This patch improves the handling of invalid ASCII characters in the Go
frontend when they appear in identifiers. Note that Go input is by
definition always UTF-8, so the ASCII-specific assumptions here are OK.
Bootstrapped and ran Go testsuite on x86_64-unknown-linux-gnu.
Committed to mainline. Will commit to 4.8 branch when it reopens.
Ian
diff -r 99af66244a26 go/lex.cc
--- a/go/lex.cc Thu Oct 10 20:11:40 2013 -0700
+++ b/go/lex.cc Fri Oct 11 10:02:05 2013 -0700
@@ -873,7 +873,28 @@
&& (cc < 'a' || cc > 'z')
&& cc != '_'
&& (cc < '0' || cc > '9'))
- break;
+ {
+ // Check for an invalid character here, as we get better
+ // error behaviour if we swallow them as part of the
+ // identifier we are building.
+ if ((cc >= ' ' && cc < 0x7f)
+ || cc == '\t'
+ || cc == '\r'
+ || cc == '\n')
+ break;
+
+ this->lineoff_ = p - this->linebuf_;
+ error_at(this->location(),
+ "invalid character 0x%x in identifier",
+ cc);
+ if (!has_non_ascii_char)
+ {
+ buf.assign(pstart, p - pstart);
+ has_non_ascii_char = true;
+ }
+ if (!Lex::is_invalid_identifier(buf))
+ buf.append("$INVALID$");
+ }
++p;
if (is_first)
{
