|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* webtrees-lib: MyArtJaub library for webtrees |
|
5
|
|
|
* |
|
6
|
|
|
* @package MyArtJaub\Webtrees |
|
7
|
|
|
* @subpackage Sosa |
|
8
|
|
|
* @author Jonathan Jaubart <[email protected]> |
|
9
|
|
|
* @copyright Copyright (c) 2009-2022, Jonathan Jaubart |
|
10
|
|
|
* @license http://www.gnu.org/licenses/gpl.html GNU General Public License, version 3 |
|
11
|
|
|
*/ |
|
12
|
|
|
|
|
13
|
|
|
declare(strict_types=1); |
|
14
|
|
|
|
|
15
|
|
|
namespace MyArtJaub\Webtrees\Module\Sosa\Services; |
|
16
|
|
|
|
|
17
|
|
|
use Fisharebest\Webtrees\Individual; |
|
18
|
|
|
use Fisharebest\Webtrees\Registry; |
|
19
|
|
|
use Fisharebest\Webtrees\Tree; |
|
20
|
|
|
use Fisharebest\Webtrees\Contracts\UserInterface; |
|
21
|
|
|
use Illuminate\Database\Capsule\Manager as DB; |
|
22
|
|
|
use Illuminate\Database\Query\Builder; |
|
23
|
|
|
use Illuminate\Database\Query\JoinClause; |
|
24
|
|
|
use Illuminate\Support\Collection; |
|
25
|
|
|
use stdClass; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Service for retrieving Sosa statistics |
|
29
|
|
|
*/ |
|
30
|
|
|
class SosaStatisticsService |
|
31
|
|
|
{ |
|
32
|
|
|
private UserInterface $user; |
|
33
|
|
|
private Tree $tree; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* Constructor for Sosa Statistics Service |
|
37
|
|
|
* |
|
38
|
|
|
* @param Tree $tree |
|
39
|
|
|
* @param UserInterface $user |
|
40
|
|
|
*/ |
|
41
|
|
|
public function __construct(Tree $tree, UserInterface $user) |
|
42
|
|
|
{ |
|
43
|
|
|
$this->tree = $tree; |
|
44
|
|
|
$this->user = $user; |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* Return the root individual for the reference tree and user. |
|
49
|
|
|
* |
|
50
|
|
|
* @return Individual|NULL |
|
51
|
|
|
*/ |
|
52
|
|
|
public function rootIndividual(): ?Individual |
|
53
|
|
|
{ |
|
54
|
|
|
$root_indi_id = $this->tree->getUserPreference($this->user, 'MAJ_SOSA_ROOT_ID'); |
|
55
|
|
|
return Registry::individualFactory()->make($root_indi_id, $this->tree); |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* Get the highest generation for the reference tree and user. |
|
60
|
|
|
* |
|
61
|
|
|
* @return int |
|
62
|
|
|
*/ |
|
63
|
|
|
public function maxGeneration(): int |
|
64
|
|
|
{ |
|
65
|
|
|
return (int) DB::table('maj_sosa') |
|
66
|
|
|
->where('majs_gedcom_id', '=', $this->tree->id()) |
|
67
|
|
|
->where('majs_user_id', '=', $this->user->id()) |
|
68
|
|
|
->max('majs_gen'); |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* Get the total count of individuals in the tree. |
|
73
|
|
|
* |
|
74
|
|
|
* @return int |
|
75
|
|
|
*/ |
|
76
|
|
|
public function totalIndividuals(): int |
|
77
|
|
|
{ |
|
78
|
|
|
return DB::table('individuals') |
|
79
|
|
|
->where('i_file', '=', $this->tree->id()) |
|
80
|
|
|
->count(); |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
/** |
|
84
|
|
|
* Get the total count of Sosa ancestors for all generations |
|
85
|
|
|
* |
|
86
|
|
|
* @return int |
|
87
|
|
|
*/ |
|
88
|
|
|
public function totalAncestors(): int |
|
89
|
|
|
{ |
|
90
|
|
|
return DB::table('maj_sosa') |
|
91
|
|
|
->where('majs_gedcom_id', '=', $this->tree->id()) |
|
92
|
|
|
->where('majs_user_id', '=', $this->user->id()) |
|
93
|
|
|
->count(); |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
/** |
|
97
|
|
|
* Get the total count of Sosa ancestors for a generation |
|
98
|
|
|
* |
|
99
|
|
|
* @return int |
|
100
|
|
|
*/ |
|
101
|
|
|
public function totalAncestorsAtGeneration(int $gen): int |
|
102
|
|
|
{ |
|
103
|
|
|
return DB::table('maj_sosa') |
|
104
|
