roo 2.7.1 → 2.8.0

data/lib/roo/constants.rb

@@ -1,5 +1,7 @@
+# frozen_string_literal: true
+
 module Roo
-  ROO_EXCEL_NOTICE   = "Excel support has been extracted to roo-xls due to its dependency on the GPL'd spreadsheet gem. Install roo-xls to use Roo::Excel.".freeze
-  ROO_EXCELML_NOTICE = "Excel SpreadsheetML support has been extracted to roo-xls. Install roo-xls to use Roo::Excel2003XML.".freeze
-  ROO_GOOGLE_NOTICE = "Google support has been extracted to roo-google. Install roo-google to use Roo::Google.".freeze
+  ROO_EXCEL_NOTICE   = "Excel support has been extracted to roo-xls due to its dependency on the GPL'd spreadsheet gem. Install roo-xls to use Roo::Excel."
+  ROO_EXCELML_NOTICE = "Excel SpreadsheetML support has been extracted to roo-xls. Install roo-xls to use Roo::Excel2003XML."
+  ROO_GOOGLE_NOTICE = "Google support has been extracted to roo-google. Install roo-google to use Roo::Google."
 end

data/lib/roo/csv.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require "csv"
 require "time"
 
@@ -63,25 +65,31 @@ module Roo
     def read_cells(sheet = default_sheet)
       sheet ||= default_sheet
       return if @cells_read[sheet]
-      set_row_count(sheet)
-      set_column_count(sheet)
-      row_num = 1
+      row_num = 0
+      max_col_num = 0
 
       each_row csv_options do |row|
-        row.each_with_index do |elem, col_num|
-          coordinate = [row_num, col_num + 1]
+        row_num += 1
+        col_num = 0
+
+        row.each do |elem|
+          col_num += 1
+          coordinate = [row_num, col_num]
           @cell[coordinate] = elem
           @cell_type[coordinate] = celltype_class(elem)
         end
-        row_num += 1
+
+        max_col_num = col_num if col_num > max_col_num
       end
 
+      set_row_count(sheet, row_num)
+      set_column_count(sheet, max_col_num)
       @cells_read[sheet] = true
     end
 
     def each_row(options, &block)
       if uri?(filename)
-        each_row_using_temp_dir(filename)
+        each_row_using_tempdir(options, &block)
       elsif is_stream?(filename_or_stream)
         ::CSV.new(filename_or_stream, options).each(&block)
       else
@@ -89,24 +97,24 @@ module Roo
       end
     end
 
-    def each_row_using_tempdir
+    def each_row_using_tempdir(options, &block)
       ::Dir.mktmpdir(Roo::TEMP_PREFIX, ENV["ROO_TMP"]) do |tmpdir|
         tmp_filename = download_uri(filename, tmpdir)
         ::CSV.foreach(tmp_filename, options, &block)
       end
     end
 
-    def set_row_count(sheet)
+    def set_row_count(sheet, last_row)
       @first_row[sheet] = 1
-      @last_row[sheet] = ::CSV.readlines(@filename, csv_options).size
+      @last_row[sheet] = last_row
       @last_row[sheet] = @first_row[sheet] if @last_row[sheet].zero?
 
       nil
     end
 
-    def set_column_count(sheet)
+    def set_column_count(sheet, last_col)
       @first_column[sheet] = 1
-      @last_column[sheet] = (::CSV.readlines(@filename, csv_options).first || []).size
+      @last_column[sheet] = last_col
       @last_column[sheet] = @first_column[sheet] if @last_column[sheet].zero?
 
       nil

data/lib/roo/excelx.rb

@@ -24,8 +24,9 @@ module Roo
     require 'roo/excelx/sheet_doc'
     require 'roo/excelx/coordinate'
     require 'roo/excelx/format'
+    require 'roo/excelx/images'
 
-    delegate [:styles, :workbook, :shared_strings, :rels_files, :sheet_files, :comments_files] => :@shared
+    delegate [:styles, :workbook, :shared_strings, :rels_files, :sheet_files, :comments_files, :image_rels, :image_files] => :@shared
     ExceedsMaxError = Class.new(StandardError)
 
     # initialization and opening of a spreadsheet file
@@ -39,7 +40,10 @@ module Roo
       sheet_options = {}
       sheet_options[:expand_merged_ranges] = (options[:expand_merged_ranges] || false)
       sheet_options[:no_hyperlinks] = (options[:no_hyperlinks] || false)
+      sheet_options[:empty_cell] = (options[:empty_cell] || false)
+      shared_options = {}
 
