|
| 1 | +// Copyright (c) 2026 ObjectStack. Licensed under the Apache-2.0 license. |
| 2 | + |
| 3 | +/** |
| 4 | + * `$contains` / `$notContains` on `translateFilter`, by the field's DECLARED |
| 5 | + * value shape — the contract `FILTER_OPERATORS`' `$contains` docblock |
| 6 | + * (`@objectstack/spec`) states: MEMBERSHIP on a `multiple: true` field or a |
| 7 | + * JSON-stored type, the SUBSTRING test on a scalar string column, as |
| 8 | + * `driver-sql` answers on every dialect and `driver-memory` on every face. |
| 9 | + * |
| 10 | + * | declared shape | `$contains: v` | `$notContains: v` | |
| 11 | + * |---|---|---| |
| 12 | + * | JSON-stored | `{ f: { $elemMatch: { $in: members, $not: { $type: 'array' } } } }` | the same test under `$not` | |
| 13 | + * | anything else, or none held | `{ f: { $regex: escaped(v) } }` | `{ f: { $not: { $regex } } }` | |
| 14 | + * |
| 15 | + * Before, every field took the `$regex`, and MongoDB applies a `$regex` to each |
| 16 | + * element of an array value: `'u1'` matched a stored `['u10']`, the cell this |
| 17 | + * file pins first, beside a scalar text column that still answers substring. |
| 18 | + * |
| 19 | + * Pinned twice, as `mongodb-20444-empty-operator.test.ts` pins `$empty`: the |
| 20 | + * emitted DOCUMENTS, and the rows they select under a server-free reading of |
| 21 | + * the MongoDB semantics those documents use. A live `mongod` suite runs the |
| 22 | + * same cases when the opt-in server is available. |
| 23 | + */ |
| 24 | + |
| 25 | +import { afterAll, beforeAll, describe, expect, it } from 'vitest'; |
| 26 | +import type { MongoMemoryServer } from 'mongodb-memory-server'; |
| 27 | +import type { FilterCondition } from '@objectstack/spec/data'; |
| 28 | +import { translateFilter, type ValueShapeResolver } from './mongodb-filter.js'; |
| 29 | +import { buildAggregationPipeline } from './mongodb-aggregation.js'; |
| 30 | +import { MongoDBDriver } from './mongodb-driver.js'; |
| 31 | +import { createTestMongod } from './test-mongod.js'; |
| 32 | + |
| 33 | +const OBJECT = 'os_contains_membership'; |
| 34 | + |
| 35 | +const FIELDS: Record<string, { type: string; multiple?: boolean; reference?: string }> = { |
| 36 | + title: { type: 'text' }, |
| 37 | + owners: { type: 'lookup', reference: OBJECT, multiple: true }, |
| 38 | + tags: { type: 'tags' }, |
| 39 | + meta: { type: 'json' }, |
| 40 | +}; |
| 41 | +const SHAPES: ValueShapeResolver = (field) => FIELDS[field]; |
| 42 | + |
| 43 | +const ROWS: Array<Record<string, unknown>> = [ |
| 44 | + { id: 'r1', title: 'u10', owners: ['u10'], tags: ['redwood'], meta: [10, 21] }, |
| 45 | + { id: 'r2', title: 'u1', owners: ['u1', 'u2'], tags: ['red'], meta: [1, 2] }, |
| 46 | + { id: 'r3', title: 'x', owners: [], tags: [], meta: [true, null] }, |
| 47 | + { id: 'r4', title: null, owners: null, tags: null, meta: null }, |
| 48 | + { id: 'r5' }, |
| 49 | + { id: 'r6', title: 'y', owners: [['u1']], tags: ['RED'], meta: { k: 'u1' } }, |
| 50 | + { id: 'r7', title: 'z', owners: 'u1', tags: ['blue'], meta: [1.5] }, |
| 51 | +]; |
| 52 | + |
| 53 | +const CASES: Array<{ name: string; where: FilterCondition; expected: string[] }> = [ |
| 54 | + { name: "'u1' against a stored ['u10'] is not a member", where: { owners: { $contains: 'u1' } }, expected: ['r2'] }, |
