khive_types/
entity.rs

1//! Entity substrate — graph nodes with typed properties and links.
2
3extern crate alloc;
4use alloc::collections::BTreeMap;
5use alloc::string::String;
6use alloc::vec::Vec;
7use core::fmt;
8use core::str::FromStr;
9
10use crate::{EdgeRelation, Header, Id128, Timestamp};
11
12/// 8 closed base kinds for graph-node classification.
13///
14/// Governed subtype values live in `Entity::entity_type`; `properties` remain
15/// metadata and must not carry ontology type strings.
16#[derive(Clone, Copy, Debug, Default, PartialEq, Eq, Hash)]
17#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
18#[cfg_attr(feature = "serde", serde(rename_all = "snake_case"))]
19pub enum EntityKind {
20    /// Algorithms, techniques, architectures, theories, models, research gaps.
21    /// The default / residual bucket.
22    #[default]
23    Concept,
24    /// Papers, preprints, technical reports, blog posts, books.
25    /// Has: title, authors, year, venue, DOI/URL.
26    Document,
27    /// Benchmarks, corpora, evaluation sets.
28    /// Has: task type, size, metrics, license.
29    Dataset,
30    /// Codebases, libraries, tools, frameworks.
31    /// Has: language, repo URL, license.
32    Project,
33    /// Researchers, engineers, authors.
34    Person,
35    /// Labs, companies, institutions.
36    Org,
37    /// Built artifacts: binaries, model checkpoints, Docker images, packages.
38    Artifact,
39    /// Running or deployable services: APIs, hosted endpoints, SaaS products.
40    Service,
41}
42
43impl EntityKind {
44    /// All 8 canonical entity kinds in taxonomy-table order.
45    pub const ALL: [Self; 8] = [
46        Self::Concept,
47        Self::Document,
48        Self::Dataset,
49        Self::Project,
50        Self::Person,
51        Self::Org,
52        Self::Artifact,
53        Self::Service,
54    ];
55
56    /// Return the canonical lowercase string for this kind, as stored on the wire.
57    pub const fn name(self) -> &'static str {
58        match self {
59            Self::Concept => "concept",
60            Self::Document => "document",
61            Self::Dataset => "dataset",
62            Self::Project => "project",
63            Self::Person => "person",
64            Self::Org => "org",
65            Self::Artifact => "artifact",
66            Self::Service => "service",
67        }
68    }
69}
70
71impl fmt::Display for EntityKind {
72    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
73        f.write_str(self.name())
74    }
75}
76
77// Canonical entity kind strings for the closed 8-kind taxonomy.
78const ENTITY_KIND_VALID: &[&str] = &[
79    "concept", "document", "dataset", "project", "person", "org", "artifact", "service",
80];
81
82impl FromStr for EntityKind {
83    type Err = crate::error::UnknownVariant;
84
85    /// Parse a string into an `EntityKind`.
86    ///
87    /// Accepts the 8 canonical kind names (case-insensitive) plus a set of
88    /// convenience aliases to aid human-authored DSL requests (e.g. `"paper"`
89    /// resolves to `Document`, `"repo"` to `Project`).
90    ///
91    /// **Note on subtype aliasing**: when `kind="paper"` is parsed here, only the
92    /// base `EntityKind::Document` is returned.  Callers that need to preserve the
93    /// `entity_type` subtoken must use the pack registry resolution path, which
94    /// returns both the base kind and the subtype string.  `from_str` is
95    /// intentionally base-kind-only for use in contexts where the subtype is
96    /// carried separately (e.g. `Entity.entity_type`).
97    fn from_str(s: &str) -> Result<Self, Self::Err> {
98        match s.trim().to_ascii_lowercase().as_str() {
99            "concept" => Ok(Self::Concept),
100            "document" | "doc" | "paper" => Ok(Self::Document),
101            "dataset" | "data" | "benchmark" => Ok(Self::Dataset),
102            "project" | "repo" | "crate" | "library" | "lib" => Ok(Self::Project),
103            "person" | "author" | "researcher" => Ok(Self::Person),
104            "org" | "organization" | "organisation" | "lab" | "company" => Ok(Self::Org),
105            "artifact" | "art" => Ok(Self::Artifact),
106            "service" | "svc" => Ok(Self::Service),
107            other => Err(crate::error::UnknownVariant::new(
108                "entity_kind",
109                other,
110                ENTITY_KIND_VALID,
111            )),
112        }
113    }
114}
115
116/// A graph node with a type, display name, and key-value properties.
117#[derive(Clone, Debug)]
118#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
119pub struct Entity {
120    /// Identity and namespace metadata shared by all substrate records.
121    #[cfg_attr(feature = "serde", serde(flatten))]
122    pub header: Header,
123    /// Closed base kind that classifies this entity.
