trustgraph/ts/packages/flow/src/storage/embeddings/qdrant-graph.ts

200 lines
6 KiB
TypeScript
Raw Normal View History

2026-04-05 22:44:45 -05:00
/**
* Qdrant graph embeddings write service.
*
* Stores entity/vector pairs in Qdrant for graph embeddings lookup.
* Collection naming: t_{user}_{collection}_{dimension}
* Collections are lazily created on first write with cosine distance.
*
* Python reference: trustgraph-flow/trustgraph/storage/graph_embeddings/qdrant/write.py
*/
import { QdrantClient } from "@qdrant/js-client-rest";
2026-06-01 16:22:25 -05:00
import { errorMessage, type Term } from "@trustgraph/base";
import { Context, Effect, Layer } from "effect";
import * as S from "effect/Schema";
2026-04-05 22:44:45 -05:00
export interface QdrantGraphEmbeddingsConfig {
url?: string;
apiKey?: string;
}
export interface GraphEmbeddingEntity {
entity: Term;
vector: number[];
chunkId?: string;
}
export interface GraphEmbeddingsMessage {
user: string;
collection: string;
entities: GraphEmbeddingEntity[];
}
function getTermValue(term: Term): string | null {
switch (term.type) {
case "IRI":
return term.iri;
case "LITERAL":
return term.value;
case "BLANK":
return term.id;
case "TRIPLE":
return null;
}
}
2026-06-01 20:26:47 -05:00
export interface QdrantGraphEmbeddingsStore {
readonly store: (message: GraphEmbeddingsMessage) => Promise<void>;
readonly deleteCollection: (user: string, collection: string) => Promise<void>;
}
2026-04-05 22:44:45 -05:00
2026-06-01 20:26:47 -05:00
export function makeQdrantGraphEmbeddingsStore(
config: QdrantGraphEmbeddingsConfig = {},
): QdrantGraphEmbeddingsStore {
const url = config.url ?? process.env.QDRANT_URL ?? "http://localhost:6333";
const apiKey = config.apiKey ?? process.env.QDRANT_API_KEY;
2026-04-05 22:44:45 -05:00
2026-06-01 20:26:47 -05:00
const client = new QdrantClient({
url,
...(apiKey !== undefined && apiKey.length > 0 ? { apiKey } : {}),
});
const knownCollections = new Set<string>();
2026-04-05 22:44:45 -05:00
2026-06-01 20:26:47 -05:00
console.log("[QdrantGraphEmbeddings] Store initialized");
2026-04-05 22:44:45 -05:00
2026-06-01 20:26:47 -05:00
const collectionName = (user: string, collection: string, dim: number): string =>
`t_${user}_${collection}_${dim}`;
2026-04-05 22:44:45 -05:00
2026-06-01 20:26:47 -05:00
const ensureCollection = async (name: string, dim: number): Promise<void> => {
if (knownCollections.has(name)) return;
2026-04-05 22:44:45 -05:00
2026-06-01 20:26:47 -05:00
const exists = await client.collectionExists(name);
2026-04-05 22:44:45 -05:00
if (!exists.exists) {
console.log(`[QdrantGraphEmbeddings] Creating collection ${name} (dim=${dim})`);
2026-06-01 20:26:47 -05:00
await client.createCollection(name, {
2026-04-05 22:44:45 -05:00
vectors: { size: dim, distance: "Cosine" },
});
}
2026-06-01 20:26:47 -05:00
knownCollections.add(name);
};
2026-04-05 22:44:45 -05:00
2026-06-01 20:26:47 -05:00
const store = async (message: GraphEmbeddingsMessage): Promise<void> => {
2026-04-05 22:44:45 -05:00
for (const entry of message.entities) {
const entityValue = getTermValue(entry.entity);
2026-05-12 08:06:58 -05:00
if (entityValue === null || entityValue.length === 0) continue;
if (entry.vector.length === 0) continue;
2026-04-05 22:44:45 -05:00
const dim = entry.vector.length;
2026-06-01 20:26:47 -05:00
const name = collectionName(message.user, message.collection, dim);
2026-04-05 22:44:45 -05:00
2026-06-01 20:26:47 -05:00
await ensureCollection(name, dim);
