Class MachineInput

java/com/google/re2j/MachineInput.java:16–267

MachineInput abstracts different representations of the input text supplied to the Machine. It provides one-character lookahead.

 * provides one-character lookahead.
 */
abstract class MachineInput {

  static final int EOF = (-1 << 3);

  static MachineInput fromUTF8(byte[] b) {
    return new UTF8Input(b);
  }

  static MachineInput fromUTF8(byte[] b, int start, int end) {
    return new UTF8Input(b, start, end);
  }

  static MachineInput fromUTF16(CharSequence s) {
    return new UTF16Input(s, 0, s.length());
  }

  static MachineInput fromUTF16(CharSequence s, int start, int end) {
    return new UTF16Input(s, start, end);
  }

  //// Interface

  // Returns the rune at the specified index; the units are
  // unspecified, but could be UTF-8 byte, UTF-16 char, or rune
  // indices.  Returns the width (in the same units) of the rune in
  // the lower 3 bits, and the rune (Unicode code point) in the high
  // bits.  Never negative, except for EOF which is represented as -1
  // << 3 | 0.
  abstract int step(int pos);

  // can we look ahead without losing info?
  abstract boolean canCheckPrefix();

  // Returns the index relative to |pos| at which |re2.prefix| is found
  // in this input stream, or a negative value if not found.
  abstract int index(RE2 re2, int pos);

  // Returns a bitmask of EMPTY_* flags.
  abstract int context(int pos);

  // Returns the end position in the same units as step().
  abstract int endPos();

  //// Implementations

  // An implementation of MachineInput for UTF-8 byte arrays.
  // |pos| and |width| are byte indices.
  private static class UTF8Input extends MachineInput {

    final byte[] b;
    final int start;
    final int end;

    UTF8Input(byte[] b) {
      this.b = b;
      start = 0;
      end = b.length;
    }