Volcano Ark Large Model Service Platform A/B Testing

https://www.volcengine.com/docs/6287/
Admin
This service provides a comprehensive platform for designing, running, managing, and analyzing A/B...
Tokens:440,382
Snippets:1,057
Trust Score:10
Update:11 months ago
Show doc for...
Tokens:
Raw
### SDK Initialization and Start

Source: https://www.volcengine.com/docs/6287/96071

Provides the core initialization and start methods for the SDK. The `start` method signals the SDK to begin processing and sending events.

```APIDOC
## POST /websites/volcengine_6287/init_and_start

### Description
This section covers the essential steps for initializing the SDK and then signaling it to start sending events. Events collected before `start` are cached and sent after `start` is called.

### Method
POST

### Endpoint
/websites/volcengine_6287

### Parameters
#### Request Body (Initialization Parameters - refer to other sections for details)
- **app_id** (number) - Required
- **channel_domain** (string) - Required
- ... other init parameters ...

### Request Example (Initialization)
```javascript
window.collectEvent('init', {
    app_id: {{APPID}},
    channel_domain: '{{REPORT_URL}}'
    // ... other initialization options
});
```

### Request Example (Start Signal)
```javascript
// Must be called after all initialization configurations are complete.
window.collectEvent('start');
```

### Response
#### Success Response (200)
SDK initialized and started successfully. Events are now being sent.

#### Response Example
(No specific response body)

### Notes
- Calling `start` is crucial for actual data transmission. Events sent before `start` are held in a cache and merged with user properties upon `start` invocation.
```

--------------------------------

### SDK Start

Source: https://www.volcengine.com/docs/6287/1408234

Instructions on starting the AppLog SDK. The `start` method should only be called after initialization. Once called, the SDK begins collecting sensitive information and reporting requests.

```APIDOC
## SDK Start

**Description**

Do not call `start` before `init`. After calling `start`, the SDK will begin collecting sensitive information and reporting requests.

### Code Example

```typescript
appLog.start()
```
```

--------------------------------

### Example Return Value for All Experiment Configurations

Source: https://www.volcengine.com/docs/6287/444915

This example demonstrates the JSON structure returned by `get_all_experiment_configs`, showcasing multiple experiments a user might be part of, each with its variant details.

```json
{
  "go": {
    "val": "v1",
    "vid": "83580"
  },
  "age": {
    "val": 12,
    "vid": "160528"
  },
  "configure": {
    "val": {
      "settings": 1
    },
    "vid": "160528"
  },
  "fromCN": {
    "val": true,
    "vid": "160528"
  },
  "name": {
    "val": "asd",
    "vid": "160528"
  }
}
```

--------------------------------

### Example Return Value for All Feature Configurations

Source: https://www.volcengine.com/docs/6287/444915

This example illustrates the JSON structure returned by `get_all_feature_configs`, showing variant details for multiple features a user might be associated with.

```json
{
  "servergo": {
    "val": false,
    "vid": "20100615"
  },
  "python_key": {
    "val": "bbb",
    "vid": "20108187"
  }
}
```

--------------------------------

### Example Return Value for Experiment Configurations

Source: https://www.volcengine.com/docs/6287/444915

This provides an example of the JSON structure returned by `get_experiment_configs` and `get_all_experiment_configs`. It illustrates how experiment variants and their values are detailed within a dictionary.

```json
{
  "age": {
    "val": 12,
    "vid": "160528"
  },
  "configure": {
    "val": {
      "settings": 1
    },
    "vid": "160528"
  },
  "fromCN": {
    "val": true,
    "vid": "160528"
  },
  "name": {
    "val": "asd",
    "vid": "160528"
  }
}
```

--------------------------------

### Example Return Value for Feature Configurations

Source: https://www.volcengine.com/docs/6287/444915

This example shows the JSON structure returned by `get_feature_configs` and `get_all_feature_configs`, detailing the variant configuration for a specific feature.

```json
{
  "servergo": {
    "val": false,
    "vid": "20100615"
  }
}
```

--------------------------------

### Start Applog SDK

Source: https://www.volcengine.com/docs/6287/1408234

This snippet demonstrates how to start the Applog SDK. The `appLog.start()` method should be called after initialization and privacy consent. Calling `start()` enables sensitive information collection and request reporting.

```typescript
appLog.start()
```

