@oscharko-dev/keiko-local-knowledge 0.2.0

package/dist/testing.js

@@ -0,0 +1,170 @@
+import { chunkDocument } from "./chunking/chunker-runner.js";
+import { createCapsule } from "./capsule-lifecycle.js";
+import { insertDocumentRow, insertParsedUnitRow } from "./discovery/persist.js";
+import { embedChunkBatch } from "./indexing/embedding-batcher.js";
+import { addSourceToCapsule } from "./source-lifecycle.js";
+const DEFAULT_EMBEDDING = {
+    provider: "openai",
+    modelId: "text-embedding-3-small",
+    vectorDimensions: 1536,
+    vectorMetric: "cosine",
+};
+export function deterministicVector(input, dimensions) {
+    const vec = new Float32Array(dimensions);
+    let hash = 0;
+    for (let i = 0; i < input.length; i += 1) {
+        hash = (hash * 31 + input.charCodeAt(i)) | 0;
+    }
+    vec[0] = input.length;
+    for (let i = 1; i < dimensions; i += 1) {
+        vec[i] = ((hash + i * 7) & 0xffff) / 0xffff;
+    }
+    return vec;
+}
+export function scriptedAdapter(options = {}) {
+    const identity = options.identity ?? DEFAULT_EMBEDDING;
+    const responder = options.responder ??
+        ((request) => ({
+            ok: true,
+            value: {
+                vector: deterministicVector(request.input, identity.vectorDimensions),
+                modelId: identity.modelId,
+                ...(identity.modelRevision !== undefined ? { modelRevision: identity.modelRevision } : {}),
+            },
+        }));
+    return {
+        endpoint: options.endpoint ?? "https://example.test/v1",
+        apiKey: options.apiKey ?? ["sk-", "test"].join(""),
+        request: async (request) => Promise.resolve(responder(request)),
+    };
+}
+function sampleCapsuleInput(overrides) {
+    return {
+        id: overrides.id,
+        displayName: overrides.displayName ?? "Engineering Capsule",
+        tags: ["alpha", "beta"],
+        retrievalEffort: "default",
+        outputMode: "answers",
+        answerGroundingPolicy: "require-citations",
+        embeddingModelIdentity: overrides.embeddingModelIdentity,
+        lifecycleState: "draft",
+        storageReference: "engineering/capsule-1",
+    };
+}
+function sampleSourceInput(id) {
+    return {
+        id,
+        displayName: `Source ${String(id)}`,
+        tags: [],
+        scope: {
+            kind: "folder",
+            rootPath: "/srv/docs",
+            recursive: true,
+        },
+    };
+}
+function composeUnit(unit, documentId, textLength) {
+    if (unit !== undefined) {
+        return { ...unit, documentId };
+    }
+    return {
+        kind: "page",
+        documentId,
+        pageNumber: 7,
+        pageLabel: "vii",
+        characterStart: 0,
+        characterEnd: textLength,
+    };
+}
+function normalizeSeedText(unit, text) {
+    const baseText = text ?? "alpha beta gamma delta epsilon zeta eta theta iota kappa lambda mu nu xi";
+    const requiredEnd = unit !== undefined && unit.kind !== "unsupported-media" ? unit.characterEnd : 0;
+    return baseText.length >= requiredEnd ? baseText : baseText.padEnd(requiredEnd, " x");
+}
+function resolveSeedOptions(options) {
+    const capsuleId = (options.capsuleId ?? "cap-1");
+    const sourceId = (options.sourceId ?? "src-1");
+    const documentId = (options.documentId ?? "doc-1");
+    const identity = options.identity ?? DEFAULT_EMBEDDING;
+    const text = normalizeSeedText(options.unit, options.text);
+    return {
+        capsuleId,
+        displayName: options.displayName ?? "Engineering Capsule",
+        sourceId,
+        documentId,
+        identity,
+        text,
+        contentHash: options.contentHash ?? "a".repeat(64),
+        safeDisplayName: options.safeDisplayName ?? "sample.txt",
+        unit: composeUnit(options.unit, documentId, text.length),
+        chunkingOptions: options.chunkingOptions ?? { maxTokens: 2, minTokens: 0, overlapTokens: 0 },
+    };
+}
+function insertSeedRows(store, options) {
+    createCapsule(store, sampleCapsuleInput({
+        id: options.capsuleId,
+        displayName: options.displayName,
+        embeddingModelIdentity: options.identity,
+    }));
+    addSourceToCapsule(store, options.capsuleId, sampleSourceInput(options.sourceId));
+    insertDocumentRow(store._internal.db, {
+        id: options.documentId,
+        capsuleId: options.capsuleId,
+        sourceId: String(options.sourceId),
+        documentPath: "docs/sample.txt",
+        sizeBytes: 1024,
+        mediaType: "text/plain",
+        contentHash: options.contentHash,
+        parserId: "text",
+        parserVersion: "1",
+        lastExtractedAt: 1_700_000_000_000,
+        status: "extracted",
+        safeDisplayName: options.safeDisplayName,
+    });
+    insertParsedUnitRow(store._internal.db, store._internal.contentCipher, options.capsuleId, `unit-${String(options.capsuleId)}`, options.unit);
+}
+function buildSeedChunks(store, options) {
+    const chunkResult = chunkDocument(store, {
+        capsuleId: options.capsuleId,
+        sourceId: options.sourceId,
+        documentId: options.documentId,
+        sourceText: options.text,
+    }, options.chunkingOptions);
+    return {
+        chunkIds: chunkResult.chunkIds,
+        chunks: chunkResult.chunkIds.map((id, index) => ({
+            id,
+            capsuleId: options.capsuleId,
+            sourceId: options.sourceId,
+            documentId: options.documentId,
+            text: `chunk-${String(index)}-${String(options.capsuleId)}`,
+        })),
+    };
+}
+async function embedSeedChunks(store, identity, chunks) {
+    let counter = 0;
+    await embedChunkBatch(chunks, {
+        adapter: scriptedAdapter({ identity }),
+        store,
+        pinnedIdentity: identity,
+        concurrency: 2,
+        now: () => 1_700_000_000_000,
+        idSource: () => {
+            counter += 1;
+            return `storage-${String(counter)}`;
+        },
+    });
+}
+export async function seedCapsuleWithVectors(store, options = {}) {
+    const resolved = resolveSeedOptions(options);
+    insertSeedRows(store, resolved);
+    const { chunkIds, chunks } = buildSeedChunks(store, resolved);
+    await embedSeedChunks(store, resolved.identity, chunks);
+    return {
+        capsuleId: resolved.capsuleId,
+        sourceId: resolved.sourceId,
+        documentId: resolved.documentId,
+        chunkIds,
+        vectorTexts: chunks.map((chunk) => chunk.text),
+    };
+}

