Skip to main content

oxilite_core/
reason.rs

1//! RDFS / OWL reasoning: the schema closure, query rewriting and OWL 2 RL materialization.
2//!
3//! Query-time reasoning rewrites each triple pattern into a derived table of *entailed*
4//! triples, computed from the asserted quads and `tbox_closure` (a small table recomputed by
5//! `optimize()` and by writes that touch schema predicates). Queries therefore stay single SQL
6//! statements. Materialization is explicit: OWL 2 RL rules run as `INSERT … SELECT`
7//! statements into `quads_inf` until nothing changes.
8//!
9// @lat: [[architecture#Reasoning]]
10
11use crate::encoding::{named_node_id, DEFAULT_GRAPH_ID, PAYLOAD_BITS};
12use crate::sql::{union_all, Statement};
13use oxrdf::vocab::{rdf, rdfs};
14use oxrdf::{QuadRef, Term, TermRef};
15use spargebra::term::{NamedNodePattern, TermPattern};
16use spargebra::{GraphUpdateOperation, Update};
17use std::collections::BTreeSet;
18
19/// Entailment regime of a query.
20#[derive(Debug, Clone, Copy, Default, PartialEq, Eq, Hash)]
21#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
22#[cfg_attr(feature = "serde", serde(rename_all = "kebab-case"))]
23pub enum Reasoning {
24    /// Asserted triples only (Oxigraph's behaviour).
25    #[default]
26    None,
27    /// RDFS: subclass, subproperty, domain and range (plus equivalent classes and properties).
28    Rdfs,
29    /// RDFS plus the OWL QL property axioms: inverse and symmetric properties, and transitive
30    /// properties.
31    OwlQl,
32}
33
34const OWL: &str = "http://www.w3.org/2002/07/owl#";
35
36fn owl(local: &str) -> i64 {
37    named_node_id(&format!("{OWL}{local}"))
38}
39
40/// Ids of the vocabulary used by the closure and the rules.
41#[derive(Debug, Clone, Copy)]
42struct Vocab {
43    ty: i64,
44    sco: i64,
45    spo: i64,
46    dom: i64,
47    rng: i64,
48    eqc: i64,
49    eqp: i64,
50    inv: i64,
51    sym: i64,
52    trans: i64,
53    same: i64,
54    func: i64,
55    ifunc: i64,
56    has_value: i64,
57    on_property: i64,
58    some_values: i64,
59    all_values: i64,
60    intersection: i64,
61    union: i64,
62    chain: i64,
63    first: i64,
64    rest: i64,
65    nil: i64,
66    thing: i64,
67}
68
69fn vocab() -> Vocab {
70    Vocab {
71        ty: named_node_id(rdf::TYPE.as_str()),
72        sco: named_node_id(rdfs::SUB_CLASS_OF.as_str()),
73        spo: named_node_id(rdfs::SUB_PROPERTY_OF.as_str()),
74        dom: named_node_id(rdfs::DOMAIN.as_str()),
75        rng: named_node_id(rdfs::RANGE.as_str()),
76        eqc: owl("equivalentClass"),
77        eqp: owl("equivalentProperty"),
78        inv: owl("inverseOf"),
79        sym: owl("SymmetricProperty"),
80        trans: owl("TransitiveProperty"),
81        same: owl("sameAs"),
82        func: owl("FunctionalProperty"),
83        ifunc: owl("InverseFunctionalProperty"),
84        has_value: owl("hasValue"),
85        on_property: owl("onProperty"),
86        some_values: owl("someValuesFrom"),
87        all_values: owl("allValuesFrom"),
88        intersection: owl("intersectionOf"),
89        union: owl("unionOf"),
90        chain: owl("propertyChainAxiom"),
91        first: named_node_id(rdf::FIRST.as_str()),
92        rest: named_node_id(rdf::REST.as_str()),
93        nil: named_node_id(rdf::NIL.as_str()),
94        thing: owl("Thing"),
95    }
96}
97
98/// `tbox_closure.kind` values.
99pub mod kind {
100    /// `sub ⊑ sup` for classes (subClassOf and equivalentClass, transitive, irreflexive rows).
101    pub const CLASS: i64 = 1;
102    /// RDFS sub-property closure (subPropertyOf and equivalentProperty).
103    pub const PROPERTY: i64 = 2;
104    /// OWL: `(x sub y)` entails `(x sup y)` (sub-properties, inverses of inverses).
105    pub const OWL_SAME: i64 = 3;
106    /// OWL: `(x sub y)` entails `(y sup x)` (inverse and symmetric properties).
107    pub const OWL_INVERSE: i64 = 4;
108    /// Transitive properties (`sub = sup`).
109    pub const TRANSITIVE: i64 = 5;
110    /// RDFS: the subject of `sub` has type `sup` (domains through sub-properties/classes).
111    pub const SUBJECT_TYPE: i64 = 6;
112    /// RDFS: the object of `sub` has type `sup`.
113    pub const OBJECT_TYPE: i64 = 7;
114    /// OWL: the subject of `sub` has type `sup` (also through inverse ranges).
115    pub const OWL_SUBJECT_TYPE: i64 = 8;
116    /// OWL: the object of `sub` has type `sup`.
117    pub const OWL_OBJECT_TYPE: i64 = 9;
118}
119
120/// Predicates whose triples change the schema closure.
121pub fn schema_predicates() -> [i64; 6] {
122    let v = vocab();
123    [v.sco, v.spo, v.dom, v.rng, v.eqc, v.eqp]
124}
125
126/// Does a triple with this predicate and object change the closure?
