webaiplanet

<p>LangRouter for TranslatePress adds routing and fallback control to TranslatePress automatic translation.</p>
<p>Instead of sending every request to a single provider, it lets you choose which translation engine should handle a request based on the current content type, target language, and fallback policy. This gives you more control over translation flow on multilingual WordPress sites.</p>
<p>This plugin is an independent extension for TranslatePress. It is not affiliated with, endorsed by, or maintained by TranslatePress.</p>
<p>It is intended for site owners, developers, agencies, and content teams that need more control than a single-engine setup can provide.</p>
<h4>Key features</h4>
<ul>
<li>Set one default engine as the baseline route.</li>
<li>Route singular content types such as posts, pages, products, and custom post types to a preferred engine.</li>
<li>Assign target languages to different engines.</li>
<li>Define fallback behavior when the current engine cannot continue.</li>
<li>Configure each engine separately in a dedicated settings page.</li>
<li>Query language support from the admin area before changing live routing rules.</li>
<li>Inspect routing behavior through built-in runtime logs.</li>
<li>Support DeepL multi-key configuration and Volcengine Ark usage tools.</li>
<li>Connect OpenAI-compatible third-party services and custom compatible endpoints.</li>
</ul>
<h4>Supported translation engines</h4>
<p>LangRouter currently supports:</p>
<ul>
<li>Volcengine Ark</li>
<li>Qwen</li>
<li>Hunyuan</li>
<li>OpenAI</li>
<li>DeepL</li>
<li>Compatible OpenAI API</li>
</ul>
<h4>How routing works</h4>
<p>LangRouter uses the following priority order:</p>
<ol>
<li>Post type route</li>
<li>Language assignment</li>
<li>Default engine</li>
</ol>
<p>If a singular content request matches a post type rule, that engine becomes the primary route.<br />
If no post type rule matches, LangRouter checks the language assignment rules.<br />
If no language rule matches, the plugin uses the default engine.</p>
<p>If the current engine cannot continue, LangRouter applies the configured fallback behavior.<br />
Depending on your settings, the request can stop immediately, jump directly to the default engine, or continue through the global fallback chain.</p>
<h4>Post type routing</h4>
<p>Post type routing is useful when different content types need different translation behavior.</p>
<p>For example, you can:</p>
<ul>
<li>route products to one engine for terminology consistency;</li>
<li>route editorial content to another engine for tone and fluency;</li>
<li>keep specific custom post types on a dedicated provider.</li>
</ul>
<p>Post type routing applies only to singular content contexts, such as a single post, page, product, or custom post type entry.<br />
Archive pages, taxonomy pages, search results, and other non-singular views do not use post type routing.</p>
<h4>Language assignment and fallback rules</h4>
<p>Language assignment selects the primary engine for a target language.<br />
Fallback rules define what to try next only after the current primary engine cannot continue.</p>
<p>This separation helps you build a clearer routing strategy:</p>
<ul>
<li>choose the preferred engine for a target language;</li>
<li>define a different engine as the fallback;</li>
<li>keep the default engine as the final baseline or safety net.</li>
</ul>
<h4>Dedicated engine settings</h4>
<p>Routing rules and engine credentials are managed separately.<br />
This makes the plugin easier to operate and maintain.</p>
<p>In the engine settings area, you can configure items such as:</p>
<ul>
<li>API keys and secrets</li>
<li>model names</li>
<li>base URLs and compatible endpoints</li>
<li>request timeouts</li>
<li>additional request JSON</li>
<li>per-engine notes and operational options</li>
<li>multi-key or account-related settings for supported engines</li>
</ul>
<h4>Logging and troubleshooting</h4>
<p>LangRouter includes a built-in runtime log viewer.<br />
You can enable file logging, view recent log content inside the admin area, and download log files for debugging.</p>
<p>This is useful when you want to confirm:</p>
<ul>
<li>which engine was selected;</li>
<li>whether a post type route was matched;</li>
<li>whether language assignment handled the request;</li>
<li>why a fallback decision was triggered;</li>
<li>whether a provider entered the execution chain successfully.</li>
</ul>
<h4>Typical use cases</h4>
<ul>
<li>Use one engine for most traffic and reroute selected languages.</li>
<li>Route products, guides, and regular posts to different engines.</li>
<li>Keep important content on stricter fallback behavior.</li>
<li>Use DeepL, Volcengine Ark, and OpenAI-compatible services in one workflow.</li>
<li>Compare providers while keeping routing and logging in one place.</li>
</ul>
<h3>External services</h3>
<p>This plugin can connect to third-party translation and AI services. These services are optional and are used only for translation, supported-language checks, usage checks, or diagnostic actions requested by an administrator.</p>
<p>No request is sent to a third-party service unless an administrator configures that engine and triggers translation or a related engine-specific action, such as a language support query, usage query, or diagnostic request.</p>
<p>The plugin may connect to the following services:</p>
<ul>
<li>
<p>OpenAI API. This service is provided by OpenAI and is used to translate content through OpenAI models.<br />
Data sent: text selected for translation, source and target language instructions, model settings, request parameters, and related request metadata required to complete the request.<br />
