mongreldb_kit_core/
schema.rs

1//! Language-neutral schema model for MongrelDB Kit.
2//!
3//! A [`Schema`] is a collection of [`Table`]s. Each table has [`Column`]s,
4//! indexes, unique constraints, foreign keys, and check constraints.
5
6use serde::{Deserialize, Serialize};
7use std::collections::HashMap;
8
9/// Storage/application types supported by Kit columns.
10#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
11#[serde(rename_all = "snake_case")]
12pub enum ColumnType {
13    Bool,
14    Int8,
15    Int16,
16    Int32,
17    Int64,
18    Float32,
19    Float64,
20    Text,
21    Bytes,
22    Json,
23    Date,
24    DateTime,
25    TimestampNanos,
26    Date64,
27    Time64,
28    Interval,
29    Decimal128,
30    /// RFC 4122 UUID (16 bytes, big-endian).
31    Uuid,
32    /// Native JSON value (stored as typed bytes).
33    JsonNative,
34    /// Variable-length array of homogeneous values.
35    Array,
36    /// A dense float32 vector for nearest-neighbour (ANN) search. The dimension
37    /// is carried on the column as `embedding_dim`.
38    Embedding,
39    /// A learned-sparse (SPLADE-style) weighted token vector, stored as a
40    /// `[[token_id, weight], ...]` list, for sparse retrieval.
41    Sparse,
42}
43
44/// How a default value is produced when a row omits a column.
45#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
46#[serde(rename_all = "snake_case")]
47pub enum DefaultKind {
48    /// A fixed JSON value written literally.
49    Static(serde_json::Value),
50    /// The current timestamp as an ISO-8601 string.
51    Now,
52    /// A fresh UUIDv4 string.
53    Uuid,
54    /// The next value from a named sequence.
55    Sequence(String),
56    /// A user-defined default registered by name (resolved at runtime).
57    CustomName(String),
58}
59
60/// Where dense embedding values for a column originate.
61///
62/// Mirrors `mongreldb_core::EmbeddingSource` in a language-neutral shape
63/// (string paths, serde-tagged). Omitting this on a column means
64/// application-supplied vectors (the engine default). Storage never calls a
65/// vendor from this field alone — generation goes through the process-local
66/// provider registry (see `Database::embed_texts` / `embedding_providers`).
67#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
68#[serde(tag = "kind", rename_all = "snake_case")]
69pub enum EmbeddingSource {
70    /// Application writes float vectors directly (default).
71    SuppliedByApplication,
72    /// Local on-disk model; a provider registered under `model_id` runs inference.
73    LocalModel {
74        /// Filesystem path to model weights / tokenizer bundle.
75        model_path: String,
76        /// Stable model identity (registry key and ANN generation stamp).
77        model_id: String,
78    },
79    /// Named provider registered on the process (`provider` registry key).
80    GeneratedColumn {
81        /// Registry key of the provider.
82        provider: String,
83    },
84}
85
86/// A column definition.
87#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
88pub struct Column {
89    /// Stable column identifier. IDs must be unique within a table.
90    pub id: u32,
91    /// Logical column name.
92    pub name: String,
93    /// Physical storage type.
94    pub storage_type: ColumnType,
95    /// Application-facing type (often the same as `storage_type`).
96    pub application_type: ColumnType,
97    /// Whether the column may contain `null`.
98    pub nullable: bool,
99    /// Whether this column is part of the primary key.
100    pub primary_key: bool,
101    /// Optional default value generator.
102    pub default: Option<DefaultKind>,
103    /// Whether the value is generated on every mutation.
104    pub generated: bool,
105    /// Permitted string values, if any.
106    #[serde(default, skip_serializing_if = "Option::is_none")]
107    pub enum_values: Option<Vec<String>>,
108    /// Minimum numeric value.
109    #[serde(default, skip_serializing_if = "Option::is_none")]
110    pub min: Option<f64>,
111    /// Maximum numeric value.
112    #[serde(default, skip_serializing_if = "Option::is_none")]
113    pub max: Option<f64>,
114    /// Minimum string/bytes length.
115    #[serde(default, skip_serializing_if = "Option::is_none")]
116    pub min_length: Option<usize>,
117    /// Maximum string/bytes length.
118    #[serde(default, skip_serializing_if = "Option::is_none")]
119    pub max_length: Option<usize>,
120    /// Regular expression a `text` value must match, stored as its source pattern.
