effspm 0.2.8__cp39-cp39-win_amd64.whl → 0.3.3__cp39-cp39-win_amd64.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (54) hide show
  1. effspm/_effspm.cp39-win_amd64.pyd +0 -0
  2. effspm/_effspm.cpp +961 -210
  3. effspm/btminer/src/build_mdd.cpp +42 -17
  4. effspm/btminer/src/build_mdd.hpp +13 -19
  5. effspm/btminer/src/freq_miner.cpp +134 -49
  6. effspm/btminer/src/freq_miner.hpp +16 -0
  7. effspm/btminer/src/load_inst.cpp +211 -126
  8. effspm/btminer/src/load_inst.hpp +22 -4
  9. effspm/btminer/src/main.cpp +83 -0
  10. effspm/btminer/src/utility.cpp +26 -41
  11. effspm/btminer/src/utility.hpp +6 -30
  12. effspm/freq_miner.hpp +2 -1
  13. effspm/htminer/src/build_mdd.cpp +46 -124
  14. effspm/htminer/src/build_mdd.hpp +56 -49
  15. effspm/htminer/src/freq_miner.cpp +341 -307
  16. effspm/htminer/src/freq_miner.hpp +39 -40
  17. effspm/htminer/src/load_inst.cpp +287 -336
  18. effspm/htminer/src/load_inst.hpp +23 -6
  19. effspm/htminer/src/main.cpp +97 -0
  20. effspm/htminer/src/utility.cpp +38 -57
  21. effspm/htminer/src/utility.hpp +9 -64
  22. effspm/largebm/src/build_mdd.cpp +69 -110
  23. effspm/largebm/src/build_mdd.hpp +22 -37
  24. effspm/largebm/src/freq_miner.cpp +241 -291
  25. effspm/largebm/src/freq_miner.hpp +25 -36
  26. effspm/largebm/src/load_inst.cpp +20 -26
  27. effspm/largebm/src/load_inst.hpp +24 -34
  28. effspm/largebm/src/main.cpp +95 -0
  29. effspm/largebm/src/utility.cpp +11 -21
  30. effspm/largebm/src/utility.hpp +7 -10
  31. effspm/largehm/src/build_mdd.cpp +75 -110
  32. effspm/largehm/src/build_mdd.hpp +53 -73
  33. effspm/largehm/src/freq_miner.cpp +134 -191
  34. effspm/largehm/src/freq_miner.hpp +37 -60
  35. effspm/largehm/src/load_inst.cpp +137 -174
  36. effspm/largehm/src/load_inst.hpp +13 -50
  37. effspm/largehm/src/main.cpp +95 -0
  38. effspm/largehm/src/utility.cpp +46 -28
  39. effspm/largehm/src/utility.hpp +18 -16
  40. effspm/largepp/src/freq_miner.cpp +184 -156
  41. effspm/largepp/src/freq_miner.hpp +11 -36
  42. effspm/largepp/src/load_inst.cpp +32 -12
  43. effspm/largepp/src/load_inst.hpp +15 -9
  44. effspm/largepp/src/main.cpp +108 -0
  45. effspm/largepp/src/pattern.hpp +31 -0
  46. effspm/load_inst.cpp +8 -8
  47. effspm/load_inst.hpp +1 -1
  48. effspm/main.cpp +103 -0
  49. {effspm-0.2.8.dist-info → effspm-0.3.3.dist-info}/METADATA +1 -1
  50. effspm-0.3.3.dist-info/RECORD +60 -0
  51. effspm-0.2.8.dist-info/RECORD +0 -53
  52. {effspm-0.2.8.dist-info → effspm-0.3.3.dist-info}/WHEEL +0 -0
  53. {effspm-0.2.8.dist-info → effspm-0.3.3.dist-info}/licenses/LICENSE +0 -0
  54. {effspm-0.2.8.dist-info → effspm-0.3.3.dist-info}/top_level.txt +0 -0
effspm/largehm/src/freq_miner.cpp CHANGED
@@ -1,129 +1,89 @@
1
-
2
- #include <cstdint>
3
- #include <vector>
4
-
5
1
  #include <iostream>
6
2
  #include <time.h>
7
- // for std::vector
8
- #include <cmath> // for std::ceil
9
-
10
3
  #include "freq_miner.hpp"
11
4
  #include "build_mdd.hpp"
12
5
  #include "utility.hpp"
13
- std::vector<std::uint64_t> ancest_base;
6
+
14
7
  namespace largehm {
15
8
 
