Auto Translate JSON

Adds a context menu item to translation files to automatically translate them into other languages using Google Translate, AWS Translate, Azure Translate, DeepL, OpenAI, or Hugging Face (cloud & local, including local AI via Ollama).

Supported File Formats

JSON - Standard JSON translation files
XML - Android strings.xml, iOS plist, generic XML
YAML - YAML/YML configuration files
ARB - Flutter Application Resource Bundle
PO/POT - GNU Gettext translation files
XLIFF - XML Localization Interchange File Format
XMB/XTB - XML Message Bundle (Angular)
Properties - Java/Spring properties files
CSV/TSV - Comma/Tab-separated values

How it works

When localizing an application, if you have a folder called something like translations, languages, or i18n that contains translation files for each language, you can use this extension to right click on your primary language file and automatically create additional translations. It uses the Google/AWS/Azure/DeepL/OpenAI/Hugging Face Translate API to perform the translations, and you must have your own API key to make the calls.

Just create empty files with the locale names as filenames and this extension will generate their translations. For example, if you want French, create a file fr.json. Right click on en.json, pick "Auto Translate" and voilà, you have a version in French.

Features

Multi-format support - Works with JSON, XML, YAML, ARB, PO, XLIFF, Properties, CSV, and more
Format auto-detection - Automatically detects file format based on extension and content
Option to keep existing translations, to cut down on data processing when adding new terms
Option to keep extra translations, if one language has additional unique terms
Supports nested elements in JSON, XML, and YAML
Supports named arguments such as: "Zip code {zip} is in {city}, {state}."
Processes all files simultaneously
Hugging Face integration - Cloud API with multiple providers or local ONNX runtime for private, offline translation
Local AI support - Use Ollama or other OpenAI-compatible local servers

Supported languages

Google: https://cloud.google.com/translate/docs/languages
AWS: https://docs.aws.amazon.com/translate/latest/dg/what-is.html#what-is-languages
Azure: https://docs.microsoft.com/en-us/azure/cognitive-services/translator/language-support
DeepL: https://www.deepl.com/docs-api/other-functions/listing-supported-languages/
OpenAI: https://platform.openai.com/
Hugging Face: https://huggingface.co/models?pipeline_tag=translation (supports many languages via different models)

Requirements

Since translation services are not free, you must provide your own API key.

For Google API key you need to provide just API key. Luckily Google gives a decent amount of translations in a trial period. Go here to set up your account and request a key: https://console.developers.google.com/apis/library/translate.googleapis.com

For AWS you need to provide your access key and secret key. Go here to set up your account and request an access key and secret key and region to use: https://aws.amazon.com/premiumsupport/knowledge-center/create-and-activate-aws-account/

For Azure you need to provide your subscription key and region. Go here to set up your account and request a subscription key: https://azure.microsoft.com/en-us/free/

For DeepL you need to provide your authentication API key. Go here to set up your account and request an API key:https://www.deepl.com/pro-api?cta=header-pro-api/

For OpenAI you need to provide just API key.Go here to set up your account and request a key: API key:https://platform.openai.com/

For Hugging Face cloud API you need to provide an API key and model name. Go here to set up your account and request a key: API key:https://huggingface.co/settings/tokens

For Hugging Face local mode you need to provide only a model name (no API key required). The model runs locally using transformers.js.

Getting Started

Request a Google/AWS/Azure/DeepL/OpenAI/Hugging Face Translate API key (or use Hugging Face local mode without API key)
Install this extension
Go to VSCode Settings>Extensions>Auto Translate JSON
Enter your Google/AWS/Azure/DeepL/OpenAI/Hugging Face API key / access key / subscription key / region / model (or use Hugging Face local mode with model name only)
(optional) Change the Source Locale setting if you don't want English
(optional) Set the Format setting or leave as "auto" for automatic detection
Create empty files for each locale you want to translate into. Locale should correspond to the language code used by the translation service. For example, if you want French, create a file fr.json (or fr.xml, fr.yaml, etc.).
- If you use Azure and want to translate into Serbian, create a file sr-Cyrl.json for Serbian Cyrillic translation or sr-Latn.json for Serbian Latin translation.
- If you use AWS or Google and want to translate into Serbian, create a file sr.json for Serbian translation.
Right click the source translation file (en.json, en.xml, en.yaml, etc.) and pick "Auto Translate"
At the prompt decide if you want to preserve previously translated values (i.e. not reprocess)
At the prompt decide if you want to keep extra translations