121    #[serde(default, skip_serializing_if = "Option::is_none")]
122    pub regex: Option<String>,
123    /// An optional check expression name for runtime evaluation.
124    #[serde(default, skip_serializing_if = "Option::is_none")]
125    pub check_expr: Option<String>,
126    /// Vector dimension for an `Embedding` column (required for ANN).
127    #[serde(default, skip_serializing_if = "Option::is_none")]
128    pub embedding_dim: Option<u32>,
129    /// How embedding values are produced. Only meaningful for
130    /// [`ColumnType::Embedding`]. `None` = application-supplied (engine default).
131    #[serde(default, skip_serializing_if = "Option::is_none")]
132    pub embedding_source: Option<EmbeddingSource>,
133    /// Encrypt this column's page payload at rest (requires an encrypted db).
134    #[serde(default, skip_serializing_if = "std::ops::Not::not")]
135    pub encrypted: bool,
136    /// Encrypt the column but keep it queryable via deterministic equality
137    /// tokens / order-preserving encoding (requires an encrypted db).
138    #[serde(default, skip_serializing_if = "std::ops::Not::not")]
139    pub encrypted_indexable: bool,
140}
141
142impl Column {
143    /// Convenience constructor for the common case.
144    pub fn new(id: u32, name: impl Into<String>, storage_type: ColumnType) -> Self {
145        Self {
146            id,
147            name: name.into(),
148            storage_type,
149            application_type: storage_type,
150            nullable: false,
151            primary_key: false,
152            default: None,
153            generated: false,
154            enum_values: None,
155            min: None,
156            max: None,
157            min_length: None,
158            max_length: None,
159            regex: None,
160            check_expr: None,
161            embedding_dim: None,
162            embedding_source: None,
163            encrypted: false,
164            encrypted_indexable: false,
165        }
166    }
167}
168
169/// The kind of secondary index the Kit declares on a column.
170#[derive(Debug, Clone, Copy, PartialEq, Eq, Default, Serialize, Deserialize)]
171#[serde(rename_all = "snake_case")]
172pub enum IndexKind {
173    /// Equality / `IN` acceleration (the default).
174    #[default]
175    Bitmap,
176    /// FM-index substring search (`contains(col, needle)` pushes to `FmContains`).
177    Fm,
178    /// HNSW approximate-nearest-neighbour index for `Embedding` columns.
179    Ann,
180    /// SPLADE-style learned-sparse retrieval index for `Sparse` columns.
181    Sparse,
182    /// MinHash/LSH set-similarity index over a JSON-array set column
183    /// (accelerates `set_similarity`).
184    MinHash,
185    /// Learned zonemap (PGM) index for ordered range predicates on numeric /
186    /// timestamp columns. Accelerates `Range`/`RangeF64` conditions.
187    LearnedRange,
188}
189
190/// An index on one or more columns.
191#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
192pub struct Index {
193    pub name: String,
194    pub columns: Vec<String>,
195    pub unique: bool,
196    /// Index kind; defaults to `Bitmap` so pre-existing schemas deserialize
197    /// unchanged.
198    #[serde(default)]
199    pub kind: IndexKind,
200}
201
202/// A uniqueness constraint over one or more columns.
203#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
204pub struct UniqueConstraint {
205    pub name: String,
206    pub columns: Vec<String>,
207}
208
209/// A foreign-key reference from child columns to parent columns.
210#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
211pub struct ForeignKey {
212    pub name: String,
213    pub columns: Vec<String>,
214    pub references_table: String,
215    pub references_columns: Vec<String>,
216    #[serde(default)]
217    pub on_delete: ForeignKeyAction,
218}
219
220/// Action taken when a referenced parent row is deleted.
221#[derive(Debug, Clone, Copy, PartialEq, Eq, Default, Serialize, Deserialize)]
222#[serde(rename_all = "snake_case")]
223pub enum ForeignKeyAction {
224    #[default]
225    Restrict,
226    Cascade,
227    SetNull,
228}
229
230/// A named table-level check constraint.
231#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
232pub struct CheckConstraint {
233    pub name: String,
234    pub expr: String,
235}
236
237/// A monotonic sequence allocator.
238#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
239pub struct Sequence {
240    pub name: String,
241    pub next_value: i64,
242}
243
244/// A table definition.
245#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
246pub struct Table {
247    /// Stable table identifier. IDs must be unique within a schema.
