Unicode properties in regexes (new S05 syntax)

Raku · Nov 27, 2011 · d4a0667 · d4a0667
1 parent 109e4aa
commit d4a0667
Show file tree

Hide file tree

Showing 3 changed files with 17 additions and 1 deletion.
diff --git a/src/QAST/Compiler.nqp b/src/QAST/Compiler.nqp
@@ -415,6 +415,16 @@ class QAST::Compiler is HLL::Compiler {
         $ops;
     }
 
+    method uniprop($node) {
+        my $ops := self.post_new('Ops', :result(%*REG<cur>));
+        my $cmpop := $node.negate ?? 'ne' !! 'eq';
+        $ops.push_pirop('assign', '$S10', '"' ~ $node[0] ~ '"');
+        $ops.push_pirop('is_uprop', '$I11', '$S10', %*REG<tgt>, %*REG<pos>);
+        $ops.push_pirop($cmpop, '$I11', 0, %*REG<fail>);
+        $ops.push_pirop('inc', %*REG<pos>) unless $node.subtype eq 'zerowidth';
+        $ops;
+    }
+
     # a :rxtype<ws> node is a normal subrule call
     method ws($node) { self.subrule($node) }
 

diff --git a/src/QRegex/P6Regex/Actions.nqp b/src/QRegex/P6Regex/Actions.nqp
@@ -409,6 +409,11 @@ class QRegex::P6Regex::Actions is HLL::Actions {
             $qast := QAST::Regex.new( PAST::Node.new($name), :rxtype<subrule>, :subtype<method>,
                                       :negate( $<sign> eq '-' ), :node($/) );
         }
+        elsif $<uniprop> {
+            my $uniprop := ~$<uniprop>;
+            $qast := QAST::Regex.new( $uniprop, :rxtype<uniprop>,
+                                      :negate( $<sign> eq '-' ), :node($/) );
+        }
         else {
             my @alts;
             for $<charspec> {

diff --git a/src/QRegex/P6Regex/Grammar.nqp b/src/QRegex/P6Regex/Grammar.nqp
@@ -193,7 +193,7 @@ grammar QRegex::P6Regex::Grammar is HLL::Grammar {
             ]?
     }
 
-    token assertion:sym<[> { <?before '['|'+'|'-'> <cclass_elem>+ }
+    token assertion:sym<[> { <?before '['|'+'|'-'|':'> <cclass_elem>+ }
 
     token cclass_elem {
         $<sign>=['+'|'-'|<?>]
@@ -209,6 +209,7 @@ grammar QRegex::P6Regex::Grammar is HLL::Grammar {
               )*
           \s* ']'
         | $<name>=[\w+]
+        | ':' $<uniprop>=[\w+]
         ]
         <.normspace>?
     }