--------------------------------

### Subscribe Did Ready Event

Source: https://www.volcengine.com/docs/6287/1408236

This section describes how to subscribe to the 'DidReady' event, which is triggered when application logs are ready. It includes the format of the 'didInfo' object and a JavaScript code example.

```APIDOC
## Subscribe Did Ready Event

### Description
Subscribes to the `Topic.DidReady` event to receive `didInfo` when application logs are ready. The `didInfo` object contains details about the device and its origin.

### Method
`appLog.topicReceiver()?.on(Topic.DidReady, (didInfo) => { ... })`

### Parameters
#### Callback Parameters
- **didInfo** (object) - Information about the device.
  - **bd_did** (string) - The device's unique identifier.
  - **install_id** (number) - The installation ID.
  - **device_info_from** (string) - Indicates if the data is from local cache (`did_from_local`) or server (`did_from_remote`).

### Request Example
```javascript
appLog.topicReceiver()?.on(Topic.DidReady, (didInfo) => {
  setTimeout(() => {
    hilog.info(0x0000, 'AppLogH', `AppLog did ready:  ${didInfo["bd_did"]}`)
  })
})
```

### Response
#### Callback Data
- **didInfo** (object) - Contains device information:
  - `bd_did` (string): Device identifier.
  - `install_id` (number): Installation identifier.
  - `device_info_from` (string): Source of the device information (`did_from_local` or `did_from_remote`).

### Response Example
```json
{
    "bd_did": "xxxx",
    "install_id": xxxx,
    "device_info_from": "did_from_remote"
}
```
```

--------------------------------

### Webhook Signature Generation Example

Source: https://www.volcengine.com/docs/6287/71443

This example shows pseudocode for generating webhook signatures using MD5, SHA1, SHA256, and HMAC-SHA256 algorithms. It explains how to encrypt the message body with a secret key for authentication.

```python
# md5/sha1/sha256摘要算法：直接将密钥拼接在消息body后面进行摘要
sign = encrypt_algorithm(body + secret_key)

# hmac sha256
sign = hmac_sha256(body, secret_key)
```

--------------------------------

### Webhook Request Body Example

Source: https://www.volcengine.com/docs/6287/71443

This is an example of the JSON array sent to the webhook endpoint for batch push notifications. It includes details about the push, task, receipt properties, parameters, user information, and trigger conditions.

```json
[
    {
        "push_id":"882f7c1c-af41-4351-9ec1-04a574597055", // 用于识别单条消息的uuid
        "app_id": "234901", // 应用ID
        "sign": "f67ebb5fac42feacdee69d77bb0314cf", // 校验签名信息
        "timestamp":1606789775, // 发送时间戳  
        "task":{
                "task_id":100, // 推送任务id
                "task_name":"test_push", // 任务名称
                "channel_id":14,         // 通道id
                "channel_name":"发优惠券", // 通道名称
                "push_experiment_id": 129103, // 推送实验id
                "push_experiment_name": "老用户年终回馈实验" // 推送实验名称
            },
        "receipt_properties": {
            "push_id": "882f7c1c-af41-4351-9ec1-04a574597055",
            "user_unique_id" : "hulan@163.com",
            "push_config_id" : 231122, 
            "push_batch_id": 2311222343112, 
            "channel_id": 14, 
            "channel_type" : "webhook", 
            "ab_version": "231892", 
            "push_experiment_id": 129103, 
            "task_unit_id":12  
         },
        "params": {
            "action_type": "老用户反馈", 
            "bonus_type": "抵用金", 
            "count" : 200, 
            "content_message":"亲爱的${user_nickname} , 您好！\n\r老客户年终回馈，送您${count}元抵用券，给咱娃新年好课选起来咯！"
         },

        "user_info":{
            "user_unique_id": "hulan@163.com", 
            "device_id": "EA7583CDA6678BC", 
            "ssid": "12345678901", 
            "user_nickname":"汤圆麻麻" 
        },
         "trigger_time":{
          "trigger_type": 1, 
          "trigger_routine": "true", 
          "trigger_cycle": "daily", 
          "trigger_weekday": "MON", 
          "trigger_monthday": "1, 10, 20", 
          "trigger_timestamp": "9:00",
         "trigger_datewindow" :"2020/01/02-2021/05/20"
        },
         "trigger_events":{
        }
    },
    {},
]

```

--------------------------------

### Webhook Receipt Event Reporting (cURL Example)

Source: https://www.volcengine.com/docs/6287/71443

This cURL command provides an example of how to send receipt events to the Volcengine DataRangers webhook endpoint. It demonstrates the POST request with JSON payload and necessary headers.

```bash
curl -XPOST -H "Content-Type: application/json" 'https://{your_sdk_domain}/datapush/receipt/api/v2/webhook/{channel_id}' -d '
[
    {
        "message_id": "abc",
        "type": "show"
    },
    {
        "message_id": "abc",
        "type": "click"
    }
]
'
```

--------------------------------

### Webhook Response Body Example

Source: https://www.volcengine.com/docs/6287/71443

This example shows the JSON array returned by the webhook endpoint indicating the success or failure of individual push messages. It includes success status, failure reasons, and relevant properties for failed messages.

```json
[
    { "success": true },
    {
        "success": false,
        "fail_reason": ".....",
        "push_id": "882f7c1c-af41-4351-9ec1-04a574597055",
        "user_unique_id" : "hulan@163.com",
        "push_config_id" : 231122, 
        "push_batch_id": 2311222343112, 
        "channel_id": 123, 
        "channel_type" : "webhook",
        "ab_version": "231892", 
        "push_experiment_id": 129103
     },
     ...
]

