From 69ec3f70fab0c1c537c68fb135cc315181b1d750 Mon Sep 17 00:00:00 2001
From: Nobuyoshi Nakada <nobu@ruby-lang.org>
Date: Mon, 11 Nov 2019 09:59:40 +0900
Subject: Warn EOF char in comment

---
 parse.y                  | 29 +++++++++++++++++++++++++----
 test/ruby/test_parse.rb  |  9 +++++++++
 test/ruby/test_syntax.rb |  6 ++++++
 3 files changed, 40 insertions(+), 4 deletions(-)

diff --git a/parse.y b/parse.y
index e84468d2a5..039abc48d7 100644
--- a/parse.y
+++ b/parse.y
@@ -6191,7 +6191,30 @@ parser_str_new(const char *ptr, long len, rb_encoding *enc, int func, rb_encodin
     return str;
 }
 
-#define lex_goto_eol(p) ((p)->lex.pcur = (p)->lex.pend)
+static const char *
+eof_char(int c)
+{
+    switch (c) {
+      case '\0': return "\\0";
+      case '\004': return "^D";
+      case '\032': return "^Z";
+    }
+    return 0;
+}
+
+static void
+lex_goto_eol(struct parser_params *p)
+{
+    const char *pcur = p->lex.pcur, *pend = p->lex.pend;
+    for (; pcur < pend; pcur++) {
+        const char *eof = eof_char(*pcur);
+        if (eof) {
+            rb_warning1("encountered %s in comment, just ignored in this version", WARN_S(eof));
+            break;
+        }
+    }
+    p->lex.pcur = pend;		/* pcur */
+}
 #define lex_eol_p(p) ((p)->lex.pcur >= (p)->lex.pend)
 #define lex_eol_n_p(p,n) ((p)->lex.pcur+(n) >= (p)->lex.pend)
 #define peek(p,c) peek_n(p, (c), 0)
@@ -7398,9 +7421,7 @@ word_match_p(struct parser_params *p, const char *word, long len)
     if (p->lex.pcur + len == p->lex.pend) return 1;
     int c = (unsigned char)p->lex.pcur[len];
     if (ISSPACE(c)) return 1;
-    switch (c) {
-      case '\0': case '\004': case '\032': return 1;
-    }
+    if (eof_char(c)) return 1;
     return 0;
 }
 
diff --git a/test/ruby/test_parse.rb b/test/ruby/test_parse.rb
index 2622ec529c..17a0c8a077 100644
--- a/test/ruby/test_parse.rb
+++ b/test/ruby/test_parse.rb
@@ -720,6 +720,15 @@ x = __ENCODING__
     assert_syntax_error("=begin", error)
   end
 
+  def test_embedded_rd_warning
+    [["\0", "\\0"], ["\C-d", "^D"], ["\C-z", "^Z"]].each do |eof, mesg|
+      mesg = /encountered #{Regexp.quote(mesg)}/
+      assert_warning(mesg) {eval("=begin\n#{eof}\n=end")}
+      assert_warning(mesg) {eval("=begin#{eof}\n=end")}
+      assert_warning(mesg) {eval("=begin\n=end#{eof}\n")}
+    end
+  end
+
   def test_float
     assert_equal(1.0/0, eval("1e10000"))
     assert_syntax_error('1_E', /trailing `_'/)
diff --git a/test/ruby/test_syntax.rb b/test/ruby/test_syntax.rb
index 5fb4dfa64f..d0748e7028 100644
--- a/test/ruby/test_syntax.rb
+++ b/test/ruby/test_syntax.rb
@@ -943,6 +943,12 @@ eom
     end
   end
 
+  def test_warning_for_eof_in_comment
+    assert_warning(/encountered \\0/) {eval("#\0")}
+    assert_warning(/encountered \^D/) {eval("#\C-d")}
+    assert_warning(/encountered \^Z/) {eval("#\C-z")}
+  end
+
   def test_unexpected_fraction
     msg = /unexpected fraction/
     assert_syntax_error("0x0.0", msg)
-- 
cgit v1.2.3