mapterm 0.1.2

A map in your terminal: vector tiles, MapLibre styles, imagery and your own data, as a ratatui widget and the mapterm viewer
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224

//! Point clustering for GeoJSON sources, like MapLibre's `cluster: true` (supercluster): for
//! each zoom level up to `max_zoom`, points closer than `radius` merge into cluster features
//! with `cluster`, `cluster_id`, `point_count` and `point_count_abbreviated` properties, built
//! level by level from the one above so clusters nest. Other geometry passes through.

use std::collections::HashMap;

use crate::{
    geojson::Feature,
    index::{BBox, Index},
    mvt::{GeomType, Value},
    render::TILE_DOTS,
};

/// Clustering settings (MapLibre's `clusterRadius` and `clusterMaxZoom`).
#[derive(Clone, Copy, Debug, PartialEq)]
pub struct ClusterOptions {
    /// How close points merge, in CSS pixels (about 4 per braille dot). MapLibre's default: 50.
    pub radius_px: f64,
    /// The last zoom with clusters; past it every point shows. MapLibre's default: 14.
    pub max_zoom: u8,
}

impl Default for ClusterOptions {
    fn default() -> Self {
        Self { radius_px: 50.0, max_zoom: 14 }
    }
}

/// Features and their index at one zoom level.
#[derive(Clone, Debug)]
pub struct Level {
    /// The features to draw at this zoom.
    pub features: Vec<Feature>,
    /// Their spatial index.
    pub index: Index,
}

impl Level {
    /// Index `features`.
    pub fn new(features: Vec<Feature>) -> Self {
        let boxes: Vec<BBox> = features.iter().map(|f| f.bbox).collect();
        Self { index: Index::new(&boxes), features }
    }
}

/// The clustered levels of a source, zoom 0 to `max_zoom`.
#[derive(Clone, Debug)]
pub struct Clusters {
    /// The settings the clusters were built with.
    pub options: ClusterOptions,
    levels: Vec<Level>,
    /// The zoom each cluster was formed at, by cluster id.
    formed: HashMap<u64, u8>,
}

/// A point being clustered: where, how many original points, and what it is.
struct Item {
    at: (f64, f64),
    count: u64,
    /// The original feature (a single point), or the cluster id.
    what: Either,
}

enum Either {
    Point(usize),
    Cluster(u64),
}

impl Clusters {
    /// Cluster the single-point features of `features`.
    pub fn build(features: &[Feature], options: ClusterOptions) -> Self {
        let is_point = |f: &Feature| f.geom == GeomType::Point && f.parts.len() == 1 && f.parts[0].len() == 1;
        let others: Vec<Feature> = features.iter().filter(|f| !is_point(f)).cloned().collect();
        let mut items: Vec<Item> =
            features.iter().enumerate().filter(|(_, f)| is_point(f)).map(|(i, f)| Item { at: f.parts[0][0], count: 1, what: Either::Point(i) }).collect();
        let mut formed = HashMap::new();
        let mut next_id = 1u64;
        let mut levels = Vec::with_capacity(options.max_zoom as usize + 1);
        for z in (0..=options.max_zoom).rev() {
            // The radius in world units at this zoom (TILE_DOTS dots per world at zoom 0).
            let r = options.radius_px / 4.0 / (TILE_DOTS * 2f64.powi(z as i32));
            items = merge(items, r, z, &mut next_id, &mut formed);
            let mut level: Vec<Feature> = items.iter().map(|it| to_feature(it, features)).collect();
            level.extend(others.iter().cloned());
            levels.push(Level::new(level));
        }
        levels.reverse();
        Self { options, levels, formed }
    }

    /// The level to draw at `zoom`, or `None` past `max_zoom` (draw the original features).
    pub fn at(&self, zoom: f64) -> Option<&Level> {
        let z = zoom.floor();
        if z > self.options.max_zoom as f64 {
            return None;
        }
        self.levels.get(z.max(0.0) as usize)
    }

    /// The zoom at which a cluster splits into its parts (MapLibre's
    /// `getClusterExpansionZoom`).
    pub fn expansion_zoom(&self, cluster_id: u64) -> Option<u8> {
        self.formed.get(&cluster_id).map(|z| z + 1)
    }
}

