This commit is contained in:
Andras Schmelczer 2026-06-18 08:00:39 +01:00
parent d34478e13c
commit 1934a38677
5 changed files with 66 additions and 9 deletions

View file

@ -827,6 +827,18 @@ impl PropertyData {
}
}
}
// Reclaim geometric-growth slack. `address_search_token_keys` and the
// posting lists are built by repeated `push`, so their capacity rounds up
// to the next doubling step; at full UK scale (88M row-token keys, 39M
// postings) that slack is ~250MB of permanently-resident waste. These
// tables live for the whole process, so shrinking once at build time is
// a pure win with no behaviour change.
address_search_token_keys.shrink_to_fit();
for postings in address_token_index.values_mut() {
postings.shrink_to_fit();
}
// Keep every distinctive token: common road words ("high", "church", "station") are
// exactly what people search, and dropping them made those roads unsearchable while a
// prefix fallback surfaced the wrong street ("Highbury" for "High"). The candidate scan
@ -838,7 +850,10 @@ impl PropertyData {
.max()
.unwrap_or(0);
let address_prefix_index = build_address_prefix_index(&address_token_index);
let address_search_interner = address_search_rodeo.into_reader();
// Resolve-only view: scoring only ever calls `.resolve(spur)`, never
// `.get(str)`, so we drop the string->key reverse map that a RodeoReader
// would keep (~tens of MB of hashmap we never touch).
let address_search_interner = address_search_rodeo.into_resolver();
let address_postings_count: usize = address_token_index.values().map(Vec::len).sum();
tracing::info!(
tokens = address_token_index.len(),
@ -861,6 +876,9 @@ impl PropertyData {
.or_default()
.push(new_row as u32);
}
for rows in postcode_row_index.values_mut() {
rows.shrink_to_fit();
}
let postcode_interner = postcode_rodeo.into_reader();
let row_to_poi_metric_idx: Vec<u32> = if poi_metrics.is_empty() {