16
- void Out_patt(std::vector<int>& seq, unsigned int freq);
9
+ using namespace std;
10
+
11
+ void Out_patt(vector<int>& seq, unsigned int freq);
17
12
  void Extend_patt(Pattern& _patt);
18
- void Mine_vec(unsigned long long int seq_ID,
19
- int pos,
20
- int num_found,
21
-
22
- std::vector<int>& items,
13
+ void Mine_vec(unsigned long long int seq_ID, int pos, int num_found,
14
+ vector<unsigned long long int>& ancest,
15
+ vector<int>& items,
23
16
  unsigned long long int inod,
24
17
  int sgn);
25
18
 
26
19
  unsigned long long int num_patt = 0;
27
20
 
28
- std::vector<bool> ilist;
29
- std::vector<bool> slist;
21
+ vector<bool> ilist;
22
+ vector<bool> slist;
30
23
 
31
- std::vector<Pattern> pot_patt;
32
- std::vector<VPattern> pot_vpatt;
33
- std::vector<unsigned long long int> last_strpnt;
24
+ vector<Pattern> pot_patt;
25
+ vector<VPattern> pot_vpatt;
26
+ vector<unsigned long long int> last_strpnt;
27
+ vector<unsigned long long int> ancest_base;
28
+ vector<int> DFS_numfound;
34
29
 
35
- std::vector<int> DFS_numfound;
36
-
37
- Pattern _patt;
30
+ Pattern _patt;
38
31
  VPattern _vpatt;
39
32
 
40
33
  int itmset_size;
41
34
  int last_neg;
42
-
43
35
  bool ilist_nempty;
44
36
 
45
-
46
37
  void Freq_miner() {
47
-
48
-
49
- // ─── Make sure DFS and VDFS are at least size L ─────────────────────────────
50
- if (DFS.size() < static_cast<size_t>(L)) {
51
- size_t old = DFS.size();
52
- DFS.resize(static_cast<size_t>(L));
53
- for (size_t i = old; i < DFS.size(); ++i) {
54
- DFS[i] = Pattern(-static_cast<int>(i) - 1);
55
- }
56
- }
57
- if (VDFS.size() < static_cast<size_t>(L)) {
58
- size_t old = VDFS.size();
59
- VDFS.resize(static_cast<size_t>(L));
60
- for (size_t i = old; i < VDFS.size(); ++i) {
61
- VDFS[i] = VPattern(static_cast<int>(i));
62
- }
63
- }
64
- // ─────────────────────────────────────────────────────────────────────────────
65
38
 
66
-
67
- if (!Tree.empty()) {
68
- // std::cout << ", Tree[0].chld=" << Tree[0].chld
69
- // << ", Tree[0].sibl=" << Tree[0].sibl
70
- // << ", Tree[0].freq=" << Tree[0].freq;
71
- }
72
- // std::cout << ", DFS.size()=" << DFS.size()
73
- // << ", theta=" << theta
74
- // << ", M=" << M
75
- // << ", E=" << E
76
- // << std::endl;
77
-
78
- std::vector<int> tmp_list;
79
- for (int i = 0; i < static_cast<int>(L); ++i) {
39
+ vector<int> list;
40
+
41
+ for (int i = 0; i < (int)L; ++i) {
80
42
  if (DFS[i].freq >= theta) {
81
- tmp_list.push_back(-i - 1);
82
- if (itmset_exists) {
83
- tmp_list.push_back(i + 1);
84
- }
43
+ list.push_back(-i - 1);
44
+ if (itmset_exists)
45
+ list.push_back(i + 1);
85
46
  }
86
47
  }
87
48
 
88
- for (int i = 0; i < static_cast<int>(DFS.size()); ++i) {
89
- DFS[i].list = tmp_list;
90
- }
49
+ for (int i = 0; i < (int)DFS.size(); ++i)
50
+ DFS[i].list = list;
91
51
 
