1use std::borrow::Cow;
2use std::collections::HashMap;
3use std::sync::RwLock;
4
5use crate::{
6 comparison::{ComparisonBatch, ComparisonVector},
7 entity::{Entity, EntityId},
8 error::ZerError,
9 record::{Record, RecordId},
10 record_pool::RecordPool,
11 schema::Schema,
12 scoring::{ModelParams, ScoredPair},
13};
14
15pub type Result<T> = std::result::Result<T, ZerError>;
16
17pub trait RecordStore: Send + Sync {
21 fn insert(&self, record: Record);
23
24 fn get(&self, id: RecordId) -> Option<Cow<'_, Record>>;
26
27 fn get_many(&self, ids: &[RecordId]) -> Vec<Option<Cow<'_, Record>>> {
30 ids.iter().map(|id| self.get(*id)).collect()
31 }
32
33 fn len(&self) -> usize;
35
36 fn is_empty(&self) -> bool {
37 self.len() == 0
38 }
39}
40
41struct VecRecordStoreInner {
44 records: Vec<Record>,
45 id_to_idx: HashMap<RecordId, usize>,
46}
47
48pub struct VecRecordStore {
50 inner: RwLock<VecRecordStoreInner>,
51}
52
53impl VecRecordStore {
54 pub fn new() -> Self {
55 Self {
56 inner: RwLock::new(VecRecordStoreInner {
57 records: Vec::new(),
58 id_to_idx: HashMap::new(),
59 }),
60 }
61 }
62}
63
64impl Default for VecRecordStore {
65 fn default() -> Self {
66 Self::new()
67 }
68}
69
70impl RecordStore for VecRecordStore {
71 fn insert(&self, record: Record) {
72 let mut inner = self.inner.write().unwrap();
73 let idx = inner.records.len();
74 inner.id_to_idx.insert(record.id, idx);
75 inner.records.push(record);
76 }
77
78 fn get(&self, id: RecordId) -> Option<Cow<'_, Record>> {
79 let inner = self.inner.read().unwrap();
80 let idx = *inner.id_to_idx.get(&id)?;
81 Some(Cow::Owned(inner.records[idx].clone()))
82 }
83
84 fn len(&self) -> usize {
85 self.inner.read().unwrap().records.len()
86 }
87}
88
89pub trait BlockIndex: Send + Sync {
96 fn insert(&mut self, record_id: RecordId, keys: Vec<String>);
98
99 fn lookup_union(&self, keys: &[String], exclude: RecordId) -> Vec<RecordId>;
102
103 fn remove(&mut self, record_id: RecordId);
105
106 fn as_any(&self) -> &dyn std::any::Any;
107 fn as_any_mut(&mut self) -> &mut dyn std::any::Any;
108}
109
110pub trait Blocker: Send + Sync {
112 fn blocking_keys(&self, record: &Record, schema: &Schema) -> Vec<String>;
113 fn index_record(&self, record: &Record, schema: &Schema, index: &mut dyn BlockIndex);
114 fn candidates(&self, record: &Record, schema: &Schema, index: &dyn BlockIndex) -> Vec<RecordId>;
115}
116
117pub trait Comparator: Send + Sync {
118 fn compare(&self, a: &Record, b: &Record, schema: &Schema) -> ComparisonVector;
120
121 fn compare_batch_from_pool(
128 &self,
129 pool: &RecordPool,
130 indices: &[(usize, usize)],
131 schema: &Schema,
132 ) -> ComparisonBatch {
133 let n_pairs = indices.len();
134 let n_fields = schema.fields.len();
135 if n_pairs == 0 {
136 return ComparisonBatch::new(0, n_fields, vec![]);
137 }
138 let pair_ids: Vec<(u64, u64)> = indices.iter()
139 .map(|&(i, j)| (pool.ids[i], pool.ids[j]))
140 .collect();
141 let mut batch = ComparisonBatch::new(n_pairs, n_fields, pair_ids);
142 for (p, &(i, j)) in indices.iter().enumerate() {
143 use crate::record::FieldValue;
144 let mut a = Record::new(pool.ids[i]);
145 let mut b = Record::new(pool.ids[j]);
146 for (f, field) in schema.fields.iter().enumerate() {
147 let va = pool.get(f, i);
148 let vb = pool.get(f, j);
149 if !va.is_empty() {
150 a = a.insert(&field.name, FieldValue::Text(va.to_string()));
151 }
152 if !vb.is_empty() {
153 b = b.insert(&field.name, FieldValue::Text(vb.to_string()));