124    pub kind: EntityKind,
125    /// Pack-governed subtype token (e.g. `"paper"`, `"snapshot"`). Never stored
126    /// raw in `properties` — queries compile this to `entities.entity_type = ?`.
127    pub entity_type: Option<String>,
128    /// Human-readable display name (required; must be non-empty).
129    pub name: String,
130    /// Optional long-form description of this entity.
131    pub description: Option<String>,
132    /// Arbitrary structured metadata as key-value pairs.
133    pub properties: BTreeMap<String, PropertyValue>,
134    /// Categorical labels for filtering and retrieval.
135    pub tags: Vec<String>,
136    /// Set when the entity is soft-deleted; absent means active.
137    pub deleted_at: Option<Timestamp>,
138}
139
140/// A directed, typed edge between two entities (or cross-substrate nodes).
141///
142/// `weight` must be finite and in `[0.0, 1.0]`. When the `serde` feature is
143/// enabled, deserialization rejects out-of-range or non-finite weights.
144#[derive(Clone, Debug)]
145#[cfg_attr(feature = "serde", derive(serde::Serialize))]
146#[cfg_attr(feature = "serde", serde(into = "LinkRaw"))]
147pub struct Link {
148    /// Unique edge identifier.
149    pub id: Id128,
150    /// Namespace that owns and isolates this edge.
151    pub namespace: String,
152    /// Source node identifier.
153    pub source: Id128,
154    /// Target node identifier.
155    pub target: Id128,
156    /// Closed relation type that semantically describes this edge.
157    pub relation: EdgeRelation,
158    /// Arbitrary structured metadata attached to this edge.
159    pub properties: BTreeMap<String, PropertyValue>,
160    /// Numeric edge weight in the range [0.0, 1.0]; 1.0 means definitional strength.
161    pub weight: f64,
162    /// Wall-clock time when this edge was created.
163    pub created_at: Timestamp,
164    /// Wall-clock time of the most recent update.
165    pub updated_at: Timestamp,
166    /// Set when the edge is soft-deleted; absent means active.
167    pub deleted_at: Option<Timestamp>,
168}
169
170/// Return whether an edge weight is finite and within `[0.0, 1.0]`.
171///
172/// Both endpoints and signed zero are accepted; NaN and infinities are rejected.
173pub fn validate_edge_weight(weight: f64) -> bool {
174    weight.is_finite() && (0.0..=1.0).contains(&weight)
175}
176
177impl Link {
178    /// Return `true` if all numeric fields carry finite, domain-valid values.
179    ///
180    /// - `weight` must be finite and in `[0.0, 1.0]`.
181    pub fn is_valid(&self) -> bool {
182        validate_edge_weight(self.weight)
183    }
184}
185
186#[cfg(feature = "serde")]
187#[derive(serde::Serialize, serde::Deserialize)]
188struct LinkRaw {
189    id: Id128,
190    namespace: String,
191    source: Id128,
192    target: Id128,
193    relation: EdgeRelation,
194    properties: BTreeMap<String, PropertyValue>,
195    weight: f64,
196    created_at: Timestamp,
197    updated_at: Timestamp,
198    deleted_at: Option<Timestamp>,
199}
200
201#[cfg(feature = "serde")]
202impl From<Link> for LinkRaw {
203    fn from(l: Link) -> Self {
204        Self {
205            id: l.id,
206            namespace: l.namespace,
207            source: l.source,
208            target: l.target,
209            relation: l.relation,
210            properties: l.properties,
211            weight: l.weight,
212            created_at: l.created_at,
213            updated_at: l.updated_at,
214            deleted_at: l.deleted_at,
215        }
216    }
217}
218
219#[cfg(feature = "serde")]
220impl TryFrom<LinkRaw> for Link {
221    type Error = String;
222
223    fn try_from(raw: LinkRaw) -> Result<Self, Self::Error> {
224        if !raw.weight.is_finite() {
225            return Err(alloc::format!(
226                "Link weight must be finite, got {}",
227                raw.weight
228            ));
229        }
230        if !(0.0..=1.0).contains(&raw.weight) {
231            return Err(alloc::format!(
232                "Link weight must be in [0.0, 1.0], got {}",
233                raw.weight
234            ));
235        }
236        Ok(Link {
237            id: raw.id,
238            namespace: raw.namespace,
239            source: raw.source,
240            target: raw.target,
241            relation: raw.relation,
242            properties: raw.properties,
243            weight: raw.weight,
244            created_at: raw.created_at,
245            updated_at: raw.updated_at,
246            deleted_at: raw.deleted_at,
247        })
248    }
249}
250
251#[cfg(feature = "serde")]
252impl<'de> serde::Deserialize<'de> for Link {
253    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
254    where
255        D: serde::Deserializer<'de>,
256    {
257        let raw = LinkRaw::deserialize(deserializer)?;
258        Link::try_from(raw).map_err(serde::de::Error::custom)
259    }
260}
261
262/// Property values stored on entities, links, and notes.
