@larkup/cli 0.1.14

package/README.md

@@ -0,0 +1,58 @@
+# Larkup CLI
+
+The official Command Line Interface for **Larkup**. 
+Build, index, and serve a Retrieval-Augmented Generation (RAG) pipeline directly from your terminal.
+
+## Installation
+
+You can install the CLI globally via npm or run it via npx:
+
+```bash
+npm install -g @larkup/cli
+# or
+npx @larkup/cli
+```
+
+## Usage
+
+The CLI provides a set of commands to manage your RAG workspaces, configure vector stores, ingest documents, and query your data.
+
+```bash
+larkup <command> [options]
+```
+
+### Core Commands
+
+- **`init [name]`**: Create a new RAG server (workspace project).
+- **`config`**: Configure vector stores and embedding models for your active workspace.
+- **`add-doc`**: Ingest documents (local files, URLs, or text) into your workspace.
+- **`index`**: Process, chunk, and index your ingested documents into the configured vector database.
+- **`serve`**: Spin up the generated RAG server locally for fast testing and API integration.
+- **`query`**: Test retrieval quality by querying your indexed data directly from the terminal.
+- **`chat`**: Start an interactive chat session with your RAG pipeline in the terminal.
+- **`generate`**: Generate server configurations or boilerplate code.
+- **`settings`**: View or update your global settings and environment variables.
+
+### Managing Workspaces (Servers)
+
+- **`use`**: Switch your active server/workspace context.
+
+## Example Workflow
+
+```bash
+# 1. Initialize a new project
+larkup init my-rag-server
+
+# 2. Add documents to the project
+larkup add-doc ./data/knowledge-base.txt
+
+# 3. Index the documents into the vector store
+larkup index
+
+# 4. Start an interactive chat session with your data
+larkup chat
+```
+
+## Documentation
+
+For more detailed guides and information, please refer to the [Larkup Documentation](https://larkup.de/docs) or visit our [Web UI](https://larkup.de).

