90 lines
2.8 KiB
Markdown
90 lines
2.8 KiB
Markdown
# d10n - Content Translation CLI Tool
|
|
|
|
`d10n` is a CLI tool for translating text content in files or directories using LLM models.
|
|
|
|
## Configuration
|
|
|
|
Create a configuration file at `~/.config/d10n.yaml` with the following format:
|
|
|
|
```yaml
|
|
api_base: "https://api.openai.com" # OpenAI-compatible API base URL
|
|
api_key: "your-api-key" # API key for the service
|
|
model: "gpt-4o" # Model to use for translation
|
|
system_prompt: "Custom prompt" # Optional: Custom system prompt
|
|
|
|
# Concurrency settings
|
|
concurrency: 3 # Number of concurrent translation tasks
|
|
|
|
# Chunked translation settings
|
|
chunk:
|
|
enabled: false # Whether to enable chunked translation
|
|
size: 10240 # Size of each chunk in tokens
|
|
prompt: "Please continue translation" # Prompt to use for continuing translation
|
|
context: 2 # Number of chunks to include as context
|
|
```
|
|
|
|
You can use the following variables in your system prompt:
|
|
- `$SOURCE_LANG`: Will be replaced with the source language code
|
|
- `$TARGET_LANG`: Will be replaced with the target language code
|
|
|
|
## Usage
|
|
|
|
```
|
|
d10n <source_path> [options]
|
|
```
|
|
|
|
### Options
|
|
|
|
- `-target <path>`: Target path for translated content (default: `<source_path>_l10n[.extension]`)
|
|
- `-source-lang <code>`: Source language code (required)
|
|
- `-target-lang <code>`: Target language code (required)
|
|
- `-model <model>`: Model to use for translation (overrides config)
|
|
- `-api-key <key>`: API key (overrides config)
|
|
- `-api-base <url>`: API base URL (overrides config)
|
|
- `-system-prompt <prompt>`: System prompt (overrides config)
|
|
- `-format <ext>`: File format to process (e.g., `md`, `txt`)
|
|
- `-concurrency <num>`: Number of concurrent translation tasks (default: 3)
|
|
- `-chunk`: Enable chunked translation for large documents
|
|
- `-chunk-size <tokens>`: Size of each chunk in tokens (default: 10240)
|
|
- `-chunk-prompt <prompt>`: Prompt for continuing translation (default: "Please continue translation")
|
|
- `-chunk-context <num>`: Number of chunks to include as context (default: 2)
|
|
|
|
### Examples
|
|
|
|
Translate a single file from English to Spanish:
|
|
```
|
|
d10n document.md -target-lang es
|
|
```
|
|
|
|
Translate a directory of Markdown files to Chinese:
|
|
```
|
|
d10n ./documents -target-lang zh -format md
|
|
```
|
|
|
|
Specify source language explicitly:
|
|
```
|
|
d10n article.txt -source-lang fr -target-lang en
|
|
```
|
|
|
|
Using chunked translation for large documents:
|
|
```
|
|
d10n large-document.md -source-lang en -target-lang es -chunk
|
|
```
|
|
|
|
Adjust concurrency for processing multiple files:
|
|
```
|
|
d10n ./documents -target-lang ja -format md -concurrency 5
|
|
```
|
|
|
|
Fine-tune chunked translation parameters:
|
|
```
|
|
d10n huge-document.md -source-lang en -target-lang zh -chunk -chunk-size 8192 -chunk-context 3
|
|
```
|
|
|
|
## Building from Source
|
|
|
|
```
|
|
go build -o d10n
|
|
```
|
|
|
|
Then move the binary to a location in your PATH.
|