|
|
->where('majs_gedcom_id', '=', $this->tree->id()) |
|
105
|
|
|
->where('majs_user_id', '=', $this->user->id()) |
|
106
|
|
|
->where('majs_gen', '=', $gen) |
|
107
|
|
|
->count(); |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
/** |
|
111
|
|
|
* Get the total count of distinct Sosa ancestors for all generations |
|
112
|
|
|
* |
|
113
|
|
|
* @return int |
|
114
|
|
|
*/ |
|
115
|
|
|
public function totalDistinctAncestors(): int |
|
116
|
|
|
{ |
|
117
|
|
|
return DB::table('maj_sosa') |
|
118
|
|
|
->where('majs_gedcom_id', '=', $this->tree->id()) |
|
119
|
|
|
->where('majs_user_id', '=', $this->user->id()) |
|
120
|
|
|
->distinct() |
|
121
|
|
|
->count('majs_i_id'); |
|
122
|
|
|
} |
|
123
|
|
|
|
|
124
|
|
|
/** |
|
125
|
|
|
* Get the mean generation time, as the slope of the linear regression of birth years vs generations |
|
126
|
|
|
* |
|
127
|
|
|
* @return float |
|
128
|
|
|
*/ |
|
129
|
|
|
public function meanGenerationTime(): float |
|
130
|
|
|
{ |
|
131
|
|
|
$row = DB::table('maj_sosa') |
|
132
|
|
|
->where('majs_gedcom_id', '=', $this->tree->id()) |
|
133
|
|
|
->where('majs_user_id', '=', $this->user->id()) |
|
134
|
|
|
->whereNotNull('majs_birth_year') |
|
135
|
|
|
->selectRaw('COUNT(majs_sosa) AS n') |
|
136
|
|
|
->selectRaw('SUM(majs_gen * majs_birth_year) AS sum_xy') |
|
137
|
|
|
->selectRaw('SUM(majs_gen) AS sum_x') |
|
138
|
|
|
->selectRaw('SUM(majs_birth_year) AS sum_y') |
|
139
|
|
|
->selectRaw('SUM(majs_gen * majs_gen) AS sum_x2') |
|
140
|
|
|
->get()->first(); |
|
141
|
|
|
|
|
142
|
|
|
return $row->n == 0 ? 0 : |
|
143
|
|
|
-($row->n * $row->sum_xy - $row->sum_x * $row->sum_y) / ($row->n * $row->sum_x2 - pow($row->sum_x, 2)); |
|
144
|
|
|
} |
|
145
|
|
|
|
|
146
|
|
|
/** |
|
147
|
|
|
* Get the statistic array detailed by generation. |
|
148
|
|
|
* Statistics for each generation are: |
|
149
|
|
|
* - The number of Sosa in generation |
|
150
|
|
|
* - The number of Sosa up to generation |
|
151
|
|
|
* - The number of distinct Sosa up to generation |
|
152
|
|
|
* - The year of the first birth in generation |
|
153
|
|
|
* - The year of the first estimated birth in generation |
|
154
|
|
|
* - The year of the last birth in generation |
|
155
|
|
|
* - The year of the last estimated birth in generation |
|
156
|
|
|
* - The average year of birth in generation |
|
157
|
|
|
* |
|
158
|
|
|
* @return array<int, array<string, int|null>> Statistics array |
|
159
|
|
|
*/ |
|
160
|
|
|
public function statisticsByGenerations(): array |
|
161
|
|
|
{ |
|
162
|
|
|
$stats_by_gen = $this->statisticsByGenerationBasicData(); |
|
163
|
|
|
$cumul_stats_by_gen = $this->statisticsByGenerationCumulativeData(); |
|
164
|
|
|
|
|
165
|
|
|
$statistics_by_gen = []; |
|
166
|
|
|
foreach ($stats_by_gen as $gen => $stats_gen) { |
|
167
|
|
|
$statistics_by_gen[(int) $stats_gen->gen] = array( |
|
168
|
|
|
'sosaCount' => (int) $stats_gen->total_sosa, |
|
169
|
|
|
'sosaTotalCount' => (int) $cumul_stats_by_gen[$gen]->total_cumul, |
|
170
|
|
|
'diffSosaTotalCount' => (int) $cumul_stats_by_gen[$gen]->total_distinct_cumul, |
|
171
|
|
|
'firstBirth' => $stats_gen->first_year, |
|
172
|
|
|
'firstEstimatedBirth' => $stats_gen->first_est_year, |
|
173
|
|
|
'lastBirth' => $stats_gen->last_year, |
|
174
|
|
|
'lastEstimatedBirth' => $stats_gen->last_est_year |
|
175
|
|
|
); |
|
176
|
|
|
} |
|
177
|
|
|
|
|
178
|
|
|
return $statistics_by_gen; |
|
179
|
|
|
} |
|
180
|
|
|
|
|
181
|
|
|
/** |
|
182
|
|
|
* Returns the basic statistics data by generation. |
|
183
|
|
|
* |
|
184
|
|
|
* @return Collection<int, \stdClass> |