248    pub id: u32,
249    pub name: String,
250    pub columns: Vec<Column>,
251    pub primary_key: Vec<String>,
252    #[serde(default, skip_serializing_if = "Vec::is_empty")]
253    pub indexes: Vec<Index>,
254    #[serde(default, skip_serializing_if = "Vec::is_empty")]
255    pub foreign_keys: Vec<ForeignKey>,
256    #[serde(default, skip_serializing_if = "Vec::is_empty")]
257    pub unique_constraints: Vec<UniqueConstraint>,
258    #[serde(default, skip_serializing_if = "Vec::is_empty")]
259    pub check_constraints: Vec<CheckConstraint>,
260}
261
262impl Table {
263    /// Find a column by name.
264    pub fn column(&self, name: &str) -> Option<&Column> {
265        self.columns.iter().find(|c| c.name == name)
266    }
267
268    /// Whether the named column is part of the primary key.
269    pub fn is_pk_column(&self, name: &str) -> bool {
270        self.primary_key.iter().any(|c| c == name)
271    }
272}
273
274/// Errors that can occur while constructing a [`Schema`].
275#[derive(Debug, Clone, PartialEq, Eq, thiserror::Error)]
276pub enum SchemaError {
277    #[error("duplicate table name \"{0}\"")]
278    DuplicateTableName(String),
279    #[error("duplicate table id {0}")]
280    DuplicateTableId(u32),
281    #[error("duplicate column name \"{1}\" in table \"{0}\"")]
282    DuplicateColumnName(String, String),
283    #[error("duplicate column id {1} in table \"{0}\"")]
284    DuplicateColumnId(String, u32),
285    #[error("primary key column \"{1}\" not found in table \"{0}\"")]
286    MissingPrimaryKeyColumn(String, String),
287    #[error("index \"{1}\" references unknown column \"{2}\" in table \"{0}\"")]
288    MissingIndexColumn(String, String, String),
289    #[error("unique constraint \"{1}\" references unknown column \"{2}\" in table \"{0}\"")]
290    MissingUniqueColumn(String, String, String),
291    #[error("foreign key \"{1}\" references unknown column \"{2}\" in table \"{0}\"")]
292    MissingForeignKeyColumn(String, String, String),
293    #[error("foreign key \"{1}\" references unknown table \"{2}\"")]
294    MissingReferencedTable(String, String, String),
295    #[error("foreign key \"{1}\" references unknown column \"{2}\" on table \"{3}\"")]
296    MissingReferencedColumn(String, String, String, String),
297    #[error(
298        "column \"{1}\" on table \"{0}\" sets embedding_source but is not an embedding column"
299    )]
300    EmbeddingSourceOnNonEmbedding(String, String),
301    #[error(
302        "embedding column \"{1}\" on table \"{0}\" with LocalModel/GeneratedColumn source requires embedding_dim > 0"
303    )]
304    EmbeddingSourceMissingDim(String, String),
305}
306
307/// A validated collection of tables.
308#[derive(Debug, Clone, PartialEq, Serialize)]
309pub struct Schema {
310    pub tables: Vec<Table>,
311    by_name: HashMap<String, usize>,
312    by_id: HashMap<u32, usize>,
313}
314
315impl<'de> serde::Deserialize<'de> for Schema {
316    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
317    where
318        D: serde::Deserializer<'de>,
319    {
320        #[derive(serde::Deserialize)]
321        struct SchemaHelper {
322            tables: Vec<Table>,
323        }
324        let helper = SchemaHelper::deserialize(deserializer)?;
325        Schema::new(helper.tables).map_err(serde::de::Error::custom)
326    }
327}
328
329/// A unique index also enforces uniqueness (guard-backed), matching SQL where a
330/// UNIQUE index is a UNIQUE constraint. Synthesize a constraint for each unique
331/// index unless an existing (or already-synthesized) unique constraint already
332/// covers exactly the same columns. Mirrors the TypeScript kit's `table()`.
333fn synthesize_unique_from_indexes(table: &mut Table) {
334    let mut synthesized: Vec<UniqueConstraint> = Vec::new();
335    for idx in &table.indexes {
336        if !idx.unique {
337            continue;
338        }
339        let covered = table
340            .unique_constraints
341            .iter()
342            .chain(synthesized.iter())
343            .any(|u| u.columns == idx.columns);
344        if !covered {
345            synthesized.push(UniqueConstraint {
346                name: idx.name.clone(),
347                columns: idx.columns.clone(),
348            });
349        }
350    }
351    table.unique_constraints.extend(synthesized);
352}
353
354impl Schema {
355    /// Build and validate a schema from a list of tables.
