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 52 namespace re2 { 53 54 class SparseSet { 55 public: SparseSet()56 SparseSet() 57 : size_(0), max_size_(0), sparse_to_dense_(NULL), dense_(NULL) {} 58 SparseSet(int max_size)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 // Don't need to zero the memory, but do so anyway 64 // to appease Valgrind. 65 if (RunningOnValgrind()) { 66 for (int i = 0; i < max_size; i++) { 67 dense_[i] = 0xababababU; 68 sparse_to_dense_[i] = 0xababababU; 69 } 70 } 71 size_ = 0; 72 } 73 ~SparseSet()74 ~SparseSet() { 75 delete[] sparse_to_dense_; 76 delete[] dense_; 77 } 78 79 typedef int* iterator; 80 typedef const int* const_iterator; 81 size()82 int size() const { return size_; } begin()83 iterator begin() { return dense_; } end()84 iterator end() { return dense_ + size_; } begin()85 const_iterator begin() const { return dense_; } end()86 const_iterator end() const { return dense_ + size_; } 87 88 // Change the maximum size of the array. 89 // Invalidates all iterators. resize(int new_max_size)90 void resize(int new_max_size) { 91 if (size_ > new_max_size) 92 size_ = new_max_size; 93 if (new_max_size > max_size_) { 94 int* a = new int[new_max_size]; 95 if (sparse_to_dense_) { 96 memmove(a, sparse_to_dense_, max_size_*sizeof a[0]); 97 if (RunningOnValgrind()) { 98 for (int i = max_size_; i < new_max_size; i++) 99 a[i] = 0xababababU; 100 } 101 delete[] sparse_to_dense_; 102 } 103 sparse_to_dense_ = a; 104 105 a = new int[new_max_size]; 106 if (dense_) { 107 memmove(a, dense_, size_*sizeof a[0]); 108 if (RunningOnValgrind()) { 109 for (int i = size_; i < new_max_size; i++) 110 a[i] = 0xababababU; 111 } 112 delete[] dense_; 113 } 114 dense_ = a; 115 } 116 max_size_ = new_max_size; 117 } 118 119 // Return the maximum size of the array. 120 // Indices can be in the range [0, max_size). max_size()121 int max_size() const { return max_size_; } 122 123 // Clear the array. clear()124 void clear() { size_ = 0; } 125 126 // Check whether i is in the array. contains(int i)127 bool contains(int i) const { 128 DCHECK_GE(i, 0); 129 DCHECK_LT(i, max_size_); 130 if (static_cast<uint>(i) >= max_size_) { 131 return false; 132 } 133 // Unsigned comparison avoids checking sparse_to_dense_[i] < 0. 134 return (uint)sparse_to_dense_[i] < (uint)size_ && 135 dense_[sparse_to_dense_[i]] == i; 136 } 137 138 // Adds i to the set. insert(int i)139 void insert(int i) { 140 if (!contains(i)) 141 insert_new(i); 142 } 143 144 // Set the value at the new index i to v. 145 // Fast but unsafe: only use if contains(i) is false. insert_new(int i)146 void insert_new(int i) { 147 if (static_cast<uint>(i) >= max_size_) { 148 // Semantically, end() would be better here, but we already know 149 // the user did something stupid, so begin() insulates them from 150 // dereferencing an invalid pointer. 151 return; 152 } 153 DCHECK(!contains(i)); 154 DCHECK_LT(size_, max_size_); 155 sparse_to_dense_[i] = size_; 156 dense_[size_] = i; 157 size_++; 158 } 159 160 // Comparison function for sorting. 161 // Can sort the sparse array so that future iterations 162 // will visit indices in increasing order using 163 // sort(arr.begin(), arr.end(), arr.less); less(int a,int b)164 static bool less(int a, int b) { return a < b; } 165 166 private: 167 int size_; 168 int max_size_; 169 int* sparse_to_dense_; 170 int* dense_; 171 172 DISALLOW_EVIL_CONSTRUCTORS(SparseSet); 173 }; 174 175 } // namespace re2 176 177 #endif // RE2_UTIL_SPARSE_SET_H__ 178