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#ifndef UTIL_SPARSE_SET_H_
6#define UTIL_SPARSE_SET_H_
7
8// DESCRIPTION
9//
10// SparseSet(m) is a set of integers in [0, m).
11// It requires sizeof(int)*m memory, but it provides
12// fast iteration through the elements in the set and fast clearing
13// of the set.
14//
15// Insertion and deletion are constant time operations.
16//
17// Allocating the set is a constant time operation
18// when memory allocation is a constant time operation.
19//
20// Clearing the set is a constant time operation (unusual!).
21//
22// Iterating through the set is an O(n) operation, where n
23// is the number of items in the set (not O(m)).
24//
25// The set iterator visits entries in the order they were first
26// inserted into the set. It is safe to add items to the set while
27// using an iterator: the iterator will visit indices added to the set
28// during the iteration, but will not re-visit indices whose values
29// change after visiting. Thus SparseSet can be a convenient
30// implementation of a work queue.
31//
32// The SparseSet implementation is NOT thread-safe. It is up to the
33// caller to make sure only one thread is accessing the set. (Typically
34// these sets are temporary values and used in situations where speed is
35// important.)
36//
37// The SparseSet interface does not present all the usual STL bells and
38// whistles.
39//
40// Implemented with reference to Briggs & Torczon, An Efficient
41// Representation for Sparse Sets, ACM Letters on Programming Languages
42// and Systems, Volume 2, Issue 1-4 (March-Dec. 1993), pp. 59-69.
43//
44// This is a specialization of sparse array; see sparse_array.h.
45
46// IMPLEMENTATION
47//
48// See sparse_array.h for implementation details.
49
50// Doing this simplifies the logic below.
51#ifndef __has_feature
52#define __has_feature(x) 0
53#endif
54
55#include <assert.h>
56#include <stdint.h>
57#if __has_feature(memory_sanitizer)
58#include <sanitizer/msan_interface.h>
59#endif
60#include <algorithm>
61#include <memory>
62#include <utility>
63
64#include "util/pod_array.h"
65
66namespace duckdb_re2 {
67
68template<typename Value>
69class SparseSetT {
70 public:
71 SparseSetT();
72 explicit SparseSetT(int max_size);
73 ~SparseSetT();
74
75 typedef int* iterator;
76 typedef const int* const_iterator;
77
78 // Return the number of entries in the set.
79 int size() const {
80 return size_;
81 }
82
83 // Indicate whether the set is empty.
84 int empty() const {
85 return size_ == 0;
86 }
87
88 // Iterate over the set.
89 iterator begin() {
90 return dense_.data();
91 }
92 iterator end() {
93 return dense_.data() + size_;
94 }
95
96 const_iterator begin() const {
97 return dense_.data();
98 }
99 const_iterator end() const {
100 return dense_.data() + size_;
101 }
102
103 // Change the maximum size of the set.
104 // Invalidates all iterators.
105 void resize(int new_max_size);
106
107 // Return the maximum size of the set.
108 // Indices can be in the range [0, max_size).
109 int max_size() const {
110 if (dense_.data() != NULL)
111 return dense_.size();
112 else
113 return 0;
114 }
115
116 // Clear the set.
117 void clear() {
118 size_ = 0;
119 }
120
121 // Check whether index i is in the set.
122 bool contains(int i) const;
123
124 // Comparison function for sorting.
125 // Can sort the sparse set so that future iterations
126 // will visit indices in increasing order using
127 // std::sort(arr.begin(), arr.end(), arr.less);
128 static bool less(int a, int b);
129
130 public:
131 // Insert index i into the set.
132 iterator insert(int i) {
133 return InsertInternal(allow_existing: true, i);
134 }
135
136 // Insert index i into the set.
137 // Fast but unsafe: only use if contains(i) is false.