127pub fn is_schema_triple(p: i64, o: i64) -> bool {
128    let v = vocab();
129    schema_predicates().contains(&p) || p == v.inv || (p == v.ty && (o == v.sym || o == v.trans))
130}
131
132fn is_schema_iri(p: &str) -> bool {
133    p == rdfs::SUB_CLASS_OF.as_str()
134        || p == rdfs::SUB_PROPERTY_OF.as_str()
135        || p == rdfs::DOMAIN.as_str()
136        || p == rdfs::RANGE.as_str()
137        || p.strip_prefix(OWL)
138            .is_some_and(|l| matches!(l, "equivalentClass" | "equivalentProperty" | "inverseOf"))
139}
140
141fn is_axiom_class(o: &str) -> bool {
142    o.strip_prefix(OWL)
143        .is_some_and(|l| matches!(l, "SymmetricProperty" | "TransitiveProperty"))
144}
145
146/// Does writing this quad change the schema closure?
147pub fn is_schema_quad(q: QuadRef<'_>) -> bool {
148    is_schema_iri(q.predicate.as_str())
149        || (q.predicate == rdf::TYPE
150            && matches!(q.object, TermRef::NamedNode(n) if is_axiom_class(n.as_str())))
151}
152
153/// Can this update change the schema closure? (Conservative: variables count as schema.)
154pub fn update_touches_schema(update: &Update) -> bool {
155    let pattern = |p: &NamedNodePattern, o: &TermPattern| match p {
156        NamedNodePattern::Variable(_) => true,
157        NamedNodePattern::NamedNode(n) if is_schema_iri(n.as_str()) => true,
158        NamedNodePattern::NamedNode(n) if *n == rdf::TYPE => match o {
159            TermPattern::NamedNode(c) => is_axiom_class(c.as_str()),
160            TermPattern::Variable(_) => true,
161            _ => false,
162        },
163        NamedNodePattern::NamedNode(_) => false,
164    };
165    update.operations.iter().any(|op| match op {
166        GraphUpdateOperation::InsertData { data } => data.iter().any(|q| {
167            is_schema_iri(q.predicate.as_str())
168                || (q.predicate == rdf::TYPE && matches!(&q.object, Term::NamedNode(n) if is_axiom_class(n.as_str())))
169        }),
170        GraphUpdateOperation::DeleteData { data } => data.iter().any(|q| {
171            is_schema_iri(q.predicate.as_str())
172                || (q.predicate == rdf::TYPE
173                    && matches!(&q.object, spargebra::term::GroundTerm::NamedNode(n) if is_axiom_class(n.as_str())))
174        }),
175        GraphUpdateOperation::DeleteInsert { delete, insert, .. } => {
176            delete.iter().any(|q| {
177                let o: TermPattern = q.object.clone().into();
178                pattern(&q.predicate, &o)
179            }) || insert.iter().any(|q| pattern(&q.predicate, &q.object))
180        }
181        GraphUpdateOperation::Create { .. } => false,
182        GraphUpdateOperation::Load { .. }
183        | GraphUpdateOperation::Clear { .. }
184        | GraphUpdateOperation::Drop { .. } => true,
185    })
186}
187
188/// SQL: the id is an IRI, a blank node or a triple term (a valid subject).
189pub(crate) fn non_literal(x: &str) -> String {
190    format!("(({x}) >> {PAYLOAD_BITS}) IN (1, 2, 9)")
191}
192
193/// Statements recomputing `tbox_closure` from the asserted quads (every graph).
194pub fn closure_statements() -> Vec<Statement> {
195    let v = vocab();
196    // Only the active registered ontology graphs contribute axioms; while none is registered,
197    // this is every graph, which is what the closure did before the registry existed.
198    let ontology_quads = crate::registry::scoped_quads(crate::registry::SchemaRole::Ontology);
199    let quads = ontology_quads.as_str();
200    let (c, p, s_, t) = (
201        kind::CLASS,
202        kind::PROPERTY,
203        kind::OWL_SAME,
204        kind::TRANSITIVE,
205    );
206    let class_edges = format!(
207        "SELECT s AS a, o AS b FROM {quads} WHERE p = {sco} UNION SELECT s, o FROM {quads} WHERE p = {eqc} UNION SELECT o, s FROM {quads} WHERE p = {eqc}",
208        sco = v.sco,
209        eqc = v.eqc
210    );
211    let prop_edges = format!(
212        "SELECT s AS a, o AS b FROM {quads} WHERE p = {spo} UNION SELECT s, o FROM {quads} WHERE p = {eqp} UNION SELECT o, s FROM {quads} WHERE p = {eqp}",
213        spo = v.spo,
214        eqp = v.eqp
215    );
216    let closure = |k: i64, edges: &str| {
217        Statement::new(format!(
218            "WITH RECURSIVE e(a, b) AS ({edges}), c(a, b) AS (SELECT a, b FROM e UNION SELECT c.a, e.b FROM c JOIN e ON e.a = c.b) \
219             INSERT OR IGNORE INTO tbox_closure(kind, sub, sup) SELECT {k}, a, b FROM c WHERE a <> b"
220        ))
221    };
222    // OWL: property edges with a direction bit (1 = inverse), composed modulo 2.