Sent only when this engine is configured and used by an administrator.<br />
Terms of Service: https://openai.com/policies/services-agreement/<br />
Privacy Policy: https://openai.com/policies/row-privacy-policy/</p>
</li>
<li>
<p>DeepL API. This service is provided by DeepL and is used to translate content through DeepL and, when needed, query supported languages.<br />
Data sent: text selected for translation, source language, target language, translation parameters, or request parameters required for supported-language queries.<br />
Sent only when this engine is configured and used by an administrator.<br />
Terms and Conditions: https://www.deepl.com/en/pro-license<br />
Privacy Policy: https://www.deepl.com/en/privacy.html</p>
</li>
<li>
<p>Tencent Cloud Hunyuan (China). This service is provided by Tencent Cloud and is used to translate content through Tencent Cloud Hunyuan endpoints configured for the China service.<br />
Data sent: text selected for translation, source language, target language, model settings, request parameters, and related metadata required to complete the request.<br />
Sent only when this engine is configured for Tencent Cloud China and used by an administrator.<br />
Terms of Service: https://cloud.tencent.com/document/product/301/97822<br />
Privacy Policy: https://cloud.tencent.com/document/product/301/11470</p>
</li>
<li>
<p>Tencent Cloud Hunyuan (International). This service is provided by Tencent Cloud International and is used to translate content through Tencent Cloud Hunyuan international endpoints.<br />
Data sent: text selected for translation, source language, target language, model settings, request parameters, and related metadata required to complete the request.<br />
Sent only when this engine is configured for Tencent Cloud international endpoints and used by an administrator.<br />
Terms of Service: https://www.tencentcloud.com/document/product/1284/75295<br />
Privacy Policy: https://www.tencentcloud.com/document/product/1284/75293</p>
</li>
<li>
<p>SiliconFlow API. This service is provided by SiliconFlow and is used when Hunyuan-compatible translation is configured through the SiliconFlow compatible endpoint.<br />
Data sent: text selected for translation, source and target language instructions, model settings, request parameters, and related metadata required to complete the request.<br />
Sent only when this compatible endpoint is configured and used by an administrator.<br />
Terms of Service: https://docs.siliconflow.cn/cn/legals/terms-of-service<br />
Privacy Policy: https://docs.siliconflow.cn/cn/legals/privacy-policy</p>
</li>
<li>
<p>Volcengine Ark. This service is provided by Volcengine and is used to translate content through Volcengine models and, when requested by the administrator, perform usage-related queries.<br />
Data sent: text selected for translation, model settings, request parameters, and, for usage-related requests, the account identifiers, date ranges, and signed request data required by the service.<br />
Sent only when this engine is configured and used by an administrator, or when the administrator manually triggers a usage-related request.<br />
Terms of Service: https://www.volcengine.com/docs/6456/70590<br />
Privacy Policy: https://www.volcengine.com/docs/6256/64902</p>
</li>
<li>
<p>Alibaba Cloud DashScope / Qwen (China). This service is provided by Alibaba Cloud and is used to translate content through Qwen-compatible endpoints configured for Alibaba Cloud China.<br />
Data sent: text selected for translation, source language, target language, model settings, request parameters, and related metadata required to complete the request.<br />
Sent only when this engine is configured for Alibaba Cloud China and used by an administrator.<br />
Terms of Service: https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20230728213935489/20230728213935489.html<br />
Privacy Policy: https://terms.aliyun.com/legal-agreement/terms/suit_bu1_ali_cloud/suit_bu1_ali_cloud202107091605_49213.html</p>
</li>
<li>
<p>Alibaba Cloud DashScope / Qwen (International). This service is provided by Alibaba Cloud International and is used to translate content through international Qwen-compatible endpoints configured by the administrator.<br />
Data sent: text selected for translation, source language, target language, model settings, request parameters, and related metadata required to complete the request.<br />
Sent only when this engine is configured for international Alibaba Cloud endpoints and used by an administrator.<br />
Terms of Service: https://www.alibabacloud.com/help/en/legal/latest/alibaba-cloud-international-website-terms-of-use-alibaba-cloud-international-website-terms-of-use<br />
Privacy Policy: https://www.alibabacloud.com/help/en/legal/latest/alibaba-cloud-international-website-privacy-policy</p>
</li>
<li>
<p>Compatible OpenAI API endpoints. These services are provided by the third-party provider selected by the administrator and are used when you configure a third-party service that implements an OpenAI-compatible API.<br />
Data sent: text selected for translation, source and target language instructions, configured model name, request headers, request parameters, and related request metadata required to complete the request.<br />
Sent only when a compatible third-party service is configured and used by an administrator.<br />
The provider, terms, privacy policy, and data handling depend on the service you choose. Review that provider before use.</p>
</li>
</ul>
<p>This plugin does not require any external service unless you choose to configure and use a supported engine.</p>