```

--------------------------------

### 5. Initialize Basic Configuration

Source: https://www.volcengine.com/docs/6287/65802

Details on initializing the SDK with basic configurations, including event tracking, picker settings, A/B testing, logging, and encryption.

```APIDOC
## 5. Initialize Basic Configuration

The following are common initialization basic configurations. `config` refers to the `InitConfig` during initialization.

### 5.1 Full-Event Tracking Switch

The full-event tracking switch is enabled by default. To disable it:
```java
// Enable the upload of full-event tracking events
config.setAutoTrackEnabled(true); // true: enable full-event tracking, false: disable full-event tracking
```

After enabling full-event tracking, page events and click events are collected by default. To customize full-event tracking collection, use the following configuration:
```java
// Default value: AutoTrackEventType.PAGE | AutoTrackEventType.CLICK
// AutoTrackEventType.ALL: All full-event tracking events
// AutoTrackEventType.PAGE: Full-event tracking page events (enabled by default)
// AutoTrackEventType.CLICK: Full-event tracking click events (enabled by default)
// AutoTrackEventType.PAGE_LEAVE: Full-event tracking page duration events (disabled by default)
// This feature is only supported in version 6.11.0 and above
config.setAutoTrackEventType(AutoTrackEventType.PAGE | AutoTrackEventType.CLICK | AutoTrackEventType.PAGE_LEAVE);
```

By default, the page view event in full-event tracking targets Activity pages. To enable collection of `bav2b_page` events for Fragments, in addition to the above switch, you also need to enable the Fragment collection switch.
```java
// Enable Fragment full-event tracking event collection
config.setAutoTrackFragmentEnabled(true);
```

### 5.2 Enable Picker-Based Tracking

Picker-based tracking is disabled by default.
```java
// Enable picker-based tracking feature. Full-event tracking must be enabled first to use this feature.
config.setPicker(new Picker(this, config));

// If you want to enable picker-based tracking for H5 pages embedded in the App, please enable the JS injection switch:
config.setH5CollectEnable(true);
// Note: H5 picker only supports native WebView and Tencent X5
```

### 5.3 A/B Testing Switch

```java
config.setAbEnable(true); // Enable A/B testing
```

### 5.4 View Log Output

**Note**
⚠️ Please note that log output is disabled by default. It is recommended to disable it for production builds.
```java
// Output logs to the console, which can be used to observe the upload status of user behavior logs. It is recommended to disable it when going live.
config.setLogEnable(true); // true: enable logs, false: disable logs
config.setLogger(new ILogger() {
    @Override
    public void log(String s, Throwable throwable) {
        Log.d("AppLog------->: ", "" + s);
    }
});
```

### 5.5 User Switch Clear A/B Information Switch

The switch is enabled by default. When switching users, cached A/B information is cleared. If you do not want to clear the cache when switching users, set this to `false`.
```java
config.clearABCacheOnUserChange(true); // Default is true, clears A/B version information when switching users
```

### 5.6 Encryption Settings Switch

Encryption settings are enabled by default. You can disable encryption during the debug phase for easier packet capture and debugging.
```java
// Encryption settings switch. It is recommended to enable it for online versions.
AppLog.setEncryptAndCompress(true); // true: enable encryption, false: disable encryption
```

### 5.7 Set APP Language and APP Region

```java
// Modify language and region, set APP Language and APP Region
config.setLanguage(...); // Set APP Language, e.g., zh-cn 
config.setRegion(...); // Set APP Region, e.g., cn
```

### 5.8 Bridge Embedded H5 Pages

If your application integrates both Web/JS SDK and Android SDK, by default, all H5 events are reported through the Web/JS SDK. If you want H5 events to be reported by the Android SDK, you can use the embedded H5 bridging feature. In this case, events generated on H5 pages will be reported through the Android SDK and not through the Web/JS SDK. This takeover reporting scenario will reuse the `user_unique_id` and common properties set on the Android side.

**Note**
Bridging requires simultaneous integration and cooperation of Web/JS SDK and Android SDK. The current version of Web/JS SDK shields H5 full-event tracking after bridging because all reporting is taken over by the Android SDK. If you still want to collect H5 full-event tracking, please refer to section 5.9: Native Side Collects H5 Full-Event Tracking. For Web/JS SDK integration, please refer to "Web/JS SDK Integration" and section 3.6.
```java
// Note: This feature only supports native WebView and Tencent X5
// The following configurations need to be enabled simultaneously. If both are not met, H5 events will continue to be reported by the Web/JS SDK:
// Enable the embedded H5 bridging switch in the Android SDK
config.setH5BridgeEnable(true);
// Set H5 embedded in native and bridged in Web/JS SDK
window.collectEvent('init', {
    // ...... other initialization configurations
    enable_native: true 
    // For versions below 5.1.3, the parameter is Native. For versions above 5.1.3, it is changed to enable_native, but the Native setting is still compatible.
})

