oscillatory.net
this repo has no description
WIP: bsky post UF data structure
+532
-65
+532
-65
indexer/src/storage/mod.rs
+532
-65
indexer/src/storage/mod.rs
···
1
-
use std::collections::HashMap;
1
+
use std::{collections::HashMap, hash::Hash};
2
+
3
+
// pub mod memory;
4
+
5
+
6
+
#[derive(Clone, Debug, Eq, Hash, PartialEq)]
7
+
struct BskyPostId {
8
+
did: String,
9
+
rkey: String,
10
+
}
11
+
12
+
impl BskyPostId {
13
+
fn from(did: &str, rkey: &str) -> BskyPostId {
14
+
BskyPostId { did: String::from(did), rkey: String::from(rkey) }
15
+
}
16
+
}
2
17
3
-
pub mod memory;
18
+
#[derive(Clone, Copy, Debug, Eq, PartialEq)]
19
+
enum SubgraphType {
20
+
Reply,
21
+
Quote,
22
+
ReplyQuote,
23
+
}
24
+
25
+
26
+
#[derive(Clone, Eq, Hash, PartialEq)]
27
+
struct BskyPostReplyTo<Id> {
28
+
target: Id,
29
+
root: Id,
30
+
}
31
+
32
+
#[derive(Clone, Eq, Hash, PartialEq)]
33
+
struct BskyPostRecord<Id> {
34
+
id: Id,
35
+
reply_to: Option<BskyPostReplyTo<Id>>,
36
+
quote_of: Option<Id>,
37
+
}
4
38
5
39
type UFIndex = u32;
6
40
type UFSize = u32;
7
41
8
-
struct UnionFind {
9
-
parents: Vec<UFIndex>,
10
-
sizes: HashMap<UFIndex, UFSize>,
42
+
#[derive(Debug)]
43
+
struct BskyPostUnionFind<Id> {
44
+
id_to_index: HashMap<Id, UFIndex>,
45
+
parents_ro: HashMap<UFIndex, UFIndex>,
46
+
parents_qo: HashMap<UFIndex, UFIndex>,
47
+
parents_rq: HashMap<UFIndex, UFIndex>,
48
+
sizes_ro: HashMap<UFIndex, UFSize>,
49
+
sizes_qo: HashMap<UFIndex, UFSize>,
50
+
sizes_rq: HashMap<UFIndex, UFSize>,
11
51
next_index: UFIndex,
12
52
}
13
53
14
-
impl UnionFind {
15
-
pub fn new() -> UnionFind {
16
-
UnionFind {
17
-
parents: Vec::new(),
18
-
sizes: HashMap::new(),
54
+
impl<Id: Clone + Eq + Hash> BskyPostUnionFind<Id> {
55
+
pub fn new() -> BskyPostUnionFind<Id> {
56
+
BskyPostUnionFind {
57
+
id_to_index: HashMap::new(),
58
+
parents_ro: HashMap::new(),
59
+
parents_qo: HashMap::new(),
60
+
parents_rq: HashMap::new(),
61
+
sizes_ro: HashMap::new(),
62
+
sizes_qo: HashMap::new(),
63
+
sizes_rq: HashMap::new(),
19
64
next_index: 0,
20
65
}
21
66
}
22
67
23
-
pub fn add(&mut self) -> UFIndex {
24
-
let idx = self.next_index;
25
-
self.parents.push(idx);
26
-
self.sizes.insert(idx, 1);
27
-
self.next_index += 1;
28
-
idx
68
+
fn get_index(&mut self, id: &Id) -> UFIndex {
69
+
match self.id_to_index.get(id) {
70
+
None => {
71
+
let idx = self.next_index;
72
+
self.id_to_index.insert(id.clone(), idx);
73
+
self.next_index += 1;
74
+
idx
75
+
},
76
+
Some(idx) => *idx,
77
+
}
78
+
}
79
+
80
+
pub fn ingest_post(&mut self, record: BskyPostRecord<Id>) -> bool {
81
+
if record.reply_to.is_none() && record.quote_of.is_none() {
82
+
return false;
83
+
}
84
+
let post_idx = self.get_index(&record.id);
85
+
86
+
if let Some(ref reply_to) = record.reply_to {
87
+
let root_idx = self.get_index(&reply_to.root);
88
+
let parent_idx = self.get_index(&reply_to.target);
89
+
90
+
// assuming that root has a parents_ro entry iff it has a sizes_ro entry.
91
+
// this is *only true* iff the root of the reply tree is always the UF root.
