feat: define internal semantic workspace schema

This commit is contained in:
2026-08-08 16:43:54 +02:00
parent 4fe4049a24
commit 2f7f923c4d
9 changed files with 463 additions and 154 deletions
+1 -1
View File
@@ -102,7 +102,7 @@ export function resolveBinding(
const supported = role === "DWH"
? canonical.dwh.supported_transports
: role === "VECTOR"
? canonical.semantic_index.vector_store.supported_transports
? (canonical.semantic_index.vector_store.supported_transports ?? [])
: ["rest_api"] as const;
const selectedValue = transportVariable ? env[transportVariable.name] : undefined;
const selectedTransport = isTransport(selectedValue) ? selectedValue : supported[0];
+1 -1
View File
@@ -131,7 +131,7 @@ export function buildInstallationContract(workspace: WorkspaceDescriptor): Insta
...connectorVariables(
namespace,
"VECTOR",
canonical.semantic_index.vector_store.supported_transports,
canonical.semantic_index.vector_store.supported_transports ?? [],
),
...(canonical.semantic_index.vector_writer
? [createVariable(namespace, "VECTOR_WRITER", "API_KEY_FILE")]
@@ -0,0 +1,28 @@
import { validateOperationalWorkspace, type WorkspaceV2, type WorkspaceV3 } from "./schema.js";
export function migrateWorkspaceV2ToV3(
legacy: WorkspaceV2,
collection: string,
): WorkspaceV3 {
return validateOperationalWorkspace({
workspace: { ...legacy.workspace, schema_version: 3 },
dwh: legacy.dwh,
semantic_index: {
vector_store: {
engine: "qdrant",
collection,
dimensions: 1024,
distance: "cosine",
},
embedding: {
provider: "ollama_internal",
model: "qwen3-embedding:0.6b",
dimensions: 1024,
},
},
llm_policy: legacy.llm_policy,
...(legacy.diagnostics?.dwh_rest
? { diagnostics: { dwh_rest: legacy.diagnostics.dwh_rest } }
: {}),
});
}
+2 -2
View File
@@ -99,8 +99,8 @@ export function renderRuntimeConfig(
};
const dwhIdentity = { database: canonical.dwh.database, schema: canonical.dwh.schema };
const vectorIdentity = {
database: canonical.semantic_index.vector_store.database,
schema: canonical.semantic_index.vector_store.schema,
database: canonical.semantic_index.vector_store.database ?? canonical.dwh.database,
schema: canonical.semantic_index.vector_store.schema ?? canonical.dwh.schema,
};
const dwhDirect = bindings.dwh.transport === "postgres_direct";
const vectorDirect = bindings.vector.transport === "pgvector_direct";
+94 -25
View File
@@ -34,7 +34,7 @@ export interface CanonicalDiagnostics {
embedding?: RestDiagnosticRequest & { response: { model: string; dimensions: string } };
}
interface WorkspaceMetadata<Version extends 1 | 2> {
interface WorkspaceMetadata<Version extends 1 | 2 | 3> {
schema_version: Version;
id: string;
name: string;
@@ -72,23 +72,52 @@ interface SemanticIndex<TVectorStore extends VectorStore> {
};
}
interface WorkspaceBase<Version extends 1 | 2, TVectorStore extends VectorStore> {
interface WorkspaceBase<Version extends 1 | 2 | 3, TVectorStore, TDiagnostics = CanonicalDiagnostics> {
workspace: WorkspaceMetadata<Version>;
dwh: WorkspaceDwh;
semantic_index: SemanticIndex<TVectorStore>;
semantic_index: {
vector_store: TVectorStore;
vector_writer?: Record<string, never>;
embedding: Version extends 3 ? {
provider: "ollama_internal";
model: "qwen3-embedding:0.6b";
dimensions: 1024;
timeout_ms?: number;
} : {