|
185
|
|
|
*/ |
|
186
|
|
|
private function statisticsByGenerationBasicData(): Collection |
|
187
|
|
|
{ |
|
188
|
|
|
return DB::table('maj_sosa') |
|
189
|
|
|
->where('majs_gedcom_id', '=', $this->tree->id()) |
|
190
|
|
|
->where('majs_user_id', '=', $this->user->id()) |
|
191
|
|
|
->groupBy('majs_gen') |
|
192
|
|
|
->orderBy('majs_gen', 'asc') |
|
193
|
|
|
->select('majs_gen AS gen') |
|
194
|
|
|
->selectRaw('COUNT(majs_sosa) AS total_sosa') |
|
195
|
|
|
->selectRaw('MIN(majs_birth_year) AS first_year') |
|
196
|
|
|
->selectRaw('MIN(majs_birth_year_est) AS first_est_year') |
|
197
|
|
|
->selectRaw('MAX(majs_birth_year) AS last_year') |
|
198
|
|
|
->selectRaw('MAX(majs_birth_year_est) AS last_est_year') |
|
199
|
|
|
->get()->keyBy('gen'); |
|
200
|
|
|
} |
|
201
|
|
|
|
|
202
|
|
|
/** |
|
203
|
|
|
* Returns the cumulative statistics data by generation |
|
204
|
|
|
* |
|
205
|
|
|
* @return Collection<int, \stdClass> |
|
206
|
|
|
*/ |
|
207
|
|
|
private function statisticsByGenerationCumulativeData(): Collection |
|
208
|
|
|
{ |
|
209
|
|
|
$list_gen = DB::table('maj_sosa')->select('majs_gen')->distinct() |
|
210
|
|
|
->where('majs_gedcom_id', '=', $this->tree->id()) |
|
211
|
|
|
->where('majs_user_id', '=', $this->user->id()); |
|
212
|
|
|
|
|
213
|
|
|
return DB::table('maj_sosa') |
|
214
|
|
|
->joinSub($list_gen, 'list_gen', function (JoinClause $join): void { |
|
215
|
|
|
$join->on('maj_sosa.majs_gen', '<=', 'list_gen.majs_gen') |
|
216
|
|
|
->where('majs_gedcom_id', '=', $this->tree->id()) |
|
217
|
|
|
->where('majs_user_id', '=', $this->user->id()); |
|
218
|
|
|
}) |
|
219
|
|
|
->groupBy('list_gen.majs_gen') |
|
220
|
|
|
->select('list_gen.majs_gen AS gen') |
|
221
|
|
|
->selectRaw('COUNT(majs_i_id) AS total_cumul') |
|
222
|
|
|
->selectRaw('COUNT(DISTINCT majs_i_id) AS total_distinct_cumul') |
|
223
|
|
|
->selectRaw('1 - COUNT(DISTINCT majs_i_id) / COUNT(majs_i_id) AS pedi_collapse_simple') |
|
224
|
|
|
->get()->keyBy('gen'); |
|
225
|
|
|
} |
|
226
|
|
|
|
|
227
|
|
|
/** |
|
228
|
|
|
* Returns the pedigree collapse improved calculation by generation. |
|
229
|
|
|
* |
|
230
|
|
|
* Format: |
|
231
|
|
|
* - key : generation |
|
232
|
|
|
* - values: |
|
233
|
|
|
* - pedi_collapse_roots : pedigree collapse of ancestor roots for the generation |
|
234
|
|
|
* - pedi_collapse_xgen : pedigree cross-generation shrinkage for the generation |
|
235
|
|
|
* |
|
236
|
|
|
* @return array<int, array<string, float>> |
|
237
|
|
|
*/ |
|
238
|
|
|
public function pedigreeCollapseByGenerationData(): array |
|
239
|
|
|
{ |
|
240
|
|
|
$table_prefix = DB::connection()->getTablePrefix(); |
|
241
|
|
|
|
|
242
|
|
|
$list_gen = DB::table('maj_sosa')->select('majs_gen')->distinct() |
|
243
|
|
|
->where('majs_gedcom_id', '=', $this->tree->id()) |
|
244
|
|
|
->where('majs_user_id', '=', $this->user->id()); |
|
245
|
|
|
|
|
246
|
|
|
/* Compute the contributions of nodes of previous generations to the current generation */ |
|
247
|
|
|
$root_ancestors_contributions = DB::table('maj_sosa AS sosa') |
|
248
|
|
|
->select(['list_gen.majs_gen AS gen', 'sosa.majs_gedcom_id', 'sosa.majs_user_id']) |
|
249
|
|
|
->addSelect(['sosa.majs_i_id', 'sosa.majs_gen']) |
|
250
|
|
|
->selectRaw( |
|
251
|
|
|
'(CASE ' . |
|
252
|
|
|
' WHEN ' . $table_prefix . 'sosa_fat.majs_i_id IS NULL' . |
|
253
|
|
|
' THEN POWER(2, ' . $table_prefix . 'list_gen.majs_gen - ' . $table_prefix . 'sosa.majs_gen - 1)' . |
|
254
|
|
|
' ELSE 0 ' . |
|
255
|
|
|
' END)' . |
|
256
|
|
|
' + (CASE ' . |
|
257
|
|
|
' WHEN ' . $table_prefix . 'sosa_mot.majs_i_id IS NULL' . |
