/
tokens_select_mt.cpp
236 lines (217 loc) · 8.66 KB
/
tokens_select_mt.cpp
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
#include "lib.h"
//#include "dev.h"
#include "recompile.h"
using namespace quanteda;
Text keep_token(Text tokens,
const std::vector<std::size_t> &spans,
const SetNgrams &set_words,
const bool &padding,
const std::pair<int, int> &window,
const std::pair<int, int> &pos){
if (tokens.size() == 0) return {}; // return empty vector for empty text
unsigned int filler = UINT_MAX; // use upper limit as a filler
bool match = false;
std::size_t start, end;
if (pos.first == 0) {
start = 0;
} else if (pos.first > 0) {
start = std::min((int)tokens.size(), pos.first - 1);
} else {
start = std::max(0, (int)tokens.size() + pos.first);
}
if (pos.second == 0) {
end = 0;
} else if (pos.second > 0) {
end = std::min((int)tokens.size(), pos.second);
} else {
end = std::max(0, (int)tokens.size() + pos.second + 1);
}
Text tokens_copy(tokens.size());
if (padding) {
std::fill(tokens_copy.begin(), tokens_copy.end(), 0);
} else {
std::fill(tokens_copy.begin(), tokens_copy.end(), filler);
}
for (std::size_t span : spans) { // substitution starts from the longest sequences
if (tokens.size() < span) continue;
for (std::size_t i = start; i < end - (span - 1); i++) {
Ngram ngram(tokens.begin() + i, tokens.begin() + i + span);
auto it = set_words.find(ngram);
if (it != set_words.end()) {
match = true;
if (window.first == 0 && window.second == 0) {
std::copy(ngram.begin(), ngram.end(), tokens_copy.begin() + i);
} else {
int from = std::max((int)i - window.first, 0);
int to = std::min((int)i + (int)span + window.second, (int)tokens.size());
std::copy(tokens.begin() + from, tokens.begin() + to, tokens_copy.begin() + from);
}
}
}
}
if (match) {
if (!padding) {
tokens_copy.erase(std::remove(tokens_copy.begin(), tokens_copy.end(), filler), tokens_copy.end());
}
} else {
if (!padding) {
tokens_copy = {};
}
}
return tokens_copy;
}
Text remove_token(Text tokens,
const std::vector<std::size_t> &spans,
const SetNgrams &set_words,
const bool &padding,
const std::pair<int, int> &window,
const std::pair<int, int> &pos){
if (tokens.size() == 0) return {}; // return empty vector for empty text
unsigned int filler = UINT_MAX; // use upper limit as a filler
Text tokens_copy = tokens;
bool match = false;
std::size_t start, end;
if (pos.first == 0) {
start = 0;
} else if (pos.first > 0) {
start = std::min((int)tokens.size(), pos.first - 1);
} else {
start = std::max(0, (int)tokens.size() + pos.first);
}
if (pos.second == 0) {
end = 0;
} else if (pos.second > 0) {
end = std::min((int)tokens.size(), pos.second);
} else {
end = std::max(0, (int)tokens.size() + pos.second + 1);
}
for (std::size_t span : spans) { // substitution starts from the longest sequences
if (tokens.size() < span) continue;
for (std::size_t i = start; i < end - (span - 1); i++) {
Ngram ngram(tokens.begin() + i, tokens.begin() + i + span);
auto it = set_words.find(ngram);
if (it != set_words.end()) {
match = true;
if (window.first == 0 && window.second == 0) {
if (padding) {
std::fill(tokens_copy.begin() + i, tokens_copy.begin() + i + span, 0);
} else {
std::fill(tokens_copy.begin() + i, tokens_copy.begin() + i + span, filler);
}
} else {
int from = std::max((int)i - window.first, 0);
int to = std::min((int)i + (int)span + window.second, (int)tokens.size());
if (padding) {
std::fill(tokens_copy.begin() + from, tokens_copy.begin() + to, 0);
} else {
std::fill(tokens_copy.begin() + from, tokens_copy.begin() + to, filler);
}
}
}
}
}
if (match && !padding) {
tokens_copy.erase(std::remove(tokens_copy.begin(), tokens_copy.end(), filler), tokens_copy.end());
}
return tokens_copy;
}
struct select_mt : public Worker{
Texts &texts;
const std::vector<std::size_t> &spans;
const SetNgrams &set_words;
const int &mode;
const bool &padding;
const std::pair<int, int> &window;
const std::pair<int, int> &pos;
// Constructor
select_mt(Texts &texts_, const std::vector<std::size_t> &spans_,
const SetNgrams &set_words_, const int &mode_, const bool &padding_,
const std::pair<int, int> &window_, const std::pair<int, int> &pos_):
texts(texts_), spans(spans_), set_words(set_words_), mode(mode_), padding(padding_),
window(window_), pos(pos_){}
// parallelFor calles this function with std::size_t
void operator()(std::size_t begin, std::size_t end){
//Rcout << "Range " << begin << " " << end << "\n";
if (mode == 1) {
for (std::size_t h = begin; h < end; h++) {
texts[h] = keep_token(texts[h], spans, set_words, padding, window, pos);
}
} else if (mode == 2) {
for (std::size_t h = begin; h < end; h++) {
texts[h] = remove_token(texts[h], spans, set_words, padding, window, pos);
}
} else {
for (std::size_t h = begin; h < end; h++) {
texts[h] = texts[h];
}
}
}
};
/*
* This function selects features in tokens object with multiple threads.
