[GR-19220] Implement IO#set_encoding_by_bom (#2372)

PullRequest: truffleruby/2704

Author: eregon

CHANGELOG.md

@@ -32,6 +32,7 @@ Compatibility:
 * Fix `ObjectSpace._id2ref` for Symbols and frozen String literals (#2358).
 * Implemented `Enumerator::Lazy#filter_map` (#2356).
 * Fix LLVM toolchain issue on macOS 10.13 (#2352, [oracle/graal#3383](https://github.com/oracle/graal/issues/3383)).
+* Implement `IO#set_encoding_by_bom` (#2372, pawandubey).
 * Implemented `Enumerator::Lazy#with_index` (#2356).
 * Implement `rb_backref_set`.
 

spec/ruby/core/io/set_encoding_by_bom_spec.rb

@@ -67,5 +67,11 @@
 
       -> { @io.set_encoding_by_bom }.should raise_error(ArgumentError, 'encoding is set to UTF-8 already')
     end
+
+    it 'returns exception if encoding conversion is already set' do
+      @io.set_encoding(Encoding::UTF_8, Encoding::UTF_16BE)
+
+      -> { @io.set_encoding_by_bom }.should raise_error(ArgumentError, 'encoding conversion is set')
+    end
   end
 end

spec/tags/core/io/set_encoding_by_bom_tags.txt

@@ -57,7 +57,6 @@ fails:Public methods on TracePoint should include parameters
 fails:Public methods on TracePoint should include raised_exception
 fails:Public methods on TracePoint should include return_value
 fails:Public methods on ENV.singleton_class should include freeze
-fails:Public methods on IO should include set_encoding_by_bom
 fails:Public methods on BasicSocket should include read_nonblock
 fails:Public methods on BasicSocket should include write_nonblock
 fails:Public methods on Socket should not include local_address

spec/tags/truffle/methods_tags.txt

@@ -2158,6 +2158,25 @@ def set_encoding(external, internal=nil, options=undefined)
     self
   end
 
+  def set_encoding_by_bom
+    unless binmode?
+      raise ArgumentError, 'ASCII incompatible encoding needs binmode'
+    end
+
+    if internal_encoding
+      raise ArgumentError, 'encoding conversion is set'
+    end
+
+    if external_encoding && external_encoding != Encoding::ASCII_8BIT
+      raise ArgumentError, "encoding is set to #{external_encoding} already"
+    end
+
+    external = strip_bom
+    if external
+      @external = Encoding.find(external)
+    end
+  end
+
   private def strip_bom
     mode = Truffle::POSIX.truffleposix_fstat_mode(Primitive.io_fd(self))
     return unless Truffle::StatOperations.file?(mode)

src/main/ruby/truffleruby/core/io.rb

@@ -147,9 +147,3 @@
 exclude :test_bom_non_reading, "needs investigation"
 exclude :test_stdin, "needs investigation"
 exclude :"test_strip_bom:UTF-16LE", "needs investigation"
-exclude :test_strip_bom_no_conv, "needs investigation"
-exclude :"test_strip_bom:UTF-16BE", "needs investigation"
-exclude :"test_strip_bom:UTF-32BE", "needs investigation"
-exclude :"test_strip_bom:UTF-32LE", "needs investigation"
-exclude :"test_strip_bom:UTF-8", "needs investigation"
-exclude :test_strip_bom_no_bom, "needs investigation"