92
- while (!DFS.empty() && give_time(std::clock() - start_time) < time_limit) {
93
- if (DFS.back().freq >= theta) {
52
+ while (!DFS.empty() && give_time(clock() - start_time) < time_limit) {
53
+ if (DFS.back().freq >= theta)
94
54
  Extend_patt(DFS.back());
95
- }
96
55
  else {
97
56
  DFS.pop_back();
98
- if (!VDFS.empty() && VDFS.back().ass_patt == static_cast<int>(DFS.size())) {
57
+ if (!VDFS.empty() && VDFS.back().ass_patt == DFS.size())
99
58
  VDFS.pop_back();
100
- }
101
59
  }
102
60
  }
103
61
  }
104
62
 
105
-
106
63
  void Extend_patt(Pattern& _pattern) {
64
+
107
65
  swap(_patt, _pattern);
108
66
  DFS.pop_back();
109
67
 
110
- slist = std::vector<bool>(L, false);
111
- ilist_nempty = false;
68
+ slist = vector<bool>(L, 0);
69
+ ilist_nempty = 0;
112
70
 
113
71
  if (itmset_exists) {
114
- ilist = std::vector<bool>(L, false);
115
- for (auto it = _patt.list.begin(); it != _patt.list.end(); ++it) {
72
+ ilist = vector<bool>(L, 0);
73
+ for (vector<int>::iterator it = _patt.list.begin();
74
+ it != _patt.list.end(); ++it) {
116
75
  if (*it < 0)
117
- slist[-(*it) - 1] = true;
76
+ slist[-(*it) - 1] = 1;
118
77
  else {
119
- ilist[(*it) - 1] = true;
120
- ilist_nempty = true;
78
+ ilist[(*it) - 1] = 1;
79
+ ilist_nempty = 1;
121
80
  }
122
81
  }
123
82
  }
124
83
  else {
125
- for (auto it = _patt.list.begin(); it != _patt.list.end(); ++it)
126
- slist[-(*it) - 1] = true;
84
+ for (vector<int>::iterator it = _patt.list.begin();
85
+ it != _patt.list.end(); ++it)
86
+ slist[-(*it) - 1] = 1;
127
87
  }
128
88
 
129
89
  last_neg = _patt.seq.size() - 1;
@@ -131,39 +91,27 @@ void Extend_patt(Pattern& _pattern) {
131
91
  --last_neg;
132
92
  itmset_size = _patt.seq.size() - last_neg;
133
93
 
134
- pot_patt = std::vector<Pattern>(L + L * (ilist_nempty ? 1 : 0));
94
+ pot_patt = vector<Pattern>(L + L * ilist_nempty);
135
95
  if (!CTree.empty())
136
- pot_vpatt = std::vector<VPattern>(L + L * (ilist_nempty ? 1 : 0));
96
+ pot_vpatt = vector<VPattern>(L + L * ilist_nempty);
137
97
 
138
- last_strpnt = std::vector<unsigned long long int>(L, 0ULL);
98
+ last_strpnt = vector<unsigned long long int>(L, 0);
139
99
 
140
- if (!VDFS.empty() && VDFS.back().ass_patt == static_cast<int>(DFS.size())) {
100
+ if (!VDFS.empty() && VDFS.back().ass_patt == DFS.size()) {
141
101
  swap(_vpatt, VDFS.back());
142
102
  VDFS.pop_back();
143
103
  for (unsigned long long int pnt = 0; pnt < _vpatt.str_pnt.size(); ++pnt) {
144
- if (_vpatt.str_pnt[pnt] < 0) {
145
- Mine_vec(_vpatt.seq_ID[pnt],
146
- -_vpatt.str_pnt[pnt],
147
- -1,
148
- ancest_base,
149
- CTree[_vpatt.seq_ID[pnt]].seq,
150
- 0,
151
- -1);
152
- }
153
- else {
154
- Mine_vec(_vpatt.seq_ID[pnt],
155
- _vpatt.str_pnt[pnt],
156
- -1,
157
- ancest_base,
158
- VTree[_vpatt.seq_ID[pnt]].seq,
159
- 0,
160
- 1);
161
- }
104
+ if (_vpatt.str_pnt[pnt] < 0)
105
+ Mine_vec(_vpatt.seq_ID[pnt], -_vpatt.str_pnt[pnt], -1,
106
+ ancest_base, CTree[_vpatt.seq_ID[pnt]].seq, 0, -1);
107
+ else
108
+ Mine_vec(_vpatt.seq_ID[pnt], _vpatt.str_pnt[pnt], -1,
109
+ ancest_base, VTree[_vpatt.seq_ID[pnt]].seq, 0, 1);
162
110
  }
163
111
  }
164
112
 
