1// Copyright 2006 The RE2 Authors.  All Rights Reserved.
2// Use of this source code is governed by a BSD-style
3// license that can be found in the LICENSE file.
4
5// DESCRIPTION
6//
7// SparseSet<T>(m) is a set of integers in [0, m).
8// It requires sizeof(int)*m memory, but it provides
9// fast iteration through the elements in the set and fast clearing
10// of the set.
11//
12// Insertion and deletion are constant time operations.
13//
14// Allocating the set is a constant time operation
15// when memory allocation is a constant time operation.
16//
17// Clearing the set is a constant time operation (unusual!).
18//
19// Iterating through the set is an O(n) operation, where n
20// is the number of items in the set (not O(m)).
21//
22// The set iterator visits entries in the order they were first
23// inserted into the array.  It is safe to add items to the set while
24// using an iterator: the iterator will visit indices added to the set
25// during the iteration, but will not re-visit indices whose values
26// change after visiting.  Thus SparseSet can be a convenient
27// implementation of a work queue.
28//
29// The SparseSet implementation is NOT thread-safe.  It is up to the
30// caller to make sure only one thread is accessing the set.  (Typically
31// these sets are temporary values and used in situations where speed is
32// important.)
33//
34// The SparseSet interface does not present all the usual STL bells and
35// whistles.
36//
37// Implemented with reference to Briggs & Torczon, An Efficient
38// Representation for Sparse Sets, ACM Letters on Programming Languages
39// and Systems, Volume 2, Issue 1-4 (March-Dec.  1993), pp.  59-69.
40//
41// For a generalization to sparse array, see sparse_array.h.
42
43// IMPLEMENTATION
44//
45// See sparse_array.h for implementation details
46
47#ifndef RE2_UTIL_SPARSE_SET_H__
48#define RE2_UTIL_SPARSE_SET_H__
49
50#include "util/util.h"
51
52namespace re2 {
53
54class SparseSet {
55 public:
56  SparseSet()
57    : size_(0), max_size_(0), sparse_to_dense_(NULL), dense_(NULL), valgrind_(RunningOnValgrind()) {}
58
59  SparseSet(int max_size) {
60    max_size_ = max_size;
61    sparse_to_dense_ = new int[max_size];
62    dense_ = new int[max_size];
63    valgrind_ = RunningOnValgrind();
64    // Don't need to zero the memory, but do so anyway
65    // to appease Valgrind.
66    if (valgrind_) {
67      for (int i = 0; i < max_size; i++) {
68        dense_[i] = 0xababababU;
69        sparse_to_dense_[i] = 0xababababU;
70      }
71    }
72    size_ = 0;
73  }
74
75  ~SparseSet() {
76    delete[] sparse_to_dense_;
77    delete[] dense_;
78  }
79
80  typedef int* iterator;
81  typedef const int* const_iterator;
82
83  int size() const { return size_; }
84  iterator begin() { return dense_; }
85  iterator end() { return dense_ + size_; }
86  const_iterator begin() const { return dense_; }
87  const_iterator end() const { return dense_ + size_; }
88
89  // Change the maximum size of the array.
90  // Invalidates all iterators.
91  void resize(int new_max_size) {
92    if (size_ > new_max_size)
93      size_ = new_max_size;
94    if (new_max_size > max_size_) {
95      int* a = new int[new_max_size];
96      if (sparse_to_dense_) {
97        memmove(a, sparse_to_dense_, max_size_*sizeof a[0]);
98        if (valgrind_) {
99          for (int i = max_size_; i < new_max_size; i++)
100            a[i] = 0xababababU;
101        }
102        delete[] sparse_to_dense_;
103      }
104      sparse_to_dense_ = a;
105
106      a = new int[new_max_size];
107      if (dense_) {
108        memmove(a, dense_, size_*sizeof a[0]);
109        if (valgrind_) {
110          for (int i = size_; i < new_max_size; i++)
111            a[i] = 0xababababU;
112        }
113        delete[] dense_;
114      }
115      dense_ = a;
116    }
117    max_size_ = new_max_size;
118  }
119
120  // Return the maximum size of the array.
121  // Indices can be in the range [0, max_size).
122  int max_size() const { return max_size_; }
123
124  // Clear the array.
125  void clear() { size_ = 0; }
126
127  // Check whether i is in the array.
128  bool contains(int i) const {
129    DCHECK_GE(i, 0);
130    DCHECK_LT(i, max_size_);
131    if (static_cast<uint>(i) >= max_size_) {
132      return false;
133    }
134    // Unsigned comparison avoids checking sparse_to_dense_[i] < 0.
135    return (uint)sparse_to_dense_[i] < (uint)size_ &&
136      dense_[sparse_to_dense_[i]] == i;
137  }
138
139  // Adds i to the set.
140  void insert(int i) {
141    if (!contains(i))
142      insert_new(i);
143  }
144
145  // Set the value at the new index i to v.
146  // Fast but unsafe: only use if contains(i) is false.
147  void insert_new(int i) {
148    if (static_cast<uint>(i) >= max_size_) {
149      // Semantically, end() would be better here, but we already know
150      // the user did something stupid, so begin() insulates them from
151      // dereferencing an invalid pointer.
152      return;
153    }
154    DCHECK(!contains(i));
155    DCHECK_LT(size_, max_size_);
156    sparse_to_dense_[i] = size_;
157    dense_[size_] = i;
158    size_++;
159  }
160
161  // Comparison function for sorting.
162  // Can sort the sparse array so that future iterations
163  // will visit indices in increasing order using
164  // sort(arr.begin(), arr.end(), arr.less);
165  static bool less(int a, int b) { return a < b; }
166
167 private:
168  int size_;
169  int max_size_;
170  int* sparse_to_dense_;
171  int* dense_;
172  bool valgrind_;
173
174  DISALLOW_EVIL_CONSTRUCTORS(SparseSet);
175};
176
177}  // namespace re2
178
179#endif  // RE2_UTIL_SPARSE_SET_H__
180