File

Apart from Path CLI parameters you can also declare some types of "files".

Tip

In most of the cases you are probably fine just using Path.

You can read and write data with Path the same way.

The difference is that these types will give you a Python file-like object instead of a Python Path.

A "file-like object" is the same type of object returned by open() as in:

with open('file.txt') as f:
    # Here f is the file-like object
    read_data = f.read()
    print(read_data)

But in some special use cases you might want to use these special types. For example if you are migrating an existing application.

FileText reading

typer.FileText gives you a file-like object for reading text, you will get str data from it.

This means that even if your file has text written in a non-english language, e.g. a text.txt file with:

la cigüeña trae al niño

You will have a str with the text inside, e.g.:

content = "la cigüeña trae al niño"

instead of having bytes, e.g.:

content = b"la cig\xc3\xbce\xc3\xb1a trae al ni\xc3\xb1o"

You will get all the correct editor support, attributes, methods, etc for the file-like object:`

Python 3.7+Python 3.7+ non-Annotated

import typer
from typing_extensions import Annotated


def main(config: Annotated[typer.FileText, typer.Option()]):
    for line in config:
        print(f"Config line: {line}")


if __name__ == "__main__":
    typer.run(main)

Tip

Prefer to use the Annotated version if possible.

import typer


def main(config: typer.FileText = typer.Option(...)):
    for line in config:
        print(f"Config line: {line}")


if __name__ == "__main__":
    typer.run(main)

Check it:

// Create a quick text config
$ echo "some settings" > config.txt

// Add another line to the config to test it
$ echo "some more settings" >> config.txt

// Now run your program
$ python main.py --config config.txt

Config line: some settings

Config line: some more settings

FileTextWrite

For writing text, you can use typer.FileTextWrite:

Python 3.7+Python 3.7+ non-Annotated

import typer
from typing_extensions import Annotated


def main(config: Annotated[typer.FileTextWrite, typer.Option()]):
    config.write("Some config written by the app")
    print("Config written")


if __name__ == "__main__":
    typer.run(main)

Tip

Prefer to use the Annotated version if possible.

import typer


def main(config: typer.FileTextWrite = typer.Option(...)):
    config.write("Some config written by the app")
    print("Config written")


if __name__ == "__main__":
    typer.run(main)

This would be for writing human text, like:

some settings
la cigüeña trae al niño

...not to write binary bytes.

Check it:

$ python main.py --config text.txt

Config written

// Check the contents of the file
$ cat text.txt

Some config written by the app

Technical Details

typer.FileTextWrite is a just a convenience class.

It's the same as using typer.FileText and setting mode="w". You will learn about mode later below.

FileBinaryRead

To read binary data you can use typer.FileBinaryRead.

You will receive bytes from it.

It's useful for reading binary files like images:

Python 3.7+Python 3.7+ non-Annotated

import typer
from typing_extensions import Annotated


def main(file: Annotated[typer.FileBinaryRead, typer.Option()]):
    processed_total = 0
    for bytes_chunk in file:
        # Process the bytes in bytes_chunk
        processed_total += len(bytes_chunk)
        print(f"Processed bytes total: {processed_total}")


if __name__ == "__main__":
    typer.run(main)

Tip

Prefer to use the Annotated version if possible.

import typer


def main(file: typer.FileBinaryRead = typer.Option(...)):
    processed_total = 0
    for bytes_chunk in file:
        # Process the bytes in bytes_chunk
        processed_total += len(bytes_chunk)
        print(f"Processed bytes total: {processed_total}")


if __name__ == "__main__":
    typer.run(main)

Check it:

$ python main.py --file lena.jpg

Processed bytes total: 512
Processed bytes total: 1024
Processed bytes total: 1536
Processed bytes total: 2048

FileBinaryWrite

To write binary data you can use typer.FileBinaryWrite.

You would write bytes to it.

It's useful for writing binary files like images.

Have in mind that you have to pass bytes to its .write() method, not str.

If you have a str, you have to encode it first to get bytes.

