cpp.texi: Update for handling of charconsts.

	* cpp.texi: Update for handling of charconsts.
	* cpplex.c (maybe_read_ucs): Don't accept D800-DFFF.  Update
	diagnostics.  Skip to the end if the UCS is too short.
	(cpp_interpret_charconst): Long charconsts issue a warning
	not an error.

	* gcc.dg/cpp/charconst.c: New tests.
	* gcc.dg/cpp/escape.c: New tests.
	* gcc.dg/cpp/escape-1.c: New tests.
	* gcc.dg/cpp/escape-2.c: New tests.
	* gcc.dg/cpp/ucs.c: New tests.

From-SVN: r42514

9 files changed, 207 insertions, 42 deletions

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index a29771e..14102e1 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,11 @@
+2001-05-23  Neil Booth  <neil@daikokuya.demon.co.uk>
+
+	* cpp.texi: Update for handling of charconsts.
+	* cpplex.c (maybe_read_ucs): Don't accept D800-DFFF.  Update
+	diagnostics.  Skip to the end if the UCS is too short.
+	(cpp_interpret_charconst): Long charconsts issue a warning
+	not an error.
+
 2001-05-23  Richard Henderson  <rth@redhat.com>
 
 	* doc/install.texi (alpha-linux): Require binutils 2.11.
diff --git a/gcc/cpp.texi b/gcc/cpp.texi
index ba50a26..a5dc889 100644
--- a/gcc/cpp.texi
+++ b/gcc/cpp.texi
@@ -2889,22 +2889,17 @@ same column as it did in the original source file.
 
 @item The numeric value of character constants in preprocessor expressions.
 
-The preprocessor interprets character constants in preprocessing
-directives on the host machine.  Expressions outside preprocessing
-directives are compiled to be interpreted on the target machine.  In the
-normal case of a native compiler, these two environments are the same
-and so character constants will be evaluated identically in both cases.
-However, in the case of a cross compiler, the values may be different.
+The preprocessor and compiler interpret character constants in the same
+way; escape sequences such as @code{\a} are given the values they would
+have on the target machine.
 
 Multi-character character constants are interpreted a character at a
 time, shifting the previous result left by the number of bits per
 character on the host, and adding the new character.  For example, 'ab'
 on an 8-bit host would be interpreted as 'a' * 256 + 'b'.  If there are
 more characters in the constant than can fit in the widest native
-integer type on the host, usually a @samp{long}, the behavior is
-undefined.
-
-Evaluation of wide character constants is not properly implemented yet.
+integer type on the host, usually a @samp{long}, the excess characters
+are ignored and a diagnostic is given.
 
 @item Source file inclusion.
 
diff --git a/gcc/cpplex.c b/gcc/cpplex.c
index d261749..3185acc 100644
--- a/gcc/cpplex.c
+++ b/gcc/cpplex.c
@@ -1706,46 +1706,48 @@ maybe_read_ucs (pfile, pstr, limit, pc)
 
   if (CPP_WTRADITIONAL (pfile))
     cpp_warning (pfile, "the meaning of '\\%c' varies with -traditional", c);
-  
-  for (length = (c == 'u' ? 4: 8); length; --length)
-    {
-      if (p >= limit)
-	{
-	  cpp_error (pfile, "incomplete universal-character-name");
-	  break;
-	}
 
-      c = *p;
-      if (ISXDIGIT (c))
-	{
-	  code = (code << 4) + hex_digit_value (c);
-	  p++;
-	}
-      else
+  length = (c == 'u' ? 4: 8);
+
+  if ((size_t) (limit - p) < length)
+    {
+      cpp_error (pfile, "incomplete universal-character-name");
+      /* Skip to the end to avoid more diagnostics.  */
+      p = limit;
+    }
+  else
+    {
+      for (; length; length--, p++)
 	{
-	  cpp_error (pfile,
-		     "non-hex digit '%c' in universal-character-name", c);
-	  break;
+	  c = *p;
+	  if (ISXDIGIT (c))
+	    code = (code << 4) + hex_digit_value (c);
+	  else
+	    {
+	      cpp_error (pfile,
+			 "non-hex digit '%c' in universal-character-name", c);
+	      /* We shouldn't skip in case there are multibyte chars.  */
+	      break;
+	    }
 	}
-
     }
 
 #ifdef TARGET_EBCDIC
   cpp_error (pfile, "universal-character-name on EBCDIC target");
   code = 0x3f;  /* EBCDIC invalid character */
 #else
-  if (code > 0x9f && !(code & 0x80000000))
-    ; /* True extended character, OK.  */
-  else if (code >= 0x20 && code < 0x7f)
-    {
-      /* ASCII printable character.  The C character set consists of all of
-	 these except $, @ and `.  We use hex escapes so that this also
-	 works with EBCDIC hosts.  */
-      if (code != 0x24 && code != 0x40 && code != 0x60)
-	cpp_error (pfile, "universal-character-name used for '%c'", code);
-    }
-  else
-    cpp_error (pfile, "invalid universal-character-name");
+ /* True extended characters are OK.  */
+  if (code >= 0xa0
+      && !(code & 0x80000000)
+      && !(code >= 0xD800 && code <= 0xDFFF))
+    ;
+  /* The standard permits $, @ and ` to be specified as UCNs.  We use
+     hex escapes so that this also works with EBCDIC hosts.  */
+  else if (code == 0x24 || code == 0x40 || code == 0x60)
+    ;
+  /* Don't give another error if one occurred above.  */
+  else if (length == 0)
+    cpp_error (pfile, "universal-character-name out of range");
 #endif
 
