Lobe CLI Toolbox

Lobe i18n is a tool designed to automate internationalization (i18n) workflows using ChatGPT. Key features include:

• Automated Translation: Leverages ChatGPT to automate the translation process.
• Large File Support: Automatically splits large files to avoid ChatGPT token limits.
• Incremental Updates: Supports incremental i18n updates by automatically extracting new content based on an entry file.
• Flexible File Structures: Supports both single-file mode (e.g., en.json) and folder mode (e.g., en/common.json), making it compatible with i18next.
• Locale Formats: Supports both flat and tree locale file structures.
• Customizable AI Configuration: Allows users to specify custom OpenAI models, API proxies, and temperature settings.

Lobe i18n is an automation tool for the i18n translation process powered by ChatGPT. Key features include:

• Automatic splitting of large translation files.
• Incremental updates.
• Customization of OpenAI models, API proxies, and temperature settings.

The mode configuration determines how Markdown content is processed for translation:

• string: The entire Markdown content is treated as a single string and sent for translation. This preserves the full structure but may use more tokens.
• mdast: Uses mdast (Markdown Abstract Syntax Tree) to parse the structure and only translates the text value nodes.
  • Pros: Significantly reduces token consumption by stripping most Markdown syntax and links.
  • Cons: May lead to less accurate translations due to the reduced context.
  • Note: If using mdast and you need to translate code blocks, you must set translateCode: true.

An automation tool for the internationalization translation process, powered by ChatGPT. Key features include:

• Automatic splitting of large files.
• Incremental updates.
• Customization options for OpenAI models, API proxies, and temperature settings.

Lobe SEO supports configuration using the cosmiconfig format. You can define settings in the following ways:

• The seo property in package.json.
• A .seorc file in JSON or YAML format.
• Specific files: .seorc.json, .seorc.yaml, .seorc.yml, .seorc.js, or .seorc.cjs.

For type safety, you can import defineConfig from @lobehub/seo-cli when using .js configuration files.

const { defineConfig } = require('@lobehub/seo-cli');

module.exports = defineConfig({
  entry: './docs/**/*.mdx',
  modelName: 'gpt-3.5-turbo-1106',
  experimental: {
    jsonMode: true,
  },
});

Lobe i18n supports two primary ways of organizing translation files:

1. Single-file structure

All languages are stored in individual JSON files within a directory.

- locales/
  - en_US.json
  - zh_CN.json

Config: Set entry to the specific source JSON file.

2. Folder structure

Each language has its own directory containing multiple JSON files (compatible with i18next).

- locales/
  - en_US/
    - common.json
  - zh_CN/
    - common.json

Config: Set entry to the source language directory.

// Single-file config example
{
  "entry": "locales/en.json",
  "entryLocale": "en_US",
  "output": "locales",
  "outputLocales": ["zh_CN", "ja_JP"]
}

// Folder-structure config example
{
  "entry": "locales/en_US",
  "entryLocale": "en_US",
  "output": "locales",
  "outputLocales": ["zh_CN", "ja_JP"]
}

The script generates a formatted packages/common/models.ts file containing three main exports:

1. LanguageModel: An enum containing all available models. The script automatically converts kebab-case model names to UPPER_SNAKE_CASE for the enum members.
2. ModelTokens: A Record<LanguageModel, number> that maps each model to its context window token count (formatted with underscores for readability).
3. defaultModel: A constant representing the recommended default model (prioritizing recent 'mini' models like o4-mini).

export enum LanguageModel {
  /**
   * o3
   */
  O3 = 'o3',
  // ... more models
}

export const ModelTokens: Record<LanguageModel, number> = {
  [LanguageModel.O3]: 200_000,
  // ... more mappings
};

export const defaultModel = LanguageModel.O4_MINI;