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),
58      valgrind_(RunningOnValgrindOrMemorySanitizer()) {}
59
60  SparseSet(int max_size) {
61    max_size_ = max_size;
62    sparse_to_dense_ = new int[max_size];
63    dense_ = new int[max_size];
64    valgrind_ = RunningOnValgrindOrMemorySanitizer();
65    // Don't need to zero the memory, but do so anyway
66    // to appease Valgrind.
67    if (valgrind_) {
68      for (int i = 0; i < max_size; i++) {
69        dense_[i] = 0xababababU;
70        sparse_to_dense_[i] = 0xababababU;
71      }
72    }
73    size_ = 0;
74  }
75
76  ~SparseSet() {
77    delete[] sparse_to_dense_;
78    delete[] dense_;
79  }
80
81  typedef int* iterator;
82  typedef const int* const_iterator;
83
84  int size() const { return size_; }
85  iterator begin() { return dense_; }
86  iterator end() { return dense_ + size_; }
87  const_iterator begin() const { return dense_; }
88  const_iterator end() const { return dense_ + size_; }
89
90  // Change the maximum size of the array.
91  // Invalidates all iterators.
92  void resize(int new_max_size) {
93    if (size_ > new_max_size)
94      size_ = new_max_size;
95    if (new_max_size > max_size_) {
96      int* a = new int[new_max_size];
97      if (sparse_to_dense_) {
98        memmove(a, sparse_to_dense_, max_size_*sizeof a[0]);
99        if (valgrind_) {
100          for (int i = max_size_; i < new_max_size; i++)
101            a[i] = 0xababababU;
102        }
103        delete[] sparse_to_dense_;
104      }
105      sparse_to_dense_ = a;
106
107      a = new int[new_max_size];
108      if (dense_) {
109        memmove(a, dense_, size_*sizeof a[0]);
110        if (valgrind_) {
111          for (int i = size_; i < new_max_size; i++)
112            a[i] = 0xababababU;
113        }
114        delete[] dense_;
115      }
116      dense_ = a;
117    }
118    max_size_ = new_max_size;
119  }
120
121  // Return the maximum size of the array.
122  // Indices can be in the range [0, max_size).
123  int max_size() const { return max_size_; }
124
125  // Clear the array.
126  void clear() { size_ = 0; }
127
128  // Check whether i is in the array.
129  bool contains(int i) const {
130    DCHECK_GE(i, 0);
131    DCHECK_LT(i, max_size_);
132    if (static_cast<uint>(i) >= max_size_) {
133      return false;
134    }
135    // Unsigned comparison avoids checking sparse_to_dense_[i] < 0.
136    return (uint)sparse_to_dense_[i] < (uint)size_ &&
137      dense_[sparse_to_dense_[i]] == i;
138  }
139
140  // Adds i to the set.
141  void insert(int i) {
142    if (!contains(i))
143      insert_new(i);
144  }
145
146  // Set the value at the new index i to v.
147  // Fast but unsafe: only use if contains(i) is false.
148  void insert_new(int i) {
149    if (static_cast<uint>(i) >= max_size_) {
150      // Semantically, end() would be better here, but we already know
151      // the user did something stupid, so begin() insulates them from
152      // dereferencing an invalid pointer.
153      return;
154    }
155    DCHECK(!contains(i));
156    DCHECK_LT(size_, max_size_);
157    sparse_to_dense_[i] = size_;
158    dense_[size_] = i;
159    size_++;
160  }
161
162  // Comparison function for sorting.
163  // Can sort the sparse array so that future iterations
164  // will visit indices in increasing order using
165  // sort(arr.begin(), arr.end(), arr.less);
166  static bool less(int a, int b) { return a < b; }
167
168 private:
169  int size_;
170  int max_size_;
171  int* sparse_to_dense_;
172  int* dense_;
173  bool valgrind_;
174
175  DISALLOW_EVIL_CONSTRUCTORS(SparseSet);
176};
177
178}  // namespace re2
179
180#endif  // RE2_UTIL_SPARSE_SET_H__
181