package/dist/chroma-GIDEKUIH.js

@@ -0,0 +1,276 @@
+import "./chunk-3RG5ZIWI.js";
+
+// ../../packages/vector-stores/src/adapters/chroma.ts
+var RRF_K = 60;
+var ChromaAdapter = class {
+  constructor(config) {
+    this.config = config;
+    this.collectionName = config.collectionName?.trim() || "documents";
+  }
+  config;
+  client = null;
+  collection = null;
+  collectionName;
+  /**
+   * Lazily connect to the Chroma server or cloud.
+   * The `chromadb` package is dynamically imported to avoid module-eval
+   * failures when the package isn't installed.
+   */
+  async getClient() {
+    if (this.client) return this.client;
+    let chromadb;
+    try {
+      chromadb = await import("chromadb");
+    } catch {
+      throw new Error(
+        'The "chromadb" package is not installed. Please install it first via the Configure page or run: pnpm add chromadb --filter @larkup/vector-stores'
+      );
+    }
+    if (this.config.mode === "cloud") {
+      if (!this.config.apiKey) {
+        throw new Error("Chroma Cloud requires an API key.");
+      }
+      const CloudClientClass = chromadb.CloudClient ?? chromadb.default?.CloudClient;
+      if (!CloudClientClass) {
+        throw new Error("Could not find CloudClient in the chromadb package.");
+      }
+      this.client = new CloudClientClass({
+        apiKey: this.config.apiKey,
+        tenant: this.config.tenant || "default_tenant",
+        database: this.config.database || "default_database"
+      });
+    } else {
+      const host = this.config.host?.trim() || "http://localhost:8000";
+      const ClientClass = chromadb.ChromaClient ?? chromadb.default?.ChromaClient;
+      if (!ClientClass) {
+        throw new Error("Could not find ChromaClient in the chromadb package.");
+      }
+      const opts = { path: host };
+      if (this.config.authToken) {
+        opts.auth = {
+          provider: "token",
+          credentials: this.config.authToken
+        };
+      }
+      this.client = new ClientClass(opts);
+    }
+    return this.client;
+  }
+  async getCollection() {
+    if (this.collection) return this.collection;
+    const client = await this.getClient();
+    this.collection = await client.getOrCreateCollection({
+      name: this.collectionName,
+      metadata: { "hnsw:space": "cosine" }
+    });
+    return this.collection;
+  }
+  // ── Lifecycle ──────────────────────────────────────────────────────────────
+  async init(_dimensions) {
+    await this.getCollection();
+  }
+  async reset() {
+    const client = await this.getClient();
+    try {
+      await client.deleteCollection({ name: this.collectionName });
+    } catch {
+    }
+    this.collection = null;
+  }
+  // ── Upsert ─────────────────────────────────────────────────────────────────
+  async upsert(records) {
+    if (records.length === 0) return;
+    const collection = await this.getCollection();
+    const ids = [];
+    const embeddings = [];
+    const documents = [];
+    const metadatas = [];
+    for (const r of records) {
+      ids.push(r.id);
+      embeddings.push(r.vector);
+      documents.push(r.text);
+      metadatas.push(this.buildMetadata(r));
+    }
+    const BATCH = 4096;
+    for (let i = 0; i < ids.length; i += BATCH) {
+      await collection.upsert({
+        ids: ids.slice(i, i + BATCH),
+        embeddings: embeddings.slice(i, i + BATCH),
+        documents: documents.slice(i, i + BATCH),
+        metadatas: metadatas.slice(i, i + BATCH)
+      });
+    }
+  }
+  buildMetadata(r) {
+    const meta = {
+      title: r.title,
+      source: r.source,
+      documentId: r.documentId,
+      chunkIndex: r.chunkIndex
+    };
+    if (r.url) meta.url = r.url;
+    if (r.metadata) {
+      for (const [k, v] of Object.entries(r.metadata)) {
+        if (v === null || v === void 0) continue;
+        if (typeof v === "string" || typeof v === "number" || typeof v === "boolean") {
+          meta[k] = v;
+        } else {
+          meta[k] = JSON.stringify(v);
+        }
+      }
+    }
+    return meta;
+  }
+  // ── Count ──────────────────────────────────────────────────────────────────
+  async count() {
+    try {
+      const collection = await this.getCollection();
+      return await collection.count();
+    } catch {
+      return null;
+    }
+  }
+  // ── Query ──────────────────────────────────────────────────────────────────
+  async query(vector, topK, queryText) {
+    const collection = await this.getCollection();
+    const isSemantic = this.config.indexType === "semantic";
+    const isLexical = this.config.indexType === "lexical";
+    if (isSemantic || !queryText?.trim()) {
+      return this.denseQuery(collection, vector, topK);
+    }
+    if (isLexical) {
