update llamacloud

Author: leehuwuj

templates/components/vectordbs/python/llamacloud/generate.py

@@ -1,20 +1,18 @@
 # flake8: noqa: E402
 import os
+
 from dotenv import load_dotenv
 
 load_dotenv()
 
-from llama_cloud import PipelineType
-
-from app.settings import init_settings
-from llama_index.core.settings import Settings
-
+import logging
 
 from app.engine.index import get_client, get_index
-
-import logging
-from llama_index.core.readers import SimpleDirectoryReader
 from app.engine.service import LLamaCloudFileService
+from app.settings import init_settings
+from llama_cloud import PipelineType
+from llama_index.core.readers import SimpleDirectoryReader
+from llama_index.core.settings import Settings
 
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger()
@@ -83,10 +81,6 @@ def generate_datasource():
                 project_id,
                 pipeline_id,
                 f,
-                custom_metadata={
-                    # Set private=false to mark the document as public (required for filtering)
-                    "private": "false",
-                },
             )
 
     logger.info("Finished generating the index")

templates/components/vectordbs/python/llamacloud/query_filter.py

@@ -5,7 +5,7 @@ def generate_filters(doc_ids):
     """
     Generate public/private document filters based on the doc_ids and the vector store.
     """
-    # Using "is_empty" filter to include the documents don't have the "private" key because they're uploaded in LlamaCloud UI
+    # public documents (ingested by "poetry run generate" or in the LlamaCloud UI) don't have the "private" field
     public_doc_filter = MetadataFilter(
         key="private",
         value=None,

templates/components/vectordbs/typescript/llamacloud/generate.ts

@@ -25,6 +25,8 @@ async function* walk(dir: string): AsyncGenerator<string> {
 
 async function loadAndIndex() {
   const index = await getDataSource();
+  // ensure the index is available or create a new one
+  await index.ensureIndex();
   const projectId = await index.getProjectId();
   const pipelineId = await index.getPipelineId();
 
@@ -37,7 +39,6 @@ async function loadAndIndex() {
         projectId,
         pipelineId,
         new File([buffer], filename),
-        { private: "false" },
       );
     } catch (error) {
       if (

templates/components/vectordbs/typescript/llamacloud/queryFilter.ts

@@ -1,7 +1,7 @@
 import { CloudRetrieveParams, MetadataFilter } from "llamaindex";
 
 export function generateFilters(documentIds: string[]) {
-  // public documents don't have the "private" field or it's set to "false"
+  // public documents (ingested by "npm run generate" or in the LlamaCloud UI) don't have the "private" field
   const publicDocumentsFilter: MetadataFilter = {
     key: "private",
     operator: "is_empty",

templates/types/streaming/fastapi/app/api/routers/chat_config.py

@@ -1,11 +1,10 @@
 import logging
 import os
 
-from fastapi import APIRouter
+from fastapi import APIRouter, HTTPException
 
 from app.api.routers.models import ChatConfig
 
-
 config_router = r = APIRouter()
 
 logger = logging.getLogger("uvicorn")
@@ -27,6 +26,10 @@ async def chat_config() -> ChatConfig:
 
     @r.get("/llamacloud")
     async def chat_llama_cloud_config():
+        if not os.getenv("LLAMA_CLOUD_API_KEY"):
+            raise HTTPException(
+                status_code=500, detail="LlamaCloud API KEY is not configured"
+            )
         projects = LLamaCloudFileService.get_all_projects_with_pipelines()
         pipeline = os.getenv("LLAMA_CLOUD_INDEX_NAME")
         project = os.getenv("LLAMA_CLOUD_PROJECT_NAME")

templates/types/streaming/nextjs/app/components/ui/chat/widgets/LlamaCloudSelector.tsx

@@ -66,7 +66,16 @@ export function LlamaCloudSelector({
   useEffect(() => {
     if (process.env.NEXT_PUBLIC_USE_LLAMACLOUD === "true" && !config) {
       fetch(`${backend}/api/chat/config/llamacloud`)
-        .then((response) => response.json())
+        .then((response) => {
+          if (!response.ok) {
+            return response.json().then((errorData) => {
+              window.alert(
+                `Error: ${JSON.stringify(errorData) || "Unknown error occurred"}`,
+              );
+            });
+          }
+          return response.json();
+        })
         .then((data) => {
           const pipeline = defaultPipeline ?? data.pipeline; // defaultPipeline will override pipeline in .env
           setConfig({ ...data, pipeline });