356    pub fn new(mut tables: Vec<Table>) -> Result<Self, SchemaError> {
357        for table in &mut tables {
358            synthesize_unique_from_indexes(table);
359        }
360
361        let mut by_name = HashMap::with_capacity(tables.len());
362        let mut by_id = HashMap::with_capacity(tables.len());
363
364        for (idx, table) in tables.iter().enumerate() {
365            if by_name.contains_key(&table.name) {
366                return Err(SchemaError::DuplicateTableName(table.name.clone()));
367            }
368            if by_id.contains_key(&table.id) {
369                return Err(SchemaError::DuplicateTableId(table.id));
370            }
371            by_name.insert(table.name.clone(), idx);
372            by_id.insert(table.id, idx);
373        }
374
375        for table in &tables {
376            Self::validate_table(table, &by_name)?;
377        }
378
379        Ok(Self {
380            tables,
381            by_name,
382            by_id,
383        })
384    }
385
386    fn validate_table(
387        table: &Table,
388        table_names: &HashMap<String, usize>,
389    ) -> Result<(), SchemaError> {
390        let mut column_names = HashMap::with_capacity(table.columns.len());
391        let mut column_ids = HashMap::with_capacity(table.columns.len());
392
393        for col in &table.columns {
394            if column_names.contains_key(&col.name) {
395                return Err(SchemaError::DuplicateColumnName(
396                    table.name.clone(),
397                    col.name.clone(),
398                ));
399            }
400            if column_ids.contains_key(&col.id) {
401                return Err(SchemaError::DuplicateColumnId(table.name.clone(), col.id));
402            }
403            if col.embedding_source.is_some() && col.storage_type != ColumnType::Embedding {
404                return Err(SchemaError::EmbeddingSourceOnNonEmbedding(
405                    table.name.clone(),
406                    col.name.clone(),
407                ));
408            }
409            if matches!(
410                col.embedding_source,
411                Some(EmbeddingSource::LocalModel { .. } | EmbeddingSource::GeneratedColumn { .. })
412            ) && col.embedding_dim.unwrap_or(0) == 0
413            {
414                return Err(SchemaError::EmbeddingSourceMissingDim(
415                    table.name.clone(),
416                    col.name.clone(),
417                ));
418            }
419            column_names.insert(col.name.clone(), col.id);
420            column_ids.insert(col.id, col.name.clone());
421        }
422
423        for pk in &table.primary_key {
424            if !column_names.contains_key(pk) {
425                return Err(SchemaError::MissingPrimaryKeyColumn(
426                    table.name.clone(),
427                    pk.clone(),
428                ));
429            }
430        }
431
432        for idx in &table.indexes {
433            for col in &idx.columns {
434                if !column_names.contains_key(col) {
435                    return Err(SchemaError::MissingIndexColumn(
436                        table.name.clone(),
437                        idx.name.clone(),
438                        col.clone(),
439                    ));
440                }
441            }
442        }
443
444        for uq in &table.unique_constraints {
445            for col in &uq.columns {
446                if !column_names.contains_key(col) {
447                    return Err(SchemaError::MissingUniqueColumn(
448                        table.name.clone(),
449                        uq.name.clone(),
450                        col.clone(),
451                    ));
452                }
453            }
454        }
455
456        for fk in &table.foreign_keys {
457            for col in &fk.columns {
458                if !column_names.contains_key(col) {
459                    return Err(SchemaError::MissingForeignKeyColumn(
460                        table.name.clone(),
461                        fk.name.clone(),
462                        col.clone(),
463                    ));
464                }
465            }
466            if !table_names.contains_key(&fk.references_table) {
467                return Err(SchemaError::MissingReferencedTable(
468                    table.name.clone(),
469                    fk.name.clone(),
470                    fk.references_table.clone(),
471                ));
472            }
473        }
474
475        Ok(())
476    }
477
478    /// Look up a table by name.
479    pub fn table(&self, name: &str) -> Option<&Table> {
480        self.by_name.get(name).map(|&idx| &self.tables[idx])
481    }
482
483    /// Look up a table by stable id.
484    pub fn table_by_id(&self, id: u32) -> Option<&Table> {
485        self.by_id.get(&id).map(|&idx| &self.tables[idx])
486    }
487
488    /// Whether the schema contains a table with the given name.