|
258
|
|
|
' THEN POWER(2, ' . $table_prefix . 'list_gen.majs_gen - ' . $table_prefix . 'sosa.majs_gen - 1)' . |
|
259
|
|
|
' ELSE 0 ' . |
|
260
|
|
|
' END) contrib' |
|
261
|
|
|
) |
|
262
|
|
|
->joinSub($list_gen, 'list_gen', function (JoinClause $join): void { |
|
263
|
|
|
$join->on('sosa.majs_gen', '<', 'list_gen.majs_gen') |
|
264
|
|
|
->where('majs_gedcom_id', '=', $this->tree->id()) |
|
265
|
|
|
->where('majs_user_id', '=', $this->user->id()); |
|
266
|
|
|
}) |
|
267
|
|
|
->leftJoin('maj_sosa AS sosa_fat', function (JoinClause $join) use ($table_prefix): void { |
|
268
|
|
|
// Link to sosa's father |
|
269
|
|
|
$join->whereRaw($table_prefix . 'sosa_fat.majs_sosa = 2 * ' . $table_prefix . 'sosa.majs_sosa') |
|
270
|
|
|
->where('sosa_fat.majs_gedcom_id', '=', $this->tree->id()) |
|
271
|
|
|
->where('sosa_fat.majs_user_id', '=', $this->user->id()); |
|
272
|
|
|
}) |
|
273
|
|
|
->leftJoin('maj_sosa AS sosa_mot', function (JoinClause $join) use ($table_prefix): void { |
|
274
|
|
|
// Link to sosa's mother |
|
275
|
|
|
$join->whereRaw($table_prefix . 'sosa_mot.majs_sosa = 2 * ' . $table_prefix . 'sosa.majs_sosa + 1') |
|
276
|
|
|
->where('sosa_mot.majs_gedcom_id', '=', $this->tree->id()) |
|
277
|
|
|
->where('sosa_mot.majs_user_id', '=', $this->user->id()); |
|
278
|
|
|
}) |
|
279
|
|
|
->where('sosa.majs_gedcom_id', '=', $this->tree->id()) |
|
280
|
|
|
->where('sosa.majs_user_id', '=', $this->user->id()) |
|
281
|
|
|
->where(function (Builder $query): void { |
|
282
|
|
|
$query->whereNull('sosa_fat.majs_i_id') |
|
283
|
|
|
->orWhereNull('sosa_mot.majs_i_id'); |
|
284
|
|
|
}); |
|
285
|
|
|
|
|
286
|
|
|
/* Identify nodes in the generations with ancestors who are also in the same generation. |
|
287
|
|
|
* This is the vertical/generational collapse that will reduce the number or roots. |
|
288
|
|
|
*/ |
|
289
|
|
|
$non_roots_ancestors = DB::table('maj_sosa AS sosa') |
|
290
|
|
|
->select(['sosa.majs_gen', 'sosa.majs_gedcom_id', 'sosa.majs_user_id', 'sosa.majs_sosa']) |
|
291
|
|
|
->selectRaw('MAX(' . $table_prefix . 'sosa_anc.majs_sosa) - MIN(' . $table_prefix . 'sosa_anc.majs_sosa)' . |
|
292
|
|
|
' AS full_ancestors') |
|
293
|
|
|
->join('maj_sosa AS sosa_anc', function (JoinClause $join) use ($table_prefix): void { |
|
294
|
|
|
$join->on('sosa.majs_gen', '<', 'sosa_anc.majs_gen') |
|
295
|
|
|
->whereRaw('FLOOR(' . $table_prefix . 'sosa_anc.majs_sosa / POWER(2, ' . |
|
296
|
|
|
$table_prefix . 'sosa_anc.majs_gen - ' . $table_prefix . 'sosa.majs_gen)) = ' . |
|
297
|
|
|
$table_prefix . 'sosa.majs_sosa') |
|
298
|
|
|
->where('sosa_anc.majs_gedcom_id', '=', $this->tree->id()) |
|
299
|
|
|
->where('sosa_anc.majs_user_id', '=', $this->user->id()); |
|
300
|
|
|
}) |
|
301
|
|
|
->where('sosa.majs_gedcom_id', '=', $this->tree->id()) |
|
302
|
|
|
->where('sosa.majs_user_id', '=', $this->user->id()) |
|
303
|
|
|
->whereIn('sosa_anc.majs_i_id', function (Builder $query) use ($table_prefix): void { |
|
304
|
|
|
$query->from('maj_sosa AS sosa_gen') |
|
305
|
|
|
->select('sosa_gen.majs_i_id')->distinct() |
|
306
|
|
|
->where('sosa_gen.majs_gedcom_id', '=', $this->tree->id()) |
|
307
|
|
|
->where('sosa_gen.majs_user_id', '=', $this->user->id()) |
|
308
|
|
|
->whereRaw($table_prefix . 'sosa_gen.majs_gen = ' . $table_prefix . 'sosa.majs_gen'); |
|
309
|
|
|
}) |
|
310
|
|
|
->groupBy(['sosa.majs_gen', 'sosa.majs_gedcom_id', 'sosa.majs_user_id', |
|
311
|
|
|
'sosa.majs_sosa', 'sosa.majs_i_id']); |
|
312
|
|
|
|
|
313
|
|
|
/* Compute the contribution of the nodes in the generation, |
|
314
|
|
|