223    let owl = Statement::new(format!(
224        "WITH RECURSIVE e(a, b, d) AS (SELECT a, b, 0 FROM ({prop_edges}) \
225           UNION SELECT s, o, 1 FROM {quads} WHERE p = {inv} UNION SELECT o, s, 1 FROM {quads} WHERE p = {inv} \
226           UNION SELECT s, s, 1 FROM {quads} WHERE p = {ty} AND o = {sym}), \
227         c(a, b, d) AS (SELECT a, b, d FROM e UNION SELECT c.a, e.b, (c.d + e.d) % 2 FROM c JOIN e ON e.a = c.b) \
228         INSERT OR IGNORE INTO tbox_closure(kind, sub, sup) SELECT {s_} + d, a, b FROM c WHERE d = 1 OR a <> b",
229        inv = v.inv,
230        ty = v.ty,
231        sym = v.sym,
232    ));
233    // Classes a domain/range class is a subclass of (reflexive).
234    let classes = format!(
235        "SELECT o AS a, o AS b FROM {quads} WHERE p IN ({dom}, {rng}) UNION SELECT sub, sup FROM tbox_closure WHERE kind = {c}",
236        dom = v.dom,
237        rng = v.rng
238    );
239    // Properties with (sub, sup) where sup is reflexive over properties having a domain/range.
240    let subs = |k: i64| {
241        format!(
242            "SELECT s AS a, s AS b FROM {quads} WHERE p IN ({dom}, {rng}) UNION SELECT sub, sup FROM tbox_closure WHERE kind = {k}",
243            dom = v.dom,
244            rng = v.rng
245        )
246    };
247    let typed = |k: i64, props: &str, axiom: i64| {
248        format!(
249            "SELECT {k}, pq.a, cd.b FROM ({props}) pq JOIN {quads} ax ON ax.p = {axiom} AND ax.s = pq.b JOIN ({classes}) cd ON cd.a = ax.o"
250        )
251    };
252    let inverse_typed = |k: i64, axiom: i64| {
253        format!(
254            "SELECT {k}, pq.sub, cd.b FROM tbox_closure pq JOIN {quads} ax ON ax.p = {axiom} AND ax.s = pq.sup JOIN ({classes}) cd ON cd.a = ax.o WHERE pq.kind = {inv}",
255            inv = kind::OWL_INVERSE
256        )
257    };
258    let insert = |sql: String| {
259        Statement::new(format!(
260            "INSERT OR IGNORE INTO tbox_closure(kind, sub, sup) {sql}"
261        ))
262    };
263    vec![
264        Statement::new("DELETE FROM tbox_closure"),
265        closure(c, &class_edges),
266        closure(p, &prop_edges),
267        owl,
268        Statement::new(format!(
269            "INSERT OR IGNORE INTO tbox_closure(kind, sub, sup) SELECT {t}, s, s FROM {quads} WHERE p = {ty} AND o = {trans}",
270            ty = v.ty,
271            trans = v.trans
272        )),
273        insert(typed(kind::SUBJECT_TYPE, &subs(p), v.dom)),
274        insert(typed(kind::OBJECT_TYPE, &subs(p), v.rng)),
275        insert(format!(
276            "{} UNION {}",
277            typed(kind::OWL_SUBJECT_TYPE, &subs(s_), v.dom),
278            inverse_typed(kind::OWL_SUBJECT_TYPE, v.rng)
279        )),
280        insert(format!(
281            "{} UNION {}",
282            typed(kind::OWL_OBJECT_TYPE, &subs(s_), v.rng),
283            inverse_typed(kind::OWL_OBJECT_TYPE, v.dom)
284        )),
285    ]
286}
287
288/// Loads the transitive properties (kept in memory with the planner statistics).
289pub fn transitive_statement(id_col: impl Fn(&str) -> String) -> Statement {
290    Statement::new(format!(
291        "SELECT {} FROM tbox_closure WHERE kind = {}",
292        id_col("sub"),
293        kind::TRANSITIVE
294    ))
295}
296
297/// Which graphs a reasoned pattern reads.
298#[derive(Debug, Clone)]
299pub enum GraphFilter {
300    /// Keep each quad's graph (the caller constrains `g`).
301    Keep,
302    /// The RDF merge of these graphs (`None`: every graph), exposed as graph 0.
303    Merge(Option<Vec<i64>>),
304}
305
306/// Builds derived tables of entailed triples.
307#[derive(Debug, Clone)]
308pub struct Entailment<'a> {
309    pub reasoning: Reasoning,
310    /// Also read materialized inferences (`quads_inf`).
311    pub inferred: bool,
312    /// Exclude every registered schema graph from the stored triples
313    /// (`QueryOptions::include_schema_graphs`).
314    pub hide_schema: bool,
315    pub transitive: &'a BTreeSet<i64>,
316    /// Maximum terms of a compound SELECT.
317    pub max_compound: usize,
318    /// Read the store as it was at this tick (see `version`), from the change log.
319    pub as_of: Option<i64>,
320}
321
322impl Entailment<'_> {
323    /// Is any rewriting needed?
324    pub fn active(&self) -> bool {
325        self.reasoning != Reasoning::None || self.inferred
326    }
327
328    /// The table of stored triples: asserted, plus materialized inferences on request, minus
329    /// the registered schema graphs when they are hidden.
330    ///
331    /// Every quad source of the compiler goes through this or [`Self::source`], so hiding is
332    /// applied once and covers patterns, paths, `OPTIONAL` and `GRAPH ?g` alike. Inferences
333    /// are conclusions rather than schema, so they are never hidden.