LangRouter for TranslatePress

Add routing, fallback control, and engine-specific settings to TranslatePress automatic translation.

Yard | Digital Agency

Mike van den Hoek

<p>This plugin registers secure API endpoints that allow you to request translations directly from DeepL without exposing your Deepl API-key. These endpoints are only accessible when a valid nonce is provided. When providing translations to website visitors, you can configure which languages are supported for translation.</p>
<h3>Caching Mechanism</h3>
<p>Each object that is translated will store its cached translation in the <code>wp_postmeta</code> table within the database. This caching mechanism ensures that translations are efficiently reused, reducing unnecessary API requests to DeepL and saving costs.</p>
<ul>
<li>Serving Cached Translations: If a cached translation is newer than the <code>post_modified</code> date of the object, the cached version is served.</li>
<li>Fetching New Translations: When the <code>post_modified</code> date of the object is more recent than the cached translation, a new translation is fetched from DeepL. Once retrieved, this translation is immediately cached for future use.</li>
<li>Cache Authorization: Only logged-in users with the <code>edit_posts</code> capability (or a custom capability configured via the <code>yard::deepl/cache_capability</code> filter) are permitted to create new cache entries. Requests from users without this capability will still return a translation from DeepL, but the result will not be stored in the cache.</li>
</ul>
<p>This approach minimizes the number of API calls to DeepL, ensuring translations are kept up to date only when necessary.</p>
<h4>Admin: Cache Metabox</h4>
<p>A metabox labeled Yard DeepL is displayed on the edit screen of supported post types (default: page). It provides two options:</p>
<ul>
<li>Disable translation cache: When checked, the cache is bypassed for this object. Useful for posts with dynamic content that should always be translated fresh.</li>
<li>Clear translation cache: When checked and the post is saved, all cached translations for this object are deleted.</li>
</ul>
<h4>Admin: Translation Cache Column</h4>
<p>A Translation cache column is added to the post list screen of all supported post types. Its purpose is to notify editors which posts should be cached as soon as possible to avoid unnecessary DeepL API costs.</p>
<ul>
<li>Green badge per language (e.g. NL, EN-US): a valid, up-to-date cached translation exists for that language. Visitor requests are served from cache at no cost.</li>
<li>Grey badge with a count per language (e.g. EN-US 42): no cache exists for that language and visitors have triggered that many live DeepL API calls. Hovering the badge shows the full count as a tooltip. These posts are the most urgent to cache.</li>
<li>—: the post has never been requested for translation, or caching is disabled for this post.</li>
</ul>
<p>The uncached call count is only incremented for visitor requests (users without cache-write capability) and is automatically reset per language once a cache entry is stored. Posts with the Disable translation cache option enabled are excluded from the column entirely.</p>
<h3>External Services</h3>
<p>This plugin connects to the DeepL API to provide translations for content.</p>
<ul>
<li><strong>Service:</strong> DeepL API (https://www.deepl.com)</li>
<li><strong>Purpose:</strong> To translate text from one language to another based on the provided target language.</li>
