From e1ef0ac0eab586c2180f133be260893dd407284a Mon Sep 17 00:00:00 2001
From: Matijs van Zuijlen <matijs@matijs.net>
Date: Fri, 24 Sep 2021 13:06:41 +0200
Subject: [PATCH] Fix squiggly heredoc line continuation handling

Firstly, this fixes an issue with non-interpolating squiggly heredocs,
where line continuation characters should be ignored. Two things were
broken:

- The line continuation backslash would be removed, as well as the
  subsequent carriage return
- The subsequent line would not be dedented.

For example, the following source:

<<~'END'
  foo \
  bar
END

Would be parsed as follows:

(dstr
  (str "foo ")
  (str "  bar\n"))

With the change, the result is:

(dstr
  (str "foo \\\n")
  (str "bar\n"))

This also fixes an issue with interpolating squiggly heredocs, where an
escaped backslash followed by a newline would be interpreted as a line
continuation.

In that case, the following source:

<<~END
  foo \\
  bar
END

Would (again) be parsed as follows:

(dstr
  (str "foo ")
  (str "  bar\n"))

With the change, the result is:

(dstr
  (str "foo \\\n")
  (str "bar\n"))
---
 lib/parser/lexer/dedenter.rb |  8 +++++++-
 test/test_parser.rb          | 16 ++++++++++++++++
 2 files changed, 23 insertions(+), 1 deletion(-)

diff --git a/lib/parser/lexer/dedenter.rb b/lib/parser/lexer/dedenter.rb
index 5f0ff2700..5ff16007a 100644
--- a/lib/parser/lexer/dedenter.rb
+++ b/lib/parser/lexer/dedenter.rb
@@ -38,7 +38,13 @@ def dedent(string)
       # Prevent the following error when processing binary encoded source.
       # "\xC0".split # => ArgumentError (invalid byte sequence in UTF-8)
       lines = string.force_encoding(Encoding::BINARY).split("\\\n")
-      lines.map! {|s| s.force_encoding(original_encoding) }
+      if lines.length == 1
+        # If the line continuation sequence was found but there is no second
+        # line, it was not really a line continuation and must be ignored.
+        lines = [string]
+      else
+        lines.map! {|s| s.force_encoding(original_encoding) }
+      end
 
       if @at_line_begin
         lines_to_dedent = lines
diff --git a/test/test_parser.rb b/test/test_parser.rb
index 7aa78c7a1..20d77d725 100644
--- a/test/test_parser.rb
+++ b/test/test_parser.rb
@@ -439,6 +439,22 @@ def test_parser_bug_640
       SINCE_2_3)
   end
 
+  def test_dedenting_non_interpolating_heredoc_line_continuation
+    assert_parses(
+      s(:dstr, s(:str, "baz\\\n"), s(:str, "qux\n")),
+      %Q{<<~'FOO'\n  baz\\\n  qux\nFOO},
+      %q{},
+      SINCE_2_3)
+  end
+
+  def test_dedenting_interpolating_heredoc_fake_line_continuation
+    assert_parses(
+      s(:dstr, s(:str, "baz\\\n"), s(:str, "qux\n")),
+      %Q{<<~'FOO'\n  baz\\\\\n  qux\nFOO},
+      %q{},
+      SINCE_2_3)
+  end
+
   # Symbols
 
   def test_symbol_plain