Add a utility to classify a file as text/binary.

For use to help automatically set the internal file attributes bits.

lib/zip/util/file_classifier.rb

@@ -0,0 +1,29 @@
+# frozen_string_literal: true
+
+module Zip
+  module Util
+    class FileClassifier
+      CNTRL_LIST = [(0..6), (14..25), (28..31)].sum([], &:to_a).freeze
+      ALLOW_LIST = ([9, 10, 13] + (32..255).to_a).freeze
+
+      def self.classify(buffer, length: 256)
+        has_allowed = false
+        buffer[0...length].each_byte do |b|
+          return :binary if CNTRL_LIST.include?(b)
+
+          has_allowed = true if ALLOW_LIST.include?(b)
+        end
+
+        has_allowed ? :text : :binary
+      end
+
+      def self.text?(buffer)
+        classify(buffer) == :text
+      end
+
+      def self.binary?(buffer)
+        classify(buffer) == :binary
+      end
+    end
+  end
+end

test/util/file_classifier_test.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+require 'test_helper'
+require 'zip/util/file_classifier'
+
+class FileClassifierTest < MiniTest::Test
+  def test_classify_data
+    assert(::Zip::Util::FileClassifier.text?('Hello, World!'))
+    assert(::Zip::Util::FileClassifier.binary?(''))
+    assert(::Zip::Util::FileClassifier.binary?("\x00"))
+    assert(::Zip::Util::FileClassifier.binary?("\x1b\x00"))
+    assert(::Zip::Util::FileClassifier.binary?("\x1b"))
+    assert(::Zip::Util::FileClassifier.text?("Hello, World\x21"))
+    assert(::Zip::Util::FileClassifier.binary?("Hello, World\x00"))
+    assert(::Zip::Util::FileClassifier.text?("Hello, World\x1b"))
+    assert(::Zip::Util::FileClassifier.text?("\xab"))
+  end
+
+  def test_detect_text_file
+    data = ::File.read('test/data/file1.txt')
+    assert_equal(:text, ::Zip::Util::FileClassifier.classify(data))
+    assert_equal(
+      :text, ::Zip::Util::FileClassifier.classify(data, length: 2_048)
+    )
+    refute(::Zip::Util::FileClassifier.binary?(data))
+    assert(::Zip::Util::FileClassifier.text?(data))
+  end
+
+  def test_detect_binary_file
+    data = ::File.binread('test/data/test.xls', 256)
+    assert_equal(:binary, ::Zip::Util::FileClassifier.classify(data))
+    assert_equal(
+      :binary, ::Zip::Util::FileClassifier.classify(data, length: 64)
+    )
+    assert(::Zip::Util::FileClassifier.binary?(data))
+    refute(::Zip::Util::FileClassifier.text?(data))
+  end
+end