roo 1.13.2 → 2.0.0beta1

data/lib/roo/spreadsheet.rb

@@ -1,32 +1,29 @@
 module Roo
   class Spreadsheet
     class << self
-      def open(file, options = {})
-        file = File === file ? file.path : file
+      def open(path, options = {})
+        path      = path.respond_to?(:path) ? path.path : path
+        extension = extension_for(path, options)
 
-        extension =
-          if options[:extension]
-            options[:file_warning] = :ignore
-            ".#{options[:extension]}".gsub(/[.]+/, ".")
-          else
-            File.extname(URI.parse(file).path)
-          end
+        begin
+          Roo::CLASS_FOR_EXTENSION.fetch(extension).new(path, options)
+        rescue KeyError
+          raise ArgumentError,
+                "Can't detect the type of #{path} - please use the :extension option to declare its type."
+        end
+      end
 
-        case extension.downcase
-        when '.xls'
-          Roo::Excel.new(file, options)
-        when '.xlsx'
-          Roo::Excelx.new(file, options)
-        when '.ods'
-          Roo::OpenOffice.new(file, options)
-        when '.xml'
-          Roo::Excel2003XML.new(file, options)
-        when ''
-          Roo::Google.new(file, options)
-        when '.csv'
-          Roo::CSV.new(file, options)
+      def extension_for(path, options)
+        case (extension = options.delete(:extension))
+        when ::Symbol
+          options[:file_warning] = :ignore
+          extension
+        when ::String
+          options[:file_warning] = :ignore
+          extension.tr('.', '').downcase.to_sym
         else
-          raise ArgumentError, "Don't know how to open file #{file}"
+          res = ::File.extname((path =~ ::URI.regexp) ? ::URI.parse(::URI.encode(path)).path : path)
+          res.tr('.', '').downcase.to_sym
         end
       end
     end

data/lib/roo/utils.rb

@@ -0,0 +1,78 @@
+module Roo
+  module Utils
+    extend self
+
+    def split_coordinate(str)
+      @split_coordinate ||= {}
+
+      @split_coordinate[str] ||= begin
+        letter, number = split_coord(str)
+        x = letter_to_number(letter)
+        y = number
+        [y, x]
+      end
+    end
+
+    alias_method :ref_to_key, :split_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]
+    end
+
+    # convert a number to something like 'AB' (1 => 'A', 2 => 'B', ...)
+    def number_to_letter(num)
+      results = []
+      num = num.to_i
+
+      while (num > 0)
+        mod = (num - 1) % 26
+        results = [(65 + mod).chr] + results
+        num = ((num - mod) / 26)
+      end
+
+      results.join
+    end
+
+    def letter_to_number(letters)
+      @letter_to_number ||= {}
+      @letter_to_number[letters] ||= begin
+         result = 0
+
+         # :bytes method returns an enumerator in 1.9.3 and an array in 2.0+
+         letters.bytes.to_a.map{|b| b > 96 ? b - 96 : b - 64 }.reverse.each_with_index{ |num, i| result += num * 26 ** i }
+
+         result
+      end
+    end
+
+    # Compute upper bound for cells in a given cell range.
+    def num_cells_in_range(str)
+      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])
+      (x2 - (x1 - 1)) * (y2 - (y1 - 1))
+    end
+
+    def load_xml(path)
+      ::File.open(path, 'rb') do |file|
+        ::Nokogiri::XML(file)
+      end
+    end
+
+    # Yield each element of a given type ('row', 'c', etc.) to caller
+    def each_element(path, 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)
+        yield Nokogiri::XML(node.outer_xml).root if block_given?
+      end
+    end
+  end
+end

data/lib/roo/version.rb

@@ -0,0 +1,3 @@
+module Roo
+  VERSION = "2.0.0beta1"
+end

data/roo.gemspec

@@ -1,209 +1,25 @@
-# Generated by jeweler
-# DO NOT EDIT THIS FILE DIRECTLY
-# Instead, edit Jeweler::Tasks in Rakefile, and run 'rake gemspec'
-# -*- encoding: utf-8 -*-
+# encoding: utf-8
+lib = File.expand_path('../lib', __FILE__)
+$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
+require 'roo/version'
 
-Gem::Specification.new do |s|
-  s.name = "roo"
-  s.version = "1.13.2"
+Gem::Specification.new do |spec|
+  spec.name          = 'roo'
+  spec.version       = Roo::VERSION
+  spec.authors       = ['Thomas Preymesser', 'Hugh McGowan', 'Ben Woosley', 'Oleksandr Simonov']
+  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'
 
