1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
|
/* SPDX-License-Identifier: MIT */
/* Copyright © 2022-present Max Bachmann */
#pragma once
#include <rapidfuzz/distance/LCSseq_impl.hpp>
#include <algorithm>
#include <limits>
namespace rapidfuzz {
template <typename InputIt1, typename InputIt2>
size_t lcs_seq_distance(InputIt1 first1, InputIt1 last1, InputIt2 first2, InputIt2 last2,
size_t score_cutoff = std::numeric_limits<size_t>::max())
{
return detail::LCSseq::distance(first1, last1, first2, last2, score_cutoff, score_cutoff);
}
template <typename Sentence1, typename Sentence2>
size_t lcs_seq_distance(const Sentence1& s1, const Sentence2& s2,
size_t score_cutoff = std::numeric_limits<size_t>::max())
{
return detail::LCSseq::distance(s1, s2, score_cutoff, score_cutoff);
}
template <typename InputIt1, typename InputIt2>
size_t lcs_seq_similarity(InputIt1 first1, InputIt1 last1, InputIt2 first2, InputIt2 last2,
size_t score_cutoff = 0)
{
return detail::LCSseq::similarity(first1, last1, first2, last2, score_cutoff, score_cutoff);
}
template <typename Sentence1, typename Sentence2>
size_t lcs_seq_similarity(const Sentence1& s1, const Sentence2& s2, size_t score_cutoff = 0)
{
return detail::LCSseq::similarity(s1, s2, score_cutoff, score_cutoff);
}
template <typename InputIt1, typename InputIt2>
double lcs_seq_normalized_distance(InputIt1 first1, InputIt1 last1, InputIt2 first2, InputIt2 last2,
double score_cutoff = 1.0)
{
return detail::LCSseq::normalized_distance(first1, last1, first2, last2, score_cutoff, score_cutoff);
}
template <typename Sentence1, typename Sentence2>
double lcs_seq_normalized_distance(const Sentence1& s1, const Sentence2& s2, double score_cutoff = 1.0)
{
return detail::LCSseq::normalized_distance(s1, s2, score_cutoff, score_cutoff);
}
template <typename InputIt1, typename InputIt2>
double lcs_seq_normalized_similarity(InputIt1 first1, InputIt1 last1, InputIt2 first2, InputIt2 last2,
double score_cutoff = 0.0)
{
return detail::LCSseq::normalized_similarity(first1, last1, first2, last2, score_cutoff, score_cutoff);
}
template <typename Sentence1, typename Sentence2>
double lcs_seq_normalized_similarity(const Sentence1& s1, const Sentence2& s2, double score_cutoff = 0.0)
{
return detail::LCSseq::normalized_similarity(s1, s2, score_cutoff, score_cutoff);
}
template <typename InputIt1, typename InputIt2>
Editops lcs_seq_editops(InputIt1 first1, InputIt1 last1, InputIt2 first2, InputIt2 last2)
{
return detail::lcs_seq_editops(detail::make_range(first1, last1), detail::make_range(first2, last2));
}
template <typename Sentence1, typename Sentence2>
Editops lcs_seq_editops(const Sentence1& s1, const Sentence2& s2)
{
return detail::lcs_seq_editops(detail::make_range(s1), detail::make_range(s2));
}
#ifdef RAPIDFUZZ_SIMD
namespace experimental {
template <int MaxLen>
struct MultiLCSseq : public detail::MultiSimilarityBase<MultiLCSseq<MaxLen>, size_t, 0,
std::numeric_limits<int64_t>::max()> {
private:
friend detail::MultiSimilarityBase<MultiLCSseq<MaxLen>, size_t, 0, std::numeric_limits<int64_t>::max()>;
friend detail::MultiNormalizedMetricBase<MultiLCSseq<MaxLen>, size_t>;
RAPIDFUZZ_CONSTEXPR_CXX14 static size_t get_vec_size()
{
# ifdef RAPIDFUZZ_AVX2
using namespace detail::simd_avx2;
# else
using namespace detail::simd_sse2;
# endif
RAPIDFUZZ_IF_CONSTEXPR (MaxLen <= 8)
return native_simd<uint8_t>::size;
else RAPIDFUZZ_IF_CONSTEXPR (MaxLen <= 16)
return native_simd<uint16_t>::size;
else RAPIDFUZZ_IF_CONSTEXPR (MaxLen <= 32)
return native_simd<uint32_t>::size;
else RAPIDFUZZ_IF_CONSTEXPR (MaxLen <= 64)
return native_simd<uint64_t>::size;
static_assert(MaxLen <= 64, "expected MaxLen <= 64");