| 55 | + { name: 'the exact complement, the rows with no value included', where: { owners: { $notContains: 'u1' } }, expected: ['r1', 'r3', 'r4', 'r5', 'r6', 'r7'] }, |
| 56 | + { name: 'the scalar control: a text column keeps the substring test', where: { title: { $contains: 'u1' } }, expected: ['r1', 'r2'] }, |
| 57 | + { name: 'the scalar control, negated', where: { title: { $notContains: 'u1' } }, expected: ['r3', 'r4', 'r5', 'r6', 'r7'] }, |
| 58 | + { name: "a tags field: 'red' is not a member of ['redwood'], and the test is case-exact", where: { tags: { $contains: 'red' } }, expected: ['r2'] }, |
| 59 | + { name: "a json field: '1' names the number 1, and not 10 or 21", where: { meta: { $contains: '1' } }, expected: ['r2'] }, |
| 60 | + { name: "a json field: '1.50' names the number 1.5", where: { meta: { $contains: '1.50' } }, expected: ['r7'] }, |
| 61 | + { name: "a json field: 'true' and 'null' name the literals", where: { $and: [{ meta: { $contains: 'true' } }, { meta: { $contains: 'null' } }] }, expected: ['r3'] }, |
| 62 | + { name: 'an object root has no member', where: { meta: { $contains: 'u1' } }, expected: [] }, |
| 63 | + { name: 'under $or beside a scalar column', where: { $or: [{ owners: { $contains: 'u2' } }, { title: { $contains: 'x' } }] }, expected: ['r2', 'r3'] }, |
| 64 | +]; |
| 65 | + |
| 66 | +// ── A server-free reading of the MongoDB vocabulary these documents use ───── |
| 67 | + |
| 68 | +/** `$elemMatch` with operator queries: some ELEMENT of an array value satisfies every one. */ |
| 69 | +function elemMatches(value: unknown, query: Record<string, unknown>): boolean { |
| 70 | + if (!Array.isArray(value)) return false; |
| 71 | + return value.some((element) => Object.entries(query).every(([op, arg]) => { |
| 72 | + switch (op) { |
| 73 | + // A scalar element equals a member of its own BSON type only: `'1'` never equals `1`. |
| 74 | + case '$in': return (arg as unknown[]).some((member) => element === member); |
| 75 | + case '$not': { |
| 76 | + const inner = arg as Record<string, unknown>; |
| 77 | + if (Object.keys(inner).length !== 1 || inner.$type !== 'array') throw new Error(`unmodelled $not ${JSON.stringify(inner)}`); |
| 78 | + return !Array.isArray(element); |
| 79 | + } |
| 80 | + default: throw new Error(`unmodelled $elemMatch operator ${op}`); |
| 81 | + } |
| 82 | + })); |
| 83 | +} |
| 84 | + |
| 85 | +/** `$regex` on a field: a string value, or ANY string element of an array value (one level). */ |
| 86 | +function regexMatches(value: unknown, pattern: string): boolean { |
| 87 | + const re = new RegExp(pattern); |
| 88 | + if (typeof value === 'string') return re.test(value); |
| 89 | + if (Array.isArray(value)) return value.some((element) => typeof element === 'string' && re.test(element)); |
| 90 | + return false; |
| 91 | +} |
| 92 | + |
| 93 | +function matchOps(value: unknown, ops: Record<string, unknown>): boolean { |
| 94 | + for (const [op, arg] of Object.entries(ops)) { |
| 95 | + switch (op) { |
| 96 | + case '$elemMatch': if (!elemMatches(value, arg as Record<string, unknown>)) return false; break; |
| 97 | + case '$regex': if (!regexMatches(value, arg as string)) return false; break; |
| 98 | + // MongoDB's field `$not` also matches a document whose field is missing. |
| 99 | + case '$not': if (matchOps(value, arg as Record<string, unknown>)) return false; break; |
