chore: replace Python examples with llama-deploy

.changeset/good-avocados-try.md

@@ -0,0 +1,5 @@
+---
+"create-llama": patch
+---
+
+chore: replace Python examples with llama-deploy

.github/workflows/e2e.yml

@@ -63,15 +63,6 @@ jobs:
         run: pnpm run pack-install
         working-directory: packages/create-llama
 
-      - name: Build and store server package
-        run: |
-          pnpm run build
-          wheel_file=$(ls dist/*.whl | head -n 1)
-          mkdir -p "${{ runner.temp }}"
-          cp "$wheel_file" "${{ runner.temp }}/"
-          echo "SERVER_PACKAGE_PATH=${{ runner.temp }}/$(basename "$wheel_file")" >> $GITHUB_ENV
-        working-directory: python/llama-index-server
-
       - name: Run Playwright tests for Python
         run: pnpm run e2e:python
         env:

packages/create-llama/e2e/python/resolve_dependencies.spec.ts

@@ -3,11 +3,8 @@ import { exec } from "child_process";
 import fs from "fs";
 import path from "path";
 import util from "util";
-import {
-  ALL_USE_CASES,
-  TemplateFramework,
-  TemplateVectorDB,
-} from "../../helpers/types";
+import { TemplateFramework, TemplateUseCase, TemplateVectorDB } from "../../helpers";
+import { ALL_PYTHON_USE_CASES } from "../../helpers/use-case";
 import { RunCreateLlamaOptions, createTestDir, runCreateLlama } from "../utils";
 
 const execAsync = util.promisify(exec);
@@ -17,11 +14,15 @@ const vectorDb: TemplateVectorDB = process.env.VECTORDB
   ? (process.env.VECTORDB as TemplateVectorDB)
   : "none";
 
+const useCases: TemplateUseCase[] = vectorDb === "llamacloud" ? [
+  "agentic_rag", "deep_research", "financial_report"
+] : ALL_PYTHON_USE_CASES
+
 test.describe("Mypy check", () => {
   test.describe.configure({ retries: 0 });
 
   test.describe("LlamaIndexServer", async () => {
-    for (const useCase of ALL_USE_CASES) {
+    for (const useCase of useCases) {
       test(`should pass mypy for use case: ${useCase}`, async () => {
         const cwd = await createTestDir();
         await createAndCheckLlamaProject({

packages/create-llama/e2e/shared/llamaindexserver_template.spec.ts

@@ -2,11 +2,11 @@ import { expect, test } from "@playwright/test";
 import { ChildProcess } from "child_process";
 import fs from "fs";
 import path from "path";
+import { type TemplateFramework, type TemplateVectorDB } from "../../helpers";
 import {
-  ALL_USE_CASES,
-  type TemplateFramework,
-  type TemplateVectorDB,
-} from "../../helpers";
+  ALL_PYTHON_USE_CASES,
+  ALL_TYPESCRIPT_USE_CASES,
+} from "../../helpers/use-case";
 import { createTestDir, runCreateLlama } from "../utils";
 
 const templateFramework: TemplateFramework = process.env.FRAMEWORK
@@ -17,10 +17,15 @@ const vectorDb: TemplateVectorDB = process.env.VECTORDB
   : "none";
 const llamaCloudProjectName = "create-llama";
 const llamaCloudIndexName = "e2e-test";
+const allUseCases =
+  templateFramework === "nextjs"
+    ? ALL_TYPESCRIPT_USE_CASES
+    : ALL_PYTHON_USE_CASES;
+const isPythonLlamaDeploy = templateFramework === "fastapi";
 
 const userMessage = "Write a blog post about physical standards for letters";
 
-for (const useCase of ALL_USE_CASES) {
+for (const useCase of allUseCases) {
   test.describe(`Test use case ${useCase} ${templateFramework} ${vectorDb}`, async () => {
     let port: number;
     let cwd: string;
@@ -35,7 +40,7 @@ for (const useCase of ALL_USE_CASES) {
         templateFramework,
         vectorDb,
         port,
-        postInstallAction: "runApp",
+        postInstallAction: isPythonLlamaDeploy ? "dependencies" : "runApp",
         useCase,
         llamaCloudProjectName,
         llamaCloudIndexName,
@@ -50,6 +55,11 @@ for (const useCase of ALL_USE_CASES) {
     });
 