provider: "ollama_compatible" | "openai_compatible";
model: string;
dimensions: number;
timeout_ms?: number;
};
};
llm_policy: {
default?: `${string}/${string}`;
allowed: `${string}/${string}`[];
};
diagnostics?: CanonicalDiagnostics;
diagnostics?: TDiagnostics;
}
export interface CanonicalWorkspace extends WorkspaceBase<2, VectorStore & { database: string; schema: string }> {}
interface QdrantVectorStore {
engine: "qdrant";
collection: string;
dimensions: 1024;
distance: "cosine";
database?: string;
schema?: string;
port?: number;
timeout_ms?: number;
supported_transports?: VectorTransport[];
}
export interface WorkspaceV3 extends WorkspaceBase<3, QdrantVectorStore> {}
export interface WorkspaceV2 extends WorkspaceBase<2, VectorStore & { database: string; schema: string }> {}
/** A readable, non-operational v1 descriptor. It must be explicitly migrated before use. */
export interface LegacyWorkspace extends WorkspaceBase<1, VectorStore & { database?: string; schema?: string }> {}
export interface WorkspaceV1 extends WorkspaceBase<1, VectorStore & { database?: string; schema?: string }> {}
export type WorkspaceDescriptor = CanonicalWorkspace | LegacyWorkspace;
export type CanonicalWorkspace = WorkspaceV3;
export type LegacyWorkspace = WorkspaceV1 | WorkspaceV2;
export type WorkspaceDescriptor = WorkspaceV1 | WorkspaceV2 | WorkspaceV3;
const workspaceId = z.string().regex(/^[a-z][a-z0-9-]{2,62}$/, {
message: "workspace id must match ^[a-z][a-z0-9-]{2,62}$",
@@ -159,6 +188,11 @@ const embeddingSchema = z.object({
dimensions,
timeout_ms: timeoutMs.optional(),
}).strict();
const internalEmbeddingSchema = z.object({
provider: z.literal("ollama_internal"),
model: z.literal("qwen3-embedding:0.6b"),
dimensions: z.literal(1024),
}).strict();
const vectorStoreShape = {
engine: z.literal("pgvector"),
collection: identifier,
@@ -178,6 +212,12 @@ const canonicalVectorStoreSchema = z.object({
database: identifier,
schema: identifier,
}).strict();
const qdrantVectorStoreSchema = z.object({
engine: z.literal("qdrant"),
collection: workspaceId,
dimensions: z.literal(1024),
distance: z.literal("cosine"),
}).strict();
const llmPolicySchema = z.object({
default: modelReference.optional(),
allowed: z.array(modelReference).min(1),
@@ -191,11 +231,13 @@ function unique<T>(values: readonly T[], context: z.RefinementCtx, path: Propert
function workspaceInvariants(workspace: any, context: z.RefinementCtx): void {
unique(workspace.dwh.supported_transports, context, ["dwh", "supported_transports"]);
unique(
workspace.semantic_index.vector_store.supported_transports,
context,
["semantic_index", "vector_store", "supported_transports"],
);
if ("supported_transports" in workspace.semantic_index.vector_store) {
unique(
workspace.semantic_index.vector_store.supported_transports,
context,
["semantic_index", "vector_store", "supported_transports"],
);
}
unique(workspace.llm_policy.allowed, context, ["llm_policy", "allowed"]);
if (workspace.semantic_index.vector_store.dimensions !== workspace.semantic_index.embedding.dimensions) {
@@ -221,7 +263,10 @@ function workspaceInvariants(workspace: any, context: z.RefinementCtx): void {
}
if (
workspace.diagnostics?.vector_rest
&& !workspace.semantic_index.vector_store.supported_transports.includes("rest_api")
&& (
