Merge pull request #23 from ruz/regexp-fixes

Fixes for character classes in regexes

Author: moritz

src/QRegex/Cursor.nqp

@@ -142,8 +142,9 @@ role NQPCursorRole {
     method wb() {
         my $cur := self."!cursor_start"();
         $cur."!cursor_pass"($!pos, "wb")
-            if $!pos == 0
-               || $!pos == nqp::chars($!target)
+            if ($!pos == 0 && nqp::iscclass(pir::const::CCLASS_WORD, $!target, $!pos))
+               || ($!pos == nqp::chars($!target)
+                   && nqp::iscclass(pir::const::CCLASS_WORD, $!target, $!pos-1))
                || nqp::iscclass(pir::const::CCLASS_WORD, $!target, $!pos-1)
                   != nqp::iscclass(pir::const::CCLASS_WORD, $!target, $!pos);
         $cur;

src/QRegex/P6Regex/Grammar.nqp

@@ -192,7 +192,7 @@ grammar QRegex::P6Regex::Grammar is HLL::Grammar {
         <.normspace>?
         [
         | '[' $<charspec>=(
-                  | \s* '-' <.obs: '- as character range','..'>
+                  | \s* '-' <!before \s* ']'> <.obs: '- as character range','.. for range, for explicit - in character class, escape it or place as last thing'>
                   | \s* [ \\ (.) | (<-[\]\\]>) ] [ \s* '..' \s* (.) ]?
               )*
           \s* ']'

t/qregex/rx_charclass

@@ -24,7 +24,15 @@
 <-[b..d]>		bbccdd		n	negated character range
 # todo :pge<reversed character range>
 <-[d..b]>		bbccdd		/parse error/	illegal character range
-<[-]>			ab-def		/Unsupported/	unescaped hyphen
+<[x-z]>			ab-def		<Unsupported>	perl5 range
+<[-]>			ab-def		y	unescaped hyphen, the only thing
+<[ - ]>			ab-def		y	unescaped hyphen, the only thing w spaces
+<[x-]>			ab-def		y	unescaped hyphen, at the end
+<[x - ]>		ab-def		y	unescaped hyphen, at the end w spaces
+# todo :pge<feature>
+<[-x]>			ab-def		y	unescaped hyphen, at the beginning
+# todo :pge<feature>
+<[ -x]>			ab-def		y	unescaped hyphen, at the beginning w spaces
 <[\-]>			ab-def		y	escaped hyphen
 <[\-]>			abcdef		n	escaped hyphen
 <-[\-]>			---x--		y	negated escaped hyphen

t/qregex/rx_subrules

@@ -5,8 +5,11 @@
 
 <?wb>def		abc\ndef\n-==\nghi	y	word boundary \W\w
 abc<?wb>		abc\ndef\n-==\nghi	y	word boundary \w\W
+<wb>			-			n	no boundary at all
 <?wb>abc		abc\ndef\n-==\nghi	y	BOS word boundary
+^^<wb>			-ab			n	BOS is not boundary
 ghi<?wb>		abc\ndef\n-==\nghi	y	EOS word boundary
+<wb>$$			ab-			n	EOS is not boundary
 a<?wb>			abc\ndef\n-==\nghi	n	\w\w word boundary
 \-<?wb>			abc\ndef\n-==\nghi	n	\W\W word boundary
 <!wb>def		abc\ndef\n-==\nghi	n	nonword boundary \W\w