-  s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
-  s.authors = ["Thomas Preymesser", "Hugh McGowan", "Ben Woosley"]
-  s.date = "2013-12-23"
-  s.description = "Roo can access the contents of various spreadsheet files. It can handle\n* OpenOffice\n* Excel\n* Google spreadsheets\n* Excelx\n* LibreOffice\n* CSV"
-  s.email = "ruby.ruby.ruby.roo@gmail.com"
-  s.extra_rdoc_files = [
-    "LICENSE",
-    "README.markdown"
-  ]
-  s.files = [
-    "CHANGELOG",
-    "Gemfile",
-    "Gemfile.lock",
-    "LICENSE",
-    "README.markdown",
-    "Rakefile",
-    "VERSION",
-    "examples/roo_soap_client.rb",
-    "examples/roo_soap_server.rb",
-    "examples/write_me.rb",
-    "lib/roo.rb",
-    "lib/roo/base.rb",
-    "lib/roo/csv.rb",
-    "lib/roo/excel.rb",
-    "lib/roo/excel2003xml.rb",
-    "lib/roo/excelx.rb",
-    "lib/roo/google.rb",
-    "lib/roo/openoffice.rb",
-    "lib/roo/roo_rails_helper.rb",
-    "lib/roo/spreadsheet.rb",
-    "lib/roo/worksheet.rb",
-    "roo.gemspec",
-    "scripts/txt2html",
-    "spec/fixtures/vcr_cassettes/google_drive.yml",
-    "spec/fixtures/vcr_cassettes/google_drive_access_token.yml",
-    "spec/fixtures/vcr_cassettes/google_drive_set.yml",
-    "spec/lib/roo/base_spec.rb",
-    "spec/lib/roo/csv_spec.rb",
-    "spec/lib/roo/excel2003xml_spec.rb",
-    "spec/lib/roo/excel_spec.rb",
-    "spec/lib/roo/excelx/format_spec.rb",
-    "spec/lib/roo/excelx_spec.rb",
-    "spec/lib/roo/google_spec.rb",
-    "spec/lib/roo/libreoffice_spec.rb",
-    "spec/lib/roo/openoffice_spec.rb",
-    "spec/lib/roo/spreadsheet_spec.rb",
-    "spec/spec_helper.rb",
-    "test/all_ss.rb",
-    "test/files/1900_base.xls",
-    "test/files/1900_base.xlsx",
-    "test/files/1904_base.xls",
-    "test/files/1904_base.xlsx",
-    "test/files/Bibelbund.csv",
-    "test/files/Bibelbund.ods",
-    "test/files/Bibelbund.xls",
-    "test/files/Bibelbund.xlsx",
-    "test/files/Bibelbund.xml",
-    "test/files/Bibelbund1.ods",
-    "test/files/Pfand_from_windows_phone.xlsx",
-    "test/files/bad_excel_date.xls",
-    "test/files/bbu.ods",
-    "test/files/bbu.xls",
-    "test/files/bbu.xlsx",
-    "test/files/bbu.xml",
-    "test/files/bode-v1.ods.zip",
-    "test/files/bode-v1.xls.zip",
-    "test/files/boolean.csv",
-    "test/files/boolean.ods",
-    "test/files/boolean.xls",
-    "test/files/boolean.xlsx",
-    "test/files/boolean.xml",
-    "test/files/borders.ods",
-    "test/files/borders.xls",
-    "test/files/borders.xlsx",
-    "test/files/borders.xml",
-    "test/files/bug-numbered-sheet-names.xlsx",
-    "test/files/bug-row-column-fixnum-float.xls",
-    "test/files/bug-row-column-fixnum-float.xml",
-    "test/files/comments.ods",
-    "test/files/comments.xls",
-    "test/files/comments.xlsx",
-    "test/files/csvtypes.csv",
-    "test/files/datetime.ods",
-    "test/files/datetime.xls",
-    "test/files/datetime.xlsx",
-    "test/files/datetime.xml",
-    "test/files/datetime_floatconv.xls",
-    "test/files/datetime_floatconv.xml",
-    "test/files/dreimalvier.ods",
-    "test/files/emptysheets.ods",
-    "test/files/emptysheets.xls",
-    "test/files/emptysheets.xlsx",
-    "test/files/emptysheets.xml",
-    "test/files/excel2003.xml",
-    "test/files/false_encoding.xls",
-    "test/files/false_encoding.xml",
-    "test/files/file_item_error.xlsx",
-    "test/files/formula.ods",
-    "test/files/formula.xls",
-    "test/files/formula.xlsx",
-    "test/files/formula.xml",
-    "test/files/formula_parse_error.xls",
-    "test/files/formula_parse_error.xml",
-    "test/files/formula_string_error.xlsx",
-    "test/files/html-escape.ods",
-    "test/files/link.xls",
-    "test/files/link.xlsx",
-    "test/files/matrix.ods",
-    "test/files/matrix.xls",
-    "test/files/named_cells.ods",
-    "test/files/named_cells.xls",
-    "test/files/named_cells.xlsx",
-    "test/files/no_spreadsheet_file.txt",
-    "test/files/numbers1.csv",
-    "test/files/numbers1.ods",
-    "test/files/numbers1.xls",
-    "test/files/numbers1.xlsx",
-    "test/files/numbers1.xml",
-    "test/files/numeric-link.xlsx",
-    "test/files/only_one_sheet.ods",
-    "test/files/only_one_sheet.xls",
-    "test/files/only_one_sheet.xlsx",
-    "test/files/only_one_sheet.xml",
-    "test/files/paragraph.ods",
-    "test/files/paragraph.xls",
-    "test/files/paragraph.xlsx",
-    "test/files/paragraph.xml",
-    "test/files/prova.xls",
-    "test/files/ric.ods",
-    "test/files/simple_spreadsheet.ods",
-    "test/files/simple_spreadsheet.xls",
-    "test/files/simple_spreadsheet.xlsx",
-    "test/files/simple_spreadsheet.xml",
-    "test/files/simple_spreadsheet_from_italo.ods",
-    "test/files/simple_spreadsheet_from_italo.xls",
-    "test/files/simple_spreadsheet_from_italo.xml",
-    "test/files/so_datetime.csv",
-    "test/files/style.ods",
-    "test/files/style.xls",
-    "test/files/style.xlsx",
-    "test/files/style.xml",
-    "test/files/time-test.csv",
-    "test/files/time-test.ods",
-    "test/files/time-test.xls",
-    "test/files/time-test.xlsx",