     test("Frontend should have a title", async ({ page }) => {
+      test.skip(
+        isPythonLlamaDeploy,
+        "Skip frontend tests for Python LllamaDeploy",
+      );
+
       await page.goto(`http://localhost:${port}`);
       await expect(page.getByText("Built by LlamaIndex")).toBeVisible({
         timeout: 5 * 60 * 1000,
@@ -60,8 +70,10 @@ for (const useCase of ALL_USE_CASES) {
       page,
     }) => {
       test.skip(
-        useCase === "financial_report" || useCase === "deep_research",
-        "Skip chat tests for financial report and deep research.",
+        useCase === "financial_report" ||
+          useCase === "deep_research" ||
+          isPythonLlamaDeploy,
+        "Skip chat tests for financial report and deep research. Also skip for Python LlamaDeploy",
       );
       await page.goto(`http://localhost:${port}`);
       await page.fill("form textarea", userMessage);

packages/create-llama/e2e/typescript/resolve_dependencies.spec.ts

@@ -4,11 +4,11 @@ import fs from "fs";
 import path from "path";
 import util from "util";
 import {
-  ALL_USE_CASES,
   TemplateFramework,
   TemplateUseCase,
   TemplateVectorDB,
 } from "../../helpers/types";
+import { ALL_TYPESCRIPT_USE_CASES } from "../../helpers/use-case";
 import { createTestDir, runCreateLlama } from "../utils";
 
 const execAsync = util.promisify(exec);
@@ -21,7 +21,7 @@ const vectorDb: TemplateVectorDB = process.env.VECTORDB
 test.describe("Test resolve TS dependencies", () => {
   test.describe.configure({ retries: 0 });
 
-  for (const useCase of ALL_USE_CASES) {
+  for (const useCase of ALL_TYPESCRIPT_USE_CASES) {
     const optionDescription = `useCase: ${useCase}, vectorDb: ${vectorDb}`;
     test.describe(`${optionDescription}`, () => {
       test(`${optionDescription}`, async () => {

packages/create-llama/helpers/env-variables.ts

@@ -1,20 +1,17 @@
 import fs from "fs/promises";
 import path from "path";
 import {
+  EnvVar,
   InstallTemplateArgs,
   ModelConfig,
   TemplateFramework,
   TemplateType,
+  TemplateUseCase,
   TemplateVectorDB,
 } from "./types";
 
 import { TSYSTEMS_LLMHUB_API_URL } from "./providers/llmhub";
-
-export type EnvVar = {
-  name?: string;
-  description?: string;
-  value?: string;
-};
+import { USE_CASE_CONFIGS } from "./use-case";
 
 const renderEnvVar = (envVars: EnvVar[]): string => {
   return envVars.reduce(
@@ -228,7 +225,15 @@ Otherwise, use CHROMA_HOST and CHROMA_PORT config above`,
   }
 };
 
-const getModelEnvs = (modelConfig: ModelConfig): EnvVar[] => {
+const getModelEnvs = (
+  modelConfig: ModelConfig,
+  framework: TemplateFramework,
+  template: TemplateType,
+  useCase: TemplateUseCase,
+): EnvVar[] => {
+  const isPythonLlamaDeploy =
+    framework === "fastapi" && template === "llamaindexserver";
+
   return [
     {
       name: "MODEL",
@@ -240,25 +245,44 @@ const getModelEnvs = (modelConfig: ModelConfig): EnvVar[] => {
       description: "Name of the embedding model to use.",
       value: modelConfig.embeddingModel,
     },
-    {
-      name: "CONVERSATION_STARTERS",
-      description: "The questions to help users get started (multi-line).",
-    },
-    ...(modelConfig.provider === "openai"
+    ...(isPythonLlamaDeploy
       ? [
           {
-            name: "OPENAI_API_KEY",
-            description: "The OpenAI API key to use.",
-            value: modelConfig.apiKey,
+            name: "NEXT_PUBLIC_STARTER_QUESTIONS",
+            description:
+              "Initial questions to display in the chat (`starterQuestions`)",
+            value: JSON.stringify(
+              USE_CASE_CONFIGS[useCase]?.starterQuestions ?? [],
+            ),
           },
+        ]
+      : [
           {
-            name: "LLM_TEMPERATURE",
-            description: "Temperature for sampling from the model.",
+            name: "CONVERSATION_STARTERS",
+            description:
+              "The questions to help users get started (multi-line).",
           },
+        ]),
+    ...(USE_CASE_CONFIGS[useCase]?.additionalEnvVars ?? []),
+    ...(modelConfig.provider === "openai"
+      ? [
           {
-            name: "LLM_MAX_TOKENS",
-            description: "Maximum number of tokens to generate.",
+            name: "OPENAI_API_KEY",
+            description: "The OpenAI API key to use.",
+            value: modelConfig.apiKey,
           },
+          ...(isPythonLlamaDeploy
+            ? []
+            : [
+                {
+                  name: "LLM_TEMPERATURE",
+                  description: "Temperature for sampling from the model.",
+                },
+                {
+                  name: "LLM_MAX_TOKENS",
+                  description: "Maximum number of tokens to generate.",
+                },
+              ]),
         ]
       : []),
     ...(modelConfig.provider === "anthropic"
@@ -367,11 +391,12 @@ const getModelEnvs = (modelConfig: ModelConfig): EnvVar[] => {
 