92
+
// but by construction it is, since we never have to call union() or find()
93
+
// for a reply tree.
94
+
95
+
let root_ro_size = match self.parents_ro.insert(root_idx, root_idx) {
96
+
None => 1,
97
+
Some(_) => *self.sizes_ro.get(&root_idx).unwrap(),
98
+
};
99
+
let mut new_root_ro_size = root_ro_size;
100
+
match self.parents_ro.insert(parent_idx, root_idx) {
101
+
None => {
102
+
new_root_ro_size += 1;
103
+
},
104
+
Some(_) => (),
105
+
}
106
+
match self.parents_ro.insert(post_idx, root_idx) {
107
+
None => {
108
+
new_root_ro_size += 1;
109
+
},
110
+
Some(_) => (),
111
+
}
112
+
113
+
if new_root_ro_size > root_ro_size {
114
+
self.sizes_ro.insert(root_idx, new_root_ro_size);
115
+
}
116
+
117
+
let mut rq_to_add = Vec::new();
118
+
if !self.parents_rq.contains_key(&post_idx) {
119
+
rq_to_add.push(post_idx);
120
+
}
121
+
if !self.parents_rq.contains_key(&root_idx) {
122
+
rq_to_add.push(root_idx);
123
+
}
124
+
if parent_idx != root_idx && !self.parents_rq.contains_key(&parent_idx) {
125
+
rq_to_add.push(parent_idx);
126
+
}
127
+
128
+
for idx in rq_to_add {
129
+
self.parents_rq.insert(idx, idx);
130
+
self.sizes_rq.insert(idx, 1);
131
+
}
132
+
if parent_idx != root_idx {
133
+
self.union_rq(root_idx, parent_idx);
134
+
}
135
+
self.union_rq(parent_idx, post_idx);
136
+
}
137
+
138
+
if let Some(ref quote_of) = record.quote_of {
139
+
let parent_idx = self.get_index("e_of);
140
+
// assuming: if it's not in parents_qo, it's also not in sizes_qo
141
+
142
+
match (self.parents_qo.get(&parent_idx), self.parents_qo.get(&post_idx)) {
143
+
(None, None) => {
144
+
self.parents_qo.insert(parent_idx, parent_idx);
145
+
self.parents_qo.insert(post_idx, parent_idx);
146
+
self.sizes_qo.insert(parent_idx, 2);
147
+
},
148
+
(None, Some(post_parent)) => {
149
+
let post_parent_idx = *post_parent;
150
+
self.parents_qo.insert(parent_idx, parent_idx);
151
+
self.sizes_qo.insert(parent_idx, 1);
152
+
self.union_qo(parent_idx, post_parent_idx);
153
+
},
154
+
(Some(parent_parent), None) => {
155
+
let parent_root_idx = self.find_qo(*parent_parent);
156
+
self.parents_qo.insert(post_idx, parent_root_idx);
157
+
self.sizes_qo.entry(parent_root_idx).and_modify(|e| *e += 1);
158
+
},
159
+
(Some(parent_parent), Some(post_parent)) => {
160
+
self.union_qo(*parent_parent, *post_parent);
161
+
},
162
+
}
163
+
164
+
let mut rq_to_add = Vec::new();
165
+
if !self.parents_rq.contains_key(&post_idx) {
166
+
rq_to_add.push(post_idx);
167
+
}
168
+
if !self.parents_rq.contains_key(&parent_idx) {
169
+
rq_to_add.push(parent_idx);
170
+
}
171
+
for idx in rq_to_add {
172
+
self.parents_rq.insert(idx, idx);
173
+
self.sizes_rq.insert(idx, 1);
174
+
}
175
+
self.union_rq(parent_idx, post_idx);
176
+
177
+
}
178
+
179
+
true
29
180
}
30
181
31
182
// find the representative node for a given index
32
-
pub fn find(&mut self, idx: UFIndex) -> UFIndex {
183
+
fn find_qo(&mut self, idx: UFIndex) -> UFIndex {
184
+
let mut curr_idx = idx;
185
+
let mut pa_idx = *self.parents_qo.get(&curr_idx).unwrap();
186
+
while curr_idx != pa_idx {
187
+
curr_idx = pa_idx;
188
+
pa_idx = *self.parents_qo.get(&curr_idx).unwrap();
189
+
}
190
+
let rep = curr_idx;
191
+
192
+
curr_idx = idx;
193
+
pa_idx = *self.parents_qo.get(&curr_idx).unwrap();
194
+
while curr_idx != pa_idx {
195
+
self.parents_qo.insert(curr_idx, rep);
196
+
curr_idx = pa_idx;
197
+
pa_idx = *self.parents_qo.get(&curr_idx).unwrap();
198
+
}
199
+
curr_idx
200
+
}
201
+
202
+
fn union_qo(&mut self, idx1: UFIndex, idx2: UFIndex) {
203
+
let pa1 = self.find_qo(idx1);
204
+
let pa2 = self.find_qo(idx2);
205
+
206
+
if pa1 == pa2 {
207
+
return;
208
+
}
209
+
210
+
let pa1_size = self.sizes_qo.get(&pa1).unwrap();