-    "test/files/time-test.xml",
-    "test/files/type_excel.ods",
-    "test/files/type_excel.xlsx",
-    "test/files/type_excelx.ods",
-    "test/files/type_excelx.xls",
-    "test/files/type_openoffice.xls",
-    "test/files/type_openoffice.xlsx",
-    "test/files/whitespace.ods",
-    "test/files/whitespace.xls",
-    "test/files/whitespace.xlsx",
-    "test/files/whitespace.xml",
-    "test/rm_sub_test.rb",
-    "test/rm_test.rb",
-    "test/test_generic_spreadsheet.rb",
-    "test/test_helper.rb",
-    "test/test_roo.rb",
-    "website/index.html",
-    "website/index.txt",
-    "website/javascripts/rounded_corners_lite.inc.js",
-    "website/stylesheets/screen.css",
-    "website/template.rhtml"
-  ]
-  s.homepage = "http://github.com/Empact/roo"
-  s.licenses = ["MIT"]
-  s.require_paths = ["lib"]
-  s.required_ruby_version = Gem::Requirement.new(">= 1.9.0")
-  s.rubygems_version = "2.0.14"
-  s.summary = "Roo can access the contents of various spreadsheet files."
-  s.test_files = ["spec/fixtures/vcr_cassettes/google_drive.yml", "spec/fixtures/vcr_cassettes/google_drive_access_token.yml", "spec/fixtures/vcr_cassettes/google_drive_set.yml", "spec/lib/roo/base_spec.rb", "spec/lib/roo/csv_spec.rb", "spec/lib/roo/excel2003xml_spec.rb", "spec/lib/roo/excel_spec.rb", "spec/lib/roo/excelx/format_spec.rb", "spec/lib/roo/excelx_spec.rb", "spec/lib/roo/google_spec.rb", "spec/lib/roo/libreoffice_spec.rb", "spec/lib/roo/openoffice_spec.rb", "spec/lib/roo/spreadsheet_spec.rb", "spec/spec_helper.rb", "test/all_ss.rb", "test/files/1900_base.xls", "test/files/1900_base.xlsx", "test/files/1904_base.xls", "test/files/1904_base.xlsx", "test/files/Bibelbund.csv", "test/files/Bibelbund.ods", "test/files/Bibelbund.xls", "test/files/Bibelbund.xlsx", "test/files/Bibelbund.xml", "test/files/Bibelbund1.ods", "test/files/Pfand_from_windows_phone.xlsx", "test/files/bad_excel_date.xls", "test/files/bbu.ods", "test/files/bbu.xls", "test/files/bbu.xlsx", "test/files/bbu.xml", "test/files/bode-v1.ods.zip", "test/files/bode-v1.xls.zip", "test/files/boolean.csv", "test/files/boolean.ods", "test/files/boolean.xls", "test/files/boolean.xlsx", "test/files/boolean.xml", "test/files/borders.ods", "test/files/borders.xls", "test/files/borders.xlsx", "test/files/borders.xml", "test/files/bug-numbered-sheet-names.xlsx", "test/files/bug-row-column-fixnum-float.xls", "test/files/bug-row-column-fixnum-float.xml", "test/files/comments.ods", "test/files/comments.xls", "test/files/comments.xlsx", "test/files/csvtypes.csv", "test/files/datetime.ods", "test/files/datetime.xls", "test/files/datetime.xlsx", "test/files/datetime.xml", "test/files/datetime_floatconv.xls", "test/files/datetime_floatconv.xml", "test/files/dreimalvier.ods", "test/files/emptysheets.ods", "test/files/emptysheets.xls", "test/files/emptysheets.xlsx", "test/files/emptysheets.xml", "test/files/excel2003.xml", "test/files/false_encoding.xls", "test/files/false_encoding.xml", "test/files/file_item_error.xlsx", "test/files/formula.ods", "test/files/formula.xls", "test/files/formula.xlsx", "test/files/formula.xml", "test/files/formula_parse_error.xls", "test/files/formula_parse_error.xml", "test/files/formula_string_error.xlsx", "test/files/html-escape.ods", "test/files/link.xls", "test/files/link.xlsx", "test/files/matrix.ods", "test/files/matrix.xls", "test/files/named_cells.ods", "test/files/named_cells.xls", "test/files/named_cells.xlsx", "test/files/no_spreadsheet_file.txt", "test/files/numbers1.csv", "test/files/numbers1.ods", "test/files/numbers1.xls", "test/files/numbers1.xlsx", "test/files/numbers1.xml", "test/files/numeric-link.xlsx", "test/files/only_one_sheet.ods", "test/files/only_one_sheet.xls", "test/files/only_one_sheet.xlsx", "test/files/only_one_sheet.xml", "test/files/paragraph.ods", "test/files/paragraph.xls", "test/files/paragraph.xlsx", "test/files/paragraph.xml", "test/files/prova.xls", "test/files/ric.ods", "test/files/simple_spreadsheet.ods", "test/files/simple_spreadsheet.xls", "test/files/simple_spreadsheet.xlsx", "test/files/simple_spreadsheet.xml", "test/files/simple_spreadsheet_from_italo.ods", "test/files/simple_spreadsheet_from_italo.xls", "test/files/simple_spreadsheet_from_italo.xml", "test/files/so_datetime.csv", "test/files/style.ods", "test/files/style.xls", "test/files/style.xlsx", "test/files/style.xml", "test/files/time-test.csv", "test/files/time-test.ods", "test/files/time-test.xls", "test/files/time-test.xlsx", "test/files/time-test.xml", "test/files/type_excel.ods", "test/files/type_excel.xlsx", "test/files/type_excelx.ods", "test/files/type_excelx.xls", "test/files/type_openoffice.xls", "test/files/type_openoffice.xlsx", "test/files/whitespace.ods", "test/files/whitespace.xls", "test/files/whitespace.xlsx", "test/files/whitespace.xml", "test/rm_sub_test.rb", "test/rm_test.rb", "test/test_generic_spreadsheet.rb", "test/test_helper.rb", "test/test_roo.rb"]
+  spec.files         = `git ls-files -z`.split("\x0")
+  spec.files.reject! { |fn| fn.include?('test/files') }
+  spec.require_paths = ['lib']
 