138 iterator insert_new(int i) {
139 return InsertInternal(allow_existing: false, i);
140 }
141
142 private:
143 iterator InsertInternal(bool allow_existing, int i) {
144 DebugCheckInvariants();
145 if (static_cast<uint32_t>(i) >= static_cast<uint32_t>(max_size())) {
146 assert(false && "illegal index");
147 // Semantically, end() would be better here, but we already know
148 // the user did something stupid, so begin() insulates them from
149 // dereferencing an invalid pointer.
150 return begin();
151 }
152 if (!allow_existing) {
153 assert(!contains(i));
154 create_index(i);
155 } else {
156 if (!contains(i))
157 create_index(i);
158 }
159 DebugCheckInvariants();
160 return dense_.data() + sparse_[i];
161 }
162
163 // Add the index i to the set.
164 // Only use if contains(i) is known to be false.
165 // This function is private, only intended as a helper
166 // for other methods.
167 void create_index(int i);
168
169 // In debug mode, verify that some invariant properties of the class
170 // are being maintained. This is called at the end of the constructor
171 // and at the beginning and end of all public non-const member functions.
172 void DebugCheckInvariants() const;
173
174 // Initializes memory for elements [min, max).
175 void MaybeInitializeMemory(int min, int max) {
176#if __has_feature(memory_sanitizer)
177 __msan_unpoison(sparse_.data() + min, (max - min) * sizeof sparse_[0]);
178#elif defined(RE2_ON_VALGRIND)
179 for (int i = min; i < max; i++) {
180 sparse_[i] = 0xababababU;
181 }
182#endif
183 }
184
185 int size_ = 0;
186 PODArray<int> sparse_;
187 PODArray<int> dense_;
188};
189
190template<typename Value>
191SparseSetT<Value>::SparseSetT() = default;
192
193// Change the maximum size of the set.
194// Invalidates all iterators.
195template<typename Value>
196void SparseSetT<Value>::resize(int new_max_size) {
197 DebugCheckInvariants();
198 if (new_max_size > max_size()) {
199 const int old_max_size = max_size();
200
201 // Construct these first for exception safety.
202 PODArray<int> a(new_max_size);
203 PODArray<int> b(new_max_size);
204
205 std::copy_n(first: sparse_.data(), n: old_max_size, result: a.data());
206 std::copy_n(first: dense_.data(), n: old_max_size, result: b.data());
207
208 sparse_ = std::move(a);
209 dense_ = std::move(b);
210
211 MaybeInitializeMemory(min: old_max_size, max: new_max_size);
212 }
213 if (size_ > new_max_size)
214 size_ = new_max_size;
215 DebugCheckInvariants();
216}
217
218// Check whether index i is in the set.
219template<typename Value>
220bool SparseSetT<Value>::contains(int i) const {
221 assert(i >= 0);
222 assert(i < max_size());
223 if (static_cast<uint32_t>(i) >= static_cast<uint32_t>(max_size())) {
224 return false;
225 }
226 // Unsigned comparison avoids checking sparse_[i] < 0.
227 return (uint32_t)sparse_[i] < (uint32_t)size_ &&
228 dense_[sparse_[i]] == i;
229}
230
231template<typename Value>
232void SparseSetT<Value>::create_index(int i) {
233 assert(!contains(i));
234 assert(size_ < max_size());
235 sparse_[i] = size_;
236 dense_[size_] = i;
237 size_++;
238}
239
240template<typename Value> SparseSetT<Value>::SparseSetT(int max_size) :
241 sparse_(max_size), dense_(max_size) {
242 MaybeInitializeMemory(min: size_, max: max_size);
243 DebugCheckInvariants();
244}
245
246template<typename Value> SparseSetT<Value>::~SparseSetT() {
247 DebugCheckInvariants();
248}
249
250template<typename Value> void SparseSetT<Value>::DebugCheckInvariants() const {
251 assert(0 <= size_);
252 assert(size_ <= max_size());
253}
254
255// Comparison function for sorting.
256template<typename Value> bool SparseSetT<Value>::less(int a, int b) {
257 return a < b;
258}
259
260typedef SparseSetT<void> SparseSet;
261
262} // namespace duckdb_re2
263
264#endif // UTIL_SPARSE_SET_H_
265