211
+
let pa2_size = self.sizes_qo.get(&pa2).unwrap();
212
+
213
+
if pa1_size >= pa2_size {
214
+
self.parents_qo.insert(pa2, pa1);
215
+
self.sizes_qo.insert(pa1, pa1_size + pa2_size);
216
+
self.sizes_qo.remove(&pa2);
217
+
} else {
218
+
self.parents_qo.insert(pa1, pa2);
219
+
self.sizes_qo.insert(pa2, pa1_size + pa2_size);
220
+
self.sizes_qo.remove(&pa1);
221
+
}
222
+
}
223
+
224
+
fn find_rq(&mut self, idx: UFIndex) -> UFIndex {
33
225
let mut curr_idx = idx;
34
-
while curr_idx != self.parents[curr_idx as usize] {
35
-
curr_idx = self.parents[curr_idx as usize];
226
+
let mut pa_idx = *self.parents_rq.get(&curr_idx).unwrap();
227
+
while curr_idx != pa_idx {
228
+
curr_idx = pa_idx;
229
+
pa_idx = *self.parents_rq.get(&curr_idx).unwrap();
36
230
}
37
231
let rep = curr_idx;
38
232
39
233
curr_idx = idx;
40
-
while curr_idx != self.parents[curr_idx as usize] {
41
-
let curr_pa = self.parents[curr_idx as usize];
42
-
self.parents[curr_idx as usize] = rep;
43
-
curr_idx = curr_pa;
234
+
pa_idx = *self.parents_rq.get(&curr_idx).unwrap();
235
+
while curr_idx != pa_idx {
236
+
self.parents_rq.insert(curr_idx, rep);
237
+
curr_idx = pa_idx;
238
+
pa_idx = *self.parents_rq.get(&curr_idx).unwrap();
44
239
}
45
-
rep
240
+
curr_idx
46
241
}
47
242
48
-
pub fn union(&mut self, idx1: UFIndex, idx2: UFIndex) {
49
-
let pa1 = self.find(idx1);
50
-
let pa2 = self.find(idx2);
243
+
fn union_rq(&mut self, idx1: UFIndex, idx2: UFIndex) {
244
+
let pa1 = self.find_rq(idx1);
245
+
let pa2 = self.find_rq(idx2);
51
246
52
-
let pa1_size = self.sizes.get(&pa1).unwrap();
53
-
let pa2_size = self.sizes.get(&pa2).unwrap();
247
+
if pa1 == pa2 {
248
+
return;
249
+
}
54
250
55
-
if pa1_size > pa2_size {
56
-
self.parents[pa2 as usize] = pa1;
57
-
self.sizes.insert(pa1, pa1_size + pa2_size);
58
-
self.sizes.remove(&pa2);
251
+
let pa1_size = self.sizes_rq.get(&pa1).unwrap();
252
+
let pa2_size = self.sizes_rq.get(&pa2).unwrap();
253
+
254
+
if pa1_size >= pa2_size {
255
+
self.parents_rq.insert(pa2, pa1);
256
+
self.sizes_rq.insert(pa1, pa1_size + pa2_size);
257
+
self.sizes_rq.remove(&pa2);
59
258
} else {
60
-
self.parents[pa1 as usize] = pa2;
61
-
self.sizes.insert(pa2, pa1_size + pa2_size);
62
-
self.sizes.remove(&pa1);
259
+
self.parents_rq.insert(pa1, pa2);
260
+
self.sizes_rq.insert(pa2, pa1_size + pa2_size);
261
+
self.sizes_rq.remove(&pa1);
63
262
}
64
263
}
65
264
66
-
pub fn component_size(&mut self, idx: UFIndex) -> UFSize {
67
-
let pa = self.find(idx);
68
-
*self.sizes.get(&pa).unwrap()
265
+
pub fn component_size(&mut self, subgraph_type: SubgraphType, id: &Id) -> Option<UFSize> {
266
+
let idx = self.id_to_index.get(id).map(|idx| *idx)?;
267
+
match subgraph_type {
268
+
SubgraphType::Reply => {
269
+
// special structure of replies means all non-root nodes are direct children of root
270
+
let pa = *self.parents_ro.get(&idx).unwrap();
271
+
self.sizes_ro.get(&pa).map(|idx| *idx)
272
+
},
273
+
SubgraphType::Quote => {
274
+
let pa = self.find_qo(idx);
275
+
self.sizes_qo.get(&pa).map(|idx| *idx)
276
+
},
277
+
SubgraphType::ReplyQuote => {
278
+
let pa = self.find_rq(idx);
279
+
self.sizes_rq.get(&pa).map(|idx| *idx)
280
+
},
281
+
}
69
282
}
70
283
}
71
284
72
285
#[cfg(test)]
73
286
mod tests {
74
-
use super::UnionFind;
287
+
use crate::storage::SubgraphType;
288
+
289
+
use super::{BskyPostId, BskyPostRecord, BskyPostUnionFind};
290
+
291
+
#[test]
292
+
fn test_simple_reply_thread_1() {
293
+
let mut uf = BskyPostUnionFind::new();
294
+
295
+
let id1 = BskyPostId::from("a", "1");
296
+
let id2 = BskyPostId::from("b", "1");
297
+
let id3 = BskyPostId::from("a", "2");
298
+
let id4 = BskyPostId::from("a", "3");
299
+