!("supported_transports" in workspace.semantic_index.vector_store)
|| !workspace.semantic_index.vector_store.supported_transports.includes("rest_api")
)
) {
context.addIssue({
code: "custom",
@@ -236,7 +281,7 @@ const workspaceShape = {
llm_policy: llmPolicySchema,
diagnostics: diagnosticsSchema,
};
const LegacyWorkspaceSchema = z.object({
const WorkspaceV1Schema = z.object({
...workspaceShape,
workspace: z.object({
schema_version: z.literal(1), id: workspaceId, name: z.string().trim().min(1),
@@ -248,7 +293,7 @@ const LegacyWorkspaceSchema = z.object({
embedding: embeddingSchema,
}).strict(),
}).strict().superRefine(workspaceInvariants);
const CanonicalWorkspaceSchema = z.object({
const WorkspaceV2Schema = z.object({
...workspaceShape,
workspace: z.object({
schema_version: z.literal(2), id: workspaceId, name: z.string().trim().min(1),
@@ -260,7 +305,22 @@ const CanonicalWorkspaceSchema = z.object({
embedding: embeddingSchema,
}).strict(),
}).strict().superRefine(workspaceInvariants);
const WorkspaceDescriptorSchema = z.union([CanonicalWorkspaceSchema, LegacyWorkspaceSchema]);
const WorkspaceV3Schema = z.object({
dwh: dwhSchema,
llm_policy: llmPolicySchema,
diagnostics: z.object({
dwh_rest: dwhRestDiagnostic.optional(),
}).strict().optional(),
workspace: z.object({
schema_version: z.literal(3), id: workspaceId, name: z.string().trim().min(1),
description: z.string().trim().min(1).optional(), language: z.enum(["en", "it"]),
}).strict(),
semantic_index: z.object({
vector_store: qdrantVectorStoreSchema,
embedding: internalEmbeddingSchema,
}).strict(),
}).strict().superRefine(workspaceInvariants);
const WorkspaceDescriptorSchema = z.union([WorkspaceV3Schema, WorkspaceV2Schema, WorkspaceV1Schema]);
export function parseWorkspaceYaml(source: string): WorkspaceDescriptor {
const documents = parseAllDocuments(source, { uniqueKeys: true });
@@ -278,14 +338,23 @@ export function validateWorkspaceDescriptor(workspace: unknown): WorkspaceDescri
}
export function isCanonicalWorkspace(workspace: WorkspaceDescriptor): workspace is CanonicalWorkspace {
return workspace.workspace.schema_version === 2;
return workspace.workspace.schema_version === 3;
}
export function isOperationalWorkspace(workspace: WorkspaceDescriptor): workspace is WorkspaceV3 {
return workspace.workspace.schema_version === 3;
}
/** Rejects readable v1 descriptors at every operational boundary until a caller migrates them. */
export function validateCanonicalWorkspace(workspace: unknown): CanonicalWorkspace {
return validateOperationalWorkspace(workspace);
}
/** Rejects readable v1/v2 descriptors at every operational boundary until a caller migrates them. */
export function validateOperationalWorkspace(workspace: unknown): WorkspaceV3 {
const descriptor = validateWorkspaceDescriptor(workspace);
if (!isCanonicalWorkspace(descriptor)) {
throw new Error("Workspace descriptor requires explicit migration to schema version 2");
if (!isOperationalWorkspace(descriptor)) {
throw new Error("Workspace descriptor requires explicit migration to schema version 3");
}
return descriptor;
}
@@ -295,19 +364,19 @@ export function validateCanonicalWorkspace(workspace: unknown): CanonicalWorkspa
* transformer never derives it from DWH identity, even where both services share a database.