* excluding the nodes with ancestors in the same generation. |
|
315
|
|
|
* Nodes with a parent missing are not excluded to cater for the missing one. |
|
316
|
|
|
*/ |
|
317
|
|
|
$known_ancestors_contributions = DB::table('maj_sosa AS sosa') |
|
318
|
|
|
->select(['sosa.majs_gen AS gen', 'sosa.majs_gedcom_id', 'sosa.majs_user_id']) |
|
319
|
|
|
->addSelect(['sosa.majs_i_id', 'sosa.majs_gen']) |
|
320
|
|
|
->selectRaw('1 AS contrib') |
|
321
|
|
|
->leftJoinSub($non_roots_ancestors, 'nonroot', function (JoinClause $join): void { |
|
322
|
|
|
$join->on('sosa.majs_gen', '=', 'nonroot.majs_gen') |
|
323
|
|
|
->on('sosa.majs_sosa', '=', 'nonroot.majs_sosa') |
|
324
|
|
|
->where('nonroot.full_ancestors', '>', 0) |
|
325
|
|
|
->where('nonroot.majs_gedcom_id', '=', $this->tree->id()) |
|
326
|
|
|
->where('nonroot.majs_user_id', '=', $this->user->id()); |
|
327
|
|
|
}) |
|
328
|
|
|
->where('sosa.majs_gedcom_id', '=', $this->tree->id()) |
|
329
|
|
|
->where('sosa.majs_user_id', '=', $this->user->id()) |
|
330
|
|
|
->whereNull('nonroot.majs_sosa'); |
|
331
|
|
|
|
|
332
|
|
|
/* Aggregate both queries, and calculate the sum of contributions by generation roots. |
|
333
|
|
|
* Exclude as well nodes that already appear in lower generations, as their branche has already been reduced. |
|
334
|
|
|
*/ |
|
335
|
|
|
$ancestors_contributions_sum = DB::connection()->query() |
|
336
|
|
|
->fromSub($root_ancestors_contributions->unionAll($known_ancestors_contributions), 'sosa_contribs') |
|
337
|
|
|
->select(['sosa_contribs.gen', 'sosa_contribs.majs_gedcom_id', 'sosa_contribs.majs_user_id']) |
|
338
|
|
|
->addSelect(['sosa_contribs.majs_i_id', 'sosa_contribs.contrib']) |
|
339
|
|
|
->selectRaw('COUNT(' . $table_prefix . 'sosa_contribs.majs_i_id) * ' . |
|
340
|
|
|
$table_prefix . 'sosa_contribs.contrib AS totalContrib') |
|
341
|
|
|
->leftJoin('maj_sosa AS sosa_low', function (JoinClause $join): void { |
|
342
|
|
|
$join->on('sosa_low.majs_gen', '<', 'sosa_contribs.majs_gen') |
|
343
|
|
|
->on('sosa_low.majs_i_id', '=', 'sosa_contribs.majs_i_id') |
|
344
|
|
|
->where('sosa_low.majs_gedcom_id', '=', $this->tree->id()) |
|
345
|
|
|
->where('sosa_low.majs_user_id', '=', $this->user->id()); |
|
346
|
|
|
}) |
|
347
|
|
|
->whereNull('sosa_low.majs_sosa') |
|
348
|
|
|
->groupBy(['sosa_contribs.gen', 'sosa_contribs.majs_gedcom_id', 'sosa_contribs.majs_user_id', |
|
349
|
|
|
'sosa_contribs.majs_i_id', 'sosa_contribs.contrib']); |
|
350
|
|
|
|
|
351
|
|
|
// Aggregate all generation roots to compute root and generation pedigree collapse |
|
352
|
|
|
$pedi_collapse_coll = DB::connection()->query()->fromSub($ancestors_contributions_sum, 'sosa_contribs_sum') |
|
353
|
|
|
->select(['gen'])->selectRaw('SUM(contrib), SUM(totalContrib)') |
|
354
|
|
|
->selectRaw('1 - SUM(contrib) / SUM(totalContrib) AS pedi_collapse_roots') // Roots/horizontal collapse |
|
355
|
|
|
->selectRaw('1 - SUM(totalContrib) / POWER ( 2, gen - 1) AS pedi_collapse_xgen') // Crossgeneration collapse |
|
356
|
|
|
->groupBy(['gen', 'majs_gedcom_id', 'majs_user_id']) |
|
357
|
|
|
->orderBy('gen') |
|
358
|
|
|
->get(); |
|
359
|
|
|
|
|
360
|
|
|
$pedi_collapse_by_gen = []; |
|
361
|
|
|
foreach ($pedi_collapse_coll as $collapse_gen) { |
|
362
|
|
|
$pedi_collapse_by_gen[(int) $collapse_gen->gen] = array( |
|
363
|
|
|
'pedi_collapse_roots' => (float) $collapse_gen->pedi_collapse_roots, |
|
364
|
|
|
'pedi_collapse_xgen' => (float) $collapse_gen->pedi_collapse_xgen |
|
365
|
|
|
); |
|
366
|
|
|
} |
|
367
|
|
|
return $pedi_collapse_by_gen; |
|
368
|
|
|
} |
|
369
|
|
|
|
|
370
|
|
|
/** |
|
371
|
|
|
