Class Regexp

java/com/google/re2j/Regexp.java:21–379

Regular expression abstract syntax tree. Produced by parser, used by compiler. NB, this corresponds to syntax.regexp in the Go implementation; Go's regexp is called RE2 in Java.

 * {@code RE2} in Java.
 */
class Regexp {

  enum Op {
    NO_MATCH, // Matches no strings.
    EMPTY_MATCH, // Matches empty string.
    LITERAL, // Matches runes[] sequence
    CHAR_CLASS, // Matches Runes interpreted as range pair list
    ANY_CHAR_NOT_NL, // Matches any character except '\n'
    ANY_CHAR, // Matches any character
    BEGIN_LINE, // Matches empty string at end of line
    END_LINE, // Matches empty string at end of line
    BEGIN_TEXT, // Matches empty string at beginning of text
    END_TEXT, // Matches empty string at end of text
    WORD_BOUNDARY, // Matches word boundary `\b`
    NO_WORD_BOUNDARY, // Matches word non-boundary `\B`
    CAPTURE, // Capturing subexpr with index cap, optional name name
    STAR, // Matches subs[0] zero or more times.
    PLUS, // Matches subs[0] one or more times.
    QUEST, // Matches subs[0] zero or one times.
    REPEAT, // Matches subs[0] [min, max] times; max=-1 => no limit.
    CONCAT, // Matches concatenation of subs[]
    ALTERNATE, // Matches union of subs[]

    // Pseudo ops, used internally by Parser for parsing stack:
    LEFT_PAREN,
    VERTICAL_BAR;

    boolean isPseudo() {
      return ordinal() >= LEFT_PAREN.ordinal();
    }
  }

  static final Regexp[] EMPTY_SUBS = {};

  Op op; // operator
  int flags; // bitmap of parse flags
  Regexp[] subs; // subexpressions, if any.  Never null.
  // subs[0] is used as the freelist.
  int[] runes; // matched runes, for LITERAL, CHAR_CLASS
  int min, max; // min, max for REPEAT
  int cap; // capturing index, for CAPTURE
  String name; // capturing name, for CAPTURE
  Map<String, Integer> namedGroups = Collections.emptyMap(); // map of group name -> capturing index
  // Do update copy ctor when adding new fields!

  Regexp(Op op) {
    this.op = op;
  }

  // Shallow copy constructor.
  Regexp(Regexp that) {
    this.op = that.op;
    this.flags = that.flags;
    this.subs = that.subs;
    this.runes = that.runes;
    this.min = that.min;
    this.max = that.max;
    this.cap = that.cap;