Fixes #75 allow Roo::Excelx to open streams

lib/roo/base.rb

@@ -429,6 +429,10 @@ def key_to_string(arr)
     "#{arr[0]},#{arr[1]}"
   end
 
+  def is_stream?(filename_or_stream)
+    filename_or_stream.respond_to?(:seek)
+  end
+
   private
 
   def clean_sheet_if_need(options)
@@ -451,6 +455,7 @@ def search_or_set_header(options)
   end
 
   def local_filename(filename, tmpdir, packed)
+    return if is_stream?(filename)
     filename = download_uri(filename, tmpdir) if uri?(filename)
     filename = unzip(filename, tmpdir) if packed == :zip
     unless File.file?(filename)

lib/roo/excelx.rb

@@ -250,20 +250,23 @@ def pad_cells(cell, last_column)
   # values for packed: :zip
   # optional cell_max (int) parameter for early aborting attempts to parse
   # enormous documents.
-  def initialize(filename, options = {})
+  def initialize(filename_or_stream, options = {})
     packed = options[:packed]
     file_warning = options.fetch(:file_warning, :error)
     cell_max = options.delete(:cell_max)
     sheet_options = {}
     sheet_options[:expand_merged_ranges] = (options[:expand_merged_ranges] || false)
 
-    file_type_check(filename,'.xlsx','an Excel-xlsx', file_warning, packed)
+    unless is_stream?(filename_or_stream)
+      file_type_check(filename_or_stream,'.xlsx','an Excel-xlsx', file_warning, packed)
+      basename = File.basename(filename_or_stream)
+    end
 
-    @tmpdir = make_tmpdir(File.basename(filename), options[:tmpdir_root])
-    @filename = local_filename(filename, @tmpdir, packed)
+    @tmpdir = make_tmpdir(basename, options[:tmpdir_root])
+    @filename = local_filename(filename_or_stream, @tmpdir, packed)
     @comments_files = []
     @rels_files = []
-    process_zipfile(@tmpdir, @filename)
+    process_zipfile(@filename || filename_or_stream)
 
     @sheet_names = workbook.sheets.map do |sheet|
       unless options[:only_visible_sheets] && sheet['state'] == 'hidden'
@@ -574,10 +577,26 @@ def extract_sheets_in_order(entries, sheet_ids, sheets, tmpdir)
   end
 
   # Extracts all needed files from the zip file
-  def process_zipfile(tmpdir, zipfilename)
+  def process_zipfile(zipfilename_or_stream)
     @sheet_files = []
-    entries = Zip::File.open(zipfilename).to_a.sort_by(&:name)
 
+    unless is_stream?(zipfilename_or_stream)
+      process_zipfile_entries Zip::File.open(zipfilename_or_stream).to_a.sort_by(&:name)
+    else
+      stream = Zip::InputStream.open zipfilename_or_stream
+      begin
+        entries = []
+        while entry = stream.get_next_entry
+          entries << entry
+        end
+        process_zipfile_entries entries
+      ensure
+        stream.close
+      end
+    end
+  end
+
+  def process_zipfile_entries entries
     # NOTE: When Google or Numbers 3.1 exports to xlsx, the worksheet filenames
     #       are not in order. With Numbers 3.1, the first sheet is always
     #       sheet.xml, not sheet1.xml. With Google, the order of the worksheets is
@@ -593,31 +612,31 @@ def process_zipfile(tmpdir, zipfilename)
     #       workbook.xml.rel:
     #         <Relationship Id="rId4" Target="worksheets/sheet5.xml" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/worksheet"/>
     #         <Relationship Id="rId3" Target="worksheets/sheet4.xml" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/worksheet"/>
-    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)
+    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)
 
     entries.each do |entry|
       path =
       case entry.name.downcase
       when /sharedstrings.xml$/
-        "#{tmpdir}/roo_sharedStrings.xml"
+        "#{@tmpdir}/roo_sharedStrings.xml"
       when /styles.xml$/
-        "#{tmpdir}/roo_styles.xml"
+        "#{@tmpdir}/roo_styles.xml"
       when /comments([0-9]+).xml$/
         # FIXME: Most of the time, The order of the comment files are the same
         #       the sheet order, i.e. sheet1.xml's comments are in comments1.xml.
         #       In some situations, this isn't true. The true location of a
         #       sheet's comment file is in the sheet1.xml.rels file. SEE
         #       ECMA-376 12.3.3 in "Ecma Office Open XML Part 1".
         nr = Regexp.last_match[1].to_i
-        @comments_files[nr - 1] = "#{tmpdir}/roo_comments#{nr}"
+        @comments_files[nr - 1] = "#{@tmpdir}/roo_comments#{nr}"
       when /sheet([0-9]+).xml.rels$/
         # FIXME: Roo seems to use sheet[\d].xml.rels for hyperlinks only, but
         #        it also stores the location for sharedStrings, comments,
         #        drawings, etc.
         nr = Regexp.last_match[1].to_i
-        @rels_files[nr - 1] = "#{tmpdir}/roo_rels#{nr}"
+        @rels_files[nr - 1] = "#{@tmpdir}/roo_rels#{nr}"
       end
 
       entry.extract(path) if path

test/test_roo.rb

@@ -18,6 +18,7 @@
 #STDERR.reopen "/dev/null","w"
 
 require 'test_helper'
+require 'stringio'
 
 class TestRoo < Minitest::Test
 
@@ -2063,5 +2064,11 @@ def test_noexpand_merged_range
     end
   end
 
-
+  def test_open_stream
+    return unless EXCELX
+    file_contents = File.read File.join(TESTDIR, fixture_filename(:numbers1, :excelx))
+    stream = StringIO.new(file_contents)
+    xlsx = Roo::Excelx.new(stream)
+    assert_equal ["Tabelle1","Name of Sheet 2","Sheet3","Sheet4","Sheet5"], xlsx.sheets
+  end
 end # class