*/
export function migrateWorkspaceV1ToV2(
workspace: LegacyWorkspace,
workspace: WorkspaceV1,
vectorIdentity: { database: string; schema: string },
): CanonicalWorkspace {
const legacy = LegacyWorkspaceSchema.parse(workspace) as LegacyWorkspace;
): WorkspaceV2 {
const legacy = WorkspaceV1Schema.parse(workspace) as WorkspaceV1;
const identity = z.object({ database: identifier, schema: identifier }).strict().parse(vectorIdentity);
return validateCanonicalWorkspace({
return WorkspaceV2Schema.parse({
...legacy,
workspace: { ...legacy.workspace, schema_version: 2 },
semantic_index: {
...legacy.semantic_index,
vector_store: { ...legacy.semantic_index.vector_store, ...identity },
},
});
}) as WorkspaceV2;
}
/** Builds a request URL only after rejecting values that can leave the declared service origin. */
@@ -320,7 +389,7 @@ export function resolveDiagnosticUrl(baseUrl: string, path: string): URL {
}
export function serializeWorkspaceYaml(workspace: CanonicalWorkspace): string {
const canonical = validateCanonicalWorkspace(workspace);
const canonical = validateOperationalWorkspace(workspace);
return stringify(canonical, { lineWidth: 0, sortMapEntries: true });
}
+100
View File
@@ -15,3 +15,103 @@ export type WorkspaceErrorCode =
| "workspace_stale" | "workspace_conflict" | "git_unavailable"
| "git_auth_failed" | "git_non_fast_forward" | "git_push_rejected"
| "connector_unavailable" | "semantic_index_incompatible";
export interface QdrantVectorStore {
engine: "qdrant";
collection: string;
dimensions: 1024;
distance: "cosine";
database?: string;
schema?: string;
port?: number;
timeout_ms?: number;
supported_transports?: ("pgvector_direct" | "rest_api" | "ssh_tunnel")[];
}
export interface InternalEmbedding {
provider: "ollama_internal";
model: "qwen3-embedding:0.6b";
dimensions: 1024;
timeout_ms?: number;
}
export interface WorkspaceV2 {
workspace: {
schema_version: 2;
id: string;
name: string;
description?: string;
language: "en" | "it";
};
dwh: {
engine: "postgres";
database: string;
schema: string;
port?: number;
timeout_ms?: number;
supported_transports: ("postgres_direct" | "rest_api" | "ssh_tunnel")[];
};
semantic_index: {
vector_store: {
engine: "pgvector";
database: string;
schema: string;
collection: string;
dimensions: number;
distance: "cosine" | "l2" | "inner_product";
port?: number;
timeout_ms?: number;
supported_transports: ("pgvector_direct" | "rest_api" | "ssh_tunnel")[];
};
vector_writer?: Record<string, never>;
embedding: {
provider: "ollama_compatible" | "openai_compatible";
model: string;
dimensions: number;
timeout_ms?: number;
};
};
llm_policy: {
default?: `${string}/${string}`;
allowed: `${string}/${string}`[];
};
diagnostics?: {
dwh_rest?: {
method: "GET" | "POST";
path: string;
auth: "none" | "bearer" | "x-api-key";
response: { database: string; schema: string };
};
vector_rest?: {
metadata: {
method: "GET" | "POST";
path: string;
auth: "none" | "bearer" | "x-api-key";
response: { collection: string; dimensions: string; distance: string };
};
reversible_probe?: {
method: "POST";
path: string;
auth: "bearer" | "x-api-key";
response: { operation: string };
};
};
embedding?: {
method: "GET" | "POST";
path: string;
auth: "none" | "bearer" | "x-api-key";
response: { model: string; dimensions: string };
};
};
}
export interface WorkspaceV3 {
workspace: WorkspaceV2["workspace"] & { schema_version: 3 };
