From df3138fd2e83f0ba079b91fe923cdeab2c319281 Mon Sep 17 00:00:00 2001
From: Stephen Paul Weber <singpolyma@singpolyma.net>
Date: Sun, 28 Apr 2019 13:18:21 -0500
Subject: [PATCH] Update text-or-binary heuristic

The `cbor` gem is a bit janky and it turns out catching NoMemoryError is
not safe, so use an encoding-based heuristic to detect binary vs utf8
data.
---
 lib/dhall.rb      | 15 +++++----------
 lib/dhall/util.rb | 19 +++++++++++++++++++
 2 files changed, 24 insertions(+), 10 deletions(-)

diff --git a/lib/dhall.rb b/lib/dhall.rb
index f28ba66..ae4b91b 100644
--- a/lib/dhall.rb
+++ b/lib/dhall.rb
@@ -22,18 +22,13 @@ module Dhall
 	end
 
 	def self.load_raw(source)
-		unless source.valid_encoding?
-			raise ArgumentError, "invalid byte sequence in #{source.encoding}"
-		end
+		source = Util.text_or_binary(source)
 
-		begin
-			return from_binary(source) if source.encoding == Encoding::BINARY
-		rescue Exception # rubocop:disable Lint/RescueException
-			# Parsing CBOR failed, so guess this is source text in standard UTF-8
-			return load_raw(source.force_encoding("UTF-8"))
+		if source.encoding == Encoding::BINARY
+			from_binary(source)
+		else
+			Parser.parse(source).value
 		end
-
-		Parser.parse(source.encode("UTF-8")).value
 	end
 
 	def self.dump(o)
diff --git a/lib/dhall/util.rb b/lib/dhall/util.rb
index f2c2b8a..1ce737f 100644
--- a/lib/dhall/util.rb
+++ b/lib/dhall/util.rb
@@ -116,5 +116,24 @@ module Dhall
 
 			Hash[hash_or_not.map { |k, v| [(yield k), v] }]
 		end
+
+		def self.utf8_if_possible(str)
+			utf8 = str.dup.force_encoding(Encoding::UTF_8)
+			utf8.valid_encoding? ? utf8 : str
+		end
+
+		def self.text_or_binary(str)
+			unless str.valid_encoding?
+				raise ArgumentError, "invalid byte sequence in #{str.encoding}"
+			end
+
+			if str.encoding == Encoding::BINARY
+				return str if str =~ /(?!\s)[[:cntrl:]]/
+
+				utf8_if_possible(str)
+			else
+				str.encode(Encoding::UTF_8)
+			end
+		end
 	end
 end
-- 
2.34.2