uf.ingest_post(BskyPostRecord {
300
+
id: id2.clone(),
301
+
reply_to: Some(super::BskyPostReplyTo { target: id1.clone(), root: id1.clone() }),
302
+
quote_of: None
303
+
});
304
+
uf.ingest_post(BskyPostRecord {
305
+
id: id3.clone(),
306
+
reply_to: Some(super::BskyPostReplyTo { target: id2.clone(), root: id1.clone() }),
307
+
quote_of: None
308
+
});
309
+
uf.ingest_post(BskyPostRecord {
310
+
id: id4.clone(),
311
+
reply_to: Some(super::BskyPostReplyTo { target: id3.clone(), root: id1.clone() }),
312
+
quote_of: None
313
+
});
314
+
315
+
assert_eq!(uf.component_size(SubgraphType::Reply, &id1), Some(4));
316
+
assert_eq!(uf.component_size(SubgraphType::Reply, &id2), Some(4));
317
+
assert_eq!(uf.component_size(SubgraphType::Reply, &id3), Some(4));
318
+
assert_eq!(uf.component_size(SubgraphType::Reply, &id4), Some(4));
319
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id1), Some(4));
320
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id2), Some(4));
321
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id3), Some(4));
322
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id4), Some(4));
323
+
324
+
}
325
+
326
+
#[test]
327
+
fn test_simple_reply_thread_2() {
328
+
let mut uf = BskyPostUnionFind::new();
329
+
330
+
let id1 = BskyPostId::from("a", "1");
331
+
let id2 = BskyPostId::from("b", "1");
332
+
let id3 = BskyPostId::from("a", "2");
333
+
let id4 = BskyPostId::from("a", "3");
334
+
uf.ingest_post(BskyPostRecord {
335
+
id: id4.clone(),
336
+
reply_to: Some(super::BskyPostReplyTo { target: id3.clone(), root: id1.clone() }),
337
+
quote_of: None
338
+
});
339
+
uf.ingest_post(BskyPostRecord {
340
+
id: id3.clone(),
341
+
reply_to: Some(super::BskyPostReplyTo { target: id2.clone(), root: id1.clone() }),
342
+
quote_of: None
343
+
});
344
+
uf.ingest_post(BskyPostRecord {
345
+
id: id2.clone(),
346
+
reply_to: Some(super::BskyPostReplyTo { target: id1.clone(), root: id1.clone() }),
347
+
quote_of: None
348
+
});
349
+
350
+
assert_eq!(uf.component_size(SubgraphType::Reply, &id1), Some(4));
351
+
assert_eq!(uf.component_size(SubgraphType::Reply, &id2), Some(4));
352
+
assert_eq!(uf.component_size(SubgraphType::Reply, &id3), Some(4));
353
+
assert_eq!(uf.component_size(SubgraphType::Reply, &id4), Some(4));
354
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id1), Some(4));
355
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id2), Some(4));
356
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id3), Some(4));
357
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id4), Some(4));
358
+
359
+
}
360
+
361
+
#[test]
362
+
fn test_simple_quote_chain_1() {
363
+
let mut uf = BskyPostUnionFind::new();
364
+
365
+
let id1 = BskyPostId::from("a", "1");
366
+
let id2 = BskyPostId::from("b", "1");
367
+
let id3 = BskyPostId::from("a", "2");
368
+
let id4 = BskyPostId::from("a", "3");
369
+
uf.ingest_post(BskyPostRecord {
370
+
id: id2.clone(),
371
+
reply_to: None,
372
+
quote_of: Some(id1.clone())
373
+
});
374
+
uf.ingest_post(BskyPostRecord {
375
+
id: id3.clone(),
376
+
reply_to: None,
377
+
quote_of: Some(id2.clone())
378
+
});
379
+
uf.ingest_post(BskyPostRecord {
380
+
id: id4.clone(),
381
+
reply_to: None,
382
+
quote_of: Some(id3.clone())
383
+
});
384
+
385
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id1), Some(4));
386
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id2), Some(4));
387
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id3), Some(4));
388
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id4), Some(4));
389
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id1), Some(4));
390
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id2), Some(4));
391
