Getting started with Streamlit in Snowflake¶

This topic walks you through deploying your first Streamlit in Snowflake app in under five minutes using a container runtime. After that, two hands-on examples show you how to build real apps that query data, personalize the experience for each viewer, and write back to Snowflake.

先决条件

Before you can create a Streamlit app, ensure that your administrator has completed the essential security setup for Streamlit apps.

您的角色必须拥有以下权限：

Privilege	Object	Notes
USAGE	Database where you create the Streamlit app
CREATE STREAMLIT, USAGE	Schema where you create the Streamlit app
USAGE	Compute pool that runs the Streamlit app	For all accounts, Snowflake configures a general-purpose compute pool that typical users will have access to. For more information, see Configuring your own preferred compute pools for Streamlit apps.
USAGE	Warehouse that runs queries in the Streamlit app

For more information, see Privileges required to create and use a Streamlit app.

Deploy your first Streamlit in Snowflake app¶

最快速的入门方式是使用默认的入门代码创建一个 Streamlit 应用。当您在不指定源文件的情况下创建应用程序时，Snowflake 会自动提供示例代码。

Sign in to Snowsight.
In the navigation menu, select Projects » Streamlit.
Select + Streamlit App.
输入应用程序的名称。
选择要在其中创建应用程序的数据库和架构。
Select Run on container.
选择计算池和查询仓库。
Select Create.

Snowsight redirects you to the app editor. Your app will be ready within a few minutes. Then, you can view and edit it immediately.

在 SQL 会话中，运行以下 SQL 命令：

CREATE STREAMLIT my_first_app
   RUNTIME_NAME = 'SYSTEM$ST_CONTAINER_RUNTIME_PY3_11'
   COMPUTE_POOL = my_compute_pool
   QUERY_WAREHOUSE = my_warehouse;

ALTER STREAMLIT my_first_app ADD LIVE VERSION FROM LAST;

To view your app, sign in to Snowsight, then In the navigation menu, select Projects » Streamlit, and select your app.

Note

Snowflake CLI version 3.14.0 or later is required. Version 3.14+ uses the modern CREATE STREAMLIT syntax by default.

初始化一个新的 Streamlit 项目：
```
snow init my_first_app --template example_streamlit
```
导航到项目目录：
```
cd my_first_app
```

Edit the snowflake.yml file to use a container runtime:

definition_version: 2
entities:
my_streamlit:
   type: streamlit
   identifier: my_first_app
   query_warehouse: my_warehouse
   compute_pool: my_compute_pool
   runtime_name: SYSTEM$ST_CONTAINER_RUNTIME_PY3_11
   main_file: streamlit_app.py
   artifacts:
   - streamlit_app.py

部署应用程序并在浏览器中打开它：
```
snow streamlit deploy --open
```

编辑应用程序

部署后，您可以编辑应用程序代码以对其进行自定义。快速测试方法如下：

Sign in to Snowsight.
In the navigation menu, select Projects » Streamlit, and then select your app.
Select Edit.
Modify the code in streamlit_app.py.
Select Run to see your changes.

获取应用程序的源位置：
```
DESCRIBE STREAMLIT my_first_app;
```

将更新后的文件复制到该位置：

COPY FILES INTO '<live_version_location_uri>' FROM @my_stage FILES = ('streamlit_app.py');

Edit streamlit_app.py in your local project directory.
重新部署：
```
snow streamlit deploy --replace
```

For more information, see Edit your Streamlit app.

下一步是什么？

现在您已经有了一个正在运行的应用程序，请尝试以下实践示例之一：

Example: Build a personalized data dashboard: Build a dashboard that queries Snowflake data and personalizes the display for each viewer using st.connection and st.user.
Example: Build a form that writes to Snowflake: Build a form that writes user input back to a Snowflake table, demonstrating st.form, dependency management, and st.user.

要了解有关特定主题的更多信息，请执行以下操作：

Create your Streamlit app: Detailed instructions for creating apps from Snowsight, SQL, or the CLI.
Manage dependencies for your Streamlit app: Add Python packages to your app.
Runtime environments for Streamlit apps: Understand container and warehouse runtimes.
External network access in Streamlit in Snowflake: Connect your app to external services.