oxilite_core/
stats.rs

1//! Statistics used by the join-order planner.
2//!
3//! Stats are refreshed explicitly (`optimize()`, or after a bulk load) rather than on every
4//! write: on D1 every index entry written is billed, and a per-write counter would be a hot
5//! row. Stale stats only degrade plan quality, never correctness.
6//!
7// @lat: [[architecture#Query planner#Statistics]]
8
9use crate::encoding::rdf_type_id;
10use crate::error::Result;
11use crate::sql::{col, expect_len, Capabilities, Request, Response, Statement};
12use std::collections::{BTreeSet, HashMap};
13
14/// Per-predicate statistics.
15#[derive(Debug, Clone, Copy, PartialEq)]
16#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
17pub struct PredicateStats {
18    pub triples: f64,
19    pub distinct_subjects: f64,
20    pub distinct_objects: f64,
21}
22
23/// Planner statistics loaded from `stats_pred` / `stats_class`.
24#[derive(Debug, Clone, Default)]
25#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
26pub struct Stats {
27    /// `true` when `optimize()` has been run at least once.
28    pub available: bool,
29    pub total: f64,
30    pub predicates: HashMap<i64, PredicateStats>,
31    /// Instance count per `rdf:type` class.
32    pub classes: HashMap<i64, f64>,
33    /// Triples per frequent (predicate, object) pair of low-cardinality predicates: values
34    /// much more common than their predicate's average (skew the average would hide).
35    pub pairs: HashMap<(i64, i64), f64>,
36    pub graph_index: bool,
37    /// The full-text index exists (`oxl:textMatch` compiles to FTS5).
38    pub text_index: bool,
39    /// Transitive properties (from `tbox_closure`), for query-time reasoning.
40    pub transitive: BTreeSet<i64>,
41}
42
43fn id_col(caps: &Capabilities, c: &str) -> String {
44    if caps.int64_as_text {
45        format!("CAST({c} AS TEXT)")
46    } else {
47        c.into()
48    }
49}
50
51impl Stats {
52    /// Statements that load statistics and store settings.
53    pub fn load_request(caps: &Capabilities) -> Request {
54        Request::read(vec![
55            Statement::new("SELECT key, value FROM oxilite_meta"),
56            Statement::new(format!(
57                "SELECT {}, triples, distinct_s, distinct_o FROM stats_pred",
58                id_col(caps, "p")
59            )),
60            Statement::new(format!(
61                "SELECT {}, instances FROM stats_class",
62                id_col(caps, "o")
63            )),
64            crate::reason::transitive_statement(|c| id_col(caps, c)),
65            Statement::new(format!(
66                "SELECT {}, {}, n FROM stats_po",
67                id_col(caps, "p"),
68                id_col(caps, "o")
69            )),
70        ])
71    }
72
73    pub fn from_response(response: &Response) -> Result<Self> {
74        expect_len(response, 5)?;
75        let mut stats = Self::default();
76        for row in &response[0].rows {
77            let key = col(row, 0)?.as_str().unwrap_or_default();
78            let value = col(row, 1)?.clone().into_string().unwrap_or_default();
79            match key {
80                "graph_index" => stats.graph_index = value == "1",
81                "text_index" => stats.text_index = value == "1",
82                "total" => {
83                    stats.total = value.parse().unwrap_or(0.0);
84                    stats.available = true;
85                }
86                _ => {}
87            }
88        }
89        for row in &response[1].rows {
90            let (Some(p), Some(t), Some(ds), Some(d_o)) = (
91                col(row, 0)?.as_i64(),
92                col(row, 1)?.as_f64(),
93                col(row, 2)?.as_f64(),
94                col(row, 3)?.as_f64(),
95            ) else {
96                continue;
97            };
98            stats.predicates.insert(
99                p,
100                PredicateStats {
101                    triples: t,
102                    distinct_subjects: ds.max(1.0),
103                    distinct_objects: d_o.max(1.0),
104                },
105            );
106        }
107        for row in &response[2].rows {
108            if let (Some(o), Some(n)) = (col(row, 0)?.as_i64(), col(row, 1)?.as_f64()) {
109                stats.classes.insert(o, n);
110            }
111        }
112        for row in &response[3].rows {
113            if let Some(p) = col(row, 0)?.as_i64() {
114                stats.transitive.insert(p);
115            }
116        }
117        for row in &response[4].rows {
118            if let (Some(p), Some(o), Some(n)) = (
119                col(row, 0)?.as_i64(),
120                col(row, 1)?.as_i64(),
121                col(row, 2)?.as_f64(),
122            ) {
123                stats.pairs.insert((p, o), n);
124            }
125        }
126        Ok(stats)
127    }
128
129    /// Statements recomputing statistics and the schema closure (run by `optimize()`).
130    pub fn refresh_request() -> Request {
131        let mut r = Self::refresh_statements();
132        r.extend(crate::reason::closure_statements());
133        Request::atomic(r)
134    }
135
136    fn refresh_statements() -> Vec<Statement> {
137        vec![
138            "DELETE FROM stats_pred".into(),
139            "INSERT INTO stats_pred(p, triples, distinct_s, distinct_o) \
140             SELECT p, COUNT(*), COUNT(DISTINCT s), COUNT(DISTINCT o) FROM quads GROUP BY p"
141                .into(),
142            "DELETE FROM stats_class".into(),
143            Statement::new(format!(
144                "INSERT INTO stats_class(o, instances) SELECT o, COUNT(*) FROM quads WHERE p = {} GROUP BY o",
145                rdf_type_id()
146            )),
147            "INSERT OR REPLACE INTO oxilite_meta(key, value) SELECT 'total', CAST(COUNT(*) AS TEXT) FROM quads"
148                .into(),
149            "DELETE FROM stats_po".into(),
150            Statement::new(format!(
151                "INSERT INTO stats_po(p, o, n) SELECT q.p, q.o, COUNT(*) FROM quads q JOIN stats_pred sp ON sp.p = q.p \
152                 WHERE sp.distinct_o <= 1024 AND q.p <> {} GROUP BY q.p, q.o, sp.triples, sp.distinct_o \
153                 HAVING COUNT(*) >= 4.0 * sp.triples / sp.distinct_o ORDER BY COUNT(*) DESC LIMIT 2000",
154                rdf_type_id()
155            )),
156        ]
157    }
158}
oxilite_core/stats.rs

oxilite_core/
stats.rs