* Return a Collection of the mean generation depth and deviation for all Sosa ancestors at a given generation. |
|
372
|
|
|
* Sosa 1 is of generation 1. |
|
373
|
|
|
* |
|
374
|
|
|
* Mean generation depth and deviation are calculated based on the works of Marie-Héléne Cazes and Pierre Cazes, |
|
375
|
|
|
* published in Population (French Edition), Vol. 51, No. 1 (Jan. - Feb., 1996), pp. 117-140 |
|
376
|
|
|
* http://kintip.net/index.php?option=com_jdownloads&task=download.send&id=9&catid=4&m=0 |
|
377
|
|
|
* |
|
378
|
|
|
* Format: |
|
379
|
|
|
* - key : sosa number of the ancestor |
|
380
|
|
|
* - values: |
|
381
|
|
|
* - root_ancestor_id : ID of the ancestor |
|
382
|
|
|
* - mean_gen_depth : Mean generation depth |
|
383
|
|
|
* - stddev_gen_depth : Standard deviation of generation depth |
|
384
|
|
|
* |
|
385
|
|
|
* @param int $gen Sosa generation |
|
386
|
|
|
* @return Collection<int, \stdClass> |
|
387
|
|
|
*/ |
|
388
|
|
|
public function generationDepthStatsAtGeneration(int $gen): Collection |
|
389
|
|
|
{ |
|
390
|
|
|
$table_prefix = DB::connection()->getTablePrefix(); |
|
391
|
|
|
$missing_ancestors_by_gen = DB::table('maj_sosa AS sosa') |
|
392
|
|
|
->selectRaw($table_prefix . 'sosa.majs_gen - ? AS majs_gen_norm', [$gen]) |
|
393
|
|
|
->selectRaw('FLOOR(((' . $table_prefix . 'sosa.majs_sosa / POW(2, ' . $table_prefix . 'sosa.majs_gen -1 )) - 1) * POWER(2, ? - 1)) + POWER(2, ? - 1) AS root_ancestor', [$gen, $gen]) //@phpcs:ignore Generic.Files.LineLength.TooLong |
|
394
|
|
|
->selectRaw('SUM(CASE WHEN ' . $table_prefix . 'sosa_fat.majs_i_id IS NULL AND ' . $table_prefix . 'sosa_mot.majs_i_id IS NULL THEN 1 ELSE 0 END) AS full_root_count') //@phpcs:ignore Generic.Files.LineLength.TooLong |
|
395
|
|
|
->selectRaw('SUM(CASE WHEN ' . $table_prefix . 'sosa_fat.majs_i_id IS NULL AND ' . $table_prefix . 'sosa_mot.majs_i_id IS NULL THEN 0 ELSE 1 END) As semi_root_count') //@phpcs:ignore Generic.Files.LineLength.TooLong |
|
396
|
|
|
->leftJoin('maj_sosa AS sosa_fat', function (JoinClause $join) use ($table_prefix): void { |
|
397
|
|
|
// Link to sosa's father |
|
398
|
|
|
$join->whereRaw($table_prefix . 'sosa_fat.majs_sosa = 2 * ' . $table_prefix . 'sosa.majs_sosa') |
|
399
|
|
|
->where('sosa_fat.majs_gedcom_id', '=', $this->tree->id()) |
|
400
|
|
|
->where('sosa_fat.majs_user_id', '=', $this->user->id()); |
|
401
|
|
|
}) |
|
402
|
|
|
->leftJoin('maj_sosa AS sosa_mot', function (JoinClause $join) use ($table_prefix): void { |
|
403
|
|
|
// Link to sosa's mother |
|
404
|
|
|
$join->whereRaw($table_prefix . 'sosa_mot.majs_sosa = 2 * ' . $table_prefix . 'sosa.majs_sosa + 1') |
|
405
|
|
|
->where('sosa_mot.majs_gedcom_id', '=', $this->tree->id()) |
|
406
|
|
|
->where('sosa_mot.majs_user_id', '=', $this->user->id()); |
|
407
|
|
|
}) |
|
408
|
|
|
->where('sosa.majs_gedcom_id', '=', $this->tree->id()) |
|
409
|
|
|
->where('sosa.majs_user_id', '=', $this->user->id()) |
|
410
|
|
|
->where('sosa.majs_gen', '>=', $gen) |
|
411
|
|
|
->where(function (Builder $query): void { |
|
412
|
|
|
$query->whereNull('sosa_fat.majs_i_id') |
|
413
|
|
|
->orWhereNull('sosa_mot.majs_i_id'); |
|
414
|
|
|
}) |
|
415
|
|
|
->groupBy(['sosa.majs_gen', 'root_ancestor']); |
|
416
|
|
|
|
|
417
|
|
|
return DB::table('maj_sosa AS sosa_list') |
|
418
|
|
|
->select(['stats_by_gen.root_ancestor AS root_ancestor_sosa', 'sosa_list.majs_i_id as root_ancestor_id']) |
|
419
|
|
|
->selectRaw('1 + SUM( (majs_gen_norm) * ( 2 * full_root_count + semi_root_count) / (2 * POWER(2, majs_gen_norm))) AS mean_gen_depth') //@phpcs:ignore Generic.Files.LineLength.TooLong |
