roo 2.3.0 → 2.10.1

data/lib/roo/csv.rb

@@ -1,5 +1,7 @@
-require 'csv'
-require 'time'
+# frozen_string_literal: true
+
+require "csv"
+require "time"
 
 # The CSV class can read csv files (must be separated with commas) which then
 # can be handled like spreadsheets. This means you can access cells like A5
@@ -9,112 +11,131 @@ require 'time'
 #
 # You can pass options to the underlying CSV parse operation, via the
 # :csv_options option.
-#
+module Roo
+  class CSV < Roo::Base
+    attr_reader :filename
+
+    # Returns an array with the names of the sheets. In CSV class there is only
+    # one dummy sheet, because a csv file cannot have more than one sheet.
+    def sheets
+      ["default"]
+    end
 
-class Roo::CSV < Roo::Base
+    def cell(row, col, sheet = nil)
+      sheet ||= default_sheet
+      read_cells(sheet)
+      @cell[normalize(row, col)]
+    end
 
-  attr_reader :filename
+    def celltype(row, col, sheet = nil)
+      sheet ||= default_sheet
+      read_cells(sheet)
+      @cell_type[normalize(row, col)]
+    end
 
-  # Returns an array with the names of the sheets. In CSV class there is only
-  # one dummy sheet, because a csv file cannot have more than one sheet.
-  def sheets
-    ['default']
-  end
+    def cell_postprocessing(_row, _col, value)
+      value
+    end
 
-  def cell(row, col, sheet=nil)
-    sheet ||= default_sheet
-    read_cells(sheet)
-    @cell[normalize(row,col)]
-  end
+    def csv_options
+      @options[:csv_options] || {}
+    end
 
-  def celltype(row, col, sheet=nil)
-    sheet ||= default_sheet
-    read_cells(sheet)
-    @cell_type[normalize(row,col)]
-  end
+    def set_value(row, col, value, _sheet)
+      @cell[[row, col]] = value
+    end
 
-  def cell_postprocessing(row,col,value)
-    value
-  end
+    def set_type(row, col, type, _sheet)
+      @cell_type[[row, col]] = type
+    end
 
-  def csv_options
-    @options[:csv_options] || {}
-  end
+    private
 
-  private
+    TYPE_MAP = {
+      String => :string,
+      Float => :float,
+      Date => :date,
+      DateTime => :datetime,
+    }
 
-  TYPE_MAP = {
-    String => :string,
-    Float => :float,
-    Date => :date,
-    DateTime => :datetime,
-  }
+    def celltype_class(value)
+      TYPE_MAP[value.class]
+    end
 
-  def celltype_class(value)
-    TYPE_MAP[value.class]
-  end
+    def read_cells(sheet = default_sheet)
+      sheet ||= default_sheet
+      return if @cells_read[sheet]
+      row_num = 0
+      max_col_num = 0
+
+      each_row csv_options do |row|
+        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
 
-  def each_row(options, &block)
-    if uri?(filename)
-      make_tmpdir do |tmpdir|
-        tmp_filename = download_uri(filename, tmpdir)
-        CSV.foreach(tmp_filename, options, &block)
+        max_col_num = col_num if col_num > max_col_num
       end
-    else
-      CSV.foreach(filename, options, &block)
+
+      set_row_count(sheet, row_num)
+      set_column_count(sheet, max_col_num)
+      @cells_read[sheet] = true
     end
-  end
 