334    pub fn base(&self) -> String {
335        let asserted = match (self.as_of, self.hide_schema) {
336            (Some(t), true) => {
337                crate::registry::without_schema_graphs(&crate::version::as_of_sql(&t.to_string()))
338            }
339            (Some(t), false) => crate::version::as_of_sql(&t.to_string()),
340            (None, true) => crate::registry::quads_without_schema_graphs().to_owned(),
341            (None, false) => "quads".to_owned(),
342        };
343        if self.inferred {
344            format!(
345                "(SELECT s, p, o, g FROM {asserted} UNION ALL SELECT s, p, o, g FROM quads_inf)"
346            )
347        } else {
348            asserted.to_string()
349        }
350    }
351
352    fn kinds(&self) -> (i64, Option<i64>, i64, i64) {
353        match self.reasoning {
354            Reasoning::OwlQl => (
355                kind::OWL_SAME,
356                Some(kind::OWL_INVERSE),
357                kind::OWL_SUBJECT_TYPE,
358                kind::OWL_OBJECT_TYPE,
359            ),
360            _ => (kind::PROPERTY, None, kind::SUBJECT_TYPE, kind::OBJECT_TYPE),
361        }
362    }
363
364    /// A `(s, p, o, g)` derived table of the entailed triples matching the constants.
365    pub fn source(
366        &self,
367        s: Option<i64>,
368        p: Option<i64>,
369        o: Option<i64>,
370        graphs: &GraphFilter,
371    ) -> String {
372        let base = self.base();
373        let g = |x: &str| match graphs {
374            GraphFilter::Keep => format!("{x}.g"),
375            GraphFilter::Merge(_) => DEFAULT_GRAPH_ID.to_string(),
376        };
377        let gw = |x: &str| match graphs {
378            GraphFilter::Merge(Some(l)) => format!(
379                " AND {x}.g IN ({})",
380                l.iter().map(i64::to_string).collect::<Vec<_>>().join(", ")
381            ),
382            _ => String::new(),
383        };
384        let eq =
385            |col: &str, v: Option<i64>| v.map(|v| format!(" AND {col} = {v}")).unwrap_or_default();
386        if self.reasoning == Reasoning::None {
387            let mut w = String::from("1");
388            w.push_str(&eq("x.s", s));
389            w.push_str(&eq("x.p", p));
390            w.push_str(&eq("x.o", o));
391            w.push_str(&gw("x"));
392            let d = if matches!(graphs, GraphFilter::Merge(_)) {
393                "DISTINCT "
394            } else {
395                ""
396            };
397            return format!(
398                "(SELECT {d}x.s AS s, x.p AS p, x.o AS o, {} AS g FROM {base} x WHERE {w})",
399                g("x")
400            );
401        }
402        let v = vocab();
403        let (same, inv, subj, obj) = self.kinds();
404        // One arm: `SELECT s AS s, p AS p, o AS o, g AS g FROM …` (compound SELECTs take their
405        // column names from the first arm, and derived tables cannot rename columns).
406        let row = |s: &str, p: &str, o: &str, gx: &str, rest: String| {
407            format!(
408                "SELECT {s} AS s, {p} AS p, {o} AS o, {} AS g FROM {rest}",
409                g(gx)
410            )
411        };
412        let ty = v.ty.to_string();
413        let type_arms = |s: Option<i64>, o: Option<i64>, asserted: bool| -> Vec<String> {
414            let mut a = Vec::new();
415            if asserted {
416                a.push(row(
417                    "x.s",
418                    &ty,
419                    "x.o",
420                    "x",
421                    format!(
422                        "{base} x WHERE x.p = {ty}{}{}{}",
423                        eq("x.s", s),
424                        eq("x.o", o),
425                        gw("x")
426                    ),
427                ));
428            }
429            // Superclasses of asserted types, then domains and ranges.
430            a.push(row("x.s", &ty, "c.sup", "x", format!(
431                "{base} x JOIN tbox_closure c ON c.kind = {} AND c.sub = x.o WHERE x.p = {ty}{}{}{}",
432                kind::CLASS, eq("x.s", s), eq("c.sup", o), gw("x")
433            )));
434            a.push(row(
435                "x.s",
436                &ty,
437                "c.sup",
438                "x",
439                format!(
440                    "tbox_closure c JOIN {base} x ON x.p = c.sub WHERE c.kind = {subj}{}{}{}",
441                    eq("x.s", s),
442                    eq("c.sup", o),
443                    gw("x")
444                ),
445            ));
446            a.push(row(
447                "x.o",
448                &ty,
449                "c.sup",
450                "x",
451                format!(
452                    "tbox_closure c JOIN {base} x ON x.p = c.sub WHERE c.kind = {obj} AND {}{}{}{}",
453                    non_literal("x.o"),
454                    eq("x.o", s),
455                    eq("c.sup", o),
456                    gw("x")
457                ),
458            ));
459            a
460        };
461        // Triples of property `pid` entailed without transitivity.
462        let prop_arms = |pid: i64, s: Option<i64>, o: Option<i64>| -> Vec<String> {
463            let pid_s = pid.to_string();
464            let mut a = vec![
465                row("x.s", &pid_s, "x.o", "x", format!("{base} x WHERE x.p = {pid}{}{}{}", eq("x.s", s), eq("x.o", o), gw("x"))),
466                row("x.s", &pid_s, "x.o", "x", format!(
467                    "tbox_closure c JOIN {base} x ON x.p = c.sub WHERE c.kind = {same} AND c.sup = {pid}{}{}{}",
468                    eq("x.s", s), eq("x.o", o), gw("x")
469                )),
470            ];
471            if let Some(inv) = inv {
472                a.push(row("x.o", &pid_s, "x.s", "x", format!(
473                    "tbox_closure c JOIN {base} x ON x.p = c.sub WHERE c.kind = {inv} AND c.sup = {pid} AND {}{}{}{}",
474                    non_literal("x.o"), eq("x.o", s), eq("x.s", o), gw("x")
475                )));
476            }
477            // The schema closure itself answers subClassOf / subPropertyOf patterns.
