Class CharClass

java/com/google/re2j/CharClass.java:18–361

A "builder"-style helper class for manipulating character classes represented as an array of pairs of runes [lo, hi], each denoting an inclusive interval. All methods mutate the internal state and return this, allowing operations to be chained.

 * All methods mutate the internal state and return {@code this}, allowing operations to be chained.
 */
class CharClass {

  private int[] r; // inclusive ranges, pairs of [lo,hi].  r.length is even.
  private int len; // prefix of |r| that is defined.  Even.

  // Constructs a CharClass with initial ranges |r|.
  // The right to mutate |r| is passed to the callee.
  CharClass(int[] r) {
    this.r = r;
    this.len = r.length;
  }

  // Constructs an empty CharClass.
  CharClass() {
    this.r = Utils.EMPTY_INTS;
    this.len = 0;
  }

  // After a call to ensureCapacity(), |r.length| is at least |newLen|.
  private void ensureCapacity(int newLen) {
    if (r.length < newLen) {
      // Expand by at least doubling, except when len == 0.
      // TODO(adonovan): opt: perhaps it would be better to allocate exactly
      // newLen, since the number of expansions is typically very small?
      if (newLen < len * 2) {
        newLen = len * 2;
      }
      int[] r2 = new int[newLen];
      System.arraycopy(r, 0, r2, 0, len);
      r = r2;
    }
  }

  // Returns the character class as an int array.  Subsequent CharClass
  // operations may mutate this array, so typically this is the last operation
  // performed on a given CharClass instance.
  int[] toArray() {
    if (this.len == r.length) {
      return r;
    } else {
      int[] r2 = new int[len];
      System.arraycopy(r, 0, r2, 0, len);
      return r2;
    }
  }

  // cleanClass() sorts the ranges (pairs of elements) of this CharClass,
  // merges them, and eliminates duplicates.
  CharClass cleanClass() {
    if (len < 4) {
      return this;
    }

    // Sort by lo increasing, hi decreasing to break ties.
    qsortIntPair(r, 0, len - 2);

    // Merge abutting, overlapping.
    int w = 2; // write index