+      shared_options[:disable_html_wrapper] = (options[:disable_html_wrapper] || false)
       unless is_stream?(filename_or_stream)
         file_type_check(filename_or_stream, %w[.xlsx .xlsm], 'an Excel 2007', file_warning, packed)
         basename = find_basename(filename_or_stream)
@@ -52,7 +56,7 @@ module Roo
       @tmpdir = self.class.make_tempdir(self, basename, options[:tmpdir_root])
       ObjectSpace.define_finalizer(self, self.class.finalize(object_id))
 
-      @shared = Shared.new(@tmpdir)
+      @shared = Shared.new(@tmpdir, shared_options)
       @filename = local_filename(filename_or_stream, @tmpdir, packed)
       process_zipfile(@filename || filename_or_stream)
 
@@ -62,10 +66,10 @@ module Roo
         end
       end.compact
       @sheets = []
-      @sheets_by_name = Hash[@sheet_names.map.with_index do |sheet_name, n|
-        @sheets[n] = Sheet.new(sheet_name, @shared, n, sheet_options)
-        [sheet_name, @sheets[n]]
-      end]
+      @sheets_by_name = {}
+      @sheet_names.each_with_index do |sheet_name, n|
+        @sheets_by_name[sheet_name] = @sheets[n] = Sheet.new(sheet_name, @shared, n, sheet_options)
+      end
 
       if cell_max
         cell_count = ::Roo::Utils.num_cells_in_range(sheet_for(options.delete(:sheet)).dimensions)
@@ -94,7 +98,12 @@ module Roo
     def sheet_for(sheet)
       sheet ||= default_sheet
       validate_sheet!(sheet)
-      @sheets_by_name[sheet]
+      @sheets_by_name[sheet] || @sheets[sheet]
+    end
+
+    def images(sheet = nil)
+      images_names = sheet_for(sheet).images.map(&:last)
+      images_names.map { |iname| image_files.find { |ifile| ifile[iname] } }
     end
 
     # Returns the content of a spreadsheet-cell.
@@ -325,7 +334,7 @@ module Roo
 
       wb.extract(path)
       workbook_doc = Roo::Utils.load_xml(path).remove_namespaces!
-      workbook_doc.xpath('//sheet').map { |s| s.attributes['id'].value }
+      workbook_doc.xpath('//sheet').map { |s| s['id'] }
     end
 
     # Internal
@@ -349,17 +358,13 @@ module Roo
 
       wb_rels.extract(path)
       rels_doc = Roo::Utils.load_xml(path).remove_namespaces!
-      worksheet_type = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/worksheet'
 
       relationships = rels_doc.xpath('//Relationship').select do |relationship|
-        relationship.attributes['Type'].value == worksheet_type
+        worksheet_types.include? relationship['Type']
       end
 
-      relationships.inject({}) do |hash, relationship|
-        attributes = relationship.attributes
-        id = attributes['Id']
-        hash[id.value] = attributes['Target'].value
-        hash
+      relationships.each_with_object({}) do |relationship, hash|
+        hash[relationship['Id']] = relationship['Target']
       end
     end
 
@@ -376,6 +381,15 @@ module Roo
       end
     end
 
+    def extract_images(entries, tmpdir)
+      img_entries = entries.select { |e| e.name[/media\/image([0-9]+)/] }
+      img_entries.each do |entry|
+        path = "#{@tmpdir}/roo#{entry.name.gsub(/xl\/|\//, "_")}"
+        image_files << path
+        entry.extract(path)
+      end
+    end
+
     # Extracts all needed files from the zip file
     def process_zipfile(zipfilename_or_stream)
       @sheet_files = []
@@ -409,6 +423,7 @@ module Roo
       sheet_ids = extract_worksheet_ids(entries, "#{@tmpdir}/roo_workbook.xml")
       sheets = extract_worksheet_rels(entries, "#{@tmpdir}/roo_workbook.xml.rels")
       extract_sheets_in_order(entries, sheet_ids, sheets, @tmpdir)
+      extract_images(entries, @tmpdir)
 
       entries.each do |entry|
         path =
@@ -435,6 +450,10 @@ module Roo
           #        drawings, etc.
           nr = Regexp.last_match[1].to_i
           rels_files[nr - 1] = "#{@tmpdir}/roo_rels#{nr}"
+        when /drawing([0-9]+).xml.rels$/
+          # Extracting drawing relationships to make images lists for each sheet
+          nr = Regexp.last_match[1].to_i
+          image_rels[nr - 1] = "#{@tmpdir}/roo_image_rels#{nr}"
         end
 