489    pub fn has_table(&self, name: &str) -> bool {
490        self.by_name.contains_key(name)
491    }
492
493    /// Rename a table in place, keeping the `by_name` index in sync. Returns
494    /// `false` if `from` is absent or `to` is already in use (no change made).
495    /// Does *not* retarget foreign keys — callers that need that should do it
496    /// before/after on the tables they own.
497    pub fn rename_table(&mut self, from: &str, to: &str) -> bool {
498        if from == to {
499            return self.has_table(from);
500        }
501        if !self.has_table(from) || self.has_table(to) {
502            return false;
503        }
504        let idx = *self.by_name.get(from).unwrap();
505        self.tables[idx].name = to.to_string();
506        self.by_name.remove(from);
507        self.by_name.insert(to.to_string(), idx);
508        true
509    }
510}
511
512#[cfg(test)]
513mod tests {
514    use super::*;
515
516    fn make_table(name: &str, id: u32) -> Table {
517        Table {
518            id,
519            name: name.into(),
520            columns: vec![Column::new(1, "id", ColumnType::Int64)],
521            primary_key: vec!["id".into()],
522            indexes: vec![],
523            foreign_keys: vec![],
524            unique_constraints: vec![],
525            check_constraints: vec![],
526        }
527    }
528
529    #[test]
530    fn schema_rejects_duplicate_table_name() {
531        let err = Schema::new(vec![make_table("a", 1), make_table("a", 2)]).unwrap_err();
532        assert!(matches!(err, SchemaError::DuplicateTableName(n) if n == "a"));
533    }
534
535    #[test]
536    fn schema_rejects_duplicate_table_id() {
537        let err = Schema::new(vec![make_table("a", 1), make_table("b", 1)]).unwrap_err();
538        assert!(matches!(err, SchemaError::DuplicateTableId(1)));
539    }
540
541    #[test]
542    fn schema_rejects_missing_pk_column() {
543        let t = Table {
544            id: 1,
545            name: "t".into(),
546            columns: vec![Column::new(1, "x", ColumnType::Text)],
547            primary_key: vec!["id".into()],
548            indexes: vec![],
549            foreign_keys: vec![],
550            unique_constraints: vec![],
551            check_constraints: vec![],
552        };
553        let err = Schema::new(vec![t]).unwrap_err();
554        assert!(matches!(err, SchemaError::MissingPrimaryKeyColumn(_, _)));
555    }
556
557    #[test]
558    fn unique_index_synthesizes_unique_constraint() {
559        let schema = Schema::new(vec![Table {
560            id: 1,
561            name: "users".into(),
562            columns: vec![
563                Column::new(1, "id", ColumnType::Int64),
564                Column::new(2, "email", ColumnType::Text),
565                Column::new(3, "handle", ColumnType::Text),
566            ],
567            primary_key: vec!["id".into()],
568            indexes: vec![
569                Index {
570                    name: "idx_email".into(),
571                    columns: vec!["email".into()],
572                    unique: true,
573                    kind: Default::default(),
574                },
575                // A non-unique index must NOT synthesize a constraint.
576                Index {
577                    name: "idx_handle".into(),
578                    columns: vec!["handle".into()],
579                    unique: false,
580                    kind: Default::default(),
581                },
582            ],
583            foreign_keys: vec![],
584            unique_constraints: vec![],
585            check_constraints: vec![],
586        }])
587        .unwrap();
588        let table = schema.table("users").unwrap();
589        assert_eq!(table.unique_constraints.len(), 1);
590        assert_eq!(
591            table.unique_constraints[0].columns,
592            vec!["email".to_string()]
593        );
594    }
595
596    #[test]
597    fn unique_index_does_not_duplicate_existing_constraint() {
598        let schema = Schema::new(vec![Table {
599            id: 1,
600            name: "users".into(),
601            columns: vec![
602                Column::new(1, "id", ColumnType::Int64),
603                Column::new(2, "email", ColumnType::Text),
604            ],
605            primary_key: vec!["id".into()],
606            indexes: vec![Index {
607                name: "idx_email".into(),
608                columns: vec!["email".into()],
609                unique: true,
610                kind: Default::default(),
611            }],
612            foreign_keys: vec![],
613            unique_constraints: vec![UniqueConstraint {
614                name: "uq_email".into(),
615                columns: vec!["email".into()],
616            }],
617            check_constraints: vec![],
618        }])
619        .unwrap();
620        // The pre-existing constraint already covers `email`; no synthesis.
