excelx.new("test/files/blank_children.xlsx") }
+
+  example "#last_row" do
+    expect(subject.last_row).to eq 3
+  end
+end