Verify your language files have been updated

Extension Settings

This extension contributes the following settings (Menu>Preferences>Settings):

General

auto-translate-json.sourceLocale: A failsafe to prevent processing the wrong file. Defaults to "en" for english. You can change this to any valid two letter locale code you wish to use.
auto-translate-json.mode: "file": files in same folder like "en.json"...; "folder": files in subfolders like "en/translation.json"
auto-translate-json.format: File format for translation. Use "auto" for automatic detection based on file extension. Supported formats: json, xml, android-xml, ios-xml, yaml, arb, po, pot, xliff, xmb, xtb, properties, csv, tsv
auto-translate-json.startDelimiter: Start delimiter for named arguments. Defaults to "{". Use "{{" for ngx-translate or transloco.
auto-translate-json.endDelimiter: End delimiter for named arguments. Defaults to "}". Use "}}" for ngx-translate or transloco.
auto-translate-json.ignorePrefix: Translation keys that start with this prefix will be ignored (e.g., "@@" for ARB metadata)

Google

auto-translate-json.googleApiKey: Enter your Google API key in this setting.

AWS

auto-translate-json.awsAccessKeyId: Enter your AWS API access key Id in this setting.
auto-translate-json.awsSecretAccessKey: Enter your AWS API secret access key in this setting.
auto-translate-json.awsRegion: Enter your AWS region in this setting.

Azure

auto-translate-json.azureSecurityKey: Enter your Azure security key in this setting.
auto-translate-json.azureRegion: Enter your Azure region in this setting.

DeepL Free

auto-translate-json.deepLFreeSecurityKey: Enter your DeepL Free security key in this setting.

DeepL Pro

auto-translate-json.deepLProSecurityKey: Enter your DeepL Pro security key in this setting.

OpenAI

auto-translate-json.openAIKey: Enter your OpenAI API key in this setting.
auto-translate-json.openAIBaseURL: Base URL for OpenAI API. Defaults to "https://api.openai.com/v1". Change for local servers.
auto-translate-json.openAIModel: Model to use. Defaults to "gpt-4.1-mini" (recommended for translations).
auto-translate-json.openAIMaxTokens: Maximum tokens per request. Defaults to 1000.
auto-translate-json.openAITemperature: Temperature (0-2). Lower values produce more consistent translations. Defaults to 0.1.

Hugging Face (Cloud)

auto-translate-json.huggingFaceApiKey: Enter your Hugging Face API key in this setting.
auto-translate-json.huggingFaceModel: Model to use (e.g., "Helsinki-NLP/opus-mt-en-fr" for English to French).
auto-translate-json.huggingFaceProvider: Optional inference provider (e.g., "hf-inference", "together", "replicate", "auto"). Leave empty for default ("auto" selects the first available provider). See the "Using Hugging Face Translation" section below for detailed explanation.

Hugging Face (Local)

auto-translate-json.huggingFaceLocalModel: Model to use for local inference (e.g., "Xenova/opus-mt-en-fr"). Runs locally using transformers.js with ONNX runtime.

Using Hugging Face Translation

This extension supports both Hugging Face cloud API and local on-device inference.

Cloud API (Requires Hugging Face API Key)

Get a Hugging Face API key:
- Create a free account at huggingface.co
- Generate an access token at huggingface.co/settings/tokens
Choose a translation model:
- Browse available models at huggingface.co/models?pipeline_tag=translation
- Popular models include:
  - Helsinki-NLP/opus-mt-en-fr (English → French)
  - Helsinki-NLP/opus-mt-en-es (English → Spanish)
  - Helsinki-NLP/opus-mt-en-de (English → German)
  - facebook/m2m100_418M (multilingual)
Configure extension settings:
- auto-translate-json.apiType: Set to "HuggingFace"
- auto-translate-json.huggingFaceApiKey: Your Hugging Face API key
- auto-translate-json.huggingFaceModel: Model name (e.g., "Helsinki-NLP/opus-mt-en-fr")
- auto-translate-json.huggingFaceProvider: Optional provider (e.g., "hf-inference", "together", "replicate", "auto"). Leave empty for default.
About the Provider Parameter: The huggingFaceProvider setting is optional and specifies which inference provider to use:
- "hf-inference": Hugging Face's official Inference API (default when available)
- "together": Together AI's inference platform
- "replicate": Replicate's hosted models
- "auto": Automatically selects the first available provider (default if left empty)
You typically don't need to set this unless:
- You have a specific provider preference (e.g., using Together AI credits)
- You encounter issues with the default provider
- You want to explicitly use a particular inference backend
Usage notes:
- Cloud API requires internet connection
- Translation quality depends on the selected model
- Check model documentation for supported language pairs
- Some models support direct language pairs (en→fr), others are multilingual