2026-04-05 22:44:45 -05:00
const payload: Record<string, unknown> = { entity: entityValue };
2026-05-12 08:06:58 -05:00
if (entry.chunkId !== undefined && entry.chunkId.length > 0) {
2026-04-05 22:44:45 -05:00
payload.chunk_id = entry.chunkId;
}
2026-06-01 20:26:47 -05:00
await client.upsert(name, {
2026-04-05 22:44:45 -05:00
points: [
{
2026-05-12 08:06:58 -05:00
id: crypto.randomUUID(),
2026-04-05 22:44:45 -05:00
vector: entry.vector,
payload,
},
],
});
}
2026-06-01 20:26:47 -05:00
};
2026-04-05 22:44:45 -05:00
2026-06-01 20:26:47 -05:00
const deleteCollection = async (user: string, collection: string): Promise<void> => {
2026-04-05 22:44:45 -05:00
const prefix = `t_${user}_${collection}_`;
2026-06-01 20:26:47 -05:00
const allCollections = await client.getCollections();
2026-04-05 22:44:45 -05:00
const matching = allCollections.collections.filter((c) =>
c.name.startsWith(prefix),
);
if (matching.length === 0) {
console.log(`[QdrantGraphEmbeddings] No collections matching prefix ${prefix}`);
return;
}
for (const coll of matching) {
2026-06-01 20:26:47 -05:00
await client.deleteCollection(coll.name);
knownCollections.delete(coll.name);
2026-04-05 22:44:45 -05:00
console.log(`[QdrantGraphEmbeddings] Deleted collection: ${coll.name}`);
}
console.log(
`[QdrantGraphEmbeddings] Deleted ${matching.length} collection(s) for ${user}/${collection}`,
);
2026-06-01 20:26:47 -05:00
};
return { store, deleteCollection };
2026-04-05 22:44:45 -05:00
}
2026-06-01 16:22:25 -05:00
export class QdrantGraphEmbeddingsStoreError extends S.TaggedErrorClass<QdrantGraphEmbeddingsStoreError>()(
"QdrantGraphEmbeddingsStoreError",
{
message: S.String,
operation: S.String,
cause: S.DefectWithStack,
},
) {}
export interface QdrantGraphEmbeddingsStoreServiceShape {
readonly store: (
message: GraphEmbeddingsMessage,
) => Effect.Effect<void, QdrantGraphEmbeddingsStoreError>;
readonly deleteCollection: (
user: string,
collection: string,
) => Effect.Effect<void, QdrantGraphEmbeddingsStoreError>;
}
export class QdrantGraphEmbeddingsStoreService extends Context.Service<
QdrantGraphEmbeddingsStoreService,
QdrantGraphEmbeddingsStoreServiceShape
>()(
"@trustgraph/flow/storage/embeddings/qdrant-graph/QdrantGraphEmbeddingsStoreService",
) {}
const qdrantGraphEmbeddingsStoreError = (operation: string, cause: unknown) =>
new QdrantGraphEmbeddingsStoreError({
operation,
message: errorMessage(cause),
cause,
});
export const makeQdrantGraphEmbeddingsStoreService = (
config: QdrantGraphEmbeddingsConfig = {},
): QdrantGraphEmbeddingsStoreServiceShape => {
2026-06-01 20:26:47 -05:00
const store = makeQdrantGraphEmbeddingsStore(config);
2026-06-01 16:22:25 -05:00
return {
store: Effect.fn("QdrantGraphEmbeddingsStore.store")(function* (message) {
return yield* Effect.tryPromise({
try: () => store.store(message),
catch: (cause) => qdrantGraphEmbeddingsStoreError("store", cause),
});
}),
deleteCollection: Effect.fn("QdrantGraphEmbeddingsStore.deleteCollection")(function* (
user,
collection,
) {
return yield* Effect.tryPromise({
try: () => store.deleteCollection(user, collection),
catch: (cause) => qdrantGraphEmbeddingsStoreError("delete-collection", cause),
});
}),
};
};
export const QdrantGraphEmbeddingsStoreLive = (
config: QdrantGraphEmbeddingsConfig = {},
): Layer.Layer<QdrantGraphEmbeddingsStoreService> =>
Layer.succeed(
QdrantGraphEmbeddingsStoreService,
QdrantGraphEmbeddingsStoreService.of(makeQdrantGraphEmbeddingsStoreService(config)),
);