   *pstr = p;
@@ -1970,7 +1972,7 @@ cpp_interpret_charconst (pfile, token, warn_multi, traditional, pchars_seen)
   else if (chars_seen > max_chars)
     {
       chars_seen = max_chars;
-      cpp_error (pfile, "character constant too long");
+      cpp_warning (pfile, "character constant too long");
     }
   else if (chars_seen > 1 && !traditional && warn_multi)
     cpp_warning (pfile, "multi-character character constant");
diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog
index f27895a..826918e 100644
--- a/gcc/testsuite/ChangeLog
+++ b/gcc/testsuite/ChangeLog
@@ -1,3 +1,11 @@
+2001-05-23  Neil Booth  <neil@daikokuya.demon.co.uk>
+
+	* gcc.dg/cpp/charconst.c: New tests.
+	* gcc.dg/cpp/escape.c: New tests.
+	* gcc.dg/cpp/escape-1.c: New tests.
+	* gcc.dg/cpp/escape-2.c: New tests.
+	* gcc.dg/cpp/ucs.c: New tests.
+
 2001-05-23  David.Billinghurst <David.Billinghurst@riotinto.com>
 
 	* gcc.misc-tests/linkage.exp:  Pass appropriate flags to 
diff --git a/gcc/testsuite/gcc.dg/cpp/charconst.c b/gcc/testsuite/gcc.dg/cpp/charconst.c
new file mode 100644
index 0000000..d8a178b
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/cpp/charconst.c
@@ -0,0 +1,33 @@
+/* Copyright (C) 2001 Free Software Foundation, Inc.  */
+
+/* { dg-do compile } */
+
+/* This tests various diagnostics about character constants, for both
+   the preprocessor and the compiler.
+
+   Neil Booth, 22 May 2001.  */
+
+#if ''			/* { dg-warning "empty" "empty charconst" } */
+#endif
+#if L''			/* { dg-warning "empty" "empty wide charconst" } */
+#endif
+#if 'very long'		/* { dg-warning "too long" "long charconst" } */
+#endif
+#if L'very long'	/* { dg-warning "too long" "long wide charconst" } */
+#endif
+/* Don't do this test for L'ab'; it depends upon sizeof (wchar_t).  */
+#if 'ab'		/* { dg-warning "multi-char" "multi-character" } */
+#endif
+
+void foo ()
+{
+  int c = '';		/* { dg-warning "empty" "empty charconst" } */
+  c = L'';		/* { dg-warning "empty" "empty wide charconst" } */
+
+  c = 'very long';	/* { dg-warning "too long" "long charconst" } */
+  c = L'very long';	/* { dg-warning "too long" "long wide charconst" } */
+
+  /* Don't do this test for L'ab'; it depends upon sizeof (wchar_t).  */
+  c = 'ab';		/* { dg-warning "multi-char" "multi-character" } */
+ 
+}
diff --git a/gcc/testsuite/gcc.dg/cpp/escape-1.c b/gcc/testsuite/gcc.dg/cpp/escape-1.c
new file mode 100644
index 0000000..52d7414
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/cpp/escape-1.c
@@ -0,0 +1,39 @@
+/* Copyright (C) 2001 Free Software Foundation, Inc.  */
+
+/* { dg-do compile } */
+
+/* This tests various diagnostics about escape sequences, for both
+   the preprocessor and the compiler.
+
+   Neil Booth, 22 May 2001.  */
+
+#if '\x'	/* { dg-error "no following" "\x with no digits" } */
+#endif
+#if '\x400'	/* { dg-warning "out of range" "\x out of range" } */
+#endif
+#if '\x0ff'	/* { dg-bogus "out of range" "\x out of range" } */
+#endif
+#if '\400'	/* { dg-warning "out of range" "\x out of range" } */
+#endif
+#if '\377'	/* { dg-bogus "out of range" "bogus \x out of range" } */
+#endif
+#if '\177' != 0x7f /* { dg-bogus "out of range" "bogus \x out of range" } */
+#error bad octal /* { dg-bogus "bad" "bad octal evaluation" } */
+#endif
+#if '\0377'	/* { dg-warning "multi" "too long octal" } */
+#endif
+#if '\p'	/* { dg-error "unknown escape" "unknown escape seq" } */
+#endif
+
+void foo ()
+{
+  int c;
+
+  c = '\x';	/* { dg-error "no following" "\x with no digits" } */
+  c = '\x100';	/* { dg-warning "out of range" "\x out of range" } */