Python 3.7+Python 3.7+ non-Annotated

import typer
from typing_extensions import Annotated


def main(file: Annotated[typer.FileBinaryWrite, typer.Option()]):
    first_line_str = "some settings\n"
    # You cannot write str directly to a binary file, you have to encode it to get bytes
    first_line_bytes = first_line_str.encode("utf-8")
    # Then you can write the bytes
    file.write(first_line_bytes)
    # This is already bytes, it starts with b"
    second_line = b"la cig\xc3\xbce\xc3\xb1a trae al ni\xc3\xb1o"
    file.write(second_line)
    print("Binary file written")


if __name__ == "__main__":
    typer.run(main)

Tip

Prefer to use the Annotated version if possible.

import typer


def main(file: typer.FileBinaryWrite = typer.Option(...)):
    first_line_str = "some settings\n"
    # You cannot write str directly to a binary file, you have to encode it to get bytes
    first_line_bytes = first_line_str.encode("utf-8")
    # Then you can write the bytes
    file.write(first_line_bytes)
    # This is already bytes, it starts with b"
    second_line = b"la cig\xc3\xbce\xc3\xb1a trae al ni\xc3\xb1o"
    file.write(second_line)
    print("Binary file written")


if __name__ == "__main__":
    typer.run(main)

$ python main.py --file binary.dat

Binary file written

// Check the binary file was created
$ ls ./binary.dat

./binary.dat

File CLI parameter configurations

You can use several configuration parameters for these types (classes) in typer.Option() and typer.Argument():

• mode: controls the "mode" to open the file with.
  • It's automatically set for you by using the classes above.
  • Read more about it below.
• encoding: to force a specific encoding, e.g. "utf-8".
• lazy: delay I/O operations. Automatic by default.
  • By default, when writing files, Click will generate a file-like object that is not yet the actual file. Once you start writing, it will go, open the file and start writing to it, but not before. This is mainly useful to avoid creating the file until you start writing to it. It's normally safe to leave this automatic. But you can overwrite it setting lazy=False. By default, it's lazy=True for writing and lazy=False for reading.
• atomic: if true, all writes will actually go to a temporal file and then moved to the final destination after completing. This is useful with files modified frequently by several users/programs.

Advanced mode

By default, Typer will configure the mode for you:

• typer.FileText: mode="r", to read text.
• typer.FileTextWrite: mode="w", to write text.
• typer.FileBinaryRead: mode="rb", to read binary data.
• typer.FileBinaryWrite: mode="wb", to write binary data.

Note about FileTextWrite

typer.FileTextWrite is actually just a convenience class. It's the same as using typer.FileText with mode="w".

But it's probably shorter and more intuitive as you can get it with autocompletion in your editor by just starting to type typer.File... just like the other classes.

Customize mode

You can override the mode from the defaults above.

For example, you could use mode="a" to write "appending" to the same file:

Python 3.7+Python 3.7+ non-Annotated

import typer
from typing_extensions import Annotated


def main(config: Annotated[typer.FileText, typer.Option(mode="a")]):
    config.write("This is a single line\n")
    print("Config line written")


if __name__ == "__main__":
    typer.run(main)

Tip

Prefer to use the Annotated version if possible.

import typer


def main(config: typer.FileText = typer.Option(..., mode="a")):
    config.write("This is a single line\n")
    print("Config line written")


if __name__ == "__main__":
    typer.run(main)

Tip

As you are manually setting mode="a", you can use typer.FileText or typer.FileTextWrite, both will work.

Check it:

$ python main.py --config config.txt

Config line written

// Run your program a couple more times to see how it appends instead of overwriting
$ python main.py --config config.txt

Config line written

$ python main.py --config config.txt

Config line written

// Check the contents of the file, it should have each of the 3 lines appended
$ cat config.txt

This is a single line
This is a single line
This is a single line

About the different types

Info

These are technical details about why the different types/classes provided by Typer.

But you don't need this information to be able to use them. You can skip it.

Typer provides you these different types (classes) because they inherit directly from the actual Python implementation that will be provided underneath for each case.

This way your editor will give you the right type checks and completion for each type.

Even if you use lazy. When you use lazy Click creates a especial object to delay writes, and serves as a "proxy" to the actual file that will be written. But this especial proxy object doesn't expose the attributes and methods needed for type checks and completion in the editor. If you access those attributes or call the methods, the "proxy" lazy object will call them in the final object and it will all work. But you wouldn't get autocompletion for them.

But because these Typer classes inherit from the actual implementation that will be provided underneath (not the lazy object), you will get all the autocompletion and type checks in the editor.