165
- std::vector<unsigned long long int> DFS_itm;
166
- std::vector<unsigned long long int> DFS_seq;
113
+ vector<unsigned long long int> DFS_itm;
114
+ vector<unsigned long long int> DFS_seq;
167
115
  if (ilist_nempty)
168
116
  DFS_numfound.clear();
169
117
 
@@ -173,23 +121,13 @@ void Extend_patt(Pattern& _pattern) {
173
121
  unsigned long long int cur_sibl = DFS_itm.back();
174
122
  DFS_itm.pop_back();
175
123
  if (Tree[cur_sibl].itmset < 0) {
176
- unsigned int carc = Tree[cur_sibl].chld;
177
- Mine_vec(carc,
178
- 0,
179
- -1,
180
- CTree[carc].ancest,
181
- CTree[carc].seq,
182
- _patt.str_pnt[pnt],
183
- -1);
124
+ unsigned long long int carc = Tree[cur_sibl].chld;
125
+ Mine_vec(carc, 0, -1, CTree[carc].ancest,
126
+ CTree[carc].seq, _patt.str_pnt[pnt], -1);
184
127
  cur_sibl = CTree[carc].ancest.back();
185
128
  while (cur_sibl != 0) {
186
- Mine_vec(cur_sibl - 1,
187
- 0,
188
- -1,
189
- CTree[carc].ancest,
190
- VTree[cur_sibl - 1].seq,
191
- _patt.str_pnt[pnt],
192
- 1);
129
+ Mine_vec(cur_sibl - 1, 0, -1, CTree[carc].ancest,
130
+ VTree[cur_sibl - 1].seq, _patt.str_pnt[pnt], 1);
193
131
  cur_sibl = VTree[cur_sibl - 1].sibl;
194
132
  }
195
133
  continue;
@@ -232,7 +170,7 @@ void Extend_patt(Pattern& _pattern) {
232
170
  last_strpnt[i] = pot_patt[i + L].str_pnt.size();
233
171
  }
234
172
  }
235
- while(!DFS_seq.empty()) {
173
+ while (!DFS_seq.empty()) {
236
174
  unsigned long long int cur_sibl = DFS_seq.back();
237
175
  DFS_seq.pop_back();
238
176
  int num_found = 0;
@@ -242,22 +180,12 @@ void Extend_patt(Pattern& _pattern) {
242
180
  }
243
181
  if (Tree[cur_sibl].itmset < 0) {
244
182
  unsigned int carc = Tree[cur_sibl].chld;
245
- Mine_vec(carc,
246
- 0,
247
- num_found,
248
- CTree[carc].ancest,
249
- CTree[carc].seq,
250
- _patt.str_pnt[pnt],
251
- -1);
183
+ Mine_vec(carc, 0, num_found, CTree[carc].ancest,
184
+ CTree[carc].seq, _patt.str_pnt[pnt], -1);
252
185
  cur_sibl = CTree[carc].ancest.back();
253
186
  while (cur_sibl != 0) {
254
- Mine_vec(cur_sibl - 1,
255
- 0,
256
- num_found,
257
- CTree[carc].ancest,
258
- VTree[cur_sibl - 1].seq,
259
- _patt.str_pnt[pnt],
260
- 1);
187
+ Mine_vec(cur_sibl - 1, 0, num_found, CTree[carc].ancest,
188
+ VTree[cur_sibl - 1].seq, _patt.str_pnt[pnt], 1);
261
189
  cur_sibl = VTree[cur_sibl - 1].sibl;
262
190
  }
263
191
  continue;