<li><strong>Data Sent:</strong> Text content for translation, the target language code, and the DeepL API key (handled securely and never exposed to users).</li>
<li><strong>Conditions:</strong> Data is sent when a request for translation is initiated.</li>
<li><strong>Privacy Policy:</strong> <a href="https://www.deepl.com/privacy" rel="nofollow ugc">DeepL Privacy Policy</a></li>
<li><strong>Terms of Service:</strong> <a href="https://www.deepl.com/pro-license" rel="nofollow ugc">DeepL Terms of Service</a></li>
</ul>
<h3>Usage</h3>
<h4>Security</h4>
<p>The API endpoints registered by this plugin are secured using a WordPress nonce. The nonce is passed to the front-end using the <code>wp_localize_script</code> function and is stored in a global JavaScript object <code>ydpl</code> which contains the following properties:</p>
<ul>
<li><code>ydpl_translate_post_id</code>: The ID of the post to be translated.</li>
<li><code>ydpl_rest_translate_url</code>: The URL of the API endpoint for translation requests.</li>
<li><code>ydpl_supported_languages</code>: The list of languages supported for translation.</li>
<li><code>ydpl_api_request_nonce</code>: The nonce used for API validation.</li>
</ul>
<p>When making requests to the API, ensure that the nonce is included in the request headers. The header should be named <code>X-WP-Nonce</code>, and it should contain a nonce created with the <code>wp_rest</code> action (available via <code>ydpl.ydpl_api_request_nonce</code> on the front-end).</p>
<p>Rate limiting: Unauthenticated requests (or requests from users without the cache capability) are limited to 3 requests per 60 seconds per IP address. Authenticated users with the required cache capability (default: edit_posts) are exempt from this rate limit.</p>
<p>Cache capability: Only users with the <code>edit_posts</code> capability can trigger cache creation. This can be customized using the <code>yard::deepl/cache_capability</code> filter.</p>
<h4>Example</h4>
<p>Request:</p>
<pre><code>var xhr = new XMLHttpRequest();
xhr.open('POST', ydpl.ydpl_rest_translate_url, true);

// Set request headers
xhr.setRequestHeader('Content-Type', 'application/json');
xhr.setRequestHeader('X-WP-Nonce', ydpl.ydpl_api_request_nonce);

// Handle response
xhr.onreadystatechange = function () {
    if (xhr.readyState === 4 && xhr.status === 200) {
        console.log('Translation:', JSON.parse(xhr.responseText));
    } else if (xhr.readyState === 4) {
        console.error('Error:', xhr.statusText);
    }
};

// Prepare and send the request body
var data = JSON.stringify({
    text: ["Look another test"],
    target_lang: "DE"
});

xhr.send(data);
</code></pre>
<p>Response:</p>
<pre><code>[
    {
        "text": "Look another test!",
        "translation": "Sehen Sie sich einen weiteren Test an!"
    }
]<h3>Filters</h3>
</code></pre>
<p>* <code>yard::deepl/cache_capability</code> (default: <code>edit_posts</code>) — The WordPress capability required to create cache entries. Users without this capability receive translations but results are not cached.<br />
* <code>yard::deepl/cache_metabox_post_types</code> (default: <code>['page']</code>) — Post types on which the Yard DeepL cache metabox is displayed.<br />
* <code>yard::deepl/disable_cache_metabox_post_types</code> — Deprecated. Use <code>yard::deepl/cache_metabox_post_types</code> instead.</p>


Yard DeepL

This plugin registers secure API endpoints that allow you to request translations directly from DeepL without exposing your DeepL API-key.

LangRouter for TranslatePress vs Yard DeepL