Class Compiler

java/com/google/re2j/Compiler.java:17–284

Compiler from Regexp (RE2 abstract syntax) to RE2 (compiled regular expression). The only entry point is #compileRegexp.

 * The only entry point is {@link #compileRegexp}.
 */
class Compiler {

  /**
   * A fragment of a compiled regular expression program.
   *
   * @see http://swtch.com/~rsc/regexp/regexp1.html
   */
  private static class Frag {
    final int i; // an instruction address (pc).
    int out; // a patch list; see explanation in Prog.java
    boolean nullable; // whether the fragment can match the empty string

    Frag() {
      this(0, 0);
    }

    Frag(int i) {
      this(i, 0);
    }

    Frag(int i, int out) {
      this(i, out, false);
    }

    Frag(int i, int out, boolean nullable) {
      this.i = i;
      this.out = out;
      this.nullable = nullable;
    }
  }

  private final Prog prog = new Prog(); // Program being built

  private Compiler() {
    newInst(Inst.FAIL); // always the first instruction
  }

  static Prog compileRegexp(Regexp re) {
    Compiler c = new Compiler();
    Frag f = c.compile(re);
    c.prog.patch(f.out, c.newInst(Inst.MATCH).i);
    c.prog.start = f.i;
    return c.prog;
  }

  private Frag newInst(int op) {
    // TODO(rsc): impose length limit.
    prog.addInst(op);
    return new Frag(prog.numInst() - 1, 0, true);
  }

  // Returns a no-op fragment.  Sometimes unavoidable.
  private Frag nop() {
    Frag f = newInst(Inst.NOP);
    f.out = f.i << 1;
    return f;
  }