From 2ea1950510003dbce5a761a712039ed69fa499f2 Mon Sep 17 00:00:00 2001
From: Kevin Newton <kddnewton@gmail.com>
Date: Mon, 16 Sep 2024 14:40:19 -0400
Subject: [ruby/prism] Do not leak explicit encoding

Fixes [Bug #20744]

https://github.com/ruby/prism/commit/f1b8b1b2a2
---
 lib/prism/translation/ruby_parser.rb           |  8 ++++-
 prism/prism.c                                  |  1 +
 test/prism/fixtures/regex_escape_encoding.txt  |  3 ++
 test/prism/locals_test.rb                      |  4 +++
 test/prism/snapshots/regex_escape_encoding.txt | 43 ++++++++++++++++++++++++++
 test/prism/test_helper.rb                      |  2 +-
 6 files changed, 59 insertions(+), 2 deletions(-)
 create mode 100644 test/prism/fixtures/regex_escape_encoding.txt
 create mode 100644 test/prism/snapshots/regex_escape_encoding.txt

diff --git a/lib/prism/translation/ruby_parser.rb b/lib/prism/translation/ruby_parser.rb
index 9cd39075ec..4ccff0b600 100644
--- a/lib/prism/translation/ruby_parser.rb
+++ b/lib/prism/translation/ruby_parser.rb
@@ -1428,7 +1428,13 @@ module Prism
         # "foo"
         # ^^^^^
         def visit_string_node(node)
-          s(node, :str, node.unescaped)
+          unescaped = node.unescaped
+
+          if node.forced_binary_encoding?
+            unescaped.force_encoding(Encoding::BINARY)
+          end
+
+          s(node, :str, unescaped)
         end
 
         # super(foo)
diff --git a/prism/prism.c b/prism/prism.c
index 6d4d2f8199..abbeb1fd36 100644
--- a/prism/prism.c
+++ b/prism/prism.c
@@ -168,6 +168,7 @@ lex_mode_push_regexp(pm_parser_t *parser, uint8_t incrementor, uint8_t terminato
         breakpoints[index++] = incrementor;
     }
 
+    parser->explicit_encoding = NULL;
     return lex_mode_push(parser, lex_mode);
 }
 
diff --git a/test/prism/fixtures/regex_escape_encoding.txt b/test/prism/fixtures/regex_escape_encoding.txt
new file mode 100644
index 0000000000..74e1647d67
--- /dev/null
+++ b/test/prism/fixtures/regex_escape_encoding.txt
@@ -0,0 +1,3 @@
+# encoding: US-ASCII
+str = "hello \xFC"
+str =~ /hello \u{fc}/
diff --git a/test/prism/locals_test.rb b/test/prism/locals_test.rb
index ea61fd9499..2c0036289c 100644
--- a/test/prism/locals_test.rb
+++ b/test/prism/locals_test.rb
@@ -9,6 +9,10 @@
 # to test on the most recent versions.
 return if !defined?(RubyVM::InstructionSequence) || RUBY_VERSION < "3.4.0"
 
+# If we're on Ruby 3.4.0 and the default parser is Prism, then there is no point
+# in comparing the locals because they will be the same.
+return if RubyVM::InstructionSequence.compile("").to_a[4][:parser] == :prism
+
 # In Ruby 3.4.0, the local table for method forwarding changed. But 3.4.0 can
 # refer to the dev version, so while 3.4.0 still isn't released, we need to
 # check if we have a high enough revision.
diff --git a/test/prism/snapshots/regex_escape_encoding.txt b/test/prism/snapshots/regex_escape_encoding.txt
new file mode 100644
index 0000000000..0f2dcc2c54
--- /dev/null
+++ b/test/prism/snapshots/regex_escape_encoding.txt
@@ -0,0 +1,43 @@
+@ ProgramNode (location: (2,0)-(3,21))
+├── flags: ∅
+├── locals: [:str]
+└── statements:
+    @ StatementsNode (location: (2,0)-(3,21))
+    ├── flags: ∅
+    └── body: (length: 2)
+        ├── @ LocalVariableWriteNode (location: (2,0)-(2,18))
+        │   ├── flags: newline
+        │   ├── name: :str
+        │   ├── depth: 0
+        │   ├── name_loc: (2,0)-(2,3) = "str"
+        │   ├── value:
+        │   │   @ StringNode (location: (2,6)-(2,18))
+        │   │   ├── flags: forced_binary_encoding
+        │   │   ├── opening_loc: (2,6)-(2,7) = "\""
+        │   │   ├── content_loc: (2,7)-(2,17) = "hello \\xFC"
+        │   │   ├── closing_loc: (2,17)-(2,18) = "\""
+        │   │   └── unescaped: "hello \xFC"
+        │   └── operator_loc: (2,4)-(2,5) = "="
+        └── @ CallNode (location: (3,0)-(3,21))
+            ├── flags: newline
+            ├── receiver:
+            │   @ LocalVariableReadNode (location: (3,0)-(3,3))
+            │   ├── flags: ∅
+            │   ├── name: :str
+            │   └── depth: 0
+            ├── call_operator_loc: ∅
+            ├── name: :=~
+            ├── message_loc: (3,4)-(3,6) = "=~"
+            ├── opening_loc: ∅
+            ├── arguments:
+            │   @ ArgumentsNode (location: (3,7)-(3,21))
+            │   ├── flags: ∅
+            │   └── arguments: (length: 1)
+            │       └── @ RegularExpressionNode (location: (3,7)-(3,21))
+            │           ├── flags: static_literal, forced_utf8_encoding
+            │           ├── opening_loc: (3,7)-(3,8) = "/"
+            │           ├── content_loc: (3,8)-(3,20) = "hello \\u{fc}"
+            │           ├── closing_loc: (3,20)-(3,21) = "/"
+            │           └── unescaped: "hello \\u{fc}"
+            ├── closing_loc: ∅
+            └── block: ∅
diff --git a/test/prism/test_helper.rb b/test/prism/test_helper.rb
index d6d0abf548..b848500283 100644
--- a/test/prism/test_helper.rb
+++ b/test/prism/test_helper.rb
@@ -209,7 +209,7 @@ module Prism
 
     private
 
-    if RUBY_ENGINE == "ruby"
+    if RUBY_ENGINE == "ruby" && RubyVM::InstructionSequence.compile("").to_a[4][:parser] != :prism
       # Check that the given source is valid syntax by compiling it with RubyVM.
       def check_syntax(source)
         ignore_warnings { RubyVM::InstructionSequence.compile(source) }
-- 
cgit v1.2.3