-  def read_cells(sheet = default_sheet)
-    sheet ||= default_sheet
-    return if @cells_read[sheet]
-    @first_row[sheet] = 1
-    @last_row[sheet] = 0
-    @first_column[sheet] = 1
-    @last_column[sheet] = 1
-    rownum = 1
-    each_row csv_options do |row|
-      row.each_with_index do |elem,i|
-        @cell[[rownum,i+1]] = cell_postprocessing rownum,i+1, elem
-        @cell_type[[rownum,i+1]] = celltype_class @cell[[rownum,i+1]]
-        if i+1 > @last_column[sheet]
-          @last_column[sheet] += 1
-        end
+    def each_row(options, &block)
+      if uri?(filename)
+        each_row_using_tempdir(options, &block)
+      else
+        csv_foreach(filename_or_stream, options, &block)
       end
-      rownum += 1
-      @last_row[sheet] += 1
     end
-    @cells_read[sheet] = true
-    #-- adjust @first_row if neccessary
-    while !row(@first_row[sheet]).any? and @first_row[sheet] < @last_row[sheet]
-      @first_row[sheet] += 1
+
+    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
-    #-- adjust @last_row if neccessary
-    while !row(@last_row[sheet]).any? and @last_row[sheet] and
-        @last_row[sheet] > @first_row[sheet]
-      @last_row[sheet] -= 1
+
+    def csv_foreach(path_or_io, options, &block)
+      if is_stream?(path_or_io)
+        ::CSV.new(path_or_io, **options).each(&block)
+      else
+        ::CSV.foreach(path_or_io, **options, &block)
+      end
     end
-    #-- adjust @first_column if neccessary
-    while !column(@first_column[sheet]).any? and
-          @first_column[sheet] and
-          @first_column[sheet] < @last_column[sheet]
-      @first_column[sheet] += 1
+
+    def set_row_count(sheet, last_row)
+      @first_row[sheet] = 1
+      @last_row[sheet] = last_row
+      @last_row[sheet] = @first_row[sheet] if @last_row[sheet].zero?
+
+      nil
     end
-    #-- adjust @last_column if neccessary
-    while !column(@last_column[sheet]).any? and
-          @last_column[sheet] and
-          @last_column[sheet] > @first_column[sheet]
-      @last_column[sheet] -= 1
+
+    def set_column_count(sheet, last_col)
+      @first_column[sheet] = 1
+      @last_column[sheet] = last_col
+      @last_column[sheet] = @first_column[sheet] if @last_column[sheet].zero?
+
+      nil
     end
-  end
 
-  def clean_sheet(sheet)
-    read_cells(sheet)
+    def clean_sheet(sheet)
+      read_cells(sheet)
+
+      @cell.each_pair do |coord, value|
+        @cell[coord] = sanitize_value(value) if value.is_a?(::String)
+      end
 
-    @cell.each_pair do |coord, value|
-      @cell[coord] = sanitize_value(value) if value.is_a?(::String)
+      @cleaned[sheet] = true
     end
 
-    @cleaned[sheet] = true
+    alias_method :filename_or_stream, :filename
   end
 end

data/lib/roo/errors.rb

@@ -6,4 +6,6 @@ module Roo
   # Raised when Roo cannot find a header row that matches the given column
   # name(s).
   class HeaderRowNotFoundError < Error; end
+
+  class FileNotFound < Error; 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
@@ -32,14 +37,9 @@ module Roo
         #
         # Returns a String representation of a cell's value.
         def formatted_value
-          date_regex = /(?<date>[dmy]+[\-\/][dmy]+([\-\/][dmy]+)?)/
-          time_regex = /(?<time>(\[?[h]\]?+:)?[m]+(:?ss|:?s)?)/
-
           formatter = @format.downcase.split(' ').map do |part|
-            if part[date_regex] == part
-              part.gsub(/#{DATE_FORMATS.keys.join('|')}/, DATE_FORMATS)
-            elsif part[time_regex]
-              part.gsub(/#{TIME_FORMATS.keys.join('|')}/, TIME_FORMATS)
+            if (parsed_format = parse_date_or_time_format(part))
+              parsed_format
             else
               warn 'Unable to parse custom format. Using "YYYY-mm-dd HH:MM:SS" format.'
               return @value.strftime('%F %T')
@@ -51,49 +51,55 @@ module Roo
 
