Files
ThothII/backend/src/workspaces/qdrant-collection.ts
T

151 lines
6.6 KiB
TypeScript

export const QDRANT_REQUIRED_INDEXES = Object.freeze([
"content_hash", "document_id", "kind", "record_key",
"record_kind", "vector_generation", "workspace_id", "workspace_revision",
]);
export type CollectionMode = "self_heal" | "require_existing" | "evidence_maintenance";
export interface CollectionCheck {
ok: boolean;
code?: "semantic_index_incompatible" | "workspace_not_activatable";
state?: "ready" | "created" | "repaired" | "upgraded";
}
export interface ReconcileCollectionOptions {
baseUrl: string;
collection: string;
dimensions: number;
distance: string;
mode: CollectionMode;
request?: typeof fetch;
signal?: AbortSignal;
}
function qdrantDistance(distance: string): string {
return distance.length === 0 ? distance : distance.charAt(0).toUpperCase() + distance.slice(1);
}
function qdrantUrl(baseUrl: string, path: string): string {
return new URL(path, baseUrl).toString();
}
async function collectionInfo(opts: ReconcileCollectionOptions, request: typeof fetch): Promise<unknown | undefined> {
const res = await request(qdrantUrl(opts.baseUrl, `/collections/${encodeURIComponent(opts.collection)}`), { method: "GET", signal: opts.signal });
if (res.status === 404) return undefined;
if (!res.ok) throw new Error("qdrant collection check failed");
return (await res.json() as any)?.result;
}
function vectorCompatibility(info: any, opts: ReconcileCollectionOptions): boolean {
const vectors = info?.config?.params?.vectors;
return Boolean(vectors && vectors.size === opts.dimensions && typeof vectors.distance === "string"
&& vectors.distance.toLowerCase() === opts.distance);
}
async function missingIndexes(opts: ReconcileCollectionOptions, info: any): Promise<string[]> {
const payloadSchema = info?.payload_schema;
if (!payloadSchema || typeof payloadSchema !== "object") return [...QDRANT_REQUIRED_INDEXES];
return QDRANT_REQUIRED_INDEXES.filter((field) => payloadSchema[field]?.data_type !== "keyword");
}
async function createCollection(opts: ReconcileCollectionOptions, request: typeof fetch): Promise<void> {
const res = await request(qdrantUrl(opts.baseUrl, `/collections/${encodeURIComponent(opts.collection)}`), {
method: "PUT",
headers: { "content-type": "application/json" },
body: JSON.stringify({
vectors: { size: opts.dimensions, distance: qdrantDistance(opts.distance) },
...(opts.mode === "evidence_maintenance" ? { sparse_vectors: { bm25: { modifier: "idf" } } } : {}),
}),
signal: opts.signal,
});
if (!res.ok && res.status !== 409) throw new Error("qdrant collection creation failed");
}
type EvidenceSparseCompatibility = "compatible" | "upgradeable" | "incompatible";
function evidenceSparseCompatibility(info: any): EvidenceSparseCompatibility {
const sparseVectors = info?.config?.params?.sparse_vectors;
if (sparseVectors === undefined) return "upgradeable";
if (!sparseVectors || typeof sparseVectors !== "object" || Array.isArray(sparseVectors)) {
return "incompatible";
}
const bm25 = sparseVectors.bm25;
if (bm25 === undefined) return "upgradeable";
return typeof bm25 === "object" && bm25 !== null && !Array.isArray(bm25)
&& typeof bm25.modifier === "string" && bm25.modifier.toLowerCase() === "idf"
? "compatible" : "incompatible";
}
async function createBm25Vector(opts: ReconcileCollectionOptions, request: typeof fetch): Promise<void> {
const res = await request(
qdrantUrl(opts.baseUrl, `/collections/${encodeURIComponent(opts.collection)}/vectors/bm25`),
{
method: "PUT",
headers: { "content-type": "application/json" },
body: JSON.stringify({ sparse: { modifier: "idf" } }),
signal: opts.signal,
},
);
if (!res.ok && res.status !== 409) throw new Error("qdrant BM25 vector creation failed");
}
async function createIndex(opts: ReconcileCollectionOptions, field: string, request: typeof fetch): Promise<void> {
const res = await request(qdrantUrl(opts.baseUrl, `/collections/${encodeURIComponent(opts.collection)}/index`), {
method: "PUT",
headers: { "content-type": "application/json" },
body: JSON.stringify({ field_name: field, field_schema: "keyword" }),
signal: opts.signal,
});
if (!res.ok && res.status !== 409) throw new Error("qdrant index creation failed");
}
/** Reconcile a Qdrant collection. Only Evidence maintenance may add the BM25 sparse vector;
* session admission remains limited to the existing dense/index self-heal behavior. */
export async function reconcileCollection(opts: ReconcileCollectionOptions): Promise<CollectionCheck> {
const request = opts.request ?? fetch;
const allowsMutation = opts.mode === "self_heal" || opts.mode === "evidence_maintenance";
let info = await collectionInfo(opts, request);
if (info === undefined) {
if (!allowsMutation) return { ok: false, code: "semantic_index_incompatible" };
await createCollection(opts, request);
// Tolerate an already-compatible concurrent creator: re-read the final state.
info = await collectionInfo(opts, request);
if (info === undefined) return { ok: false, code: "workspace_not_activatable" };
}
if (!vectorCompatibility(info, opts)) {
return { ok: false, code: "semantic_index_incompatible" };
}
let bm25Added = false;
if (opts.mode === "evidence_maintenance") {
const sparse = evidenceSparseCompatibility(info);
if (sparse === "incompatible") return { ok: false, code: "semantic_index_incompatible" };
if (sparse === "upgradeable") {
await createBm25Vector(opts, request);
info = await collectionInfo(opts, request);
if (!vectorCompatibility(info, opts) || evidenceSparseCompatibility(info) !== "compatible") {
return { ok: false, code: "semantic_index_incompatible" };
}
bm25Added = true;
}
}
const missing = await missingIndexes(opts, info);
if (missing.length > 0) {
if (!allowsMutation) return { ok: false, code: "semantic_index_incompatible" };
for (const field of missing) await createIndex(opts, field, request);
// Qdrant payload indexes become visible asynchronously: poll until the
// contract is complete or a bounded deadline passes (fail closed).
const deadline = Date.now() + 15000;
let current: any = info;
while (Date.now() < deadline) {
current = await collectionInfo(opts, request);
if (!vectorCompatibility(current, opts)) break;
if ((await missingIndexes(opts, current)).length === 0) {
return { ok: true, state: "repaired" };
}
await new Promise((resolve) => setTimeout(resolve, 500));
}
return { ok: false, code: "semantic_index_incompatible" };
}
return { ok: true, state: bm25Added ? "upgraded" : "ready" };
}