dwh: WorkspaceV2["dwh"];
semantic_index: {
vector_store: QdrantVectorStore;
embedding: InternalEmbedding;
};
llm_policy: WorkspaceV2["llm_policy"];
diagnostics?: WorkspaceV2["diagnostics"];
}
@@ -8,6 +8,7 @@ import {
migrateLegacyWorkspace,
writeMigratedWorkspace,
} from "../src/workspaces/migrate-legacy.js";
import * as workspaceSchema from "../src/workspaces/schema.js";
import { parseWorkspaceYaml } from "../src/workspaces/schema.js";
const temporaryRoots: string[] = [];
@@ -30,10 +31,15 @@ test("migrates the current local PSD descriptor without copying secret values",
test("keeps an incomplete legacy vector identity readable and explicitly migration-required", () => {
const result = migrateLegacyWorkspace(readFixture("tht.example.yaml"), { id: "example" });
const isOperationalWorkspace = (workspaceSchema as { isOperationalWorkspace?: unknown }).isOperationalWorkspace;
expect(result.state).toBe("migration_required");
expect(result.workspace.workspace.schema_version).toBe(1);
expect(parseWorkspaceYaml(result.source).workspace.schema_version).toBe(1);
expect(isOperationalWorkspace).toBeTypeOf("function");
expect((isOperationalWorkspace as (workspace: ReturnType<typeof parseWorkspaceYaml>) => boolean)(
parseWorkspaceYaml(result.source),
)).toBe(false);
});
test("writes versioned repository artifacts atomically without replacing a prior migration", async () => {
@@ -0,0 +1,101 @@
import { expect, test } from "vitest";
import { migrateWorkspaceV2ToV3 } from "../src/workspaces/migrate-v2-qdrant.js";
import { parseWorkspaceYaml } from "../src/workspaces/schema.js";
const workspaceV2Yaml = `workspace:
schema_version: 2
id: psd-clinical
name: Policlinico San Donato
description: Clinical data warehouse workspace
language: it
dwh:
engine: postgres
database: postgres
schema: datawarehouse
supported_transports:
- postgres_direct
- rest_api
semantic_index:
vector_store:
engine: pgvector
database: postgres
schema: vectors
collection: clinical_documents
dimensions: 768
distance: inner_product
supported_transports:
- pgvector_direct
- rest_api
embedding:
provider: openai_compatible
model: text-embedding-3-large
dimensions: 768
llm_policy:
default: zai/glm-5.2
allowed:
- zai/glm-5.2
diagnostics:
dwh_rest:
method: POST
path: /rpc/dwh
auth: bearer
response:
database: database
schema: schema
vector_rest:
metadata:
method: GET
path: /vector
auth: bearer
response:
collection: collection
dimensions: dimensions
distance: distance
embedding:
method: GET
path: /models
auth: none
response:
model: model
dimensions: dimensions
`;
test("migrates a schema v2 workspace to the internal qdrant schema v3 shape", () => {
const legacy = parseWorkspaceYaml(workspaceV2Yaml);
const migrated = migrateWorkspaceV2ToV3(legacy, "psd-clinical");
expect(migrated).toMatchObject({
workspace: { schema_version: 3, id: "psd-clinical", name: "Policlinico San Donato" },
dwh: legacy.dwh,
semantic_index: {
vector_store: {
engine: "qdrant",
collection: "psd-clinical",
dimensions: 1024,
distance: "cosine",
},
embedding: {
provider: "ollama_internal",
model: "qwen3-embedding:0.6b",
dimensions: 1024,
},
},
llm_policy: legacy.llm_policy,
diagnostics: {
dwh_rest: legacy.diagnostics?.dwh_rest,
},
});
});
test("drops vector and embedding diagnostics and transports during v2 to v3 migration", () => {
const legacy = parseWorkspaceYaml(workspaceV2Yaml);
const migrated = migrateWorkspaceV2ToV3(legacy, "psd-clinical");
