From 0ca5a41298796ba6a8ec56c15f87747f218472a5 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20B=C3=B6hmer?= <mail@jan-boehmer.de>
Date: Sun, 26 Apr 2026 22:11:27 +0200
Subject: [PATCH] Added option for translating AI extracted output

---
 .../Providers/AIInfoExtractor.php             | 11 ++++++++-
 .../AIExtractorSettings.php                   |  8 +++++++
 translations/messages.en.xlf                  | 24 +++++++++++++++++++
 3 files changed, 42 insertions(+), 1 deletion(-)

diff --git a/src/Services/InfoProviderSystem/Providers/AIInfoExtractor.php b/src/Services/InfoProviderSystem/Providers/AIInfoExtractor.php
index 92ed4e19..2032f634 100644
--- a/src/Services/InfoProviderSystem/Providers/AIInfoExtractor.php
+++ b/src/Services/InfoProviderSystem/Providers/AIInfoExtractor.php
@@ -35,6 +35,7 @@ use League\HTMLToMarkdown\HtmlConverter;
 use Symfony\AI\Platform\Message\Message;
 use Symfony\AI\Platform\Message\MessageBag;
 use Symfony\Component\DomCrawler\Crawler;
+use Symfony\Component\Intl\Languages;
 use Symfony\Contracts\HttpClient\HttpClientInterface;
 
 use function Symfony\Component\String\u;
@@ -210,7 +211,7 @@ final class AIInfoExtractor implements InfoProviderInterface
 
     private function buildSystemPrompt(): string
     {
-        return <<<'PROMPT'
+        $tmp = <<<'PROMPT'
 You are an expert at extracting electronic component information from web pages. Extract structured data in JSON format, from markdown extracted from a product page.
 Focus on the main content of the page, such as product descriptions, specifications, and tables. Ignore navigation menus, footers, and sidebars.
 
@@ -224,6 +225,14 @@ Rules:
 
 For parameters, combine name, value, and unit. The unit should be separate if possible.
 PROMPT;
+
+        if ($this->settings->outputLanguage === null) {
+            $tmp .= "\n\nProvide the response in the same language of the webpage.";
+        } else {
+            $tmp .= "\n\nThe response must be in ". Languages::getName($this->settings->outputLanguage, 'en') ." language. Translate texts if needed.";
+        }
+
+        return $tmp;
     }
 
 }
diff --git a/src/Settings/InfoProviderSystem/AIExtractorSettings.php b/src/Settings/InfoProviderSystem/AIExtractorSettings.php
index 876c687f..4ef9a1fa 100644
--- a/src/Settings/InfoProviderSystem/AIExtractorSettings.php
+++ b/src/Settings/InfoProviderSystem/AIExtractorSettings.php
@@ -32,7 +32,9 @@ use Jbtronics\SettingsBundle\Settings\Settings;
 use Jbtronics\SettingsBundle\Settings\SettingsParameter;
 use Jbtronics\SettingsBundle\Settings\SettingsTrait;
 use Symfony\AI\Platform\Capability;
+use Symfony\Component\Form\Extension\Core\Type\LanguageType;
 use Symfony\Component\Translation\TranslatableMessage as TM;
+use Symfony\Component\Validator\Constraints\Language;
 
 #[Settings(name: "ai_extractor", label: new TM("settings.ips.ai_extractor"), description: new TM("settings.ips.ai_extractor.description"))]
 #[SettingsIcon("fa-plug")]
@@ -56,4 +58,10 @@ class AIExtractorSettings
         description: new TM("settings.ips.ai_extractor.max_content_length.description"),
     )]
     public int $maxContentLength = 50000;
+
+    #[Language]
+    #[SettingsParameter(label: new TM("settings.ips.ai_extractor.output_language"), description: new TM("settings.ips.ai_extractor.output_language.description"),
+        formType: LanguageType::class,
+    )]
+    public ?string $outputLanguage = null;
 }
diff --git a/translations/messages.en.xlf b/translations/messages.en.xlf
index e16a6d69..c0bf8b60 100644
--- a/translations/messages.en.xlf
+++ b/translations/messages.en.xlf
@@ -13103,5 +13103,29 @@ Buerklin-API Authentication server:
         <target>The AI model that should be used for extraction. Must support structured output.</target>
       </segment>
     </unit>
+    <unit id="H1SYgGs" name="settings.ips.ai_extractor.max_content_length">
+      <segment>
+        <source>settings.ips.ai_extractor.max_content_length</source>
+        <target>Max. Website Content length</target>
+      </segment>
+    </unit>
+    <unit id="SZWiZE3" name="settings.ips.ai_extractor.max_content_length.description">
+      <segment>
+        <source>settings.ips.ai_extractor.max_content_length.description</source>
+        <target>The maximum number of characters of the website that are sent to the AI service.</target>
+      </segment>
+    </unit>
+    <unit id="pCsAHOv" name="settings.ips.ai_extractor.output_language">
+      <segment>
+        <source>settings.ips.ai_extractor.output_language</source>
+        <target>Output language</target>
+      </segment>
+    </unit>
+    <unit id="NVHHgpD" name="settings.ips.ai_extractor.output_language.description">
+      <segment>
+        <source>settings.ips.ai_extractor.output_language.description</source>
+        <target>By default, the providers returns information in the same language as the website. With that option you can ask the AI to translate it for you. Might only work with certain models.</target>
+      </segment>
+    </unit>
   </file>
 </xliff>