/// Merge items closer than `r` (greedily, in order, using a grid of `r`-sized cells).
fn merge(items: Vec<Item>, r: f64, z: u8, next_id: &mut u64, formed: &mut HashMap<u64, u8>) -> Vec<Item> {
    let cell = |p: (f64, f64)| ((p.0 / r).floor() as i64, (p.1 / r).floor() as i64);
    let mut grid: HashMap<(i64, i64), Vec<usize>> = HashMap::new();
    for (i, it) in items.iter().enumerate() {
        grid.entry(cell(it.at)).or_default().push(i);
    }
    let mut taken = vec![false; items.len()];
    let mut out = Vec::new();
    let mut items: Vec<Option<Item>> = items.into_iter().map(Some).collect();
    for i in 0..items.len() {
        if taken[i] {
            continue;
        }
        taken[i] = true;
        let at = items[i].as_ref().map(|it| it.at).unwrap_or_default();
        let (cx, cy) = cell(at);
        let mut group = vec![i];
        for gx in cx - 1..=cx + 1 {
            for gy in cy - 1..=cy + 1 {
                for &j in grid.get(&(gx, gy)).into_iter().flatten() {
                    if !taken[j] && items[j].as_ref().is_some_and(|o| (o.at.0 - at.0).hypot(o.at.1 - at.1) <= r) {
                        taken[j] = true;
                        group.push(j);
                    }
                }
            }
        }
        if group.len() == 1 {
            out.extend(items[i].take());
            continue;
        }
        // A cluster at the weighted center of its parts.
        let (mut sx, mut sy, mut n) = (0.0, 0.0, 0u64);
        for &j in &group {
            if let Some(it) = items[j].take() {
                sx += it.at.0 * it.count as f64;
                sy += it.at.1 * it.count as f64;
                n += it.count;
            }
        }
        let id = *next_id;
        *next_id += 1;
        formed.insert(id, z);
        out.push(Item { at: (sx / n as f64, sy / n as f64), count: n, what: Either::Cluster(id) });
    }
    out
}

fn to_feature(it: &Item, originals: &[Feature]) -> Feature {
    match it.what {
        Either::Point(i) => originals[i].clone(),
        Either::Cluster(id) => {
            let n = it.count;
            let abbreviated = if n >= 10_000 {
                format!("{}k", (n as f64 / 1000.0).round())
            } else if n >= 1000 {
                format!("{}k", (n as f64 / 100.0).round() / 10.0)
            } else {
                n.to_string()
            };
            Feature {
                id: Some(Value::Num(id as f64)),
                geom: GeomType::Point,
                parts: vec![vec![it.at]],
                bbox: [it.at.0, it.at.1, it.at.0, it.at.1],
                props: vec![
                    ("cluster".into(), Value::Bool(true)),
                    ("cluster_id".into(), Value::Num(id as f64)),
                    ("point_count".into(), Value::Num(n as f64)),
                    ("point_count_abbreviated".into(), Value::Str(abbreviated)),
                ],
                label: None,
            }
        }
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::render::to_world;

    fn point(lon: f64, lat: f64, name: &str) -> Feature {
        let p = to_world(lon, lat);
        Feature { id: None, geom: GeomType::Point, parts: vec![vec![p]], bbox: [p.0, p.1, p.0, p.1], props: vec![("name".into(), Value::Str(name.into()))], label: Some(name.into()) }
    }

    #[test]
    fn nearby_points_cluster_by_zoom() {
        // Two tight groups 1 km apart within DC, and one point in Baltimore.
        let mut fs = Vec::new();
        for i in 0..30 {
            fs.push(point(-77.03 + i as f64 * 1e-5, 38.89, "a"));
            fs.push(point(-77.02 + i as f64 * 1e-5, 38.89, "b"));
        }
        fs.push(point(-76.61, 39.29, "baltimore"));
        let c = Clusters::build(&fs, ClusterOptions::default());
        let count = |z: f64| c.at(z).map(|l| l.features.iter().map(|f| match f.props.iter().find(|(k, _)| k == "point_count") {
            Some((_, Value::Num(n))) => *n as u64,
            _ => 1,
        }).collect::<Vec<_>>());
        let total = |z: f64| count(z).unwrap().iter().sum::<u64>();
        for z in [2.0, 8.0, 12.0, 14.0] {
            assert_eq!(total(z), 61, "every point counted once at z{z}");
        }
        assert_eq!(count(8.0).unwrap().len(), 2, "DC (60) and Baltimore at z8: {:?}", count(8.0));
        assert!(count(14.0).unwrap().contains(&30), "two groups of 30 at z14: {:?}", count(14.0));
        assert!(c.at(15.0).is_none(), "past max zoom: the original points");
        // Abbreviated counts and expansion zooms.
        let dc = c.at(8.0).unwrap().features.iter().find(|f| f.props.iter().any(|(k, v)| k == "point_count" && *v == Value::Num(60.0))).unwrap();
        assert!(dc.props.iter().any(|(k, v)| k == "point_count_abbreviated" && *v == Value::Str("60".into())));
        let Some(Value::Num(id)) = dc.id.clone() else { panic!() };
        let z = c.expansion_zoom(id as u64).unwrap();
        assert!(z > 8 && z <= 15, "{z}");
    }
}