// Supplement: Since the Web/JS SDK shields H5 full-event tracking after bridging,
// the full-event tracking switch in the Web/JS SDK becomes invalid.
// If you want to maintain H5 full-event tracking, please refer to section 4.7
window.collectEvent('init', {
    // Full-event tracking switch 
    autotrack: invalid
})
```

### 5.8.1 Pre-work for Bridging Embedded H5 Pages Whitelist

After the switch is enabled, a whitelist must be configured. Only domains configured in the whitelist will take effect for bridging. Wildcards can be used in the whitelist, where `*` represents a wildcard.
```java
// Embedded H5 page whitelist configuration
// Example: To bridge www.volcengine.com and www.bytedance.com H5 pages
config.setH5BridgeAllowlist(Arrays.asList("www.volcengine.com", "*.bytedance.*"));
// Multiple whitelist configurations for www.bytedance.com are available, please configure according to your business scenario, e.g.:
// www.bytedance.com or .bytedance. or *.*.* can all achieve the same result.
```

If you want to bridge all H5 pages without filtering any domains or local resource files, you can configure to allow all H5 bridging. This configuration has higher priority than `setH5BridgeAllowlist`.
```java
// Allow all H5 bridging
// This feature is only supported in version 6.9.7 and above
config.setH5BridgeAllowAll(true);
```

### 5.8.2 Plugin Injection Considerations for Bridging Embedded H5 Pages

In general, H5 bridging requires you to integrate the plugin as described in section 1.2. At this time, the Bridge injection is completed by the Plugin instrumentation. However, in some scenarios, instrumentation may not take effect (scenarios where instrumentation fails may include obfuscation, overwriting, etc.). You can choose to manually disable the SDK's automatic Bridge injection capability and then manually call the injection. At this time, you can add the configuration `autoInjectWebViewBridge = false` in the full-event tracking Plugin and then manually inject the Bridge before loading the H5 in the WebView that needs to be bridged.

#### Gradle 8.0 Below: Disable Plugin Automatic Injection
```groovy
// Disable automatic injection through the plugin
teaExtension {
    // Disable automatic injection of the bridge for webview integration
    autoInjectWebViewBridge = false
    // Omit other plugin usages
}
```

#### Gradle 8.0 Above: Disable Plugin Automatic Injection
```groovy
// AGP8 version plugin 6.16.3 supports
appLog {
    // Disable automatic injection of the bridge for webview integration
    autoInjectWebViewBridge = false
    // Omit other plugin usages
}
```

#### Manual Bridge Injection
```java
// This feature is only supported in version 6.13.0 and above
WebView myWebView = findViewById(R.id.webview);
// Inject Bridge before loadUrl
AppLog.initWebViewBridge(myWebView, url);
myWebView.loadUrl(url);
```
```

--------------------------------

### Webhook Push Request Data Example (JSON)

Source: https://www.volcengine.com/docs/6287/71443

An example of the JSON payload sent by the Volcengine platform to a webhook service when a single push notification is triggered. This includes details about the push, task, parameters, user info, and trigger events.

```json
{
        "push_id":"882f7c1c-af41-4351-9ec1-04a574597055", // 用于识别单条消息的uuid
        "app_id": "234901", // 应用ID
        "sign": "f67ebb5fac42feacdee69d77bb0314cf", // 校验签名信息
        "timestamp":1606789775, // 发送时间戳  
        "task":{
                "task_id":100, // 推送任务id
                "task_name":"test_push", // 任务名称
                "channel_id":14,         // 通道id
                "channel_name":"发优惠券", // 通道名称
                "push_experiment_id": 129103, // 推送实验id
                "push_experiment_name": "老用户年终回馈实验" // 推送实验名称
            },
        "receipt_properties": {
            "push_id": "882f7c1c-af41-4351-9ec1-04a574597055",
            "user_unique_id" : "hulan@163.com",
            "push_config_id" : 231122, 
            "push_batch_id": 2311222343112, 
            "channel_id": 14, 
            "channel_type" : "webhook", 
            "ab_version": "231892", 
            "push_experiment_id": 129103, 
            "task_unit_id":12  
         },       
        "params": { 
            "action_type": "老用户反馈", 
            "bonus_type": "抵用金", 
            "count" : 200, 
            "content_message":"亲爱的${user_nickname} , 您好！\n\r老客户年终回馈，送您${count}元抵用券，给咱娃新年好课选起来咯！"
         },

        "user_info":{ 
            "user_unique_id": "hulan@163.com", 
            "device_id": "EA7583CDA6678BC", 
            "ssid": "12345678901", 
            "user_nickname":"汤圆麻麻" 
        },
         "trigger_time":{
          "trigger_type": 1, 
          "trigger_routine": "true", 
          "trigger_cycle": "daily", 
          "trigger_weekday": "MON", 
            "trigger_monthday": "1, 10, 20", 
          "trigger_timestamp": "9:00", 
         "trigger_datewindow" :"2020/01/02-2021/05/20"
        }
         "trigger_events":{
        }
}
```

