compiler: fix parsing issue with non-ASCII first package char

Fix a bug in the parser code that decides whether a given name should
be considered exported or not. The function Lex::is_exported_name
(which assumes that its input is a mangled name) was being called on
non-mangled (raw utf-8) names in various places. For the bug in
question this caused an imported package to be registered under the
wrong name. To fix the issue, rename 'Lex::is_exported_name' to
'Lex::is_exported_mangled_name', and add a new 'Lex::is_exported_name'
that works on utf-8 strings.

Fixes golang/go#27836.

Change-Id: I2d3b923c9f507348a7f0f2c8161d3cc2dc68d02f
Reviewed-on: https://go-review.googlesource.com/137736
Reviewed-by: Ian Lance Taylor <[email protected]>

Author: thanm

go/import.cc

@@ -983,8 +983,6 @@ Import::read_name()
   std::string ret = this->read_identifier();
   if (ret == "?")
     ret.clear();
-  else if (!Lex::is_exported_name(ret))
-    ret = '.' + this->package_->pkgpath() + '.' + ret;
   return ret;
 }
 

go/lex.cc

@@ -2764,7 +2764,7 @@ Lex::is_unicode_uppercase(unsigned int c)
 // mangled name which includes only ASCII characters.
 
 bool
-Lex::is_exported_name(const std::string& name)
+Lex::is_exported_mangled_name(const std::string& name)
 {
   unsigned char c = name[0];
   if (c != '.')
@@ -2791,6 +2791,18 @@ Lex::is_exported_name(const std::string& name)
     }
 }
 
+// Return whether the identifier NAME should be exported.  NAME is a
+// an unmangled utf-8 string and may contain non-ASCII characters.
+
+bool
+Lex::is_exported_name(const std::string& name)
+{
+  unsigned int uchar;
+  if (Lex::fetch_char(name.c_str(), &uchar) != 0)
+    return Lex::is_unicode_letter(uchar) && Lex::is_unicode_uppercase(uchar);
+  return false;
+}
+
 // Return whether the identifier NAME contains an invalid character.
 // This is based on how we handle invalid characters in
 // gather_identifier.

go/lex.h

@@ -408,6 +408,11 @@ class Lex
   // Return whether the identifier NAME should be exported.  NAME is a
   // mangled name which includes only ASCII characters.
   static bool
+  is_exported_mangled_name(const std::string& name);
+
+  // Return whether the identifier NAME should be exported.  NAME is
+  // an unmangled utf-8 string and may contain non-ASCII characters.
+  static bool
   is_exported_name(const std::string& name);
 
   // Return whether the identifier NAME is invalid.  When we see an