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id3), Some(4));
392
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id4), Some(4));
393
+
394
+
}
75
395
76
396
#[test]
77
-
fn test_1() {
78
-
let mut uf = UnionFind::new();
79
-
let idx1 = uf.add();
80
-
let idx2 = uf.add();
81
-
let idx3 = uf.add();
82
-
uf.union(idx3, idx2);
83
-
assert_eq!(uf.component_size(idx1), 1);
84
-
assert_eq!(uf.component_size(idx2), 2);
85
-
assert_eq!(uf.component_size(idx3), 2);
86
-
uf.union(idx1, idx3);
87
-
assert_eq!(uf.component_size(idx1), 3);
88
-
assert_eq!(uf.component_size(idx2), 3);
89
-
assert_eq!(uf.component_size(idx3), 3);
90
-
let idx4 = uf.add();
91
-
let idx5 = uf.add();
92
-
let idx6 = uf.add();
93
-
uf.union(idx4, idx5);
94
-
uf.union(idx5, idx6);
95
-
assert_eq!(uf.component_size(idx1), 3);
96
-
assert_eq!(uf.component_size(idx4), 3);
97
-
assert_eq!(uf.component_size(idx5), 3);
98
-
assert_eq!(uf.component_size(idx6), 3);
99
-
uf.union(idx1, idx6);
100
-
assert_eq!(uf.component_size(idx3), 6);
101
-
assert_eq!(uf.component_size(idx4), 6);
397
+
fn test_simple_quote_chain_2() {
398
+
let mut uf = BskyPostUnionFind::new();
399
+
400
+
let id1 = BskyPostId::from("a", "1");
401
+
let id2 = BskyPostId::from("b", "1");
402
+
let id3 = BskyPostId::from("a", "2");
403
+
let id4 = BskyPostId::from("a", "3");
404
+
uf.ingest_post(BskyPostRecord {
405
+
id: id2.clone(),
406
+
reply_to: None,
407
+
quote_of: Some(id1.clone())
408
+
});
409
+
uf.ingest_post(BskyPostRecord {
410
+
id: id4.clone(),
411
+
reply_to: None,
412
+
quote_of: Some(id3.clone())
413
+
});
414
+
uf.ingest_post(BskyPostRecord {
415
+
id: id3.clone(),
416
+
reply_to: None,
417
+
quote_of: Some(id2.clone())
418
+
});
419
+
420
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id1), Some(4));
421
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id2), Some(4));
422
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id3), Some(4));
423
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id4), Some(4));
424
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id1), Some(4));
425
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id2), Some(4));
426
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id3), Some(4));
427
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id4), Some(4));
428
+
429
+
}
430
+
431
+
#[test]
432
+
fn test_simple_quote_chain_3() {
433
+
let mut uf = BskyPostUnionFind::new();
434
+
435
+
let id1 = BskyPostId::from("a", "1");
436
+
let id2 = BskyPostId::from("b", "1");
437
+
let id3 = BskyPostId::from("a", "2");
438
+
let id4 = BskyPostId::from("a", "3");
439
+
uf.ingest_post(BskyPostRecord {
440
+
id: id3.clone(),
441
+
reply_to: None,
442
+
quote_of: Some(id2.clone())
443
+
});
444
+
uf.ingest_post(BskyPostRecord {
445
+
id: id2.clone(),
446
+
reply_to: None,
447
+
quote_of: Some(id1.clone())
448
+
});
449
+
uf.ingest_post(BskyPostRecord {
450
+
id: id4.clone(),
451
+
reply_to: None,
452
+
quote_of: Some(id3.clone())
453
+
});
454
+
455
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id1), Some(4));
456
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id2), Some(4));
457
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id3), Some(4));
458
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id4), Some(4));
459
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id1), Some(4));
460
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id2), Some(4));
461
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id3), Some(4));
462
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id4), Some(4));
463
+
464
+
}
465
+
466
+
#[test]
467
+
fn test_simple_quote_chain_4() {
468
+
let mut uf = BskyPostUnionFind::new();
469
+
470
+
