snowykami/meilisearch
1
0
Fork 0
mirror of https://github.com/meilisearch/meilisearch.git synced 2025-09-05 20:26:31 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add geo bounding box filter

Browse Source
This commit is contained in:
Mubelotix
2025-08-19 11:57:04 +02:00
parent 47f1e707d5
commit 3ee8a75065
6 changed files with 379 additions and 48 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
crates/meilisearch/tests/common/mod.rs
View File

@ -522,6 +522,26 @@ pub async fn shared_index_with_geo_documents() -> &'static Index<'static, Shared
.await
}
pub async fn shared_index_geojson_documents() -> &'static Index<'static, Shared> {
static INDEX: OnceCell<Index<'static, Shared>> = OnceCell::const_new();
INDEX
.get_or_init(|| async {
// Retrieved from https://gitlab-forge.din.developpement-durable.gouv.fr/pub/geomatique/descartes/d-map/-/blob/main/demo/examples/commons/countries.geojson?ref_type=heads
let server = Server::new_shared();
let index = server._index("SHARED_GEOJSON_DOCUMENTS").to_shared();
let countries = include_str!("../documents/geojson/assets/countries.geojson");
let lille = serde_json::from_str::<serde_json::Value>(countries).unwrap();
let (response, _code) = index._add_documents(Value(lille), Some("name")).await;
server.wait_task(response.uid()).await.succeeded();
let (response, _code) =
index._update_settings(json!({"filterableAttributes": ["_geojson"]})).await;
server.wait_task(response.uid()).await.succeeded();
index
})
.await
}
pub async fn shared_index_for_fragments() -> Index<'static, Shared> {
static INDEX: OnceCell<(Server<Shared>, String)> = OnceCell::const_new();
let (server, uid) = INDEX

180
crates/meilisearch/tests/documents/geojson/assets/countries.geojson Normal file
View File

File diff suppressed because one or more lines are too long

101
crates/meilisearch/tests/documents/geojson/mod.rs
View File

@ -1,4 +1,7 @@
use crate::{common::Server, json};
use crate::{
common::{shared_index_geojson_documents, Server},
json,
};
use meili_snap::{json_string, snapshot};
const LILLE: &str = include_str!("assets/lille.geojson");
@ -281,3 +284,99 @@ async fn partial_update_geojson() {
index.search_get("?filter=_geoPolygon([0.9,0.9],[2,0.9],[2,2],[0.9,2])").await;
assert_eq!(response.get("hits").unwrap().as_array().unwrap().len(), 0);
}
#[actix_rt::test]
async fn geo_bounding_box() {
let index = shared_index_geojson_documents().await;
// The bounding box is a polygon over middle Europe
let (response, code) =
index.search_get("?filter=_geoBoundingBox([21.43443989912143,50.53987503447863],[0.54979129195425,43.76393151539099])&attributesToRetrieve=name").await;
snapshot!(code, @"200 OK");
snapshot!(response, @r#"
{
"hits": [
{
"name": "Austria"
},
{
"name": "Belgium"
},
{
"name": "Bosnia_and_Herzegovina"
},
{
"name": "Switzerland"
},
{
"name": "Czech_Republic"
},
{
"name": "Germany"
},
{
"name": "France"
},
{
"name": "Croatia"
},
{
"name": "Hungary"
},
{
"name": "Italy"
},
{
"name": "Luxembourg"
},
{
"name": "Netherlands"
},
{
"name": "Poland"
},
{
"name": "Romania"
},
{
"name": "Republic_of_Serbia"
},
{
"name": "Slovakia"
},
{
"name": "Slovenia"
}
],
"query": "",
"processingTimeMs": "[duration]",
"limit": 20,
"offset": 0,
"estimatedTotalHits": 17
}
"#);
// Between Russia and Alaska
// WARNING: This test doesn't pass, the countries are those I imagine being found but maybe there is also Canada or something
let (response, code) = index
.search_get("?filter=_geoBoundingBox([70,-148],[63,152])&attributesToRetrieve=name")
.await;
snapshot!(code, @"200 OK");
snapshot!(response, @r#"
{
"hits": [
{
"name": "Russia"
},
{
"name": "United_States_of_America"
}
],
"query": "",
"processingTimeMs": "[duration]",
"limit": 20,
"offset": 0,
"estimatedTotalHits": 2
}
"#);
}

2
crates/milli/src/index.rs
View File

@ -1031,7 +1031,7 @@ impl Index {
}
/// Returns true if the geo sorting feature is enabled.
pub fn is_geojson_enabled(&self, rtxn: &RoTxn<'_>) -> Result<bool> {
pub fn is_geojson_filtering_enabled(&self, rtxn: &RoTxn<'_>) -> Result<bool> {
let geojson_filter =
self.filterable_attributes_rules(rtxn)?.iter().any(|field| field.has_geojson());
Ok(geojson_filter)