478            let schema_kind = if pid == v.sco {
479                Some(kind::CLASS)
480            } else if pid == v.spo {
481                Some(kind::PROPERTY)
482            } else {
483                None
484            };
485            if let Some(k) = schema_kind {
486                a.push(format!(
487                    "SELECT sub AS s, {pid} AS p, sup AS o, {DEFAULT_GRAPH_ID} AS g FROM tbox_closure WHERE kind = {k}{}{}",
488                    eq("sub", s),
489                    eq("sup", o)
490                ));
491            }
492            a
493        };
494        // A transitive property: the closure of its (otherwise) entailed triples, walked from a
495        // constant endpoint when there is one.
496        let transitive = |pid: i64, s: Option<i64>, o: Option<i64>| -> String {
497            let u = union_all(prop_arms(pid, None, None), self.max_compound);
498            let rec = match (s, o) {
499                (Some(s), _) => format!(
500                    "r(n, g) AS (SELECT o, g FROM u WHERE s = {s} UNION SELECT u.o, u.g FROM r JOIN u ON u.s = r.n AND u.g = r.g) \
501                     SELECT {s} AS s, {pid} AS p, n AS o, g FROM r{}",
502                    o.map(|o| format!(" WHERE n = {o}")).unwrap_or_default()
503                ),
504                (None, Some(o)) => format!(
505                    "r(n, g) AS (SELECT s, g FROM u WHERE o = {o} UNION SELECT u.s, u.g FROM r JOIN u ON u.o = r.n AND u.g = r.g) \
506                     SELECT n AS s, {pid} AS p, {o} AS o, g FROM r"
507                ),
508                (None, None) => format!(
509                    "r(s, o, g) AS (SELECT s, o, g FROM u UNION SELECT r.s, u.o, r.g FROM r JOIN u ON u.s = r.o AND u.g = r.g) \
510                     SELECT s, {pid} AS p, o, g FROM r"
511                ),
512            };
513            format!("SELECT * FROM (WITH RECURSIVE u(s, p, o, g) AS ({u}), {rec})")
514        };
515        let mut arms = Vec::new();
516        match p {
517            Some(pid) if pid == v.ty => arms.extend(type_arms(s, o, true)),
518            Some(pid) if self.reasoning == Reasoning::OwlQl && self.transitive.contains(&pid) => {
519                arms.push(transitive(pid, s, o));
520            }
521            Some(pid) => arms.extend(prop_arms(pid, s, o)),
522            None => {
523                // Every entailed triple: asserted, through property axioms, the schema closure,
524                // types, and transitive properties.
525                arms.push(row(
526                    "x.s",
527                    "x.p",
528                    "x.o",
529                    "x",
530                    format!(
531                        "{base} x WHERE 1{}{}{}",
532                        eq("x.s", s),
533                        eq("x.o", o),
534                        gw("x")
535                    ),
536                ));
537                arms.push(row(
538                    "x.s",
539                    "c.sup",
540                    "x.o",
541                    "x",
542                    format!(
543                        "tbox_closure c JOIN {base} x ON x.p = c.sub WHERE c.kind = {same}{}{}{}",
544                        eq("x.s", s),
545                        eq("x.o", o),
546                        gw("x")
547                    ),
548                ));
549                if let Some(inv) = inv {
550                    arms.push(row("x.o", "c.sup", "x.s", "x", format!(
551                        "tbox_closure c JOIN {base} x ON x.p = c.sub WHERE c.kind = {inv} AND {}{}{}{}",
552                        non_literal("x.o"), eq("x.o", s), eq("x.s", o), gw("x")
553                    )));
554                }
555                for (k, pid) in [(kind::CLASS, v.sco), (kind::PROPERTY, v.spo)] {
556                    arms.push(format!(
557                        "SELECT sub AS s, {pid} AS p, sup AS o, {DEFAULT_GRAPH_ID} AS g FROM tbox_closure WHERE kind = {k}{}{}",
558                        eq("sub", s),
559                        eq("sup", o)
560                    ));
561                }
562                arms.extend(type_arms(s, o, false));
563                if self.reasoning == Reasoning::OwlQl {
564                    for &pid in self.transitive {
565                        arms.push(transitive(pid, s, o));
566                    }
567                }
568            }
569        }
570        format!(
571            "(SELECT DISTINCT s, p, o, g FROM ({}))",
572            union_all(arms, self.max_compound)
573        )
574    }
575}
576
577/// Statements of one OWL 2 RL round: every rule inserts its new conclusions into
578/// `quads_inf` (graph 0), reading asserted and inferred triples of every graph.