--------------------------------

### Python SDK Examples for A/B Testing API

Source: https://www.volcengine.com/docs/6287/65829

This Python code demonstrates how to use the rangersdk to interact with the A/B Testing open API. It includes examples for making GET and POST requests, handling parameters, and processing JSON responses. Replace placeholders with your actual AK, SK, and request details.

```python
from rangersdk.client import RangersClient
import ujson

ak = '{使用AK替换}'
sk = '{使用SK替换}'
bc = RangersClient(ak, sk)

# Get Method Example
def test1():
    params = {'{key}': '{value}'} # 参数，选填
    resp = bc.data_tester('/openapi/v1/164287/details', params=params, method='get') # get请求方法
    print(resp.content) # 打印返回结果
    res = resp.json() # 将结果转化为json格式
    self.assertEqual(200, res['code'], 'err message: {}'.format(res['message']))

# Post Method Example
def test2():
    params = {'{key}': '{value}'} # 参数，选填
    resp = bc.data_tester('/openapi/v1/openapi-test', body=ujson.dumps(params), method='post') # post请求方法
    print(resp.content) # 打印返回结果
    res = resp.json() # 将结果转化为json格式
    self.assertEqual(200, res['code'], 'err message: {}'.format(res['message']))

```

--------------------------------

### Get Umeng DeviceToken for DataRangers (Android)

Source: https://www.volcengine.com/docs/6287/71443

This snippet shows how to obtain the DeviceToken from Umeng Push SDK on Android and pass it as a header to DataRangers SDK. It covers both SaaS and private deployment environments.

```java
String resistrationId = mPushAgent.getRegistrationId();
HashMap<String,Object> param = new HashMap<String,Object> ();
param.put ("upush_device_token",resistrationId);
//saas环境
AppLog.setHeaderInfo (param);
//私有化部署
AppLog.profileSet(params);
```

--------------------------------

### 私有化环境 SDK 初始化与启动

Source: https://www.volcengine.com/docs/6287/1408234

配置 HarmonyOS Next SDK 以适应私有化部署环境。初始化时使用 AppLog 和 InitConfig，设置 App ID、Channel 和日志启用，并通过 setUri() 指定私有化数据上送地址。SDK 初始化后需要手动调用 start()。

```JavaScript
import { AppLog, InitConfig, createByDomain } from '@volcengine/applog'

let appLog = new AppLog()
let config = new InitConfig('yourAppId', 'yourChannel')
  .setLogEnabled(true)
  .setUri(createByDomain("your_REPORT_URL"))

appLog.init(context, config)
appLog.start()
```

--------------------------------

### Frequently Asked Questions (FAQ)

Source: https://www.volcengine.com/docs/6287/65805

Answers to common questions regarding decision_id, track_id, metaHost, trackHost, data storage, anonymous user bucketing, and more.

```APIDOC
### Frequently Asked Questions (FAQ)

#### How to pass values for `decision_id` and `track_id` in the Server-side SDK?

*   `decision_id`: Local bucketing user identifier, not used for event reporting.
*   `track_id`/`trackId`: Event reporting user identifier, used for event reporting, requires replacement with the customer's actual user identifier.

**Usage Scenarios:**

*   If all server-side experiments involve real-name users, it's recommended to use the user ID (UUID) for both `decision_id` and `track_id`.
*   If anonymous users are involved:
    *   If client-side SDK is integrated, it's recommended to use the `ssid` passed from the client-side SDK for `decision_id`.
    *   If only integrating the server-side SDK for server-side experiments, `decision_id` can be any unique identifier for the user, such as the device ID.

#### How to get `metaHost` and `trackHost` for the Server-side SDK, and what do they represent?

*   `metaHost`: Address for retrieving meta information. For private deployments, it's usually the system access address.
*   `trackHost`: Data reporting address. These addresses are typically provided in the deployment documentation after system setup. If you cannot find the corresponding documentation, please contact your project manager or Volcengine support personnel.

#### Does the server-side meta information get stored? If so, where?

Yes, it is stored in the memory of the server-side SDK process. The meta information (including experiment name, version allocation, target audience, whitelist, etc.) is usually small and does not consume much memory. Server-side meta information is updated every 60 seconds by default. This interval can be modified using the `setInterval` API.

#### Does the Server-side SDK support bucketing for anonymous users?

Only the Java, Go, and PHP versions of the SDK support bucketing for anonymous users. Ensure you strictly follow the anonymous user integration instructions in the help documentation. For example, in Java, when dealing with anonymous users, `trackId` should be an empty string, and you need to set the device ID using `setDeviceId` or `setWebId`.

**Note:** SaaS environments require using the `setBdDid` method.