122
crates/milli/src/search/facet/filter.rs
View File

@ -51,14 +51,17 @@ impl Display for BadGeoError {
}
Self::Lat(lat) => write!(
f,
"Bad latitude `{}`. Latitude must be contained between -90 and 90 degrees. ",
"Bad latitude `{}`. Latitude must be contained between -90 and 90 degrees.",
lat
),
Self::Lng(lng) => write!(
f,
"Bad longitude `{}`. Longitude must be contained between -180 and 180 degrees. ",
lng
),
Self::Lng(lng) => {
let normalized = (lng + 180.0).rem_euclid(360.0) - 180.0;
write!(
f,
"Bad longitude `{}`. Longitude must be contained between -180 and 180 degrees. Hint: try using `{normalized}` instead.",
lng
)
}
}
}
}
@ -686,39 +689,42 @@ impl<'a> Filter<'a> {
}
}
FilterCondition::GeoBoundingBox { top_right_point, bottom_left_point } => {
if index.is_geo_filtering_enabled(rtxn)? {
let top_right: [f64; 2] = [
top_right_point[0].parse_finite_float()?,
top_right_point[1].parse_finite_float()?,
];
let bottom_left: [f64; 2] = [
bottom_left_point[0].parse_finite_float()?,
bottom_left_point[1].parse_finite_float()?,
];
if !(-90.0..=90.0).contains(&top_right[0]) {
return Err(
top_right_point[0].as_external_error(BadGeoError::Lat(top_right[0]))
)?;
}
if !(-180.0..=180.0).contains(&top_right[1]) {
return Err(
top_right_point[1].as_external_error(BadGeoError::Lng(top_right[1]))
)?;
}
if !(-90.0..=90.0).contains(&bottom_left[0]) {
return Err(bottom_left_point[0]
.as_external_error(BadGeoError::Lat(bottom_left[0])))?;
}
if !(-180.0..=180.0).contains(&bottom_left[1]) {
return Err(bottom_left_point[1]
.as_external_error(BadGeoError::Lng(bottom_left[1])))?;
}
if top_right[0] < bottom_left[0] {
return Err(bottom_left_point[1].as_external_error(
BadGeoError::BoundingBoxTopIsBelowBottom(top_right[0], bottom_left[0]),
))?;
}
let top_right: [f64; 2] = [
top_right_point[0].parse_finite_float()?,
top_right_point[1].parse_finite_float()?,
];
let bottom_left: [f64; 2] = [
bottom_left_point[0].parse_finite_float()?,
bottom_left_point[1].parse_finite_float()?,
];
if !(-90.0..=90.0).contains(&top_right[0]) {
return Err(
top_right_point[0].as_external_error(BadGeoError::Lat(top_right[0]))
)?;
}
if !(-180.0..=180.0).contains(&top_right[1]) {
return Err(
top_right_point[1].as_external_error(BadGeoError::Lng(top_right[1]))
)?;
}
if !(-90.0..=90.0).contains(&bottom_left[0]) {
return Err(
bottom_left_point[0].as_external_error(BadGeoError::Lat(bottom_left[0]))
)?;
}
if !(-180.0..=180.0).contains(&bottom_left[1]) {
return Err(
bottom_left_point[1].as_external_error(BadGeoError::Lng(bottom_left[1]))
)?;
}
if top_right[0] < bottom_left[0] {
return Err(bottom_left_point[1].as_external_error(
BadGeoError::BoundingBoxTopIsBelowBottom(top_right[0], bottom_left[0]),
))?;
}
let mut r1 = None;
if index.is_geo_filtering_enabled(rtxn)? {
// Instead of writing a custom `GeoBoundingBox` filter we're simply going to re-use the range
// filter to create the following filter;
// `_geo.lat {top_right[0]} TO {bottom_left[0]} AND _geo.lng {top_right[1]} TO {bottom_left[1]}`
@ -813,9 +819,34 @@ impl<'a> Filter<'a> {
)?
};
Ok(selected_lat & selected_lng)
} else {
Err(top_right_point[0].as_external_error(
r1 = Some(selected_lat & selected_lng);
}
let mut r2 = None;
if index.is_geojson_filtering_enabled(rtxn)? {
let polygon = geo_types::Polygon::new(
geo_types::LineString(vec![
geo_types::Coord { x: top_right[0], y: top_right[1] },
geo_types::Coord { x: bottom_left[0], y: top_right[1] },
geo_types::Coord { x: bottom_left[0], y: bottom_left[1] },
geo_types::Coord { x: top_right[0], y: bottom_left[1] },
]),
Vec::new(),
);
let result = index
.cellulite
.in_shape(rtxn, &polygon, &mut |_| ())
.map_err(InternalError::CelluliteError)?;
r2 = Some(RoaringBitmap::from_iter(result)); // TODO: Remove once we update roaring
}
match (r1, r2) {
(Some(r1), Some(r2)) => Ok(r1 | r2),
(Some(r1), None) => Ok(r1),
(None, Some(r2)) => Ok(r2),
(None, None) => Err(top_right_point[0].as_external_error(
FilterError::AttributeNotFilterable {
attribute: RESERVED_GEO_FIELD_NAME,
filterable_patterns: filtered_matching_patterns(
@ -823,11 +854,11 @@ impl<'a> Filter<'a> {
&|features| features.is_filterable(),
),
},
))?
))?,
}
}
FilterCondition::GeoPolygon { points } => {
if index.is_geojson_enabled(rtxn)? {
if index.is_geojson_filtering_enabled(rtxn)? {
let polygon = geo_types::Polygon::new(
geo_types::LineString(
points
@ -846,8 +877,9 @@ impl<'a> Filter<'a> {
.cellulite
.in_shape(rtxn, &polygon, &mut |_| ())
.map_err(InternalError::CelluliteError)?;
// TODO: Remove once we update roaring
let result = roaring::RoaringBitmap::from_iter(result);
let result = roaring::RoaringBitmap::from_iter(result); // TODO: Remove once we update roaring
Ok(result)
} else {
Err(points[0][0].as_external_error(FilterError::AttributeNotFilterable {

2
crates/milli/src/update/new/extract/geo/cellulite.rs
View File

@ -31,7 +31,7 @@ impl GeoJsonExtractor {
index: &Index,
grenad_parameters: GrenadParameters,
) -> Result<Option<Self>> {
if index.is_geojson_enabled(rtxn)? {
if index.is_geojson_filtering_enabled(rtxn)? {
Ok(Some(GeoJsonExtractor { grenad_parameters }))
} else {
Ok(None)