@@ -268,19 +196,18 @@ void Extend_patt(Pattern& _pattern) {
268
196
  if (cur_itm > 0) {
269
197
  if (num_found == itmset_size &&
270
198
  ilist[cur_itm - 1] &&
271
- (std::abs(Tree[Tree[cur_sibl].anct].itmset) < std::abs(Tree[_patt.str_pnt[pnt]].itmset)
272
- || !check_parent(Tree[cur_sibl].anct,
273
- _patt.str_pnt[pnt],
274
- last_strpnt[cur_itm - 1],
275
- pot_patt[cur_itm + L - 1].str_pnt)))
276
- {
199
+ (abs(Tree[Tree[cur_sibl].anct].itmset) <
200
+ abs(Tree[_patt.str_pnt[pnt]].itmset) ||
201
+ !check_parent(Tree[cur_sibl].anct, _patt.str_pnt[pnt],
202
+ last_strpnt[cur_itm - 1],
203
+ pot_patt[cur_itm + L - 1].str_pnt))) {
277
204
  pot_patt[cur_itm + L - 1].freq += Tree[cur_sibl].freq;
278
205
  if (Tree[cur_sibl].chld != 0 || Tree[cur_sibl].itmset < 0)
279
206
  pot_patt[cur_itm + L - 1].str_pnt.push_back(cur_sibl);
280
207
  }
281
208
  if (slist[cur_itm - 1] &&
282
- std::abs(Tree[Tree[cur_sibl].anct].itmset) <= std::abs(Tree[_patt.str_pnt[pnt]].itmset))
283
- {
209
+ abs(Tree[Tree[cur_sibl].anct].itmset) <=
210
+ abs(Tree[_patt.str_pnt[pnt]].itmset)) {
284
211
  pot_patt[cur_itm - 1].freq += Tree[cur_sibl].freq;
285
212
  if (Tree[cur_sibl].chld != 0 || Tree[cur_sibl].itmset < 0)
286
213
  pot_patt[cur_itm - 1].str_pnt.push_back(cur_sibl);
@@ -288,8 +215,9 @@ void Extend_patt(Pattern& _pattern) {
288
215
  if (Tree[cur_sibl].chld != 0 || Tree[cur_sibl].itmset < 0) {
289
216
  DFS_seq.push_back(cur_sibl);
290
217
  if (ilist_nempty) {
291
- if (num_found < itmset_size
292
- && cur_itm == std::abs(_patt.seq[last_neg + num_found]))
218
+ if (num_found < itmset_size &&
219
+ cur_itm ==
220
+ abs(_patt.seq[last_neg + num_found]))
293
221
  DFS_numfound.push_back(num_found + 1);
294
222
  else
295
223
  DFS_numfound.push_back(num_found);
@@ -299,8 +227,8 @@ void Extend_patt(Pattern& _pattern) {
299
227
  else {
300
228
  cur_itm = -cur_itm;
301
229
  if (slist[cur_itm - 1] &&
302
- std::abs(Tree[Tree[cur_sibl].anct].itmset) <= std::abs(Tree[_patt.str_pnt[pnt]].itmset))
303
- {
230
+ abs(Tree[Tree[cur_sibl].anct].itmset) <=
231
+ abs(Tree[_patt.str_pnt[pnt]].itmset)) {
304
232
  pot_patt[cur_itm - 1].freq += Tree[cur_sibl].freq;
305
233
  if (Tree[cur_sibl].chld != 0 || Tree[cur_sibl].itmset < 0)
306
234
  pot_patt[cur_itm - 1].str_pnt.push_back(cur_sibl);
@@ -320,126 +248,141 @@ void Extend_patt(Pattern& _pattern) {
320
248
  }
321
249
  }
322
250
 