expect(migrated.diagnostics).toEqual({
dwh_rest: legacy.diagnostics?.dwh_rest,
});
expect(migrated.semantic_index.vector_store).not.toHaveProperty("supported_transports");
expect(migrated.semantic_index.embedding).not.toHaveProperty("timeout_ms");
});
+130 -125
View File
@@ -1,9 +1,14 @@
import { expect, test } from "vitest";
import * as workspaceSchema from "../src/workspaces/schema.js";
import { parseWorkspaceYaml, serializeWorkspaceYaml, validateCanonicalWorkspace } from "../src/workspaces/schema.js";
import {
parseWorkspaceYaml,
serializeWorkspaceYaml,
validateCanonicalWorkspace,
type WorkspaceDescriptor,
} from "../src/workspaces/schema.js";
export const validYaml = `workspace:
schema_version: 2
schema_version: 3
id: psd-clinical
name: Policlinico San Donato
description: Clinical data warehouse workspace
@@ -20,23 +25,14 @@ dwh:
- ssh_tunnel
semantic_index:
vector_store:
engine: pgvector
database: postgres
schema: vectors
collection: clinical_documents
dimensions: 768
engine: qdrant
collection: psd-clinical
dimensions: 1024
distance: cosine
port: 5432
timeout_ms: 5000
supported_transports:
- pgvector_direct
- rest_api
- ssh_tunnel
embedding:
provider: ollama_compatible
model: nomic-embed-text-v2-moe
dimensions: 768
timeout_ms: 5000
provider: ollama_internal
model: qwen3-embedding:0.6b
dimensions: 1024
llm_policy:
default: zai/glm-5.2
allowed:
@@ -45,7 +41,7 @@ llm_policy:
`;
test("rejects a workspace whose embedding dimensions differ from its collection", () => {
expect(() => parseWorkspaceYaml(validYaml.replace("dimensions: 768", "dimensions: 1536")))
expect(() => parseWorkspaceYaml(validYaml.replace("dimensions: 1024", "dimensions: 1536")))
.toThrow(/dimensions/i);
});
@@ -76,113 +72,134 @@ test("accepts optional connection ports and timeouts but rejects unsafe values",
.toThrow(/port/i);
expect(() => parseWorkspaceYaml(validYaml.replace("timeout_ms: 5000", "timeout_ms: 0")))
.toThrow(/timeout/i);
expect(() => parseWorkspaceYaml(validYaml.replace("dimensions: 1024", "dimensions: 2048")))
.toThrow(/1024|dimensions/i);
});
test("requires explicit vector database and schema identities with strict diagnostic declarations", () => {
const diagnosticWorkspace = validYaml.replace(
" engine: pgvector\n database: postgres",
" engine: pgvector\n database: vector_database",
).replace(
"llm_policy:\n",
"diagnostics:\n"
+ " dwh_rest:\n"
+ " method: POST\n"
+ " path: /rpc/ping\n"
+ " auth: bearer\n"
+ " response:\n"
+ " database: database\n"
+ " schema: schema\n"
+ " vector_rest:\n"
+ " metadata:\n"
+ " method: GET\n"
+ " path: /metadata\n"
+ " auth: bearer\n"
+ " response:\n"
+ " collection: collection\n"
+ " dimensions: dimensions\n"
+ " distance: distance\n"
+ " reversible_probe:\n"
+ " method: POST\n"
+ " path: /rpc/diagnostic_vector_probe\n"
+ " auth: bearer\n"
+ " response: { operation: operation }\n"
+ " embedding:\n"
+ " method: GET\n"
+ " path: /models\n"
+ " auth: none\n"
+ " response:\n"
+ " model: model\n"
+ " dimensions: dimensions\n"
+ "llm_policy:\n",
);
expect(parseWorkspaceYaml(diagnosticWorkspace).semantic_index.vector_store).toMatchObject({
database: "vector_database",
schema: "vectors",
test("accepts only the schema v3 internal qdrant semantic shape", () => {
expect(parseWorkspaceYaml(validYaml)).toMatchObject({
workspace: { schema_version: 3, id: "psd-clinical" },
semantic_index: {
vector_store: {