+  c = '\x0ff';	/* { dg-bogus "out of range" "\x out of range" } */
+  c = '\400';	/* { dg-warning "out of range" "\x out of range" } */
+  c = '\377';	/* { dg-bogus "out of range" "bogus \x out of range" } */
+  c = '\0377';	/* { dg-warning "multi" "too long octal" } */
+  c = '\p';	/* { dg-error "unknown escape" "unknown escape seq" } */
+}
diff --git a/gcc/testsuite/gcc.dg/cpp/escape-2.c b/gcc/testsuite/gcc.dg/cpp/escape-2.c
new file mode 100644
index 0000000..31bf882
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/cpp/escape-2.c
@@ -0,0 +1,20 @@
+/* Copyright (C) 2001 Free Software Foundation, Inc.  */
+
+/* { dg-do compile } */
+/* { dg-options "-pedantic -std=c99 -fno-show-column" } */
+
+/* This tests various diagnostics with -pedantic about escape
+   sequences, for both the preprocessor and the compiler.
+
+   Neil Booth, 22 May 2001.  */
+
+#if '\e'		/* { dg-warning "non-ISO" "non-ISO \\e" } */
+#endif
+#if '\u00a0'		/* { dg-bogus "unknown" "\\u is known in C99" } */
+#endif
+
+void foo ()
+{
+  int c = '\E';		/* { dg-warning "non-ISO" "non-ISO \\E" } */
+  c = '\u00a0';		/* { dg-bogus "unknown" "\\u is known in C99" } */
+}
diff --git a/gcc/testsuite/gcc.dg/cpp/escape.c b/gcc/testsuite/gcc.dg/cpp/escape.c
new file mode 100644
index 0000000..44ad4c1
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/cpp/escape.c
@@ -0,0 +1,25 @@
+/* Copyright (C) 2001 Free Software Foundation, Inc.  */
+
+/* { dg-do compile } */
+/* { dg-options "-Wtraditional -std=c89 -fno-show-column" } */
+
+/* This tests various diagnostics with -Wtraditioanl about escape
+   sequences, for both the preprocessor and the compiler.
+
+   Neil Booth, 22 May 2001.  */
+
+#if '\a'		/* { dg-warning "traditional" "traditional bell" } */
+#endif
+#if '\x1a' != 26	/* { dg-warning "traditional" "traditional hex" } */
+ #error bad hex		/* { dg-bogus "bad" "bad hexadecimal evaluation" } */
+#endif
+#if '\u'		/* { dg-warning "unknown" "\u is unknown in C89" } */
+#endif
+
+void foo ()
+{
+  int c = '\a';		/* { dg-warning "traditional" "traditional bell" } */
+
+  c = '\xa1';		/* { dg-warning "traditional" "traditional hex" } */
+  c = '\u';		/* { dg-warning "unknown" "\u is unknown in C89" } */
+}
diff --git a/gcc/testsuite/gcc.dg/cpp/ucs.c b/gcc/testsuite/gcc.dg/cpp/ucs.c
new file mode 100644
index 0000000..d131b0b
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/cpp/ucs.c
@@ -0,0 +1,35 @@
+/* Copyright (C) 2001 Free Software Foundation, Inc.  */
+
+/* { dg-do compile } */
+/* { dg-options "-std=c99" } */
+
+/* This tests universal character sequences.
+
+   Neil Booth, 22 May 2001.  */
+
+#if L'\u1234' != 0x1234
+#error bad short ucs	/* { dg-bogus "bad" "bad \u1234 evaluation" } */
+#endif
+#if L'\U1234abcd' != 0x1234abcd
+#error bad long ucs	/* { dg-bogus "bad" "bad \U1234abcd evaluation" } */
+#endif
+
+void foo ()
+{
+  int c;
+
+  c = L'\ubad';		/* { dg-error "incomplete" "incompete UCN 1" } */
+  c = L"\U1234"[0];	/* { dg-error "incomplete" "incompete UCN 2" } */
+
+  c = L'\u000x';	/* { dg-error "non-hex" "non-hex digit in UCN" } */
+			/* { dg-warning "too long" "" { target *-*-* } 24 } */
+
+  c = '\u0024';		/* { dg-bogus "invalid" "0024 is a valid UCN" } */
+  c = "\u0040"[0];	/* { dg-bogus "invalid" "0040 is a valid UCN" } */
+  c = '\u00a0';		/* { dg-bogus "invalid" "00a0 is a valid UCN" } */
+  c = '\U00000060';	/* { dg-bogus "invalid" "0060 is a valid UCN" } */
+
+  c = '\u0025';		/* { dg-error "range" "0025 is an invalid UCN" } */
+  c = L"\uD800"[0];	/* { dg-error "range" "D800 is an invalid UCN" } */
+  c = L'\U0000DFFF';	/* { dg-error "range" "DFFF is an invalid UCN" } */
+}