579pub fn materialize_round() -> Vec<Statement> {
580    let v = vocab();
581    let a = "(SELECT s, p, o FROM quads UNION ALL SELECT s, p, o FROM quads_inf)";
582    let nl = |x: &str| non_literal(x);
583    let lists = format!(
584        "RECURSIVE l(head, node, item) AS (SELECT s, s, o FROM {a} WHERE p = {first} \
585           UNION SELECT l.head, r.o, f.o FROM l JOIN {a} r ON r.s = l.node AND r.p = {rest} JOIN {a} f ON f.s = r.o AND f.p = {first})",
586        first = v.first,
587        rest = v.rest
588    );
589    let rules: Vec<String> = vec![
590        // eq-sym, eq-trans
591        format!("SELECT o, {same}, s FROM {a} WHERE p = {same}", same = v.same),
592        format!("SELECT x.s, {same}, y.o FROM {a} x JOIN {a} y ON y.s = x.o AND y.p = {same} WHERE x.p = {same}", same = v.same),
593        // eq-rep-s, eq-rep-p, eq-rep-o
594        format!("SELECT e.o, t.p, t.o FROM {a} e JOIN {a} t ON t.s = e.s WHERE e.p = {same} AND {}", nl("e.o"), same = v.same),
595        format!("SELECT t.s, e.o, t.o FROM {a} e JOIN {a} t ON t.p = e.s WHERE e.p = {same} AND ((e.o) >> {PAYLOAD_BITS}) = 1", same = v.same),
596        format!("SELECT t.s, t.p, e.o FROM {a} e JOIN {a} t ON t.o = e.s WHERE e.p = {same}", same = v.same),
597        // prp-dom, prp-rng
598        format!("SELECT t.s, {ty}, d.o FROM {a} d JOIN {a} t ON t.p = d.s WHERE d.p = {dom}", ty = v.ty, dom = v.dom),
599        format!("SELECT t.o, {ty}, d.o FROM {a} d JOIN {a} t ON t.p = d.s WHERE d.p = {rng} AND {}", nl("t.o"), ty = v.ty, rng = v.rng),
600        // prp-fp, prp-ifp
601        format!(
602            "SELECT t1.o, {same}, t2.o FROM {a} f JOIN {a} t1 ON t1.p = f.s JOIN {a} t2 ON t2.p = f.s AND t2.s = t1.s \
603             WHERE f.p = {ty} AND f.o = {func} AND t1.o <> t2.o AND {} AND {}",
604            nl("t1.o"), nl("t2.o"), same = v.same, ty = v.ty, func = v.func
605        ),
606        format!(
607            "SELECT t1.s, {same}, t2.s FROM {a} f JOIN {a} t1 ON t1.p = f.s JOIN {a} t2 ON t2.p = f.s AND t2.o = t1.o \
608             WHERE f.p = {ty} AND f.o = {ifunc} AND t1.s <> t2.s",
609            same = v.same, ty = v.ty, ifunc = v.ifunc
610        ),
611        // prp-symp, prp-trp
612        format!("SELECT t.o, t.p, t.s FROM {a} f JOIN {a} t ON t.p = f.s WHERE f.p = {ty} AND f.o = {sym} AND {}", nl("t.o"), ty = v.ty, sym = v.sym),
613        format!(
614            "SELECT t1.s, t1.p, t2.o FROM {a} f JOIN {a} t1 ON t1.p = f.s JOIN {a} t2 ON t2.p = f.s AND t2.s = t1.o WHERE f.p = {ty} AND f.o = {trans}",
615            ty = v.ty, trans = v.trans
616        ),
617        // prp-spo1, prp-eqp1, prp-eqp2
618        format!("SELECT t.s, x.o, t.o FROM {a} x JOIN {a} t ON t.p = x.s WHERE x.p = {spo}", spo = v.spo),
619        format!("SELECT t.s, x.o, t.o FROM {a} x JOIN {a} t ON t.p = x.s WHERE x.p = {eqp}", eqp = v.eqp),
620        format!("SELECT t.s, x.s, t.o FROM {a} x JOIN {a} t ON t.p = x.o WHERE x.p = {eqp}", eqp = v.eqp),
621        // prp-inv1, prp-inv2
622        format!("SELECT t.o, x.o, t.s FROM {a} x JOIN {a} t ON t.p = x.s WHERE x.p = {inv} AND {}", nl("t.o"), inv = v.inv),
623        format!("SELECT t.o, x.s, t.s FROM {a} x JOIN {a} t ON t.p = x.o WHERE x.p = {inv} AND {}", nl("t.o"), inv = v.inv),
624        // cax-sco, cax-eqc1, cax-eqc2
625        format!("SELECT t.s, {ty}, x.o FROM {a} x JOIN {a} t ON t.p = {ty} AND t.o = x.s WHERE x.p = {sco}", ty = v.ty, sco = v.sco),
626        format!("SELECT t.s, {ty}, x.o FROM {a} x JOIN {a} t ON t.p = {ty} AND t.o = x.s WHERE x.p = {eqc}", ty = v.ty, eqc = v.eqc),
627        format!("SELECT t.s, {ty}, x.s FROM {a} x JOIN {a} t ON t.p = {ty} AND t.o = x.o WHERE x.p = {eqc}", ty = v.ty, eqc = v.eqc),
628        // scm-sco (transitive subclasses)
629        format!("SELECT x.s, {sco}, y.o FROM {a} x JOIN {a} y ON y.s = x.o AND y.p = {sco} WHERE x.p = {sco}", sco = v.sco),
630        // scm-eqc1 (the other scm-* rules only restate schema triples whose instance-level
631        // consequences the rules above already derive; `reasonable` omits them too)
632        format!("SELECT s, {sco}, o FROM {a} WHERE p = {eqc} UNION ALL SELECT o, {sco}, s FROM {a} WHERE p = {eqc}", sco = v.sco, eqc = v.eqc),
633        // Every subject is an owl:Thing; owl:Thing and owl:Nothing are classes (cls-thing, cls-nothing1)
634        format!("SELECT s, {ty}, {thing} FROM {a} WHERE {} UNION SELECT {thing}, {ty}, {class} UNION SELECT {nothing}, {ty}, {class}", nl("s"), ty = v.ty, thing = v.thing, class = owl("Class"), nothing = owl("Nothing")),