|
420
|
|
|
->selectRaw(' SQRT(' . |
|
421
|
|
|
' SUM(POWER(majs_gen_norm, 2) * ( 2 * full_root_count + semi_root_count) / (2 * POWER(2, majs_gen_norm)))' . //@phpcs:ignore Generic.Files.LineLength.TooLong |
|
422
|
|
|
' - POWER( SUM( (majs_gen_norm) * ( 2 * full_root_count + semi_root_count) / (2 * POWER(2, majs_gen_norm))), 2)' . //@phpcs:ignore Generic.Files.LineLength.TooLong |
|
423
|
|
|
' ) AS stddev_gen_depth') |
|
424
|
|
|
->joinSub($missing_ancestors_by_gen, 'stats_by_gen', function (JoinClause $join): void { |
|
425
|
|
|
$join->on('sosa_list.majs_sosa', '=', 'stats_by_gen.root_ancestor') |
|
426
|
|
|
->where('sosa_list.majs_gedcom_id', '=', $this->tree->id()) |
|
427
|
|
|
->where('sosa_list.majs_user_id', '=', $this->user->id()); |
|
428
|
|
|
}) |
|
429
|
|
|
->groupBy(['stats_by_gen.root_ancestor', 'sosa_list.majs_i_id']) |
|
430
|
|
|
->orderBy('stats_by_gen.root_ancestor') |
|
431
|
|
|
->get()->keyBy('root_ancestor_sosa'); |
|
432
|
|
|
} |
|
433
|
|
|
|
|
434
|
|
|
/** |
|
435
|
|
|
* Return a collection of the most duplicated root Sosa ancestors. |
|
436
|
|
|
* The number of ancestors to return is limited by the parameter $limit. |
|
437
|
|
|
* If several individuals are tied when reaching the limit, none of them are returned, |
|
438
|
|
|
* which means that there can be less individuals returned than requested. |
|
439
|
|
|
* |
|
440
|
|
|
* Format: |
|
441
|
|
|
* - value: |
|
442
|
|
|
* - sosa_i_id : sosa individual |
|
443
|
|
|
* - sosa_count: number of duplications of the ancestor (e.g. 3 if it appears 3 times) |
|
444
|
|
|
* |
|
445
|
|
|
* @param int $limit |
|
446
|
|
|
* @return Collection<\stdClass> |
|
447
|
|
|
*/ |
|
448
|
|
|
public function topMultipleAncestorsWithNoTies(int $limit): Collection |
|
449
|
|
|
{ |
|
450
|
|
|
$table_prefix = DB::connection()->getTablePrefix(); |
|
451
|
|
|
$multiple_ancestors = DB::table('maj_sosa AS sosa') |
|
452
|
|
|
->select('sosa.majs_i_id AS sosa_i_id') |
|
453
|
|
|
->selectRaw('COUNT(' . $table_prefix . 'sosa.majs_sosa) AS sosa_count') |
|
454
|
|
|
->leftJoin('maj_sosa AS sosa_fat', function (JoinClause $join) use ($table_prefix): void { |
|
455
|
|
|
// Link to sosa's father |
|
456
|
|
|
$join->whereRaw($table_prefix . 'sosa_fat.majs_sosa = 2 * ' . $table_prefix . 'sosa.majs_sosa') |
|
457
|
|
|
->where('sosa_fat.majs_gedcom_id', '=', $this->tree->id()) |
|
458
|
|
|
->where('sosa_fat.majs_user_id', '=', $this->user->id()); |
|
459
|
|
|
}) |
|
460
|
|
|
->leftJoin('maj_sosa AS sosa_mot', function (JoinClause $join) use ($table_prefix): void { |
|
461
|
|
|
// Link to sosa's mother |
|
462
|
|
|
$join->whereRaw($table_prefix . 'sosa_mot.majs_sosa = 2 * ' . $table_prefix . 'sosa.majs_sosa + 1') |
|
463
|
|
|
->where('sosa_mot.majs_gedcom_id', '=', $this->tree->id()) |
|
464
|
|
|
->where('sosa_mot.majs_user_id', '=', $this->user->id()); |
|
465
|
|
|
}) |
|
466
|
|
|
->where('sosa.majs_gedcom_id', '=', $this->tree->id()) |
|
467
|
|
|
->where('sosa.majs_user_id', '=', $this->user->id()) |
|
468
|
|
|
->whereNull('sosa_fat.majs_sosa') // We keep only root individuals, i.e. those with no father or mother |
|
469
|
|
|
->whereNull('sosa_mot.majs_sosa') |
|
470
|
|
|
->groupBy('sosa.majs_i_id') |
|
471
|
|
|
->havingRaw('COUNT(' . $table_prefix . 'sosa.majs_sosa) > 1') // Limit to the duplicate sosas. |
|
472
|
|
|
->orderByRaw('COUNT(' . $table_prefix . 'sosa.majs_sosa) DESC, MIN(' . $table_prefix . 'sosa.majs_sosa) ASC') //@phpcs:ignore Generic.Files.LineLength.TooLong |
|
473
|
|
|
->limit($limit + 1) // We want to select one more than required, for ties |
|
474
|
|