         private
 
+        def parse_date_or_time_format(part)
+          date_regex = /(?<date>[dmy]+[\-\/][dmy]+([\-\/][dmy]+)?)/
+          time_regex = /(?<time>(\[?[h]\]?+:)?[m]+(:?ss|:?s)?)/
+
+          if part[date_regex] == part
+            formats = DATE_FORMATS
+          elsif part[time_regex]
+            formats = TIME_FORMATS
+          else
+            return false
+          end
+
+          part.gsub(/#{formats.keys.join('|')}/, formats)
+        end
+
         DATE_FORMATS = {
-          'yyyy'.freeze => '%Y'.freeze,  # Year: 2000
-          'yy'.freeze => '%y'.freeze,    # Year: 00
+          'yyyy' => '%Y',  # Year: 2000
+          'yy' => '%y',    # Year: 00
           # mmmmm => J-D
-          'mmmm'.freeze => '%B'.freeze,  # Month: January
-          'mmm'.freeze => '%^b'.freeze,   # Month: JAN
-          'mm'.freeze => '%m'.freeze,    # Month: 01
-          'm'.freeze => '%-m'.freeze,    # Month: 1
-          'dddd'.freeze => '%A'.freeze,  # Day of the Week: Sunday
-          'ddd'.freeze => '%^a'.freeze,   # Day of the Week: SUN
-          'dd'.freeze => '%d'.freeze,    # Day of the Month: 01
-          'd'.freeze => '%-d'.freeze,    # Day of the Month: 1
+          'mmmm' => '%B',  # Month: January
+          'mmm' => '%^b',   # Month: JAN
+          'mm' => '%m',    # Month: 01
+          'm' => '%-m',    # Month: 1
+          'dddd' => '%A',  # Day of the Week: Sunday
+          'ddd' => '%^a',   # Day of the Week: SUN
+          'dd' => '%d',    # Day of the Month: 01
+          'd' => '%-d'    # Day of the Month: 1
           # '\\\\'.freeze => ''.freeze,  # NOTE: Fixes a custom format's output.
         }
 
         TIME_FORMATS = {
-          'hh'.freeze => '%H'.freeze,    # Hour (24): 01
-          'h'.freeze => '%-k'.freeze,    # Hour (24): 1
+          'hh' => '%H',    # Hour (24): 01
+          'h' => '%-k',    # Hour (24): 1
           # 'hh'.freeze => '%I'.freeze,    # Hour (12): 08
           # 'h'.freeze => '%-l'.freeze,    # Hour (12): 8
-          'mm'.freeze => '%M'.freeze,    # Minute: 01
+          'mm' => '%M',    # Minute: 01
           # FIXME: is this used? Seems like 'm' is used for month, not minute.
-          'm'.freeze => '%-M'.freeze,    # Minute: 1
-          'ss'.freeze => '%S'.freeze,    # Seconds: 01
-          's'.freeze => '%-S'.freeze,    # Seconds: 1
-          'am/pm'.freeze => '%p'.freeze, # Meridian: AM
-          '000'.freeze => '%3N'.freeze,  # Fractional Seconds: thousandth.
-          '00'.freeze => '%2N'.freeze,   # Fractional Seconds: hundredth.
-          '0'.freeze => '%1N'.freeze,    # Fractional Seconds: tenths.
+          'm' => '%-M',    # Minute: 1
+          'ss' => '%S',    # Seconds: 01
+          's' => '%-S',    # Seconds: 1
+          'am/pm' => '%p', # Meridian: AM
+          '000' => '%3N',  # Fractional Seconds: thousandth.
+          '00' => '%2N',   # Fractional Seconds: hundredth.
+          '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.to_i, mm.to_i, dd.to_i, hh.to_i, mi.to_i, 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

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

@@ -3,10 +3,11 @@ module Roo
   class Excelx
     class Cell
       class Empty < Cell::Base
-        attr_reader :value, :formula, :format, :cell_type, :cell_value, :hyperlink, :coordinate
+        attr_reader :value, :formula, :format, :cell_type, :cell_value, :coordinate
+
+        attr_reader_with_default default_type: nil, style: nil
 
         def initialize(coordinate)
-          @value = @formula = @format = @cell_type = @cell_value = @hyperlink = nil
           @coordinate = coordinate
         end