| 100 | + default: throw new Error(`unmodelled field operator ${op}`); |
| 101 | + } |
| 102 | + } |
| 103 | + return true; |
| 104 | +} |
| 105 | + |
| 106 | +/** MongoDB implicit equality: a scalar matches itself, and an array value matches by element. */ |
| 107 | +function equals(value: unknown, comparand: unknown): boolean { |
| 108 | + if (Array.isArray(value)) return value.some((element) => element === comparand); |
| 109 | + return value === comparand; |
| 110 | +} |
| 111 | + |
| 112 | +function matchDoc(row: Record<string, unknown>, doc: Record<string, unknown>): boolean { |
| 113 | + for (const [key, value] of Object.entries(doc)) { |
| 114 | + if (key === '$and') { if (!(value as Array<Record<string, unknown>>).every((d) => matchDoc(row, d))) return false; continue; } |
| 115 | + if (key === '$or') { if (!(value as Array<Record<string, unknown>>).some((d) => matchDoc(row, d))) return false; continue; } |
| 116 | + if (key.startsWith('$')) throw new Error(`unmodelled document operator ${key}`); |
| 117 | + const cond = value as Record<string, unknown>; |
| 118 | + const isOps = cond !== null && typeof cond === 'object' && !Array.isArray(cond) |
| 119 | + && Object.keys(cond).every((k) => k.startsWith('$')); |
| 120 | + if (isOps ? !matchOps(row[key], cond) : !equals(row[key], value)) return false; |
| 121 | + } |
| 122 | + return true; |
| 123 | +} |
| 124 | + |
| 125 | +const select = (doc: Record<string, unknown>) => ROWS.filter((r) => matchDoc(r, doc)).map((r) => String(r.id)).sort(); |
| 126 | + |
| 127 | +describe('translateFilter — $contains asks membership on a declared JSON-stored field', () => { |
| 128 | + it('emits the membership test on a JSON-stored field and the substring pattern on a scalar column', () => { |
| 129 | + const membership = (members: unknown[]) => ({ $elemMatch: { $in: members, $not: { $type: 'array' } } }); |
| 130 | + expect(translateFilter({ owners: { $contains: 'u1' } }, undefined, SHAPES)).toEqual({ owners: membership(['u1']) }); |
| 131 | + expect(translateFilter({ owners: { $notContains: 'u1' } }, undefined, SHAPES)).toEqual({ owners: { $not: membership(['u1']) } }); |
| 132 | + expect(translateFilter({ meta: { $contains: '1' } }, undefined, SHAPES)).toEqual({ meta: membership(['1', 1]) }); |
| 133 | + expect(translateFilter({ meta: { $contains: '1.50' } }, undefined, SHAPES)).toEqual({ meta: membership(['1.50', 1.5]) }); |
| 134 | + expect(translateFilter({ meta: { $contains: 'true' } }, undefined, SHAPES)).toEqual({ meta: membership(['true', true]) }); |
| 135 | + expect(translateFilter({ meta: { $contains: 'null' } }, undefined, SHAPES)).toEqual({ meta: membership(['null', null]) }); |
| 136 | + expect(translateFilter({ title: { $contains: 'u1' } }, undefined, SHAPES)).toEqual({ title: { $regex: 'u1' } }); |
| 137 | + expect(translateFilter({ title: { $notContains: 'u1' } }, undefined, SHAPES)).toEqual({ title: { $not: { $regex: 'u1' } } }); |
| 138 | + }); |
| 139 | + |
| 140 | + it('a member is matched literally: no regex metacharacter reaches the membership test', () => { |
| 141 | + expect(translateFilter({ tags: { $contains: 'a.b+c' } }, undefined, SHAPES)) |
| 142 | + .toEqual({ tags: { $elemMatch: { $in: ['a.b+c'], $not: { $type: 'array' } } } }); |
| 143 | + }); |
| 144 | + |
