disallow creating Uni with out-of-range codepoints

rakudo · Sep 2, 2021 · c56c578 · c56c578
1 parent ffde2ba
commit c56c578
Show file tree

Hide file tree

Showing 3 changed files with 23 additions and 3 deletions.
diff --git a/src/core.c/Exception.pm6 b/src/core.c/Exception.pm6
@@ -3327,6 +3327,15 @@ my class X::CompUnit::UnsatisfiedDependency is Exception {
     }
 }
 
+my class X::InvalidCodepoint is Exception {
+    has $.code;
+    method message() {
+        $.code > 0x10ffff
+        ?? "Invalid codepoint $.code; must not exceed 0x10ffff (1114111)"
+        !! "Invalid surrogate codepoint $.code"
+    }
+}
+
 my class Exceptions::JSON {
     method process($ex) {
         $*ERR.print: Rakudo::Internals::JSON.to-json($ex);

diff --git a/src/core.c/Uni.pm6 b/src/core.c/Uni.pm6
@@ -3,18 +3,24 @@ my class NFD is repr('VMArray') is array_type(uint32) { ... }
 my class NFKC is repr('VMArray') is array_type(uint32) { ... }
 my class NFKD is repr('VMArray') is array_type(uint32) { ... }
 
+my class X::InvalidCodepoint { ... }
+
 my class Uni does Positional[uint32] does Stringy is repr('VMArray') is array_type(uint32) {
 
     multi method new(Uni:) { nqp::create(self) }
     multi method new(Uni: *@codes) {
         @codes.elems;  # reify and test for lazy sequences
         my $uni        := nqp::create(self);
         my $codepoints := nqp::getattr(@codes,List,'$!reified');
+        my $code;
 
         nqp::while(
           nqp::elems($codepoints),
-          nqp::push_i($uni,nqp::shift($codepoints))
-        );
+          nqp::if(nqp::isgt_i($code = nqp::shift($codepoints), 0x10ffff)
+                  || (nqp::isle_i(0xd800, $code) && nqp::isle_i($code, 0xdfff))
+                  || nqp::islt_i($code, 0),
+            X::InvalidCodepoint.new(:$code).throw,
+            nqp::push_i($uni,$code)));
 
         $uni
     }

diff --git a/src/core.c/core_epilogue.pm6 b/src/core.c/core_epilogue.pm6
@@ -67,10 +67,15 @@ augment class Uni {
         my $uni      := nqp::create(self);
         my int $elems = nqp::elems(codepoints);
         my int $i = -1;
+        my int $code;
 
         nqp::while(
           nqp::islt_i(($i = nqp::add_i($i,1)),$elems),
-          nqp::push_i($uni,nqp::atpos_i(codepoints,$i))
+          nqp::if(nqp::isgt_i($code = nqp::atpos_i(codepoints,$i), 0x10ffff)
+                  || (nqp::isle_i(0xd800, $code) && nqp::isle_i($code, 0xdfff))
+                  || nqp::islt_i($code, 0),
+            X::InvalidCodepoint.new(:$code).throw,
+            nqp::push_i($uni,$code))
         );
 
         $uni