}
static size_t find_block_count(size_t count)
{
size_t vec_size = get_vec_size();
size_t simd_vec_count = detail::ceil_div(count, vec_size);
return detail::ceil_div(simd_vec_count * vec_size * MaxLen, 64);
}
public:
MultiLCSseq(size_t count) : input_count(count), pos(0), PM(find_block_count(count) * 64)
{
str_lens.resize(result_count());
}
/**
* @brief get minimum size required for result vectors passed into
* - distance
* - similarity
* - normalized_distance
* - normalized_similarity
*
* @return minimum vector size
*/
size_t result_count() const
{
size_t vec_size = get_vec_size();
size_t simd_vec_count = detail::ceil_div(input_count, vec_size);
return simd_vec_count * vec_size;
}
template <typename Sentence1>
void insert(const Sentence1& s1_)
{
insert(detail::to_begin(s1_), detail::to_end(s1_));
}
template <typename InputIt1>
void insert(InputIt1 first1, InputIt1 last1)
{
auto len = std::distance(first1, last1);
int block_pos = static_cast<int>((pos * MaxLen) % 64);
auto block = (pos * MaxLen) / 64;
assert(len <= MaxLen);
if (pos >= input_count) throw std::invalid_argument("out of bounds insert");
str_lens[pos] = static_cast<size_t>(len);
for (; first1 != last1; ++first1) {
PM.insert(block, *first1, block_pos);
block_pos++;
}
pos++;
}
private:
template <typename InputIt2>
void _similarity(size_t* scores, size_t score_count, const detail::Range<InputIt2>& s2,
size_t score_cutoff = 0) const
{
if (score_count < result_count())
throw std::invalid_argument("scores has to have >= result_count() elements");
auto scores_ = detail::make_range(scores, scores + score_count);
RAPIDFUZZ_IF_CONSTEXPR (MaxLen == 8)
detail::lcs_simd<uint8_t>(scores_, PM, s2, score_cutoff);
else RAPIDFUZZ_IF_CONSTEXPR (MaxLen == 16)
detail::lcs_simd<uint16_t>(scores_, PM, s2, score_cutoff);
else RAPIDFUZZ_IF_CONSTEXPR (MaxLen == 32)
detail::lcs_simd<uint32_t>(scores_, PM, s2, score_cutoff);
else RAPIDFUZZ_IF_CONSTEXPR (MaxLen == 64)
detail::lcs_simd<uint64_t>(scores_, PM, s2, score_cutoff);
}
template <typename InputIt2>
size_t maximum(size_t s1_idx, const detail::Range<InputIt2>& s2) const
{
return std::max(str_lens[s1_idx], s2.size());
}
size_t get_input_count() const noexcept
{
return input_count;
}
size_t input_count;
size_t pos;
detail::BlockPatternMatchVector PM;
std::vector<size_t> str_lens;
};
} /* namespace experimental */
#endif
template <typename CharT1>
struct CachedLCSseq
: detail::CachedSimilarityBase<CachedLCSseq<CharT1>, size_t, 0, std::numeric_limits<int64_t>::max()> {
template <typename Sentence1>
explicit CachedLCSseq(const Sentence1& s1_) : CachedLCSseq(detail::to_begin(s1_), detail::to_end(s1_))
{}
template <typename InputIt1>
CachedLCSseq(InputIt1 first1, InputIt1 last1) : s1(first1, last1), PM(detail::make_range(first1, last1))
{}
private:
friend detail::CachedSimilarityBase<CachedLCSseq<CharT1>, size_t, 0, std::numeric_limits<int64_t>::max()>;
friend detail::CachedNormalizedMetricBase<CachedLCSseq<CharT1>>;
template <typename InputIt2>
size_t maximum(const detail::Range<InputIt2>& s2) const
{
return std::max(s1.size(), s2.size());
}
template <typename InputIt2>
size_t _similarity(const detail::Range<InputIt2>& s2, size_t score_cutoff, size_t) const
{
return detail::lcs_seq_similarity(PM, detail::make_range(s1), s2, score_cutoff);
}
std::vector<CharT1> s1;
detail::BlockPatternMatchVector PM;
};
#ifdef RAPIDFUZZ_DEDUCTION_GUIDES
template <typename Sentence1>
explicit CachedLCSseq(const Sentence1& s1_) -> CachedLCSseq<char_type<Sentence1>>;
template <typename InputIt1>
CachedLCSseq(InputIt1 first1, InputIt1 last1) -> CachedLCSseq<iter_value_t<InputIt1>>;
#endif
} // namespace rapidfuzz
|