-  if s.respond_to? :specification_version then
-    s.specification_version = 4
+  spec.add_dependency 'nokogiri', '~> 1'
+  spec.add_dependency 'rubyzip', '~> 1.1', '< 2.0.0'
 
-    if Gem::Version.new(Gem::VERSION) >= Gem::Version.new('1.2.0') then
-      s.add_runtime_dependency(%q<spreadsheet>, ["> 0.6.4"])
-      s.add_runtime_dependency(%q<nokogiri>, [">= 0"])
-      s.add_runtime_dependency(%q<rubyzip>, [">= 0"])
-      s.add_development_dependency(%q<google_drive>, [">= 0"])
-      s.add_development_dependency(%q<jeweler>, [">= 0"])
-    else
-      s.add_dependency(%q<spreadsheet>, ["> 0.6.4"])
-      s.add_dependency(%q<nokogiri>, [">= 0"])
-      s.add_dependency(%q<rubyzip>, [">= 0"])
-      s.add_dependency(%q<google_drive>, [">= 0"])
-      s.add_dependency(%q<jeweler>, [">= 0"])
-    end
-  else
-    s.add_dependency(%q<spreadsheet>, ["> 0.6.4"])
-    s.add_dependency(%q<nokogiri>, [">= 0"])
-    s.add_dependency(%q<rubyzip>, [">= 0"])
-    s.add_dependency(%q<google_drive>, [">= 0"])
-    s.add_dependency(%q<jeweler>, [">= 0"])
-  end
+  spec.add_development_dependency 'rake', '~> 10.1'
+  spec.add_development_dependency 'minitest', '~> 5.4', '>= 5.4.3'
 end
-