engine: "qdrant",
collection: "psd-clinical",
dimensions: 1024,
distance: "cosine",
},
embedding: {
provider: "ollama_internal",
model: "qwen3-embedding:0.6b",
dimensions: 1024,
},
},
});
expect(() => parseWorkspaceYaml(diagnosticWorkspace.replace("database: vector_database", 'database: " "')))
.toThrow(/database/i);
expect(() => parseWorkspaceYaml(diagnosticWorkspace.replace("schema: vectors", 'schema: " "')))
.toThrow(/schema/i);
expect(() => parseWorkspaceYaml(diagnosticWorkspace.replace("method: POST", "method: PATCH")))
.toThrow(/method/i);
expect(() => parseWorkspaceYaml(diagnosticWorkspace.replace(" distance: distance", " distance: distance\n extra: ignored")))
.toThrow(/unrecognized key/i);
for (const unsafePath of [
"//diagnostic.invalid/rpc", "'/\\\\diagnostic'", "'/rpc\\\\diagnostic'", "'/rpc/%5Cdiagnostic'", "'/rpc/\u0001'",
]) {
expect(() => parseWorkspaceYaml(diagnosticWorkspace.replace("path: /rpc/ping", `path: ${unsafePath}`)))
.toThrow(/origin-relative|path/i);
}
expect(() => parseWorkspaceYaml(diagnosticWorkspace.replace("auth: bearer", "auth: basic")))
.toThrow(/auth/i);
expect(() => parseWorkspaceYaml(diagnosticWorkspace.replace(" schema: schema", " schema: schema\n status: status")))
.toThrow(/unrecognized key/i);
expect(() => parseWorkspaceYaml(diagnosticWorkspace.replace(" schema: schema", " schema: bad field")))
.toThrow(/response field/i);
});
test("requires a reversible writer probe to declare the response operation it verifies", () => {
const writerProbe = validYaml.replace("llm_policy:\n", `diagnostics:
vector_rest:
metadata:
method: GET
path: /metadata
auth: bearer
response: { collection: collection, dimensions: dimensions, distance: distance }
reversible_probe:
method: POST
path: /diagnostic-probe
auth: bearer
llm_policy:
`);
expect(() => parseWorkspaceYaml(writerProbe)).toThrow(/response|operation/i);
test("rejects pgvector semantic stores in schema v3", () => {
expect(() => parseWorkspaceYaml(validYaml.replace("engine: qdrant", "engine: pgvector")))
.toThrow(/qdrant|pgvector/i);
});
test("keeps v1 descriptors readable but requires explicit migration before v2 operations", () => {
const v1WithoutVectorIdentity = validYaml.replace("schema_version: 2", "schema_version: 1").replace(
" database: postgres\n schema: vectors\n", "",
test("rejects supported_transports inside schema v3 semantic identity", () => {
const withTransport = validYaml.replace(
" distance: cosine\n",
" distance: cosine\n supported_transports:\n - rest_api\n",
);
expect(() => parseWorkspaceYaml(v1WithoutVectorIdentity)).not.toThrow();
expect(() => parseWorkspaceYaml(validYaml)).not.toThrow();
const v1 = parseWorkspaceYaml(v1WithoutVectorIdentity);
const v2 = parseWorkspaceYaml(validYaml);
expect(() => parseWorkspaceYaml(withTransport)).toThrow(/unrecognized key|supported_transports/i);
});
test("rejects external embedding providers in schema v3", () => {
expect(() => parseWorkspaceYaml(validYaml.replace("provider: ollama_internal", "provider: openai_compatible")))
.toThrow(/ollama_internal|provider/i);
});
test("rejects non-cosine distance in schema v3", () => {
expect(() => parseWorkspaceYaml(validYaml.replace("distance: cosine", "distance: l2")))
.toThrow(/cosine|distance/i);
});
test("rejects unknown fields in schema v3 semantic identity", () => {
const withUnknownField = validYaml.replace(
" collection: psd-clinical\n",
