DeepL CLI is a fast and lightweight command-line tool for the DeepL API, written in Crystal.

• Supports document translation pdf, docx, txt, etc.
• Supports glossaries
• Precompiled binaries available

• Download the binary from the Releases

brew install kojix2/brew/deepl-cli

export HTTP_PROXY=https://[IP]:[port]
export HTTPS_PROXY=https://[IP]:[port]

You will need an API key for DeepL. Create one here and set it as an environment variable:

export DEEPL_AUTH_KEY=your_api_key_here

deepl [options] <file>

deepl [options] <file>

Options:

    -i, --input TEXT                 Input text
    -f, --from [LANG]                Source language [AUTO]
    -t, --to [LANG]                  Target language [EN]
    -p, --paste                      Input text from clipboard
    -g, --glossary NAME              Glossary name
    -F, --formality OPT              Formality (default more less)
    -C, --context TEXT               Context (experimental)
    -S, --split-sentences OPT        Split sentences
    -A, --ansi                       Do not remove ANSI escape codes

Note: ANSI escape sequences are removed by default.

To translate a document, use the doc subcommand:

deepl doc [options] <file>

Options for document translation:

    -f, --from [LANG]                Source language [AUTO]
    -t, --to [LANG]                  Target language [EN]
    -g, --glossary NAME              Glossary name
    -F, --formality OPT              Formality (default more less)
    -o, --output FILE                Output file
    -O, --output-format FORMAT       Output file format

Supported file formats.

• docx - Microsoft Word Document
• pptx - Microsoft PowerPoint Document
• xlsx - Microsoft Excel Document
• pdf - Portable Document Format
• htm / html - HTML Document
• txt - Plain Text Document
• xlf / xliff - XLIFF Document, version 2.1

For glossary management, use the glossary subcommand:

deepl glossary [options]

Options for glossary management:

    list                             List glossaries
    create                           Create a glossary
    delete                           Delete glossaries
    edit                             Edit glossaries
    view                             View glossaries
    -l, --list                       List glossaries
    -p, --language-pairs             List language pairs

Use the rephrase subcommand to rewrite text in different styles or tones.

Usage: deepl rephrase [options] <file>

Options for rephrase:

    -i, --input TEXT                 Input text
    -s, --writing-style STYLE        academic business casual default simple
    -t, --tone TONE                  confident diplomatic enthusiastic friendly

Below are examples for translating text, translating documents, and working with glossaries.

To translate the text "Hola mundo" from Spanish (ES) to English (EN):

deepl -i "Hola mundo" -t en        # Translation: Hello world

Or, using standard input:

echo "Hola mundo" | deepl -t en    # Translation: Hello world

Standard input translation is useful for quick references.

git --help | deepl -t fr | less

The man command can also be translated (by removing ANSI escape sequences):

man git | deepl -t de | less

To translate multiple lines, press Ctrl+D when you have finished typing. This is particularly useful when copying and pasting from the clipboard.

deepl -f es
# Hola
# mundo
# Ctrl + D

Translate text from the clipboard:

deepl --paste

You can also pass a text file as an argument:

deepl -t tr foo.txt

It's possible to pass multiple text files:

deepl -t nl foo.txt bar.txt

If you are translating multiple files, you might want to add the filename to the header:

bat --style header *.txt | deepl -t it

To use a glossary for translation:

deepl -g myglossary -f ru

To refer to the original text, you can use tee dev/stderr:

fortune | tee /dev/stderr | deepl

You can directly translate documents:

deepl doc your.pdf -t pt
# The translated document will be saved as your_PT.pdf

To use a glossary for translation:

deepl doc -g myglossary -f pl

To translate a PDF document and save it in docx format:

deepl doc input.pdf -O docx -o output.docx

To translate multiple files, use Unix commands such as find, xargs, fd:

find . -name "*.pdf" -exec deepl doc -t ja {} +

ls -1 *.docx | xargs -L1 deepl doc -t ko

fd -e pdf -e docx -x deepl doc -t zh

The DeepL API supports glossaries. See here for the format of the glossary file. The glossary API is v2. It does not yet support v3.

To create a glossary:

deepl glossary create -n mydic -f en -t pt mydict.tsv

To list glossaries:

deepl glossary list

To list only the names of the glossary:

deepl glossary -l

To use a glossary for text translation:

deepl -g mydict -f en deep.txt

To use a glossary for document translation:

deepl doc -g mydict -f en deep.pdf

To display the contents of the glossary:

deepl glossary view mydict

To list the languages in which glossaries can be created:

deepl glossary -p

Change writing style:

bin/deepl rephrase -i "Thanks" -s academic
# Expressions of gratitude are extended.

Change tone:

bin/deepl rephrase -i "Thanks" -t friendly
# Thank you so much! I really appreciate it.

To display a list of available source languages:

deepl -f

To display a list of available target languages:

deepl -t

To output usage information:

deepl -u
# https://api.deepl.com/v2
# character_count: 614842
# character_limit: 1000000000000

DeepL CLI automatically detects the target language, but if this does not work, the environment variable DEEPL_TARGET_LANG can be set.

• Fork this repository
• Report bugs
• Fix bugs and submit pull requests
• Write, clarify, or fix documentation
• Suggest or add new features

Compilation from source code

git clone https://github.com/kojix2/deepl-cli
cd deepl-cli
shards build --release

A compiled binary file will be created in the bin directory. Installation is simply copying the generated binary.

sudo cp bin/deepl /usr/local/bin

If you encounter a bug, follow the command with the -d option and run it. You can view the backtrace.

deepl doc -d a.pdf

• https://github.com/kojix2/deepl.cr/

This project is licensed under the MIT License.

Happy translating!