635        // cls-hv1, cls-hv2
636        format!(
637            "SELECT t.s, op.o, hv.o FROM {a} hv JOIN {a} op ON op.s = hv.s AND op.p = {onp} JOIN {a} t ON t.p = {ty} AND t.o = hv.s WHERE hv.p = {hv}",
638            onp = v.on_property, ty = v.ty, hv = v.has_value
639        ),
640        format!(
641            "SELECT t.s, {ty}, hv.s FROM {a} hv JOIN {a} op ON op.s = hv.s AND op.p = {onp} JOIN {a} t ON t.p = op.o AND t.o = hv.o WHERE hv.p = {hv}",
642            onp = v.on_property, ty = v.ty, hv = v.has_value
643        ),
644        // cls-svf1, cls-svf2
645        format!(
646            "SELECT t.s, {ty}, sv.s FROM {a} sv JOIN {a} op ON op.s = sv.s AND op.p = {onp} JOIN {a} t ON t.p = op.o \
647             JOIN {a} vt ON vt.s = t.o AND vt.p = {ty} AND vt.o = sv.o WHERE sv.p = {svf}",
648            onp = v.on_property, ty = v.ty, svf = v.some_values
649        ),
650        format!(
651            "SELECT t.s, {ty}, sv.s FROM {a} sv JOIN {a} op ON op.s = sv.s AND op.p = {onp} JOIN {a} t ON t.p = op.o WHERE sv.p = {svf} AND sv.o = {thing}",
652            onp = v.on_property, ty = v.ty, svf = v.some_values, thing = v.thing
653        ),
654        // cls-avf
655        format!(
656            "SELECT t.o, {ty}, av.o FROM {a} av JOIN {a} op ON op.s = av.s AND op.p = {onp} JOIN {a} xt ON xt.p = {ty} AND xt.o = av.s \
657             JOIN {a} t ON t.s = xt.s AND t.p = op.o WHERE av.p = {avf} AND {}",
658            nl("t.o"), onp = v.on_property, ty = v.ty, avf = v.all_values
659        ),
660        // cls-int1, cls-int2, cls-uni (list members through rdf:first / rdf:rest)
661        format!(
662            "WITH {lists}, m(c, item) AS (SELECT x.s, l.item FROM {a} x JOIN l ON l.head = x.o WHERE x.p = {int}) \
663             SELECT t.s, {ty}, m.c FROM m JOIN {a} t ON t.p = {ty} AND t.o = m.item GROUP BY m.c, t.s \
664             HAVING COUNT(DISTINCT m.item) = (SELECT COUNT(DISTINCT m2.item) FROM m m2 WHERE m2.c = m.c)",
665            ty = v.ty, int = v.intersection
666        ),
667        format!(
668            "WITH {lists} SELECT t.s, {ty}, l.item FROM {a} x JOIN l ON l.head = x.o JOIN {a} t ON t.p = {ty} AND t.o = x.s WHERE x.p = {int}",
669            ty = v.ty, int = v.intersection
670        ),
671        format!(
672            "WITH {lists} SELECT t.s, {ty}, x.s FROM {a} x JOIN l ON l.head = x.o JOIN {a} t ON t.p = {ty} AND t.o = l.item WHERE x.p = {uni}",
673            ty = v.ty, uni = v.union
674        ),
675        // prp-spo2 (property chains)
676        format!(
677            "WITH {lists}, \
678               ix(head, node, i) AS (SELECT s, s, 1 FROM {a} WHERE p = {first} UNION SELECT ix.head, r.o, ix.i + 1 FROM ix JOIN {a} r ON r.s = ix.node AND r.p = {rest} WHERE r.o <> {nil}), \
679               steps(chain, i, prop) AS (SELECT x.s, ix.i, f.o FROM {a} x JOIN ix ON ix.head = x.o JOIN {a} f ON f.s = ix.node AND f.p = {first} WHERE x.p = {chain}), \
680               w(chain, start, i, cur) AS (SELECT st.chain, t.s, 1, t.o FROM steps st JOIN {a} t ON t.p = st.prop WHERE st.i = 1 \
681                 UNION SELECT w.chain, w.start, w.i + 1, t.o FROM w JOIN steps st ON st.chain = w.chain AND st.i = w.i + 1 JOIN {a} t ON t.s = w.cur AND t.p = st.prop) \
682             SELECT w.start, w.chain, w.cur FROM w WHERE w.i = (SELECT MAX(i) FROM steps s2 WHERE s2.chain = w.chain)",
683            first = v.first, rest = v.rest, nil = v.nil, chain = v.chain
684        ),
685    ];
686    let mut statements: Vec<Statement> = rules
687        .into_iter()
688        .map(|r| {
689            let (with, body) = match r.strip_prefix("WITH ") {
690                Some(rest) => split_with(rest),
691                None => (String::new(), r),
692            };
693            let body = name_columns(&body);
694            Statement::new(format!(
695                "{with}INSERT OR IGNORE INTO quads_inf(s, p, o, g) SELECT DISTINCT r.s, r.p, r.o, {DEFAULT_GRAPH_ID} FROM ({body}) AS r \
696                 WHERE {} AND NOT EXISTS (SELECT 1 FROM quads a WHERE a.s = r.s AND a.p = r.p AND a.o = r.o AND a.g = {DEFAULT_GRAPH_ID})",
697                non_literal("r.s")
698            ))
699        })
700        .collect();
701    statements.push(inference_attribute(OWL_PRODUCER));
702    statements
703}
704
705/// Renames the three result columns of a rule's first SELECT to `s, p, o`.