| 145 | + for (const c of CASES) { |
| 146 | + it(`${c.name}: ${JSON.stringify(c.where)} selects ${JSON.stringify(c.expected)}`, () => { |
| 147 | + const doc = translateFilter(c.where, undefined, SHAPES) as Record<string, unknown>; |
| 148 | + expect(select(doc), JSON.stringify(doc)).toEqual(c.expected); |
| 149 | + }); |
| 150 | + } |
| 151 | + |
| 152 | + it('beside $startsWith on the same field, both constraints survive — $elemMatch contests no key', () => { |
| 153 | + expect(translateFilter({ tags: { $contains: 'red', $startsWith: 'r' } }, undefined, SHAPES)) |
| 154 | + .toEqual({ tags: { $elemMatch: { $in: ['red'], $not: { $type: 'array' } }, $regex: '^r' } }); |
| 155 | + }); |
| 156 | + |
| 157 | + it('the aggregate $match translates it the way find() does', () => { |
| 158 | + const pipeline = buildAggregationPipeline({ |
| 159 | + where: { owners: { $contains: 'u1' } }, |
| 160 | + aggregations: [{ function: 'count', alias: 'n' }] as never, |
| 161 | + valueShape: SHAPES, |
| 162 | + }); |
| 163 | + expect(pipeline[0]).toEqual({ $match: { owners: { $elemMatch: { $in: ['u1'], $not: { $type: 'array' } } } } }); |
| 164 | + }); |
| 165 | + |
| 166 | + it('with no declaration held, every field keeps the substring reading, as driver-sql does for a table it was never told about', () => { |
| 167 | + const doc = translateFilter({ owners: { $contains: 'u1' } }) as Record<string, unknown>; |
| 168 | + expect(doc).toEqual({ owners: { $regex: 'u1' } }); |
| 169 | + // The per-element substring the declaration exists to replace: 'u1' answers |
| 170 | + // ['u10']. (Whether a `$regex` reaches into the NESTED array of r6 is left |
| 171 | + // unmodelled here; the membership test above excludes it by construction.) |
| 172 | + expect(select(doc)).toEqual(expect.arrayContaining(['r1', 'r2', 'r7'])); |
| 173 | + expect(translateFilter({ nope: { $contains: 'u1' } }, undefined, SHAPES)).toEqual({ nope: { $regex: 'u1' } }); |
| 174 | + }); |
| 175 | +}); |
| 176 | + |
| 177 | +const sharedMongod: MongoMemoryServer | undefined = await createTestMongod('$contains membership'); |
| 178 | + |
| 179 | +describe.skipIf(!sharedMongod)('MongoDBDriver — $contains membership against a live mongod', () => { |
| 180 | + const mongod = sharedMongod as MongoMemoryServer; |
| 181 | + let driver: MongoDBDriver; |
| 182 | + |
| 183 | + beforeAll(async () => { |
| 184 | + driver = new MongoDBDriver({ url: mongod.getUri(), database: OBJECT }); |
| 185 | + await driver.connect(); |
| 186 | + await driver.syncSchema(OBJECT, { name: OBJECT, fields: FIELDS }); |
| 187 | + for (const row of ROWS) await driver.create(OBJECT, { ...row }); |
| 188 | + }, 90_000); |
| 189 | + |
| 190 | + afterAll(async () => { |
| 191 | + if (driver) await driver.disconnect(); |
| 192 | + if (sharedMongod) await sharedMongod.stop(); |
| 193 | + }); |
| 194 | + |
| 195 | + for (const c of CASES) { |
| 196 | + it(`${c.name}: selects ${JSON.stringify(c.expected)}`, async () => { |
| 197 | + const rows = (await driver.find(OBJECT, { where: c.where })) as Array<Record<string, unknown>>; |
| 198 | + expect(rows.map((r) => String(r.id)).sort()).toEqual(c.expected); |
| 199 | + }); |
| 200 | + } |
| 201 | +}); |
![objectstack-fleet[bot]](https://avatars.githubusercontent.com/u/254823548?v=4&size=40){kind=avatar}
0 commit comments