Add support for ASCII/Unicode modifiers

[chrisseaton]

This JRuby spec https://github.com/jruby/jruby/blob/master/spec/ruby/language/regexp/modifiers_spec.rb#L103-L108 is tagged https://github.com/jruby/jruby/blob/master/spec/tags/ruby/language/regexp/modifiers_tags.txt#L1 because Joni doesn't yet have support for ASCII/Unicode modifiers. Onigmo commit: https://github.com/k-takata/Onigmo/blob/master/regparse.c#L5170.

[headius]

I started filling in the missing group flags but did not get everything in place.

My work is in ac5ed3b.

I tracked the 'a' flag back to this commit in Onigmo:k-takata/Onigmo@52b242c

Ping @lopex!

Untested, incomplete diff that starts to fill in more blanks:

diff --git a/src/org/joni/Analyser.java b/src/org/joni/Analyser.java
index 6a4e418..9682d95 100644
--- a/src/org/joni/Analyser.java
+++ b/src/org/joni/Analyser.java
@@ -51,6 +51,7 @@ import org.joni.ast.StringNode;
 import org.joni.constants.AnchorType;
 import org.joni.constants.EncloseType;
 import org.joni.constants.NodeType;
+import org.joni.constants.OPCode;
 import org.joni.constants.StackPopLevel;
 import org.joni.constants.TargetInfo;

@@ -2100,6 +2101,38 @@ final class Analyser extends Parser {
                 opt.anchor.add(an.type);
                 break;

+            case AnchorType.WORD_BOUND:
+                if (an.asciiRange) {
+                    opt.anchor.add(OPCode.ASCII_WORD_BOUND);
+                } else {
+                    opt.anchor.add(OPCode.WORD_BOUND);
+                }
+                break;
+
+            case AnchorType.NOT_WORD_BOUND:
+                if (an.asciiRange) {
+                    opt.anchor.add(OPCode.NOT_ASCII_WORD_BOUND);
+                } else {
+                    opt.anchor.add(OPCode.NOT_WORD_BOUND);
+                }
+                break;
+
+            case AnchorType.WORD_BEGIN:
+                if (an.asciiRange) {
+                    opt.anchor.add(OPCode.ASCII_WORD_BEGIN);
+                } else {
+                    opt.anchor.add(OPCode.WORD_BEGIN);
+                }
+                break;
+
+            case AnchorType.WORD_END:
+                if (an.asciiRange) {
+                    opt.anchor.add(OPCode.ASCII_WORD_END);
+                } else {
+                    opt.anchor.add(OPCode.WORD_END);
+                }
+                break;
+
             case AnchorType.PREC_READ:
                 NodeOptInfo nopt = new NodeOptInfo();
                 optimizeNodeLeft(an.target, nopt, oenv);

[lopex]

@headius, this is going to be a tough one. This and dependent features were plagued by bugs across many commits. Now picking related ones, it's going to be a multi day effort (already have some code ported).

[lopex]

Fixed in 416f5f9 and 4e3b5cc