Local Mode (No API Key Required)

Choose a local ONNX model:
- Models must be compatible with @huggingface/transformers JavaScript runtime
- Popular local models use the "Xenova/" prefix:
  - Xenova/opus-mt-en-fr (English → French)
  - Xenova/m2m100_418M (multilingual)

Note: Helsinki-NLP models only work with the cloud API and will fail with "Could not locate file: tokenizer.json" error when used locally.

Configure extension settings:
- auto-translate-json.apiType: Set to "HuggingFaceLocal"
- auto-translate-json.huggingFaceLocalModel: Model name (e.g., "Xenova/opus-mt-en-fr")
First run behavior:
- Model weights (~300MB) are downloaded and cached automatically
- Subsequent runs use the cached model
- No internet required after initial download
Performance considerations:
- Local inference runs on your CPU (or GPU if available)
- Translation speed depends on model size and hardware
- Larger models provide better quality but are slower
- Recommended for batch translations or privacy-sensitive projects

Local AI with Ollama

Install Ollama
Pull a recommended model: ollama pull qwen2.5:14b
Configure extension settings:
- auto-translate-json.openAIKey: "ollama" (placeholder)
- auto-translate-json.openAIBaseURL: "http://localhost:11434/v1"
- auto-translate-json.openAIModel: "qwen2.5:14b"
- auto-translate-json.openAIMaxTokens: 512

Other Local AI Servers

Compatible with any OpenAI-compatible server:

Limitations

Files must be named with the locale code that may be different depending on the translation service that you use. Please see the supported languages above.

Format-Specific Notes

ARB (Flutter)

Keys starting with @@ (like @@locale) are treated as metadata and not translated
Set ignorePrefix to @@ to skip metadata keys

Android XML

Automatically detects <resources> structure
Preserves translatable="false" attributes

PO/POT

Preserves headers and metadata
Handles plural forms

XLIFF

Supports XLIFF 1.2 and 2.0 formats

Prices

Google

https://cloud.google.com/translate/pricing

AWS

https://aws.amazon.com/translate/pricing/

Azure

https://azure.microsoft.com/en-us/pricing/details/cognitive-services/translator/

DeepL

https://www.deepl.com/pro?cta=header-prices

OpenAI

https://openai.com/pricing

Hugging Face

https://huggingface.co/pricing (cloud API) | Local mode is free but requires local resources

WARNING

Keep your keys safe!

Name		Name	Last commit message	Last commit date
Latest commit History 109 Commits
.vscode		.vscode
images		images
src		src
.eslintrc.json		.eslintrc.json
.gitignore		.gitignore
.prettierrc.json		.prettierrc.json
.vscodeignore		.vscodeignore
CHANGELOG.md		CHANGELOG.md
LICENSE.md		LICENSE.md
PUBLISH.md		PUBLISH.md
README.md		README.md
package-lock.json		package-lock.json
package.json		package.json
tsconfig.json		tsconfig.json
vsc-extension-quickstart.md		vsc-extension-quickstart.md

Folders and files

Latest commit

History

Repository files navigation

Auto Translate JSON

Supported File Formats

How it works

Features

Supported languages

Requirements

Getting Started

Extension Settings

General

Google

AWS

Azure

DeepL Free

DeepL Pro

OpenAI

Hugging Face (Cloud)

Hugging Face (Local)

Using Hugging Face Translation

Cloud API (Requires Hugging Face API Key)

Local Mode (No API Key Required)

Local AI with Ollama

Other Local AI Servers

Limitations

Format-Specific Notes

ARB (Flutter)

Android XML

PO/POT

XLIFF

Prices

WARNING

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases 3

Packages 0

Uh oh!

Uh oh!

Contributors

Uh oh!

Languages

Packages