let id1 = BskyPostId::from("a", "1");
471
+
let id2 = BskyPostId::from("b", "1");
472
+
let id3 = BskyPostId::from("a", "2");
473
+
let id4 = BskyPostId::from("a", "3");
474
+
uf.ingest_post(BskyPostRecord {
475
+
id: id3.clone(),
476
+
reply_to: None,
477
+
quote_of: Some(id2.clone())
478
+
});
479
+
uf.ingest_post(BskyPostRecord {
480
+
id: id4.clone(),
481
+
reply_to: None,
482
+
quote_of: Some(id3.clone())
483
+
});
484
+
uf.ingest_post(BskyPostRecord {
485
+
id: id2.clone(),
486
+
reply_to: None,
487
+
quote_of: Some(id1.clone())
488
+
});
489
+
490
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id1), Some(4));
491
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id2), Some(4));
492
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id3), Some(4));
493
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id4), Some(4));
494
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id1), Some(4));
495
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id2), Some(4));
496
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id3), Some(4));
497
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id4), Some(4));
498
+
499
+
}
500
+
501
+
#[test]
502
+
fn test_simple_quote_chain_5() {
503
+
let mut uf = BskyPostUnionFind::new();
504
+
505
+
let id1 = BskyPostId::from("a", "1");
506
+
let id2 = BskyPostId::from("b", "1");
507
+
let id3 = BskyPostId::from("a", "2");
508
+
let id4 = BskyPostId::from("a", "3");
509
+
uf.ingest_post(BskyPostRecord {
510
+
id: id4.clone(),
511
+
reply_to: None,
512
+
quote_of: Some(id3.clone())
513
+
});
514
+
uf.ingest_post(BskyPostRecord {
515
+
id: id2.clone(),
516
+
reply_to: None,
517
+
quote_of: Some(id1.clone())
518
+
});
519
+
uf.ingest_post(BskyPostRecord {
520
+
id: id3.clone(),
521
+
reply_to: None,
522
+
quote_of: Some(id2.clone())
523
+
});
524
+
525
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id1), Some(4));
526
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id2), Some(4));
527
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id3), Some(4));
528
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id4), Some(4));
529
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id1), Some(4));
530
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id2), Some(4));
531
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id3), Some(4));
532
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id4), Some(4));
533
+
534
+
}
535
+
536
+
#[test]
537
+
fn test_simple_quote_chain_6() {
538
+
let mut uf = BskyPostUnionFind::new();
539
+
540
+
let id1 = BskyPostId::from("a", "1");
541
+
let id2 = BskyPostId::from("b", "1");
542
+
let id3 = BskyPostId::from("a", "2");
543
+
let id4 = BskyPostId::from("a", "3");
544
+
uf.ingest_post(BskyPostRecord {
545
+
id: id4.clone(),
546
+
reply_to: None,
547
+
quote_of: Some(id3.clone())
548
+
});
549
+
uf.ingest_post(BskyPostRecord {
550
+
id: id3.clone(),
551
+
reply_to: None,
552
+
quote_of: Some(id2.clone())
553
+
});
554
+
uf.ingest_post(BskyPostRecord {
555
+
id: id2.clone(),
556
+
reply_to: None,
557
+
quote_of: Some(id1.clone())
558
+
});
559
+
560
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id1), Some(4));
561
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id2), Some(4));
562
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id3), Some(4));
563
+
assert_eq!(uf.component_size(SubgraphType::Quote, &id4), Some(4));
564
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id1), Some(4));
565
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id2), Some(4));
566
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id3), Some(4));
567
+
assert_eq!(uf.component_size(SubgraphType::ReplyQuote, &id4), Some(4));
568
+
102
569
}
103
570
}