| 1 |
|
| 2 |
|
| 3 |
|
| 4 |
|
| 5 |
|
| 6 |
|
| 7 |
|
| 8 |
|
| 9 |
|
| 10 |
|
| 11 |
|
| 12 |
|
| 13 |
|
| 14 |
use std::path::PathBuf; |
| 15 |
|
| 16 |
use audiofiles_core::db::Database; |
| 17 |
use audiofiles_core::worker_runtime::{WorkerCtx, WorkerHandle, spawn_worker}; |
| 18 |
use audiofiles_core::{fingerprint, similarity}; |
| 19 |
|
| 20 |
|
| 21 |
pub enum SearchCommand { |
| 22 |
|
| 23 |
FindSimilar { hash: String, limit: usize }, |
| 24 |
|
| 25 |
FindNearDuplicates { hash: String, limit: usize }, |
| 26 |
|
| 27 |
Invalidate, |
| 28 |
} |
| 29 |
|
| 30 |
|
| 31 |
pub enum SearchEvent { |
| 32 |
|
| 33 |
SimilarResults { source: String, hashes: Vec<String> }, |
| 34 |
|
| 35 |
NearDuplicateResults { source: String, hashes: Vec<String> }, |
| 36 |
|
| 37 |
Error { error: String }, |
| 38 |
} |
| 39 |
|
| 40 |
|
| 41 |
|
| 42 |
pub type SearchWorkerHandle = WorkerHandle<SearchCommand, SearchEvent>; |
| 43 |
|
| 44 |
|
| 45 |
|
| 46 |
|
| 47 |
|
| 48 |
|
| 49 |
|
| 50 |
|
| 51 |
|
| 52 |
|
| 53 |
|
| 54 |
|
| 55 |
struct SearchState { |
| 56 |
db: Database, |
| 57 |
similarity_index: Option<similarity::SimilarityIndex>, |
| 58 |
fingerprint_index: Option<fingerprint::FingerprintIndex>, |
| 59 |
} |
| 60 |
|
| 61 |
|
| 62 |
|
| 63 |
pub fn spawn_search_worker(db_path: PathBuf) -> std::io::Result<SearchWorkerHandle> { |
| 64 |
spawn_worker( |
| 65 |
"search-worker", |
| 66 |
move || { |
| 67 |
Database::open(&db_path).map(|db| SearchState { |
| 68 |
db, |
| 69 |
similarity_index: None, |
| 70 |
fingerprint_index: None, |
| 71 |
}) |
| 72 |
}, |
| 73 |
|e| SearchEvent::Error { |
| 74 |
error: e.to_string(), |
| 75 |
}, |
| 76 |
|_state| SearchEvent::Error { |
| 77 |
error: "search worker panicked (internal error)".to_string(), |
| 78 |
}, |
| 79 |
search_step, |
| 80 |
) |
| 81 |
} |
| 82 |
|
| 83 |
fn search_step(state: &mut SearchState, cmd: SearchCommand, ctx: &WorkerCtx<SearchEvent>) { |
| 84 |
match cmd { |
| 85 |
SearchCommand::Invalidate => { |
| 86 |
state.similarity_index = None; |
| 87 |
state.fingerprint_index = None; |
| 88 |
} |
| 89 |
SearchCommand::FindSimilar { hash, limit } => { |
| 90 |
|
| 91 |
|
| 92 |
|
| 93 |
state.fingerprint_index = None; |
| 94 |
let result: Result<Vec<String>, audiofiles_core::error::CoreError> = (|| { |
| 95 |
if state.similarity_index.is_none() { |
| 96 |
let data = similarity::SimilarityIndex::load_data(&state.db)?; |
| 97 |
state.similarity_index = |
| 98 |
Some(similarity::SimilarityIndex::build_from_data(data)); |
| 99 |
} |
| 100 |
let built = state.similarity_index.as_ref().expect("set above"); |
| 101 |
let features = similarity::load_features(&state.db, &hash)?; |
| 102 |
Ok(built |
| 103 |
.find_similar(&hash, &features, limit) |
| 104 |
.into_iter() |
| 105 |
.map(|r| r.hash) |
| 106 |
.collect()) |
| 107 |
})(); |
| 108 |
ctx.emit(match result { |
| 109 |
Ok(hashes) => SearchEvent::SimilarResults { |