706fn name_columns(select: &str) -> String {
707    let rest = select.strip_prefix("SELECT ").unwrap_or(select);
708    let from = top_level(rest, " FROM ").unwrap_or(rest.len());
709    let cols: Vec<&str> = split_top(&rest[..from], ',');
710    if cols.len() != 3 {
711        return select.to_string();
712    }
713    format!(
714        "SELECT {} AS s, {} AS p, {} AS o{}",
715        cols[0].trim(),
716        cols[1].trim(),
717        cols[2].trim(),
718        &rest[from..]
719    )
720}
721
722/// Byte offset of the first top-level (outside parentheses) occurrence of `pat`.
723fn top_level(s: &str, pat: &str) -> Option<usize> {
724    let mut depth = 0i32;
725    for (i, c) in s.char_indices() {
726        match c {
727            '(' => depth += 1,
728            ')' => depth -= 1,
729            _ if depth == 0 && s[i..].starts_with(pat) => return Some(i),
730            _ => {}
731        }
732    }
733    None
734}
735
736fn split_top(s: &str, sep: char) -> Vec<&str> {
737    let (mut depth, mut start, mut out) = (0i32, 0usize, Vec::new());
738    for (i, c) in s.char_indices() {
739        match c {
740            '(' => depth += 1,
741            ')' => depth -= 1,
742            c if c == sep && depth == 0 => {
743                out.push(&s[start..i]);
744                start = i + 1;
745            }
746            _ => {}
747        }
748    }
749    out.push(&s[start..]);
750    out
751}
752
753/// Splits `ctes SELECT …` (after `WITH `) into the `WITH … ` prefix and the final SELECT.
754fn split_with(rest: &str) -> (String, String) {
755    // The final SELECT is the last top-level "SELECT" (outside parentheses).
756    let bytes = rest.as_bytes();
757    let (mut depth, mut last) = (0i32, 0usize);
758    for (i, &c) in bytes.iter().enumerate() {
759        match c {
760            b'(' => depth += 1,
761            b')' => depth -= 1,
762            b'S' if depth == 0 && rest[i..].starts_with("SELECT ") => last = i,
763            _ => {}
764        }
765    }
766    (
767        format!("WITH {} ", rest[..last].trim_end()),
768        rest[last..].to_string(),
769    )
770}
771
772/// The producer id of OWL 2 RL materialization (SQL rules and `reasonable` alike).
773pub const OWL_PRODUCER: &str = "owl2rl";
774
775/// The id a producer name is stored under in `quads_inf_src` and `inf_producers`.
776pub fn producer_id(name: &str) -> i64 {
777    // A positive 62-bit hash: stable across runs and backends, computed without a lookup.
778    (xxhash_rust::xxh3::xxh3_64(name.as_bytes()) >> 2) as i64
779}
780
781/// Discards one producer's inferences: its attributions go, and so does every inferred quad no
782/// other producer derived. The producer is (re)registered under its name.
783pub fn inference_reset(name: &str) -> Vec<Statement> {
784    let id = producer_id(name);
785    vec![
786        Statement::new(format!(
787            "INSERT OR REPLACE INTO inf_producers(id, name) VALUES ({id}, '{}')",
788            name.replace('\'', "''")
789        )),
790        Statement::new(format!("DELETE FROM quads_inf_src WHERE src = {id}")),
791        Statement::new(
792            "DELETE FROM quads_inf WHERE NOT EXISTS (SELECT 1 FROM quads_inf_src q \
793             WHERE q.s = quads_inf.s AND q.p = quads_inf.p AND q.o = quads_inf.o AND q.g = quads_inf.g)",
794        ),
795    ]
796}
797
798/// Attributes every inferred quad no producer claims yet to `name`. Run after a producer
799/// writes, so what it derived (and nobody had derived before) is recorded as its own.
800pub fn inference_attribute(name: &str) -> Statement {
801    Statement::new(format!(
802        "INSERT OR IGNORE INTO quads_inf_src(src, s, p, o, g) SELECT {}, i.s, i.p, i.o, i.g FROM quads_inf i \
803         WHERE NOT EXISTS (SELECT 1 FROM quads_inf_src q WHERE q.s = i.s AND q.p = i.p AND q.o = i.o AND q.g = i.g)",
804        producer_id(name)
805    ))
806}
807
808/// Statements run before OWL 2 RL materialization (see [`materialize_reset_for`]).
809pub fn materialize_reset(caps: &crate::sql::Capabilities) -> Vec<Statement> {
810    materialize_reset_for(caps, OWL_PRODUCER)
811}
812
813/// Statements run before a producer materializes: its previous inferences are discarded, and
814/// the vocabulary that rule conclusions use (`owl:sameAs`, `rdfs:subClassOf`, …) gets its term
815/// rows, since it may not appear in the data.
816pub fn materialize_reset_for(caps: &crate::sql::Capabilities, producer: &str) -> Vec<Statement> {
817    let mut rows = crate::encoding::EncodedRows::default();
818    for iri in [
819        rdf::TYPE.as_str(),
820        rdfs::SUB_CLASS_OF.as_str(),
821        rdfs::SUB_PROPERTY_OF.as_str(),
822        rdfs::DOMAIN.as_str(),
823        rdfs::RANGE.as_str(),
824    ] {
825        rows.iri(iri);
826    }
827    for local in [
828        "sameAs",
829        "equivalentClass",
830        "equivalentProperty",
831        "Thing",
832        "Nothing",
833        "Class",
834    ] {
835        rows.iri(&format!("{OWL}{local}"));
836    }
837    rows.dedup();
838    let mut out = inference_reset(producer);
839    out.extend(crate::writer::term_statements(&rows, caps));
840    out
841}