         entry.extract(path) if path
@@ -442,7 +461,14 @@ module Roo
     end
 
     def safe_send(object, method, *args)
-      object.send(method, *args) if object && object.respond_to?(method)
+      object.send(method, *args) if object&.respond_to?(method)
+    end
+
+    def worksheet_types
+      [
+        'http://schemas.openxmlformats.org/officeDocument/2006/relationships/worksheet', # OOXML Transitional
+        'http://purl.oclc.org/ooxml/officeDocument/relationships/worksheet' # OOXML Strict
+      ]
     end
   end
 end

data/lib/roo/excelx/cell.rb

@@ -40,19 +40,23 @@ module Roo
       end
 
       def self.create_cell(type, *values)
+        cell_class(type)&.new(*values)
+      end
+
+      def self.cell_class(type)
         case type
         when :string
-          Cell::String.new(*values)
+          Cell::String
         when :boolean
-          Cell::Boolean.new(*values)
+          Cell::Boolean
         when :number
-          Cell::Number.new(*values)
+          Cell::Number
         when :date
-          Cell::Date.new(*values)
+          Cell::Date
         when :datetime
-          Cell::DateTime.new(*values)
+          Cell::DateTime
         when :time
-          Cell::Time.new(*values)
+          Cell::Time
         end
       end
 

data/lib/roo/excelx/cell/base.rb

@@ -1,13 +1,18 @@
+# frozen_string_literal: true
+
+require "roo/helpers/default_attr_reader"
+
 module Roo
   class Excelx
     class Cell
       class Base
+        extend Roo::Helpers::DefaultAttrReader
         attr_reader :cell_type, :cell_value, :value
 
         # FIXME: I think style should be deprecated. Having a style attribute
         #        for a cell doesn't really accomplish much. It seems to be used
         #        when you want to export to excelx.
-        attr_reader :style
+        attr_reader_with_default default_type: :base, style: 1
 
 
         # FIXME: Updating a cell's value should be able tochange the cell's type,
@@ -34,14 +39,12 @@ module Roo
         attr_writer :value
 
         def initialize(value, formula, excelx_type, style, link, coordinate)
-          @link = !!link
           @cell_value = value
-          @cell_type = excelx_type
-          @formula = formula
-          @style = style
+          @cell_type = excelx_type if excelx_type
+          @formula = formula if formula
+          @style = style unless style == 1
           @coordinate = coordinate
-          @type = :base
-          @value = link? ? Roo::Link.new(link, value) : value
+          @value = link ? Roo::Link.new(link, value) : value
         end
 
         def type
@@ -50,16 +53,16 @@ module Roo
           elsif link?
             :link
           else
-            @type
+            default_type
           end
         end
 
         def formula?
-          !!@formula
+          !!(defined?(@formula) && @formula)
         end
 
         def link?
-          !!@link
+          Roo::Link === @value
         end
 
         alias_method :formatted_value, :value
@@ -68,9 +71,16 @@ module Roo
           formatted_value
         end
 
-        # DEPRECATED: Please use link instead.
+        # DEPRECATED: Please use link? instead.
         def hyperlink
-          warn '[DEPRECATION] `hyperlink` is deprecated.  Please use `link` instead.'
+          warn '[DEPRECATION] `hyperlink` is deprecated.  Please use `link?` instead.'
+          link?
+        end
+
+        # DEPRECATED: Please use link? instead.
+        def link
+          warn '[DEPRECATION] `link` is deprecated.  Please use `link?` instead.'
+          link?
         end
 
         # DEPRECATED: Please use cell_value instead.
@@ -88,6 +98,10 @@ module Roo
         def empty?
           false
         end
+
+        def presence
+          empty? ? nil : self
+        end
       end
     end
   end

data/lib/roo/excelx/cell/boolean.rb

@@ -1,17 +1,20 @@
+# frozen_string_literal: true
+
 module Roo
   class Excelx
     class Cell
       class Boolean < Cell::Base
-        attr_reader :value, :formula, :format, :cell_type, :cell_value, :link, :coordinate
+        attr_reader :value, :formula, :format, :cell_value, :coordinate
+
+        attr_reader_with_default default_type: :boolean, cell_type: :boolean
 
         def initialize(value, formula, style, link, coordinate)
-          super(value, formula, nil, style, link, coordinate)
-          @type = @cell_type = :boolean
-          @value = link? ? Roo::Link.new(link, value) : create_boolean(value)
+          super(value, formula, nil, style, nil, coordinate)
+          @value = link ? Roo::Link.new(link, value) : create_boolean(value)
         end
 