data/spec/lib/roo/base_spec.rb

@@ -1,7 +1,4 @@
 require 'spec_helper'
 
-describe Roo::GenericSpreadsheet do
-  it 'is an alias of Base' do
-    expect(Roo::GenericSpreadsheet).to eq(Roo::Base)
-  end
+describe Roo::Base do
 end

data/spec/lib/roo/csv_spec.rb

@@ -11,11 +11,25 @@ describe Roo::CSV do
   end
 
   describe '#parse' do
-    subject {
+    subject do
       csv.parse(options)
-    }
+    end
     context 'with headers: true' do
-      let(:options) { {headers: true} }
+      let(:options) { { headers: true } }
+
+      it "doesn't blow up" do
+        expect { subject }.to_not raise_error
+      end
+    end
+  end
+
+  describe '#parse_with_clean_option' do
+    subject do
+      csv.parse(options)
+    end
+    context 'with clean: true' do
+      let(:options) { {clean: true} }
+      let(:path) { 'test/files/parse_with_clean_option.csv' }
 
       it "doesn't blow up" do
         expect { subject }.to_not raise_error
@@ -25,30 +39,24 @@ describe Roo::CSV do
 
   describe '#csv_options' do
     context 'when created with the csv_options option' do
-      let(:options) {
+      let(:options) do
         {
           col_sep: '\t',
           quote_char: "'"
         }
-      }
+      end
 
       it 'returns the csv options' do
         csv = Roo::CSV.new(path, csv_options: options)
-        csv.csv_options.should == options
+        expect(csv.csv_options).to eq(options)
       end
     end
 
     context 'when created without the csv_options option' do
       it 'returns a hash' do
         csv = Roo::CSV.new(path)
-        csv.csv_options.should == {}
+        expect(csv.csv_options).to eq({})
       end
     end
   end
 end
-
-describe Roo::Csv do
-  it 'is an alias of LibreOffice' do
-    expect(Roo::Csv).to eq(Roo::CSV)
-  end
-end

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

@@ -11,8 +11,8 @@ describe Roo::Excelx::Format do
       '0%' => :percentage,
       '0.00%' => :percentage,
       '0.00E+00' => :float,
-      '# ?/?' => :float, #??? TODO:
-      '# ??/??' => :float, #??? TODO:
+      '# ?/?' => :float, # ??? TODO:
+      '# ??/??' => :float, # ??? TODO:
       'mm-dd-yy' => :date,
       'd-mmm-yy' => :date,
       'd-mmm' => :date,
@@ -33,17 +33,18 @@ describe Roo::Excelx::Format do
       '##0.0E+0' => :float,
       '@' => :float,
       #-- zusaetzliche Formate, die nicht standardmaessig definiert sind:
-      "yyyy\\-mm\\-dd" => :date,
+      'yyyy\\-mm\\-dd' => :date,
       'dd/mm/yy' => :date,
       'hh:mm:ss' => :time,
-      "dd/mm/yy\\ hh:mm" => :datetime,
+      'dd/mm/yy\\ hh:mm' => :datetime,
       'dd/mmm/yy\\ hh:mm' => :datetime,
       'dd/mmm/yy' => :date, # 2011-05-21
       'yyyy-mm-dd' => :date, # 2011-09-16
-      'yyyy-mm-dd;@' => :date
+      'yyyy-mm-dd;@' => :date,
+      '#0_);[Red]\(0\)' => :float
     }.each do |format, type|
       it "translates #{format} to #{type}" do
-        Roo::Excelx::Format.to_type(format).should == type
+        expect(Roo::Excelx::Format.to_type(format)).to eq(type)
       end
     end
   end