```java
User user = new User.UserBuilder().create("decisionID", "") // Set trackId if it has a value, otherwise set an empty string
        .setDeviceId(6981329701821561868L) // Use this method for private deployment
        .setWebId(7018215618686981329L) // Use this method for Web and Mini Programs
        .setBdDid("H5PPHTEHXLMEGCGROGLY2343242M2LMFIMQE4VIRW76HKYWH4Q01") // Use this method for SaaS
        .build();
```

#### What data does the `getAllExperimentConfigs` method of the Server-side SDK fetch (meta information)?

The Server-side SDK periodically fetches the latest experiment meta information from the remote Meta service, including experiment name, version allocation, target audience, whitelist, etc.

#### What is the frequency of the Server-side SDK requesting the meta service to pull metadata?

It makes one request during initialization, and then one request every 10 seconds. This interval can be customized via the `setInterval` API, as shown in the integration documentation.

#### What is the bucketing method of the Server-side SDK?

The Server-side SDK uses local bucketing. It fetches the latest experiment meta information (including experiment name, version allocation, target audience, whitelist, etc.) from the Meta service during initialization and then performs real-time bucketing when the `activate` interface is called.

### How to implement the logic for entering and exiting groups using Redis?

Refer to the integration example below for implementing entry/exit logic with Redis.

```java
/**
 * Implements the UserAbInfoHandler interface and its declared methods
 * for entry/exit logic using Redis.
 */
public class NotOutGroupRedisImplement implements UserAbInfoHandler {
    /**
     * Stores bucketing results.
     * @param userId   User ID
     * @param hashMap Bucketing result parameters and values, key=bucketing parameter, value=identifier for the corresponding version.
     * @return boolean
     */
    @Override
    public boolean createOrUpdate(String userId, HashMap<String, String> hashMap) {
        long res = 0;
        for (Map.Entry<String, String> entry : hashMap.entrySet()) {
             res = JedisUtils.getJedis().hset(userId, entry.getKey(), entry.getValue());
        }
        return res > 0;
    }

    /**
     * Queries the user's corresponding group.
     * @param userId
     * @return HashMap of user's bucketing results: key=bucketing parameter, value=identifier for the corresponding version.
     */
    @Override
    public HashMap<String, String> query(String userId) {
        Map<String, String> resMap = JedisUtils.getJedis().hgetAll(userId);
        return (HashMap<String, String>) resMap;
    }
}
```

Configure the entry/exit logic in the AB bucketing code:

```java
AbClient abClient = new AbClient.Builder(appKey).build();
NotOutGroupRedisImplement notOutGroupHandler = new NotOutGroupRedisImplement();
abClient.setUserAbInfoHandler(notOutGroupHandler);
Variable variable1 = abClient.activate("variantKey", user, defaultValue);
```

#### What is the impact of network interruptions to `metahost` and `trackhost`?

*   `metahost` network interruption prevents synchronization of the latest experiment information. Previously fetched information is unaffected (local cache exists).
*   `trackHost` network interruption causes event reporting failures. Bucketing may succeed, but the server will not receive group entry information.

#### Can IP addresses be configured if `metahost` and `trackhost` do not have domain names?

Yes, IP addresses can be configured. When configuring IP addresses, the port number must also be included.

#### When uploading event data through the server-side API, if data from the 13th is sent on the 14th, the group entry data is visible that day, but the metric data is not. Why?

When uploading data through the server-side API, if `local_time_ms` is for the current day, the data is real-time. If it is for a previous day, the data is non-real-time and only writes metadata, not real-time tables. You need to check the next day.

#### The `getAllExperimentConfigs` method in the Server-side SDK is described as "Gets the variant details of all experiments the user has hit." Does "all experiments" include both client-side and server-side experiments?

The APIs provided by the server-side SDK can only process server-side experiments and retrieve related data.

#### How to test if the data reporting address/`trackhost` is accessible?

Use `curl` commands for verification. You need to modify the `appkey` (to your application's appkey), the domain name `xxx.com`, and `local_time_ms` to the current timestamp.

*   SaaS: `https://mcs.ctobsnssdk.com`
*   Private Deployment: Use the domain name of the data reporting address.

**Note:** When configuring `trackHost`, only provide the domain name. Do not end the domain name with `/`.

```bash
curl -X POST -H "Content-Type: application/json" -H "X-MCS-AppKey: appkey" -d '{"user":{"user_unique_id":"XXX"},"header":{"app_name":"xxx"},"events":[{"event":"predefine_pageview","params":"{}","local_time_ms":1630467848888}]}' https://XXXX.com/v2/event/json
```
```

--------------------------------

### Get Umeng DeviceToken for DataRangers (iOS)

Source: https://www.volcengine.com/docs/6287/71443

This snippet demonstrates how to retrieve the DeviceToken from Umeng Push SDK on iOS and configure it for DataRangers SDK. It includes logic for iOS 13 and above, and setting custom headers for both SaaS and private deployment.

```objectivec
#include <arpa/inet.h>
- (void)application:(UIApplication *)application didRegisterForRemoteNotificationsWithDeviceToken:(NSData *)deviceToken
{
    if (![deviceToken isKindOfClass:[NSData class]]) return;
    const unsigned *tokenBytes = (const unsigned *)[deviceToken bytes];
    NSString *hexToken = [NSString stringWithFormat:@"%08x%08x%08x%08x%08x%08x%08x%08x",
                          ntohl(tokenBytes[0]), ntohl(tokenBytes[1]), ntohl(tokenBytes[2]),
                          ntohl(tokenBytes[3]), ntohl(tokenBytes[4]), ntohl(tokenBytes[5]),
                          ntohl(tokenBytes[6]), ntohl(tokenBytes[7])];
    NSLog(@"deviceToken:%@",hexToken);
}
//saas环境
[BDAutoTrack setCustomHeaderBlock:^NSDictionary<NSString *,id> * _Nonnull{
        return @{@"upush_device_token":hexToken};
}];
//私有化部署
[BDAutoTrack profileSet:^NSDictionary<NSString *,id> * _Nonnull{
        return @{@"upush_device_token":hexToken};
}];
```