+      return this.documentQuery(collection, queryText, topK);
+    }
+    const denseHits = await this.denseQuery(collection, vector, topK);
+    try {
+      const textHits = await this.documentQuery(collection, queryText, topK);
+      if (textHits.length > 0) {
+        return rrfMerge(denseHits, textHits, topK);
+      }
+    } catch {
+    }
+    return denseHits;
+  }
+  async denseQuery(collection, vector, topK) {
+    const result = await collection.query({
+      queryEmbeddings: [vector],
+      nResults: topK,
+      include: ["documents", "metadatas", "distances"]
+    });
+    return this.parseQueryResult(result);
+  }
+  async documentQuery(collection, queryText, topK) {
+    const result = await collection.get({
+      whereDocument: { "$contains": queryText },
+      limit: topK,
+      include: ["documents", "metadatas"]
+    });
+    return this.parseGetResult(result);
+  }
+  parseGetResult(result) {
+    if (!result.ids || result.ids.length === 0) return [];
+    const ids = result.ids;
+    const documents = result.documents ?? [];
+    const metadatas = result.metadatas ?? [];
+    return ids.map((id, i) => {
+      const meta = metadatas[i] ?? {};
+      const score = 1;
+      const {
+        title,
+        url,
+        source,
+        documentId,
+        chunkIndex,
+        ...customMetadata
+      } = meta;
+      return {
+        id,
+        score,
+        text: documents[i] ?? "",
+        title: title ?? "Untitled",
+        url: url || void 0,
+        documentId: documentId ?? "",
+        metadata: Object.keys(customMetadata).length > 0 ? customMetadata : void 0
+      };
+    });
+  }
+  parseQueryResult(result) {
+    if (!result.ids?.[0]) return [];
+    const ids = result.ids[0];
+    const documents = result.documents?.[0] ?? [];
+    const metadatas = result.metadatas?.[0] ?? [];
+    const distances = result.distances?.[0] ?? [];
+    return ids.map((id, i) => {
+      const meta = metadatas[i] ?? {};
+      const distance = distances[i] ?? 0;
+      const score = 1 / (1 + distance);
+      const {
+        title,
+        url,
+        source,
+        documentId,
+        chunkIndex,
+        ...customMetadata
+      } = meta;
+      return {
+        id,
+        score,
+        text: documents[i] ?? "",
+        title: title ?? "Untitled",
+        url: url || void 0,
+        documentId: documentId ?? "",
+        metadata: Object.keys(customMetadata).length > 0 ? customMetadata : void 0
+      };
+    });
+  }
+  // ── Connection test ────────────────────────────────────────────────────────
+  async testConnection(_dimensions) {
+    let client;
+    try {
+      client = await this.getClient();
+    } catch (err) {
+      throw new Error(`Failed to connect to Chroma: ${err.message}`);
+    }
+    try {
+      await client.heartbeat();
+    } catch (err) {
+      const msg = err.message ?? "";
+      if (msg.includes("401") || msg.includes("Unauthorized") || msg.includes("403") || msg.includes("Forbidden")) {
+        throw new Error(
+          "Invalid Chroma credentials. Please check your auth token or API key."
+        );
+      }
+      if (msg.includes("ECONNREFUSED") || msg.includes("fetch failed")) {
+        throw new Error(
+          "Could not reach the Chroma server. Is it running and accessible at the configured URL?"
+        );
+      }
+      throw new Error(`Failed to reach Chroma server: ${msg}`);
+    }
+    try {
+      await client.listCollections();
+    } catch (err) {
+      throw new Error(
+        `Connected to Chroma but failed to access the database or tenant. Please verify your tenant and database names. Error: ${err.message}`
+      );
+    }
+  }
+};
+function rrfMerge(denseHits, sparseHits, topK) {
+  const scores = /* @__PURE__ */ new Map();
+  const addList = (hits) => hits.forEach((hit, rank) => {
+    const contrib = 1 / (RRF_K + rank + 1);
+    const prev = scores.get(hit.id);
+    if (prev) {
+      prev.score += contrib;
+    } else {
+      scores.set(hit.id, { hit, score: contrib });
+    }
+  });
+  addList(denseHits);
+  addList(sparseHits);
+  return [...scores.values()].sort((a, b) => b.score - a.score).slice(0, topK).map(({ hit, score }) => ({ ...hit, score }));
+}
+export {
+  ChromaAdapter
+};

package/dist/chunk-3RG5ZIWI.js

@@ -0,0 +1,10 @@
+var __require = /* @__PURE__ */ ((x) => typeof require !== "undefined" ? require : typeof Proxy !== "undefined" ? new Proxy(x, {
+  get: (a, b) => (typeof require !== "undefined" ? require : a)[b]
+}) : x)(function(x) {
+  if (typeof require !== "undefined") return require.apply(this, arguments);
+  throw Error('Dynamic require of "' + x + '" is not supported');
+});
+
+export {
+  __require
+};