Regex: add support for character classes

Now we support regular expression patterns a la '[0-9]'.

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

test/Regex.java

@@ -54,5 +54,7 @@ public class Regex {
     expectGroups("a(bb)*a", "abbbba", "bb");
     expectGroups("a(bb)?(bb)+a", "abba", null, "bb");
     expectFind(" +", "Hello  ,   world! ", "  ", "   ", " ");
+    expectMatch("[0-9A-Fa-f]+", "08ef");
+    expectNoMatch("[0-9A-Fa-f]+", "08@ef");
   }
 }

test/regex/Compiler.java

@@ -27,6 +27,7 @@ class Compiler implements PikeVMOpcodes {
     private int offset;
     private int groupCount = -1;
     private int findPreambleSize;
+    private ArrayList<CharacterMatcher> classes;
 
     public Output(Expression expr) {
       // try-run to determine the code size
@@ -34,6 +35,7 @@ class Compiler implements PikeVMOpcodes {
       program = new int[offset];
       offset = 0;
       groupCount = -1;
+      classes = new ArrayList<CharacterMatcher>();
       // write it out!
       expr.writeCode(this);
     }
@@ -60,15 +62,38 @@ class Compiler implements PikeVMOpcodes {
     }
 
     public PikeVM toVM() {
-      return new PikeVM(program, findPreambleSize, groupCount);
+      CharacterMatcher[] classes = new CharacterMatcher[this.classes.size()];
+      this.classes.toArray(classes);
+      return new PikeVM(program, findPreambleSize, groupCount, classes);
     }
 
+    public int addClass(CharacterMatcher characterClass) {
+      if (program == null) {
+        return -1;
+      }
+      int result = classes.size();
+      classes.add(characterClass);
+      return result;
+    }
   }
 
   private abstract class Expression {
     protected abstract void writeCode(Output output);
   }
 
+  private class CharacterRange extends Expression {
+    private final CharacterMatcher characterClass;
+
+    public CharacterRange(CharacterMatcher characterClass) {
+      this.characterClass = characterClass;
+    }
+
+    protected void writeCode(Output output) {
+      output.add(CHARACTER_CLASS);
+      output.add(output.addClass(characterClass));
+    }
+  }
+
   private class Repeat extends Expression {
     private Expression expr;
     private int minCount, maxCount;
@@ -177,6 +202,8 @@ class Compiler implements PikeVMOpcodes {
 
   public Pattern compile(String regex) {
     char[] array = regex.toCharArray();
+    CharacterMatcher.Parser characterClassParser =
+      new CharacterMatcher.Parser(array);
     for (int index = 0; index < array.length; ++ index) {
       char c = array[index];
       Group current = groups.peek();
@@ -214,6 +241,15 @@ class Compiler implements PikeVMOpcodes {
         }
         groups.pop();
         continue;
+      case '[': {
+        CharacterMatcher matcher = characterClassParser.parseClass(index);
+        if (matcher == null) {
+          throw new RuntimeException("Invalid range @" + index + ": " + regex);
+        }
+        current.push(new CharacterRange(matcher));
+        index = characterClassParser.getEndOffset() - 1;
+        continue;
+      }
       default:
         throw new RuntimeException("Parse error @" + index + ": " + regex);
       }

test/regex/PikeVM.java

@@ -26,16 +26,20 @@ class PikeVM implements PikeVMOpcodes {
    * findPrefixLength} instead.
    */
   private final int findPrefixLength;
+  private final CharacterMatcher[] classes;
 
   public interface Result {
     void set(int[] start, int[] end);
   }
 
-  protected PikeVM(int[] program, int findPrefixLength, int groupCount) {
+  protected PikeVM(int[] program, int findPrefixLength, int groupCount,
+    CharacterMatcher[] classes)
+  {
     this.program = program;
     this.findPrefixLength = findPrefixLength;
     this.groupCount = groupCount;
     offsetsCount = 2 * groupCount + 2;
+    this.classes = classes;
   }
 
   /**
@@ -329,6 +333,11 @@ class PikeVM implements PikeVMOpcodes {
         case DOTALL:
           current.queueNext(pc, pc + 1, next);
           break;
+        case CHARACTER_CLASS:
+          if (classes[program[pc + 1]].matches(c)) {
+            current.queueNext(pc, pc + 2, next);
+          }
+          break;
         /* immediate opcodes, i.e. thread continues within the same step */
         case SAVE_OFFSET:
           int index = program[pc + 1];

test/regex/PikeVMOpcodes.java

@@ -22,6 +22,8 @@ interface PikeVMOpcodes {
   final static int DOT = -1;
   final static int DOTALL = -2;
 
+  final static int CHARACTER_CLASS = -20;
+
   final static int SAVE_OFFSET = -40;
 
   final static int SPLIT = -50;