         def formatted_value
-          value ? 'TRUE'.freeze : 'FALSE'.freeze
+          value ? 'TRUE' : 'FALSE'
         end
 
         private
@@ -19,7 +22,7 @@ module Roo
         def create_boolean(value)
           # FIXME: Using a boolean will cause methods like Base#to_csv to fail.
           #       Roo is using some method to ignore false/nil values.
-          value.to_i == 1 ? true : false
+          value.to_i == 1
         end
       end
     end

data/lib/roo/excelx/cell/date.rb

@@ -4,23 +4,23 @@ module Roo
   class Excelx
     class Cell
       class Date < Roo::Excelx::Cell::DateTime
-        attr_reader :value, :formula, :format, :cell_type, :cell_value, :link, :coordinate
+        attr_reader :value, :formula, :format, :cell_type, :cell_value, :coordinate
+
+        attr_reader_with_default default_type: :date
 
         def initialize(value, formula, excelx_type, style, link, base_date, coordinate)
           # NOTE: Pass all arguments to the parent class, DateTime.
           super
-          @type = :date
           @format = excelx_type.last
-          @value = link? ? Roo::Link.new(link, value) : create_date(base_date, value)
+          @value = link ? Roo::Link.new(link, value) : create_date(base_date, value)
         end
 
         private
 
-        def create_date(base_date, value)
-          date = base_date + value.to_i
-          yyyy, mm, dd = date.strftime('%Y-%m-%d').split('-')
+        def create_datetime(_,_);  end
 
-          ::Date.new(yyyy.to_i, mm.to_i, dd.to_i)
+        def create_date(base_date, value)
+          base_date + value.to_i
         end
       end
     end

data/lib/roo/excelx/cell/datetime.rb

@@ -1,16 +1,21 @@
+# frozen_string_literal: true
+
 require 'date'
 
 module Roo
   class Excelx
     class Cell
       class DateTime < Cell::Base
-        attr_reader :value, :formula, :format, :cell_value, :link, :coordinate
+        SECONDS_IN_DAY = 60 * 60 * 24
+
+        attr_reader :value, :formula, :format, :cell_value, :coordinate
 
-        def initialize(value, formula, excelx_type, style, link, base_date, coordinate)
-          super(value, formula, excelx_type, style, link, coordinate)
-          @type = :datetime
+        attr_reader_with_default default_type: :datetime
+
+        def initialize(value, formula, excelx_type, style, link, base_timestamp, coordinate)
+          super(value, formula, excelx_type, style, nil, coordinate)
           @format = excelx_type.last
-          @value = link? ? Roo::Link.new(link, value) : create_datetime(base_date, value)
+          @value = link ? Roo::Link.new(link, value) : create_datetime(base_timestamp, value)
         end
 
         # Public: Returns formatted value for a datetime. Format's can be an
@@ -78,7 +83,7 @@ module Roo
 
         TIME_FORMATS = {
           'hh' => '%H',    # Hour (24): 01
-          'h' => '%-k'.freeze,    # Hour (24): 1
+          'h' => '%-k',    # Hour (24): 1
           # 'hh'.freeze => '%I'.freeze,    # Hour (12): 08
           # 'h'.freeze => '%-l'.freeze,    # Hour (12): 8
           'mm' => '%M',    # Minute: 01
@@ -92,18 +97,9 @@ module Roo
           '0' => '%1N'    # Fractional Seconds: tenths.
         }
 
-        def create_datetime(base_date, value)
-          date = base_date + value.to_f.round(6)
-          datetime_string = date.strftime('%Y-%m-%d %H:%M:%S.%N')
-          t = round_datetime(datetime_string)
-
-          ::DateTime.civil(t.year, t.month, t.day, t.hour, t.min, t.sec)
-        end
-
-        def round_datetime(datetime_string)
-          /(?<yyyy>\d+)-(?<mm>\d+)-(?<dd>\d+) (?<hh>\d+):(?<mi>\d+):(?<ss>\d+.\d+)/ =~ datetime_string
-
-          ::Time.new(yyyy, mm, dd, hh, mi, ss.to_r).round(0)
+        def create_datetime(base_timestamp, value)
+          timestamp = (base_timestamp + (value.to_f.round(6) * SECONDS_IN_DAY)).round(0)
+          ::Time.at(timestamp).utc.to_datetime
         end
       end
     end