263///
264/// Recursive: supports arrays and nested objects for free-form JSON properties
265/// (e.g. `entity_ids[]`, `alternatives_considered[]`).
266#[derive(Clone, Debug, PartialEq)]
267#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
268#[cfg_attr(feature = "serde", serde(untagged))]
269pub enum PropertyValue {
270    String(String),
271    Integer(i64),
272    Float(f64),
273    Boolean(bool),
274    Array(Vec<PropertyValue>),
275    Object(BTreeMap<String, PropertyValue>),
276    Null,
277}
278
279impl fmt::Display for PropertyValue {
280    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
281        match self {
282            Self::String(s) => f.write_str(s),
283            Self::Integer(n) => write!(f, "{n}"),
284            Self::Float(n) => write!(f, "{n}"),
285            Self::Boolean(b) => write!(f, "{b}"),
286            Self::Array(arr) => write!(f, "[{} items]", arr.len()),
287            Self::Object(obj) => write!(f, "{{{} keys}}", obj.len()),
288            Self::Null => f.write_str("null"),
289        }
290    }
291}
292
293#[cfg(test)]
294mod tests {
295    use super::*;
296    use crate::{Namespace, Timestamp};
297    #[cfg(feature = "serde")]
298    use alloc::string::ToString;
299
300    #[test]
301    fn edge_weight_validation_covers_numeric_boundaries() {
302        for weight in [-0.0, 0.0, f64::from_bits(1), 0.5, 1.0] {
303            assert!(crate::validate_edge_weight(weight), "{weight:?}");
304        }
305        for weight in [
306            f64::NAN,
307            f64::INFINITY,
308            f64::NEG_INFINITY,
309            -f64::from_bits(1),
310            1.0 + f64::EPSILON,
311        ] {
312            assert!(!crate::validate_edge_weight(weight), "{weight:?}");
313        }
314    }
315
316    #[test]
317    fn entity_with_properties() {
318        let mut props = BTreeMap::new();
319        props.insert("role".into(), PropertyValue::String("engineer".into()));
320        props.insert("age".into(), PropertyValue::Integer(30));
321
322        let entity = Entity {
323            header: Header::new(
324                Id128::from_u128(1),
325                Namespace::local(),
326                Timestamp::from_secs(1700000000),
327            ),
328            kind: EntityKind::Person,
329            entity_type: Some("researcher".into()),
330            name: "Operator".into(),
331            description: None,
332            properties: props,
333            tags: alloc::vec![],
334            deleted_at: None,
335        };
336        assert_eq!(entity.kind, EntityKind::Person);
337        assert_eq!(entity.kind.name(), "person");
338        assert_eq!(entity.entity_type.as_deref(), Some("researcher"));
339        assert_eq!(entity.properties.len(), 2);
340    }
341
342    #[test]
343    fn entity_kind_default_is_concept() {
344        assert_eq!(EntityKind::default(), EntityKind::Concept);
345    }
346
347    #[test]
348    fn entity_kind_display_roundtrip() {
349        for kind in EntityKind::ALL {
350            let s = alloc::format!("{kind}");
351            let parsed = EntityKind::from_str(&s).unwrap();
352            assert_eq!(parsed, kind);
353        }
354    }
355
356    #[test]
357    fn entity_kind_from_str_aliases() {
358        assert_eq!(EntityKind::from_str("doc").unwrap(), EntityKind::Document);
359        assert_eq!(EntityKind::from_str("paper").unwrap(), EntityKind::Document);
360        assert_eq!(
361            EntityKind::from_str("benchmark").unwrap(),
362            EntityKind::Dataset
363        );
364        assert_eq!(EntityKind::from_str("repo").unwrap(), EntityKind::Project);
365        assert_eq!(EntityKind::from_str("author").unwrap(), EntityKind::Person);
366        assert_eq!(EntityKind::from_str("lab").unwrap(), EntityKind::Org);
367        assert_eq!(EntityKind::from_str("art").unwrap(), EntityKind::Artifact);
368        assert_eq!(EntityKind::from_str("svc").unwrap(), EntityKind::Service);
369    }
370
371    #[test]
372    fn entity_kind_artifact_and_service_roundtrip() {
373        assert_eq!(EntityKind::Artifact.name(), "artifact");
374        assert_eq!(EntityKind::Service.name(), "service");
375        assert_eq!(
376            EntityKind::from_str("artifact").unwrap(),
377            EntityKind::Artifact
378        );
379        assert_eq!(
380            EntityKind::from_str("service").unwrap(),
381            EntityKind::Service
382        );
383    }
384
385    #[test]
386    fn entity_kind_all_has_eight_variants() {
387        assert_eq!(EntityKind::ALL.len(), 8);
388        assert!(EntityKind::ALL.contains(&EntityKind::Artifact));