package/dist/version.d.ts

@@ -0,0 +1,2 @@
+export declare const KEIKO_LOCAL_KNOWLEDGE_VERSION: "0.1.0";
+//# sourceMappingURL=version.d.ts.map

package/dist/version.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"version.d.ts","sourceRoot":"","sources":["../src/version.ts"],"names":[],"mappings":"AAGA,eAAO,MAAM,6BAA6B,EAAG,OAAgB,CAAC"}

package/dist/version.js

@@ -0,0 +1,4 @@
+// Package version literal for @oscharko-dev/keiko-local-knowledge. Bumped lockstep with the
+// package.json "version" field so a value-import gives consumers a single source of truth
+// without parsing JSON at runtime.
+export const KEIKO_LOCAL_KNOWLEDGE_VERSION = "0.1.0";

package/package.json

@@ -0,0 +1,43 @@
+{
+  "name": "@oscharko-dev/keiko-local-knowledge",
+  "version": "0.2.0",
+  "type": "module",
+  "license": "Apache-2.0",
+  "description": "Internal local-knowledge package: on-disk capsule store (node:sqlite + WAL) and capsule/source/CapsuleSet lifecycle CRUD for the Local Knowledge Connector (Epic #189, Issue #193). Not published independently.",
+  "main": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.js"
+    },
+    "./testing": {
+      "types": "./dist/testing.d.ts",
+      "import": "./dist/testing.js"
+    }
+  },
+  "scripts": {
+    "build": "tsc -b tsconfig.json",
+    "typecheck": "tsc -b tsconfig.json",
+    "test": "vitest run"
+  },
+  "files": [
+    "dist"
+  ],
+  "sideEffects": false,
+  "engines": {
+    "node": ">=22"
+  },
+  "dependencies": {
+    "@oscharko-dev/keiko-contracts": "0.2.0",
+    "@oscharko-dev/keiko-model-gateway": "0.2.0",
+    "@oscharko-dev/keiko-security": "0.2.0",
+    "@oscharko-dev/keiko-workspace": "0.2.0",
+    "@types/yauzl": "^3.3.0",
+    "pdfjs-dist": "^6.0.227",
+    "yauzl": "^3.3.2"
+  },
+  "optionalDependencies": {
+    "@napi-rs/canvas": "^1.0.0"
+  }
+}