--------------------------------

### SaaS-非云原生 SDK 初始化与启动

Source: https://www.volcengine.com/docs/6287/1408234

集成 HarmonyOS Next SDK 到 SaaS-非云原生环境。SDK 通过 AppLog 和 InitConfig 初始化，支持设置 App ID、Channel 和日志启用。初始化后需手动调用 start() 方法启动数据采集。

```TypeScript
import { AppLog, InitConfig } from '@volcengine/applog'

let appLog = new AppLog()
let config = new InitConfig('yourAppId', 'yourChannel')
  .setLogEnabled(true)

appLog.init(context, config)
appLog.start()
```

--------------------------------

### Track Component Exposure with JavaScript

Source: https://www.volcengine.com/docs/6287/1408236

This code demonstrates how to track component exposure events using the appLog SDK. It involves marking a component with an ID, starting observation, and optionally disposing of the observer. The 'exposureCallback' can be used to add custom properties or filter events.

```javascript
//  需要给组件添加 ID
Button('Test', { stateEffect: true, type: ButtonType.Capsule })
  .id("test_btn"))

//  请在页面 build 后开始监控，如果传入过早，无法通过 NodeId 找到对应组件，就无法处理曝光
//  开始曝光监控，曝光时给 SDK 传给 ID，来确定需要监控哪个组件的曝光
appLog.exposure()?.observeByNodeId("test_btn", this.getUIContext(), {
    //  Optional，默认为 "$bav2b_exposure"
    eventName: "custom_exposure",
    //  Optional，自定义曝光事件属性
    properties: {
        "test": "test"
    },
    //  Optional, 曝光配置，有效曝光面积和曝光事件回调
    //  有效曝光面积: 有效曝光的 View 显示面积的比例，默认是 0，取值 0-1
    //  曝光事件回调: 触发曝光事件时回调，会携带当前曝光事件相关属性，可以进行自定义追加以及支持过滤
    config: {
        exposureCallback: (param) => {
          //  支持在回调中添加自定义属性
          //  true 保留曝光事件
          return true
        }
    }
})

//  在适当实际停止曝光监控，如离开页面或者其他逻辑
appLog.exposure()?.disposeByNodeId("test_btn")
```

--------------------------------

### Collect Duration Events in Java

Source: https://www.volcengine.com/docs/6287/1271145

This section explains how to collect events with duration attributes, such as video playback time. It provides Java code examples for starting, pausing, resuming, and stopping duration events using the `AppLog` class. Custom event names are supported for `stopDurationEvent` from version 6.16.10.

```java
// 在视频开始播放时调用
AppLog.startDurationEvent("play");

// 在视频暂停播放时调用
AppLog.pauseDurationEvent("play");

// 在视频继续播放时调用
AppLog.resumeDurationEvent("play");

// 在结束播放时调用，此时会上报一个 play 事件，且带有 $event_duration 属性（记录了播放时长，单位毫秒） 
JsonObject params = new JsonObject();
// 可以在 params 填入自己的自定义参数
AppLog.stopDurationEvent("play", params);
// 6.16.10 后自持自定义时长事件事件名，默认情况事件名为方法第一个参数
AppLog.stopDurationEvent("play", params, "custom_play");
```