389        assert!(EntityKind::ALL.contains(&EntityKind::Service));
390    }
391
392    #[test]
393    fn entity_kind_unknown_valid_list_includes_new_kinds() {
394        let err = EntityKind::from_str("gadget").unwrap_err();
395        assert!(err.valid.contains(&"artifact"));
396        assert!(err.valid.contains(&"service"));
397    }
398
399    #[test]
400    fn entity_kind_from_str_case_insensitive() {
401        assert_eq!(
402            EntityKind::from_str("CONCEPT").unwrap(),
403            EntityKind::Concept
404        );
405        assert_eq!(EntityKind::from_str("Person").unwrap(), EntityKind::Person);
406    }
407
408    #[test]
409    fn entity_kind_from_str_unknown_errors() {
410        let err = EntityKind::from_str("gadget").unwrap_err();
411        assert_eq!(err.domain, "entity_kind");
412        assert_eq!(err.value, "gadget");
413        assert!(err.valid.contains(&"concept"));
414    }
415
416    #[test]
417    fn link_construction() {
418        let ts = Timestamp::from_secs(1700000000);
419        let link = Link {
420            id: Id128::from_u128(100),
421            namespace: "default".into(),
422            source: Id128::from_u128(1),
423            target: Id128::from_u128(2),
424            relation: EdgeRelation::Extends,
425            properties: BTreeMap::new(),
426            weight: 1.0,
427            created_at: ts,
428            updated_at: ts,
429            deleted_at: None,
430        };
431        assert_eq!(link.relation, EdgeRelation::Extends);
432        assert!(link.is_valid());
433    }
434
435    #[test]
436    fn link_is_valid_rejects_out_of_range() {
437        let ts = Timestamp::from_secs(1700000000);
438        let link = Link {
439            id: Id128::from_u128(100),
440            namespace: "default".into(),
441            source: Id128::from_u128(1),
442            target: Id128::from_u128(2),
443            relation: EdgeRelation::Extends,
444            properties: BTreeMap::new(),
445            weight: 2.0,
446            created_at: ts,
447            updated_at: ts,
448            deleted_at: None,
449        };
450        assert!(!link.is_valid());
451    }
452
453    #[cfg(feature = "serde")]
454    #[test]
455    fn link_serde_rejects_weight_above_one() {
456        let json = serde_json::json!({
457            "id": "00000000-0000-0000-0000-000000000064",
458            "namespace": "default",
459            "source": "00000000-0000-0000-0000-000000000001",
460            "target": "00000000-0000-0000-0000-000000000002",
461            "relation": "extends",
462            "properties": {},
463            "weight": 2.0,
464            "created_at": 1700000000000000_u64,
465            "updated_at": 1700000000000000_u64,
466            "deleted_at": null
467        });
468        let result: Result<Link, _> = serde_json::from_value(json);
469        assert!(result.is_err());
470        let err = result.unwrap_err().to_string();
471        assert!(
472            err.contains("[0.0, 1.0]"),
473            "error should mention range: {err}"
474        );
475    }
476
477    #[cfg(feature = "serde")]
478    #[test]
479    fn link_serde_rejects_negative_weight() {
480        let json = serde_json::json!({
481            "id": "00000000-0000-0000-0000-000000000064",
482            "namespace": "default",
483            "source": "00000000-0000-0000-0000-000000000001",
484            "target": "00000000-0000-0000-0000-000000000002",
485            "relation": "extends",
486            "properties": {},
487            "weight": -0.1,
488            "created_at": 1700000000000000_u64,
489            "updated_at": 1700000000000000_u64,
490            "deleted_at": null
491        });
492        let result: Result<Link, _> = serde_json::from_value(json);
493        assert!(result.is_err());
494    }
495
496    #[cfg(feature = "serde")]
497    #[test]
498    fn link_serde_accepts_valid_weight() {
499        let json = serde_json::json!({
500            "id": "00000000-0000-0000-0000-000000000064",
501            "namespace": "default",
502            "source": "00000000-0000-0000-0000-000000000001",
503            "target": "00000000-0000-0000-0000-000000000002",
504            "relation": "extends",
505            "properties": {},
506            "weight": 0.75,
507            "created_at": 1700000000000000_u64,
508            "updated_at": 1700000000000000_u64,
509            "deleted_at": null
510        });
511        let link: Link = serde_json::from_value(json).expect("valid weight should deserialize");
512        assert_eq!(link.weight, 0.75);
513    }
514}
khive_types/entity.rs

khive_types/
entity.rs