" collection: psd-clinical\n namespace: psd\n",
);
expect(() => parseWorkspaceYaml(withUnknownField)).toThrow(/unrecognized key/i);
});
test("keeps v1 and v2 descriptors parseable but non-operational", () => {
const v1Yaml = `workspace:
schema_version: 1
id: psd-clinical
name: Policlinico San Donato
language: it
dwh:
engine: postgres
database: postgres
schema: datawarehouse
supported_transports:
- postgres_direct
semantic_index:
vector_store:
engine: pgvector
collection: clinical_documents
dimensions: 1024
distance: cosine
supported_transports:
- pgvector_direct
embedding:
provider: ollama_compatible
model: nomic-embed-text-v2-moe
dimensions: 1024
llm_policy:
allowed:
- zai/glm-5.2
`;
const v2Yaml = `workspace:
schema_version: 2
id: psd-clinical
name: Policlinico San Donato
language: it
dwh:
engine: postgres
database: postgres
schema: datawarehouse
supported_transports:
- postgres_direct
semantic_index:
vector_store:
engine: pgvector
database: postgres
schema: vectors
collection: clinical_documents
dimensions: 1024
distance: cosine
supported_transports:
- pgvector_direct
embedding:
provider: ollama_compatible
model: nomic-embed-text-v2-moe
dimensions: 1024
llm_policy:
allowed:
- zai/glm-5.2
`;
const v1 = parseWorkspaceYaml(v1Yaml);
const v2 = parseWorkspaceYaml(v2Yaml);
const isOperationalWorkspace = (workspaceSchema as { isOperationalWorkspace?: unknown }).isOperationalWorkspace;
expect(v1.workspace.schema_version).toBe(1);
expect(() => validateCanonicalWorkspace(v1)).toThrow(/migrat/i);
expect(v2.workspace.schema_version).toBe(2);
const migrate = (workspaceSchema as { migrateWorkspaceV1ToV2?: unknown }).migrateWorkspaceV1ToV2;
expect(migrate).toBeTypeOf("function");
const migrated = (migrate as (workspace: typeof v1, identity: { database: string; schema: string }) => unknown)(v1, {
database: "vector_database",
schema: "vectors",
});
expect(validateCanonicalWorkspace(migrated)).toMatchObject({
workspace: { schema_version: 2 },
semantic_index: { vector_store: { database: "vector_database", schema: "vectors" } },
expect(validateCanonicalWorkspace(parseWorkspaceYaml(validYaml))).toMatchObject({
workspace: { schema_version: 3 },
});
expect(() => validateCanonicalWorkspace(v1)).toThrow(/schema version 3|migration/i);
expect(() => validateCanonicalWorkspace(v2)).toThrow(/schema version 3|migration/i);
expect(isOperationalWorkspace).toBeTypeOf("function");
expect((isOperationalWorkspace as (workspace: WorkspaceDescriptor) => boolean)(v1)).toBe(false);
expect((isOperationalWorkspace as (workspace: WorkspaceDescriptor) => boolean)(v2)).toBe(false);
});
test("constructs diagnostic URLs only when the resolved URL remains on the service origin", () => {
@@ -205,23 +222,11 @@ test("rejects REST diagnostic declarations without their matching connector tran
response:
database: database
schema: schema
vector_rest:
metadata:
method: GET
path: /metadata
auth: bearer
response:
collection: collection
dimensions: dimensions
distance: distance
llm_policy:
`;
const declared = validYaml.replace("llm_policy:\n", diagnostics);
expect(() => parseWorkspaceYaml(declared.replace(" - rest_api\n", ""))).toThrow(/dwh_rest/i);
expect(() => parseWorkspaceYaml(declared.replace(" - rest_api\n", " - rest_api\n", 1).replace(
" - rest_api\n", "",
))).toThrow(/vector_rest/i);
});
test("serializes canonical YAML that parses back to the same workspace", () => {