| 110 |
source: hash, |
| 111 |
hashes, |
| 112 |
}, |
| 113 |
Err(e) => SearchEvent::Error { |
| 114 |
error: e.to_string(), |
| 115 |
}, |
| 116 |
}); |
| 117 |
} |
| 118 |
SearchCommand::FindNearDuplicates { hash, limit } => { |
| 119 |
|
| 120 |
state.similarity_index = None; |
| 121 |
let result: Result<Vec<String>, audiofiles_core::error::CoreError> = (|| { |
| 122 |
if state.fingerprint_index.is_none() { |
| 123 |
let entries = fingerprint::FingerprintIndex::load_data(&state.db)?; |
| 124 |
state.fingerprint_index = |
| 125 |
Some(fingerprint::FingerprintIndex::build_from_data(entries)); |
| 126 |
} |
| 127 |
let built = state.fingerprint_index.as_ref().expect("set above"); |
| 128 |
let reference = fingerprint::load_fingerprint(&state.db, &hash)?; |
| 129 |
Ok(built |
| 130 |
.find_near_duplicates(&hash, &reference.envelope, limit) |
| 131 |
.into_iter() |
| 132 |
.map(|r| r.hash) |
| 133 |
.collect()) |
| 134 |
})(); |
| 135 |
ctx.emit(match result { |
| 136 |
Ok(hashes) => SearchEvent::NearDuplicateResults { |
| 137 |
source: hash, |
| 138 |
hashes, |
| 139 |
}, |
| 140 |
Err(e) => SearchEvent::Error { |
| 141 |
error: e.to_string(), |
| 142 |
}, |
| 143 |
}); |
| 144 |
} |
| 145 |
} |
| 146 |
} |
| 147 |
|
| 148 |
#[cfg(test)] |
| 149 |
mod tests { |
| 150 |
use super::*; |
| 151 |
|
| 152 |
|
| 153 |
|
| 154 |
|
| 155 |
|
| 156 |
fn recv_one(handle: &SearchWorkerHandle) -> SearchEvent { |
| 157 |
let deadline = std::time::Instant::now() + std::time::Duration::from_secs(10); |
| 158 |
while std::time::Instant::now() < deadline { |
| 159 |
if let Some(ev) = handle.try_recv() { |
| 160 |
return ev; |
| 161 |
} |
| 162 |
std::thread::sleep(std::time::Duration::from_millis(1)); |
| 163 |
} |
| 164 |
panic!("search worker produced no event"); |
| 165 |
} |
| 166 |
|
| 167 |
#[test] |
| 168 |
fn worker_responds_on_empty_db() { |
| 169 |
let dir = tempfile::tempdir().unwrap(); |
| 170 |
let db_path = dir.path().join("audiofiles.db"); |
| 171 |
|
| 172 |
let _ = Database::open(&db_path).unwrap(); |
| 173 |
|
| 174 |
let handle = spawn_search_worker(db_path).unwrap(); |
| 175 |
|
| 176 |
|
| 177 |
assert!(handle.send(SearchCommand::FindSimilar { |
| 178 |
hash: "deadbeef".to_string(), |
| 179 |
limit: 10, |
| 180 |
})); |
| 181 |
match recv_one(&handle) { |
| 182 |
SearchEvent::SimilarResults { source, hashes } => { |
| 183 |
assert_eq!(source, "deadbeef"); |
| 184 |
assert!(hashes.is_empty()); |
| 185 |
} |
| 186 |
SearchEvent::Error { .. } => {} |
| 187 |
SearchEvent::NearDuplicateResults { .. } => panic!("wrong event kind"), |
| 188 |
} |
| 189 |
|
| 190 |
|
| 191 |
assert!(handle.send(SearchCommand::Invalidate)); |
| 192 |
assert!(handle.send(SearchCommand::FindNearDuplicates { |
| 193 |
hash: "deadbeef".to_string(), |
| 194 |
limit: 10, |
| 195 |
})); |
| 196 |
let _ = recv_one(&handle); |
| 197 |
drop(handle); |
| 198 |
} |
| 199 |
} |
| 200 |
|