From fd406fc7046f427385b644759265ae06ed741d6b Mon Sep 17 00:00:00 2001
From: Owen Avery <powerboat9.gamer@gmail.com>
Date: Sun, 8 Jan 2023 17:19:12 -0500
Subject: [PATCH] gccrs: Implemented UTF-8 checking for include_str!()

gcc/rust/ChangeLog:

	* expand/rust-macro-builtins.cc
	(MacroBuiltin::include_str_handler): Add check for valid UTF-8.

gcc/testsuite/ChangeLog:

	* rust/compile/builtin_macro_include_str.rs:
	Include test of invalid UTF-8.
	* rust/compile/invalid_utf8: File with invalid UTF-8.

Signed-off-by: Owen Avery <powerboat9.gamer@gmail.com>
---
 gcc/rust/expand/rust-macro-builtins.cc             | 51 +++++++++++++++++++++-
 .../rust/compile/builtin_macro_include_str.rs      |  1 +
 gcc/testsuite/rust/compile/invalid_utf8            |  1 +
 3 files changed, 51 insertions(+), 2 deletions(-)
 create mode 100644 gcc/testsuite/rust/compile/invalid_utf8

diff --git a/gcc/rust/expand/rust-macro-builtins.cc b/gcc/rust/expand/rust-macro-builtins.cc
index e594a25..3b6f69b 100644
--- a/gcc/rust/expand/rust-macro-builtins.cc
+++ b/gcc/rust/expand/rust-macro-builtins.cc
@@ -389,8 +389,55 @@ MacroBuiltin::include_str_handler (Location invoc_locus,
 
   std::vector<uint8_t> bytes = load_file_bytes (target_filename.c_str ());
 
-  /* FIXME: Enforce that the file contents are valid UTF-8.  */
-  std::string str ((const char *) &bytes[0], bytes.size ());
+  /* FIXME: reuse lexer */
+  int expect_single = 0;
+  for (uint8_t b : bytes)
+    {
+      if (expect_single)
+	{
+	  if ((b & 0xC0) != 0x80)
+	    /* character was truncated, exit with expect_single != 0 */
+	    break;
+	  expect_single--;
+	}
+      else if (b & 0x80)
+	{
+	  if (b >= 0xF8)
+	    {
+	      /* more than 4 leading 1s */
+	      expect_single = 1;
+	      break;
+	    }
+	  else if (b >= 0xF0)
+	    {
+	      /* 4 leading 1s */
+	      expect_single = 3;
+	    }
+	  else if (b >= 0xE0)
+	    {
+	      /* 3 leading 1s */
+	      expect_single = 2;
+	    }
+	  else if (b >= 0xC0)
+	    {
+	      /* 2 leading 1s */
+	      expect_single = 1;
+	    }
+	  else
+	    {
+	      /* only 1 leading 1 */
+	      expect_single = 1;
+	      break;
+	    }
+	}
+    }
+
+  std::string str;
+  if (expect_single)
+    rust_error_at (invoc_locus, "%s was not a valid utf-8 file",
+		   target_filename.c_str ());
+  else
+    str = std::string ((const char *) &bytes[0], bytes.size ());
 
   auto node = AST::SingleASTNode (make_string (invoc_locus, str));
   auto str_tok = make_token (Token::make_string (invoc_locus, std::move (str)));
diff --git a/gcc/testsuite/rust/compile/builtin_macro_include_str.rs b/gcc/testsuite/rust/compile/builtin_macro_include_str.rs
index 38f5e3b..8092193 100644
--- a/gcc/testsuite/rust/compile/builtin_macro_include_str.rs
+++ b/gcc/testsuite/rust/compile/builtin_macro_include_str.rs
@@ -10,4 +10,5 @@ fn main () {
   include_str! ("foo.txt", "bar.txt"); // { dg-error "macro takes 1 argument" "" }
   include_str! ("builtin_macro_include_str.rs"); // ok
   include_str! ("builtin_macro_include_str.rs",); // trailing comma ok
+  include_str! ("invalid_utf8"); // { dg-error "invalid_utf8 was not a valid utf-8 file" "" }
 }
diff --git a/gcc/testsuite/rust/compile/invalid_utf8 b/gcc/testsuite/rust/compile/invalid_utf8
new file mode 100644
index 0000000..29e181e
--- /dev/null
+++ b/gcc/testsuite/rust/compile/invalid_utf8
@@ -0,0 +1 @@
+ÿ
-- 
2.7.4