323
- std::vector<int> ilistp;
324
- std::vector<int> slistp;
325
- for (auto it = _patt.list.begin(); it != _patt.list.end(); ++it) {
326
- if (*it > 0 && pot_patt[(*it) + L - 1].freq >= theta)
251
+ vector<int> ilistp;
252
+ vector<int> slistp;
253
+ for (vector<int>::iterator it = _patt.list.begin();
254
+ it != _patt.list.end(); ++it) {
255
+ if (*it > 0 && pot_patt[(*it) + L - 1].freq >= theta)
327
256
  ilistp.push_back(*it);
328
257
  else if (*it < 0 && pot_patt[-(*it) - 1].freq >= theta) {
329
- if (itmset_exists)
258
+ if (itmset_exists)
330
259
  slistp.push_back(-(*it));
331
260
  ilistp.push_back(*it);
332
261
  slistp.push_back(*it);
333
262
  }
334
- }
263
+ }
335
264
 
336
- for (auto it = ilistp.begin(); it != ilistp.end(); ++it) {
265
+ for (vector<int>::iterator it = ilistp.begin(); it != ilistp.end(); ++it) {
337
266
  int p;
338
- if (*it < 0)
267
+ if (*it < 0)
339
268
  p = -(*it) - 1;
340
269
  else
341
270
  p = (*it) - 1 + L;
271
+
342
272
  pot_patt[p].str_pnt.shrink_to_fit();
343
273
  DFS.push_back(pot_patt[p]);
274
+
275
+ // Build the full pattern sequence
344
276
  DFS.back().seq = _patt.seq;
345
277
  DFS.back().seq.push_back(*it);
278
+
279
+ // Update candidate list
346
280
  if (*it < 0)
347
281
  DFS.back().list = slistp;
348
282
  else
349
283
  DFS.back().list = ilistp;
284
+
285
+ // Attach VPatterns if needed
350
286
  if (!CTree.empty() && !pot_vpatt[p].str_pnt.empty()) {
351
- pot_vpatt[p].ass_patt = static_cast<int>(DFS.size()) - 1;
287
+ pot_vpatt[p].ass_patt = DFS.size() - 1;
352
288
  VDFS.push_back(pot_vpatt[p]);
353
289
  }
354
- if (b_disp || b_write)
290
+
291
+ // ✅ Always record the pattern for Python, independent of b_disp / b_write.
292
+ collectedPatterns.push_back(DFS.back().seq);
293
+
294
+ // Original output behavior (only if requested)
295
+ if (b_disp || b_write)
355
296
  Out_patt(DFS.back().seq, DFS.back().freq);
297
+
356
298
  ++num_patt;
357
299
  }
300
+
358
301
  }
359
302
 
303
+ void Mine_vec(unsigned long long int seq_ID, int pos, int num_found,
304
+ vector<unsigned long long int>& ancest,
305
+ vector<int>& items,
306
+ unsigned long long int pnt,
307
+ int sgn) {
360
308
 
361
- void Mine_vec(std::uint64_t seq_ID,
362
- int pos,
363
- int num_found,
364
- std::vector<std::uint64_t>& ancest,
365
- std::vector<int>& items,
366
- std::uint64_t pnt,
367
- int sgn)
368
- {
369
- std::vector<bool> found(L + L * (ilist_nempty ? 1 : 0), false);
309
+ vector<bool> found(L + L * ilist_nempty, 0);
370
310
 
371
311
  if (num_found == -1) {
372
- while (pos < static_cast<int>(items.size()) && items[pos] > 0) {
312
+ while (pos < (int)items.size() && items[pos] > 0) {
373
313
  int cur_itm = items[pos];
374
314
  if (ilist[cur_itm - 1] && !found[cur_itm + L - 1]) {
375
- if (pos + 1 < static_cast<int>(items.size())) {
315
+ if (pos + 1 < (int)items.size()) {
376
316
  pot_vpatt[cur_itm + L - 1].seq_ID.push_back(seq_ID);
377
317
  pot_vpatt[cur_itm + L - 1].str_pnt.push_back(sgn * (pos + 1));
378
318
  }
379
319
  ++pot_patt[cur_itm + L - 1].freq;
380
- found[cur_itm + L - 1] = true;
320
+ found[cur_itm + L - 1] = 1;
381
321
  }
382
322
  ++pos;
383
323
  }
384
324
  }
385
325
 