|
->get(); |
|
475
|
|
|
|
|
476
|
|
|
if ($multiple_ancestors->count() > $limit) { |
|
477
|
|
|
$last_count = $multiple_ancestors->last()->sosa_count; |
|
478
|
|
|
$multiple_ancestors = $multiple_ancestors->reject( |
|
479
|
|
|
fn (stdClass $element): bool => $element->sosa_count === $last_count |
|
480
|
|
|
); |
|
481
|
|
|
} |
|
482
|
|
|
return $multiple_ancestors; |
|
483
|
|
|
} |
|
484
|
|
|
|
|
485
|
|
|
/** |
|
486
|
|
|
* Return a computed array of statistics about the dispersion of ancestors across the ancestors |
|
487
|
|
|
* at a specified generation. |
|
488
|
|
|
* |
|
489
|
|
|
* Format: |
|
490
|
|
|
* - key : rank of the ancestor in generation G for which exclusive ancestors have been found |
|
491
|
|
|
* For instance 3 represent the maternal grand father |
|
492
|
|
|
* 0 is used for shared ancestors |
|
493
|
|
|
* - values: |
|
494
|
|
|
* - branches: same as key |
|
495
|
|
|
* - majs_i_id: xref of the ancestor at rank key in generation G, or null for shared ancestors |
|
496
|
|
|
* - count_indi: number of ancestors exclusively in the ancestors of the ancestor at rank key |
|
497
|
|
|
* |
|
498
|
|
|
* For instance a result at generation 3 could be : |
|
499
|
|
|
* [ |
|
500
|
|
|
* 0 => { branches: 0, majs_i_id: X1, count_indi: 12 } -> 12 ancestors are shared by the grand-parents |
|
501
|
|
|
* 1 => { branches: 1, majs_i_id: X2, count_indi: 32 } -> 32 ancestors are exclusive to the paternal grand-father |
|
502
|
|
|
* 2 => { branches: 2, majs_i_id: X3, count_indi: 25 } -> 25 ancestors are exclusive to the paternal grand-mother |
|
503
|
|
|
* 3 => { branches: 3, majs_i_id: X4, count_indi: 12 } -> 12 ancestors are exclusive to the maternal grand-father |
|
504
|
|
|
* 4 => { branches: 4, majs_i_id: X5, count_indi: 30 } -> 30 ancestors are exclusive to the maternal grand-mother |
|
505
|
|
|
* ] |
|
506
|
|
|
* |
|
507
|
|
|
* @param int $gen |
|
508
|
|
|
* @return Collection<int, \stdClass> |
|
509
|
|
|
*/ |
|
510
|
|
|
public function ancestorsDispersionForGeneration(int $gen): Collection |
|
511
|
|
|
{ |
|
512
|
|
|
$ancestors_branches = DB::table('maj_sosa') |
|
513
|
|
|
->select('majs_i_id AS i_id') |
|
514
|
|
|
->selectRaw('FLOOR(majs_sosa / POW(2, (majs_gen - ?))) - POW(2, ? -1) + 1 AS branch', [$gen, $gen]) |
|
515
|
|
|
->where('majs_gedcom_id', '=', $this->tree->id()) |
|
516
|
|
|
->where('majs_user_id', '=', $this->user->id()) |
|
517
|
|
|
->where('majs_gen', '>=', $gen) |
|
518
|
|
|
->groupBy('majs_i_id', 'branch'); |
|
519
|
|
|
|
|
520
|
|
|
$consolidated_ancestors_branches = DB::table('maj_sosa') |
|
521
|
|
|
->fromSub($ancestors_branches, 'indi_branch') |
|
522
|
|
|
->select('i_id') |
|
523
|
|
|
->selectRaw('CASE WHEN COUNT(branch) > 1 THEN 0 ELSE MIN(branch) END AS branches') |
|
524
|
|
|
->groupBy('i_id'); |
|
525
|
|
|
|
|
526
|
|
|
return DB::table('maj_sosa') |
|
527
|
|
|
->rightJoinSub( |
|
528
|
|
|
$consolidated_ancestors_branches, |
|
529
|
|
|
'indi_branch_consolidated', |
|
530
|
|
|
function (JoinClause $join) use ($gen): void { |
|
531
|
|
|
$join->where('maj_sosa.majs_gedcom_id', '=', $this->tree->id()) |
|
532
|
|
|
->where('maj_sosa.majs_user_id', '=', $this->user->id()) |
|
533
|
|
|
->where('branches', '>', 0) |
|
534
|
|
|
->whereRaw('majs_sosa = POW(2, ? - 1) + branches - 1', [$gen]); |
|
535
|
|
|
} |
|
536
|
|
|
) |
|
537
|
|
|
->select(['branches', 'majs_i_id']) |
|
538
|
|
|
->selectRaw('COUNT(i_id) AS count_indi') |
|
539
|
|
|
->groupBy(['branches', 'majs_i_id']) |
|
540
|
|
|
->get()->keyBy('branches'); |
|
541
|
|
|
} |
|
542
|
|
|
} |
|
543
|
|
|
|
jon48 /
webtrees-lib