* The number of threads is set by RcppParallel::setThreadOptions()
* @used tokens_select()
* @creator Kohei Watanabe
* @param texts_ tokens ojbect
* @param words_ list of features to remove or keep
* @param mode_ 1: keep; 2: remove
* @param padding_ fill places where features are removed with zero
*
*/
// [[Rcpp::export]]
List qatd_cpp_tokens_select(const List &texts_,
const CharacterVector types_,
const List &words_,
int mode,
bool padding,
int window_left,
int window_right,
int pos_from,
int pos_to){
Texts texts = Rcpp::as<Texts>(texts_);
Types types = Rcpp::as<Types>(types_);
std::pair<int, int> window(window_left, window_right);
std::pair<int, int> pos(pos_from, pos_to);
SetNgrams set_words;
std::vector<std::size_t> spans = register_ngrams(words_, set_words);
// dev::Timer timer;
// dev::start_timer("Token select", timer);
#if QUANTEDA_USE_TBB
select_mt select_mt(texts, spans, set_words, mode, padding, window, pos);
parallelFor(0, texts.size(), select_mt);
#else
if (mode == 1) {
for (std::size_t h = 0; h < texts.size(); h++) {
texts[h] = keep_token(texts[h], spans, set_words, padding, window, pos);
}
} else if(mode == 2) {
for (std::size_t h = 0; h < texts.size(); h++) {
texts[h] = remove_token(texts[h], spans, set_words, padding, window, pos);
}
} else {
for (std::size_t h = 0; h < texts.size(); h++){
texts[h] = texts[h];
}
}
#endif
// dev::stop_timer("Token select", timer);
return recompile(texts, types, true, false, is_encoded(types_));
}
/***R
toks <- list(rep(1:10, 1))
#toks <- list(rep(1:10, 1), rep(5:15, 1))
#dict <- as.list(1:100000)
dict <- list(c(1, 2), c(5, 6), 10, 15, 20)
#dict <- list(c(99))
#qatd_cpp_tokens_select(toks, letters, dict, 1, TRUE, 1, 1)
qatd_cpp_tokens_select(toks, letters, dict, 1, TRUE, 0, 0, 1, 100)
qatd_cpp_tokens_select(toks, letters, dict, 1, TRUE, 0, 1, 1, 100)
qatd_cpp_tokens_select(toks, letters, dict, 1, FALSE, 0, 0, 1, 2)
qatd_cpp_tokens_select(toks, letters, dict, 2, TRUE, 0, 0, 1, 5)
qatd_cpp_tokens_select(toks, letters, dict, 2, TRUE, 0, 1, 1, 2)
qatd_cpp_tokens_select(toks, letters, dict, 2, TRUE, 0, 2, 1, 3)
qatd_cpp_tokens_select(toks, letters, as.list(1:10), 1, FALSE, 0, 0, 1, 2)
qatd_cpp_tokens_select(toks, letters, as.list(1:10), 1, TRUE, 0, 0, 1, 2)
qatd_cpp_tokens_select(toks, letters, as.list(1:10), 2, FALSE, 0, 0, 1, 1)
qatd_cpp_tokens_select(toks, letters, as.list(1:10), 2, TRUE, 0, 0, 1, 1)
*/