386
326
  for (unsigned int k = pos; k < items.size(); ++k) {
387
- int cur_itm = std::abs(items[k]);
388
- if (items[k] < 0)
327
+ int cur_itm = abs(items[k]);
328
+ if (items[k] < 0)
389
329
  num_found = 0;
390
330
  if (slist[cur_itm - 1] && !found[cur_itm - 1]) {
391
- if (ancest.empty() || std::abs(Tree[ancest[cur_itm - 1]].itmset) <= std::abs(Tree[pnt].itmset)) {
392
- if (k + 1 < static_cast<int>(items.size())) {
331
+ if (ancest.empty() ||
332
+ abs(Tree[ancest[cur_itm - 1]].itmset) <= abs(Tree[pnt].itmset)) {
333
+ if (k + 1 < (int)items.size()) {
393
334
  pot_vpatt[cur_itm - 1].seq_ID.push_back(seq_ID);
394
335
  pot_vpatt[cur_itm - 1].str_pnt.push_back(sgn * (k + 1));
395
336
  }
396
337
  ++pot_patt[cur_itm - 1].freq;
397
338
  }
398
- found[cur_itm - 1] = true;
339
+ found[cur_itm - 1] = 1;
399
340
  }
400
341
  if (num_found == itmset_size) {
401
342
  if (ilist[cur_itm - 1] && !found[cur_itm + L - 1]) {
402
- if (ancest.empty() ||
403
- std::abs(Tree[ancest[cur_itm - 1]].itmset) < std::abs(Tree[pnt].itmset)
404
- || !check_parent(ancest[cur_itm - 1], pnt, last_strpnt[cur_itm - 1], pot_patt[cur_itm + L - 1].str_pnt))
405
- {
406
- if (k + 1 < static_cast<int>(items.size())) {
343
+ if (ancest.empty() ||
344
+ abs(Tree[ancest[cur_itm - 1]].itmset) < abs(Tree[pnt].itmset) ||
345
+ !check_parent(ancest[cur_itm - 1], pnt,
346
+ last_strpnt[cur_itm - 1],
347
+ pot_patt[cur_itm + L - 1].str_pnt)) {
348
+ if (k + 1 < (int)items.size()) {
407
349
  pot_vpatt[cur_itm + L - 1].seq_ID.push_back(seq_ID);
408
350
  pot_vpatt[cur_itm + L - 1].str_pnt.push_back(sgn * (k + 1));
409
351
  }
410
352
  ++pot_patt[cur_itm + L - 1].freq;
411
353
  }
412
- found[cur_itm + L - 1] = true;
354
+ found[cur_itm + L - 1] = 1;
413
355
  }
414
356
  }
415
- else if (cur_itm == std::abs(_patt.seq[last_neg + num_found])) {
357
+ else if (cur_itm == abs(_patt.seq[last_neg + num_found]))
416
358
  ++num_found;
417
- }
418
359
  }
419
360
  }
420
361
 
362
+ void Out_patt(vector<int>& seq, unsigned int freq) {
421
363
 
422
- void Out_patt(std::vector<int>& seq, unsigned int freq) {
423
- largehm::collected.push_back(seq);
424
- std::ofstream file_o;
425
- if (b_write)
364
+ ofstream file_o;
365
+ if (b_write)
426
366
  file_o.open(out_file, std::ios::app);
427
367
 
428
- for (int ii = 0; ii < static_cast<int>(seq.size()); ii++) {
368
+ for (int ii = 0; ii < (int)seq.size(); ii++) {
429
369
  if (b_disp)
430
- std::cout << seq[ii] << " ";
431
- if (b_write)
370
+ cout << seq[ii] << " ";
371
+ if (b_write)
432
372
  file_o << seq[ii] << " ";
433
373
  }
434
374
  if (b_disp)
435
- std::cout << std::endl;
375
+ cout << endl;
376
+ if (b_write)
377
+ file_o << endl;
378
+
379
+ if (b_disp)
380
+ cout << "************** Freq: " << freq << endl;
436
381
  if (b_write) {
437
- file_o << std::endl;
438
- file_o << "************** Freq: " << freq << std::endl;
382
+ file_o << "************** Freq: " << freq << endl;
439
383
  file_o.close();
440
384
  }
441
- if (b_disp)
442
- std::cout << "************** Freq: " << freq << std::endl;
443
385
  }
444
386
 