621        let table = schema.table("users").unwrap();
622        assert_eq!(table.unique_constraints.len(), 1);
623        assert_eq!(table.unique_constraints[0].name, "uq_email");
624    }
625
626    #[test]
627    fn schema_roundtrips_json() {
628        let schema = Schema::new(vec![Table {
629            id: 1,
630            name: "users".into(),
631            columns: vec![
632                Column::new(1, "id", ColumnType::Int64),
633                Column {
634                    nullable: true,
635                    ..Column::new(2, "email", ColumnType::Text)
636                },
637            ],
638            primary_key: vec!["id".into()],
639            indexes: vec![Index {
640                name: "idx_email".into(),
641                columns: vec!["email".into()],
642                unique: true,
643                kind: Default::default(),
644            }],
645            foreign_keys: vec![],
646            unique_constraints: vec![],
647            check_constraints: vec![CheckConstraint {
648                name: "chk_id_positive".into(),
649                expr: "id > 0".into(),
650            }],
651        }])
652        .unwrap();
653
654        let json = serde_json::to_string(&schema).unwrap();
655        let decoded: Schema = serde_json::from_str(&json).unwrap();
656        assert_eq!(decoded.tables.len(), 1);
657        assert_eq!(decoded.table("users").unwrap().columns.len(), 2);
658    }
659
660    #[test]
661    fn embedding_source_roundtrips_json() {
662        let mut emb = Column::new(2, "vec", ColumnType::Embedding);
663        emb.embedding_dim = Some(4);
664        emb.embedding_source = Some(EmbeddingSource::LocalModel {
665            model_path: "/models/kit-mini".into(),
666            model_id: "kit-mini".into(),
667        });
668        let schema = Schema::new(vec![Table {
669            id: 1,
670            name: "docs".into(),
671            columns: vec![Column::new(1, "id", ColumnType::Int64), emb],
672            primary_key: vec!["id".into()],
673            indexes: vec![],
674            foreign_keys: vec![],
675            unique_constraints: vec![],
676            check_constraints: vec![],
677        }])
678        .unwrap();
679        let json = serde_json::to_string(&schema).unwrap();
680        assert!(json.contains("local_model"));
681        assert!(json.contains("kit-mini"));
682        let decoded: Schema = serde_json::from_str(&json).unwrap();
683        let col = decoded.table("docs").unwrap().column("vec").unwrap();
684        assert_eq!(
685            col.embedding_source,
686            Some(EmbeddingSource::LocalModel {
687                model_path: "/models/kit-mini".into(),
688                model_id: "kit-mini".into(),
689            })
690        );
691    }
692
693    #[test]
694    fn embedding_source_rejected_on_non_embedding() {
695        let mut col = Column::new(2, "name", ColumnType::Text);
696        col.embedding_source = Some(EmbeddingSource::SuppliedByApplication);
697        let err = Schema::new(vec![Table {
698            id: 1,
699            name: "t".into(),
700            columns: vec![Column::new(1, "id", ColumnType::Int64), col],
701            primary_key: vec!["id".into()],
702            indexes: vec![],
703            foreign_keys: vec![],
704            unique_constraints: vec![],
705            check_constraints: vec![],
706        }])
707        .unwrap_err();
708        assert!(matches!(
709            err,
710            SchemaError::EmbeddingSourceOnNonEmbedding(_, _)
711        ));
712    }
713
714    #[test]
715    fn generated_embedding_requires_dim() {
716        let mut emb = Column::new(2, "vec", ColumnType::Embedding);
717        emb.embedding_source = Some(EmbeddingSource::GeneratedColumn {
718            provider: "local-test".into(),
719        });
720        let err = Schema::new(vec![Table {
721            id: 1,
722            name: "t".into(),
723            columns: vec![Column::new(1, "id", ColumnType::Int64), emb],
724            primary_key: vec!["id".into()],
725            indexes: vec![],
726            foreign_keys: vec![],
727            unique_constraints: vec![],
728            check_constraints: vec![],
729        }])
730        .unwrap_err();
731        assert!(matches!(err, SchemaError::EmbeddingSourceMissingDim(_, _)));
732    }
733}
mongreldb_kit_core/schema.rs

mongreldb_kit_core/
schema.rs
