Regex: support lookaheads

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

test/Regex.java

@@ -59,5 +59,6 @@ public class Regex {
     expectGroups("(?:a)", "a");
     expectGroups("a|(b|c)", "a", (String)null);
     expectGroups("a|(b|c)", "c", "c");
+    expectGroups("(?=a)a", "a");
   }
 }

test/regex/Compiler.java

@@ -28,6 +28,7 @@ class Compiler implements PikeVMOpcodes {
     private int groupCount = -1;
     private int findPreambleSize;
     private ArrayList<CharacterMatcher> classes;
+    private ArrayList<PikeVM> lookaheads;
 
     public Output(Expression expr) {
       // try-run to determine the code size
@@ -36,6 +37,7 @@ class Compiler implements PikeVMOpcodes {
       offset = 0;
       groupCount = -1;
       classes = new ArrayList<CharacterMatcher>();
+      lookaheads = new ArrayList<PikeVM>();
       // write it out!
       expr.writeCode(this);
     }
@@ -64,7 +66,10 @@ class Compiler implements PikeVMOpcodes {
     public PikeVM toVM() {
       CharacterMatcher[] classes = new CharacterMatcher[this.classes.size()];
       this.classes.toArray(classes);
-      return new PikeVM(program, findPreambleSize, groupCount, classes);
+      PikeVM[] lookaheads = new PikeVM[this.lookaheads.size()];
+      this.lookaheads.toArray(lookaheads);
+      return new PikeVM(program, findPreambleSize, groupCount, classes,
+        lookaheads);
     }
 
     public int addClass(CharacterMatcher characterClass) {
@@ -75,6 +80,15 @@ class Compiler implements PikeVMOpcodes {
       classes.add(characterClass);
       return result;
     }
+
+    public int addLookahead(PikeVM lookahead) {
+      if (program == null) {
+        return -1;
+      }
+      int result = lookaheads.size();
+      lookaheads.add(lookahead);
+      return result;
+    }
   }
 
   private abstract class Expression {
@@ -212,6 +226,17 @@ class Compiler implements PikeVMOpcodes {
     }
   }
 
+  private class Lookahead extends Expression {
+    private final Group group = new Group(false, null);
+
+    @Override
+    protected void writeCode(Output output) {
+      PikeVM vm = new Output(group).toVM();
+      output.add(LOOKAHEAD);
+      output.add(output.addLookahead(vm));
+    }
+  }
+
   private class Group0 extends Expression {
     private final Group group;
 
@@ -271,10 +296,24 @@ class Compiler implements PikeVMOpcodes {
       case '(': {
         boolean capturing = true;
         if (index + 1 < array.length && array[index + 1] == '?') {
-          if (index + 2 < array.length && array[index + 2] == ':') {
+          index += 2;
-            index += 2;
+          if (index >= array.length) {
+            throw new RuntimeException("Short pattern @" + index + ": "
+              + regex);
+          }
+          c = array[index];
+          switch (c) {
+          case ':':
             capturing = false;
-          } else {
+            break;
+          case '=': {
+            capturing = false;
+            Lookahead lookahead = new Lookahead();
+            current.push(lookahead);
+            groups.push(lookahead.group);
+            continue;
+          }
+          default:
             throw new UnsupportedOperationException("Not yet supported: "
               + regex.substring(index));
           }

test/regex/PikeVM.java

@@ -27,19 +27,21 @@ class PikeVM implements PikeVMOpcodes {
    */
   private final int findPrefixLength;
   private final CharacterMatcher[] classes;
+  private final PikeVM[] lookaheads;
 
   public interface Result {
     void set(int[] start, int[] end);
   }
 
   protected PikeVM(int[] program, int findPrefixLength, int groupCount,
-    CharacterMatcher[] classes)
+    CharacterMatcher[] classes, PikeVM[] lookaheads)
   {
     this.program = program;
     this.findPrefixLength = findPrefixLength;
     this.groupCount = groupCount;
     offsetsCount = 2 * groupCount + 2;
     this.classes = classes;
+    this.lookaheads = lookaheads;
   }
 
   /**
@@ -313,7 +315,12 @@ class PikeVM implements PikeVMOpcodes {
           if (anchorEnd && i < end) {
             continue;
           }
+          if (result == null) {
+            // only interested in a match, no need to go on
+            return true;
+          }
           current.setResult(result);
+
           // now that we found a match, even higher-priority matches must match
           // at the same start offset
           if (!anchorStart) {
@@ -338,10 +345,18 @@ class PikeVM implements PikeVMOpcodes {
             current.queueNext(pc, pc + 2, next);
           }
           break;
+        case LOOKAHEAD:
+          if (lookaheads[program[pc + 1]].matches(characters,
+              i, characters.length, true, false, null)) {
+            current.queueImmediately(pc, pc + 2, false);
+          }
+          break;
         /* immediate opcodes, i.e. thread continues within the same step */
         case SAVE_OFFSET:
-          int index = program[pc + 1];
+          if (result != null) {
-          current.saveOffset(pc, index, i);
+            int index = program[pc + 1];
+            current.saveOffset(pc, index, i);
+          }
           current.queueImmediately(pc, pc + 2, false);
           break;
         case SPLIT:

test/regex/PikeVMOpcodes.java

@@ -24,6 +24,8 @@ interface PikeVMOpcodes {
 
   final static int CHARACTER_CLASS = -20;
 
+  final static int LOOKAHEAD = -30;
+
   final static int SAVE_OFFSET = -40;
 
   final static int SPLIT = -50;