 const getFrameworkEnvs = (
   framework: TemplateFramework,
+  template?: TemplateType,
   port?: number,
 ): EnvVar[] => {
   const sPort = port?.toString() || "8000";
   const result: EnvVar[] = [];
-  if (framework === "fastapi") {
+  if (framework === "fastapi" && template !== "llamaindexserver") {
     result.push(
       ...[
         {
@@ -403,6 +428,7 @@ export const createBackendEnvFile = async (
     | "template"
     | "port"
     | "useLlamaParse"
+    | "useCase"
   >,
 ) => {
   // Init env values
@@ -418,11 +444,27 @@ export const createBackendEnvFile = async (
         ]
       : []),
     ...getVectorDBEnvs(opts.vectorDb, opts.framework, opts.template),
-    ...getFrameworkEnvs(opts.framework, opts.port),
-    ...getModelEnvs(opts.modelConfig),
+    ...getFrameworkEnvs(opts.framework, opts.template, opts.port),
+    ...getModelEnvs(
+      opts.modelConfig,
+      opts.framework,
+      opts.template,
+      opts.useCase,
+    ),
   ];
   // Render and write env file
   const content = renderEnvVar(envVars);
-  await fs.writeFile(path.join(root, envFileName), content);
+
+  const isPythonLlamaDeploy =
+    opts.framework === "fastapi" && opts.template === "llamaindexserver";
+
+  // each llama-deploy service will need a .env inside its directory
+  // this .env will be copied along with workflow code when service is deployed
+  // so that we need to put the .env file inside src/ instead of root
+  const envPath = isPythonLlamaDeploy
+    ? path.join(root, "src", envFileName)
+    : path.join(root, envFileName);
+
+  await fs.writeFile(envPath, content);
   console.log(`Created '${envFileName}' file. Please check the settings.`);
 };

packages/create-llama/helpers/index.ts

@@ -117,8 +117,13 @@ const downloadFile = async (url: string, destPath: string) => {
 const prepareContextData = async (
   root: string,
   dataSources: TemplateDataSource[],
+  isPythonLlamaDeploy: boolean,
 ) => {
-  await makeDir(path.join(root, "data"));
+  const dataDir = isPythonLlamaDeploy
+    ? path.join(root, "ui", "data")
+    : path.join(root, "data");
+
+  await makeDir(dataDir);
   for (const dataSource of dataSources) {
     const dataSourceConfig = dataSource?.config as FileSourceConfig;
     // If the path is URLs, download the data and save it to the data directory
@@ -128,20 +133,15 @@ const prepareContextData = async (
         dataSourceConfig.url.toString(),
       );
       const destPath = path.join(
-        root,
-        "data",
+        dataDir,
         dataSourceConfig.filename ??
           path.basename(dataSourceConfig.url.toString()),
       );
       await downloadFile(dataSourceConfig.url.toString(), destPath);
     } else {
       // Copy local data
       console.log("Copying data from path:", dataSourceConfig.path);
-      const destPath = path.join(
-        root,
-        "data",
-        path.basename(dataSourceConfig.path),
-      );
+      const destPath = path.join(dataDir, path.basename(dataSourceConfig.path));
       await fsExtra.copy(dataSourceConfig.path, destPath);
     }
   }
@@ -156,6 +156,9 @@ export const installTemplate = async (props: InstallTemplateArgs) => {
     await installTSTemplate(props);
   }
 
+  const isPythonLlamaDeploy =
+    props.framework === "fastapi" && props.template === "llamaindexserver";
+
   // This is a backend, so we need to copy the test data and create the env file.
 
   // Copy the environment file to the target directory.
@@ -164,6 +167,7 @@ export const installTemplate = async (props: InstallTemplateArgs) => {
   await prepareContextData(
     props.root,
     props.dataSources.filter((ds) => ds.type === "file"),
+    isPythonLlamaDeploy,
   );
 
   if (
@@ -183,10 +187,12 @@ export const installTemplate = async (props: InstallTemplateArgs) => {
     );
   }
 
-  // Create outputs directory
-  await makeDir(path.join(props.root, "output/tools"));
-  await makeDir(path.join(props.root, "output/uploaded"));
-  await makeDir(path.join(props.root, "output/llamacloud"));
+  if (!isPythonLlamaDeploy) {
+    // Create outputs directory (llama-deploy doesn't need this)
+    await makeDir(path.join(props.root, "output/tools"));
+    await makeDir(path.join(props.root, "output/uploaded"));
+    await makeDir(path.join(props.root, "output/llamacloud"));
+  }
 };
 
 export * from "./types";