387
+
445
388
  } // namespace largehm
effspm/largehm/src/freq_miner.hpp CHANGED
@@ -1,77 +1,54 @@
1
- #ifndef LARGEHM_FREQ_MINER_HPP
2
- #define LARGEHM_FREQ_MINER_HPP
3
- #include <cstdint>
4
- #include <vector>
5
- #include <string>
6
- #include <fstream>
7
- #include <ctime> // for clock_t
8
- extern std::vector<std::uint64_t> ancest_base;
1
+ #pragma once
2
+
3
+ #include "load_inst.hpp"
4
+ #include "build_mdd.hpp"
5
+
9
6
  namespace largehm {
10
7
 
11
- //
12
- // ─── Pattern & VPattern ──────────────────────────────────────────────────────
13
- //
8
+ using namespace std;
9
+
10
+ void Freq_miner();
14
11
 
15
12
  class Pattern {
16
13
  public:
17
- std::vector<int> seq;
18
- unsigned int freq;
19
- std::vector<int> list;
20
- std::vector<unsigned long long int> str_pnt;
14
+ vector<int> seq;
15
+ vector<unsigned long long int> str_pnt;
16
+ vector<int> list;
17
+ unsigned long long int freq;
18
+
19
+ Pattern(vector<int>& _seq, int item) {
20
+ seq.swap(_seq);
21
+ seq.push_back(item);
22
+ freq = 0;
23
+ }
21
24
 
22
- Pattern(int start_code = 0) : freq(0) {
23
- if (start_code != 0)
24
- seq.push_back(start_code);
25
+ Pattern(int item) {
26
+ seq.push_back(item);
27
+ freq = 0;
28
+ }
29
+
30
+ Pattern() {
31
+ freq = 0;
25
32
  }
26
33
  };
27
34
 
28
35
  class VPattern {
29
36
  public:
30
- std::vector<unsigned long long int> str_pnt;
31
- std::vector<unsigned long long int> seq_ID;
32
- int ass_patt;
37
+ unsigned long long int ass_patt;
38
+ vector<int> str_pnt;
39
+ vector<unsigned long long int> seq_ID;
33
40
 
34
- VPattern(int assoc = -1) : ass_patt(assoc) {}
35
- };
41
+ VPattern(unsigned long long int _patt) {
42
+ ass_patt = _patt;
43
+ }
36
44
 
37
- //
38
- // ─── Globals used by Freq_miner ──────────────────────────────────────────────
39
- //
40
- extern std::vector<Pattern> DFS;
41
- extern std::vector<VPattern> VDFS;
45
+ VPattern() {
46
+ ass_patt = 0;
47
+ }
48
+ };
42
49
 
43
50
  extern unsigned long long int num_patt;
44
-
45
- extern std::vector<bool> ilist;
46
- extern std::vector<bool> slist;
47
-
48
- extern std::vector<Pattern> pot_patt;
49
- extern std::vector<VPattern> pot_vpatt;
50
- extern std::vector<unsigned long long int> last_strpnt;
51
-
52
- extern std::vector<int> DFS_numfound;
53
-
54
- extern Pattern _patt;
55
- extern VPattern _vpatt;
56
-
57
- extern int itmset_size;
58
- extern int last_neg;
59
- extern bool ilist_nempty;
60
-
61
- //
62
- // ─── Function Prototypes ─────────────────────────────────────────────────────
63
- //
64
- void Freq_miner();
65
- void Extend_patt(Pattern& _patt);
66
- void Mine_vec(std::uint64_t seq_ID,
67
- int pos,
68
- int num_found,
69
- std::vector<std::uint64_t>& ancest,
70
- std::vector<int>& items,
71
- std::uint64_t pnt,
72
- int sgn);
73
- void Out_patt(std::vector<int>& seq, unsigned int freq);
51
+ extern vector<Pattern> DFS;
52
+ extern vector<VPattern> VDFS;
74
53
 
75
54
  } // namespace largehm
76
-
77
- #endif // LARGEHM_FREQ_MINER_HPP