154 }
155 }
156 let v = self.compare(&a, &b, schema);
157 for (f, &level) in v.levels.iter().enumerate() {
158 batch.set_level(f, p, level);
159 }
160 }
161 batch
162 }
163
164}
165
166pub trait Scorer: Send + Sync {
167 fn score(&self, vector: &ComparisonVector, params: &ModelParams) -> ScoredPair;
169
170 fn score_batch(&self, batch: &ComparisonBatch, params: &ModelParams) -> Vec<ScoredPair> {
172 (0..batch.n_pairs)
173 .map(|p| self.score(&batch.pair_as_vector(p), params))
174 .collect()
175 }
176
177 fn estimate_params(
178 &self,
179 batch: &ComparisonBatch,
180 init: Option<ModelParams>,
181 max_iter: usize,
182 ) -> Result<ModelParams>;
183}
184
185pub trait Clusterer: Send + Sync {
187 fn cluster(&self, pairs: &[ScoredPair], params: &ModelParams) -> Vec<Entity>;
188}
189
190pub trait EntityStore: Send + Sync {
192 fn upsert_entity(&self, entity: &Entity) -> Result<EntityId>;
193 fn get_entity(&self, id: EntityId) -> Result<Entity>;
194 fn record_to_entity(&self, record_id: RecordId) -> Result<Option<EntityId>>;
195 fn all_entities(&self) -> Result<Vec<Entity>>;
196}
197
198pub trait IntoRecord {
204 fn into_record(self, id: RecordId) -> Record;
205}
206
207impl IntoRecord for Record {
208 fn into_record(self, _id: RecordId) -> Record {
209 self
210 }
211}
212
213pub trait Judge: Send + Sync {
215 fn adjudicate(&self, pairs: &[ScoredPair]) -> Result<Vec<JudgeVerdict>>;
216}
217
218impl<J: Judge + ?Sized> Judge for Box<J> {
219 fn adjudicate(&self, pairs: &[ScoredPair]) -> Result<Vec<JudgeVerdict>> {
220 (**self).adjudicate(pairs)
221 }
222}
223
224#[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
225pub enum JudgeVerdict {
226 IncreaseConfidence,
227 DecreaseConfidence,
228 NoChange,
229}
230
231#[cfg(test)]
232mod tests {
233 use super::*;
234
235 #[test]
236 fn traits_are_object_safe() {
237 let _: Box<dyn super::BlockIndex>;
238 let _: Box<dyn super::Blocker>;
239 let _: Box<dyn super::Comparator>;
240 let _: Box<dyn super::Scorer>;
241 let _: Box<dyn super::Clusterer>;
242 let _: Box<dyn super::EntityStore>;
243 let _: Box<dyn super::Judge>;
244 let _: Box<dyn super::RecordStore>;
245 }
246
247 #[test]
248 fn vec_record_store_insert_and_get() {
249 use crate::record::{FieldValue, Record};
250 let store = VecRecordStore::new();
251 assert!(store.is_empty());
252
253 let r = Record::new(42).insert("name", FieldValue::Text("Alice".into()));
254 store.insert(r);
255
256 assert_eq!(store.len(), 1);
257 let fetched = store.get(42).expect("record 42 must exist");
258 assert_eq!(fetched.id, 42);
259 }
260
261 #[test]
262 fn vec_record_store_get_missing_returns_none() {
263 let store = VecRecordStore::new();
264 assert!(store.get(999).is_none());
265 }
266
267 #[test]
268 fn vec_record_store_get_many() {
269 use crate::record::Record;
270 let store = VecRecordStore::new();
271 store.insert(Record::new(1));
272 store.insert(Record::new(2));
273 store.insert(Record::new(3));
274
275 let results = store.get_many(&[1, 3, 99]);
276 assert!(results[0].is_some());
277 assert!(results[1].is_some());
278 assert!(results[2].is_none());
279 }
280
281 #[test]
282 fn vec_record_store_is_sendable() {
283 use std::sync::Arc;
284 let store: Arc<dyn RecordStore> = Arc::new(VecRecordStore::new());
285 let store2 = Arc::clone(&store);
286 let handle = std::thread::spawn(move || {
287 store2.insert(crate::record::Record::new(7));
288 });
289 handle.join().unwrap();
290 assert_eq!(store.len(), 1);
291 }
292}