--------------------------------

### Initialization Configuration Parameters

Source: https://www.volcengine.com/docs/6287/96071

This section details the various parameters available for initializing the Web/JS SDK, covering aspects like data reporting, A/B testing, and user identification.

```APIDOC
## Initialization Configuration Parameters

### Description
This endpoint provides a detailed list of configuration parameters for initializing the Web/JS SDK. These parameters control various aspects of data collection, reporting, and experimental features.

### Method
Not Applicable (Configuration Table)

### Endpoint
Not Applicable (Configuration Table)

### Parameters
#### Initialization Parameters
- **app_id** (number) - Required - Unique identifier for the business product. Must be pre-applied.
- **channel** (string) - Optional - Reporting channel identifier. Possible values: 'cn', 'sg', 'va'. Mutually exclusive with `channel_domain`.
- **channel_domain** (string) - Optional - Custom reporting domain and path prefix. If set, the `channel` parameter is ignored. Must be a valid domain.
- **ab_channel_domain** (string) - Optional - Private domain for A/B testing. Used by A/B experiments to fetch configurations.
- **log** (boolean) - Optional - If true, detailed log information will be printed to the console. Defaults to `false`. Recommended to enable.
- **enable_ab_test** (boolean) - Required - Enables or disables the A/B testing functionality. Defaults to `false`.
- **enable_ab_visual** (boolean) - Optional - If true, enables visual A/B experiment logic when visual configurations are available. Requires `enable_ab_test` to be true. Defaults to `false`.
- **enable_multilink** (boolean) - Optional - Enables multi-link A/B experiments. Requires `enable_ab_test` to be true. Defaults to `false`.
- **multilink_timeout_ms** (number) - Optional - Default timeout in milliseconds for closing the overlay in multi-link experiments. Defaults to 500ms. Requires `enable_ab_test` and `enable_multilink` to be true.
- **disable_auto_pv** (boolean) - Optional - If true, disables the SDK from automatically reporting default page views. Defaults to `false`.
- **disable_sdk_monitor** (boolean) - Optional - If true, disables the SDK's self-monitoring event reporting upon startup. Does not disable error log reporting. Defaults to `false`.
- **autotrack** (boolean) - Optional - Enables automatic event tracking (no manual coding required). Refer to the full tracking documentation for details. Defaults to `false`.
- **enable_stay_duration** (boolean) - Optional - Enables tracking of user stay duration on the page. Refer to the stay duration documentation for details. Defaults to `false`.
- **cross_subdomain** (boolean) - Optional - If true, automatically recognizes users across subdomains (e.g., `a.yourdomain.com` and `b.yourdomain.com` will be treated as the same user). Defaults to `false`.
- **cookie_domain** (string) - Optional - Domain for storing cookies containing tokens or UTM parameters when `cross_subdomain` is enabled. If not provided, cookies are stored under the current domain.
- **cookie_expire** (number) - Optional - Cookie expiration time in milliseconds. Defaults to 7 days.
- **max_report** (number) - Optional - Maximum number of custom tracked events to merge in a single report. Defaults to 20.
- **reportTime** (number) - Optional - Interval in milliseconds for the asynchronous event reporting queue. Defaults to 30ms.
- **configPersist** (number) - Optional - Determines whether configuration parameters should be persisted. `0`: No persistence; `1`: Session-level storage; `2`: LocalStorage persistence. Defaults to `0`.
- **timeout** (number) - Optional - Timeout in milliseconds for event reporting. Defaults to 100000ms.
- **enable_native** (boolean) - Optional - If true, all tracking and configuration will be handled by the native SDK (for versions below 5.1.3, this was `Native`). Defaults to `false`.
- **spa** (boolean) - Optional - Enables monitoring of route changes in Single Page Applications (SPAs). When enabled, PV and stay duration events are re-reported upon route changes. Defaults to `false`.
- **disable_track_event** (boolean) - Optional - If true, all custom tracked events will not be reported. Defaults to `false`.
- **enable_custom_webid** (boolean) - Optional - If true, enables the use of a custom web ID. Defaults to `false`.
- **allow_hash** (boolean) - Optional - Supports listening to hash-based routing. Available in version 5.1.3 and above. Defaults to `false`.
- **ab_cross** (boolean) - Optional - Enables cross-domain storage for A/B test data. Available in version 5.1.3 and above. Defaults to `false`.
- **ab_cookie_domain** (string) - Optional - The domain used for storing A/B test data when `ab_cross` is enabled. Available in version 5.1.3 and above.
- **disable_ab_reset** (boolean) - Optional - Prevents A/B reset when switching UUIDs. Switching UUIDs normally clears A/B caches and re-sends A/B bucketing requests, which might lead to VID non-reporting (UUID mismatch). Available in version 5.1.4 and above. Defaults to `false`.
```

--------------------------------

### Full Automation (Beta)

Source: https://www.volcengine.com/docs/6287/1408236

This section describes the full automation feature (beta), available from version 1.2.0. It focuses on leveraging the Navigation component for route tracking, as HarmonyOS Next discourages direct page route monitoring. It also notes limitations regarding Tab switching.

```APIDOC
## Full Automation (Beta)

### Description
Enables automatic event collection, primarily focusing on Navigation component navigation since HarmonyOS Next recommends against direct page route monitoring. This feature is in beta and available from version 1.2.0.

### Method
`appLog.bav()?.enablePlugin(uiContext, options)`

### Parameters
#### `enablePlugin` Parameters
- **uiContext** (object) - The UI context obtained from the window stage.
- **options** (object) - Configuration options for enabling plugins:
  - **enablePageLeaveEvent** (boolean, Optional) - Set to `true` to collect page leave events.
  - **enableClickEvent** (boolean, Optional) - Set to `true` to collect component click events.

### Request Example
```javascript
onWindowStageCreate(windowStage: window.WindowStage): void {
  // Recommended to enable at EntryAbility loadContent for comprehensive lifecycle tracking
  windowStage.loadContent('pages/Index', (err, data) => {
    appLog.bav()?.enablePlugin(windowStage.getMainWindowSync().getUIContext(), {
      enablePageLeaveEvent: true,
      enableClickEvent: true
    })
  })
}
```

### Limitations
- Does not currently support Tab switching monitoring due to limitations in system Tab monitoring.
- Only Tab changes can be detected, not Tab entry or exit.
```
Tokens:
Raw