robinhilliard / pipes Goto Github PK

Hello Robin,

Thank you so much for this project. I have long felt that Python could benefit from a pipe operator, and to see it actually be implemented is just awesome. I was so pleased with it that I decided to fork your project and take it to its logical extent (arguably, I took it too far). In my fork, which I'm calling Superpipe, each segment of the pipe is much more generalized. You can insert the piped value into any position of a function call, construct new collections on-the-fly, and even call methods on the piped value.

Some examples:

[2, 3] >> [1, *_, 4, 5] >> f"list: {_}" >> print

5 >> _ + 99 >> [_+1, 3] >> sum >> _.bit_length() >> print("Bit length:", _)

I had initially intended to open a pull request here, however as I kept adding more and more to it I began to feel like I had diverged too much from the original vision, and so am deciding to keep my fork independent.

This is the first time I've forked a public project like this, and I want to ensure that I'm respecting your license and you properly.

I am keeping the fork MIT-licensed, so I think I should be in the clear there, however I also intend to upload the fork as a new package on PyPi. I think this should be okay? The only question then left is how do I appropriately credit you. The Github repository correctly attributes you as the author of the fork, and I will include an additional link in the readme, however I am uncertain what to do with regards to authors of the PyPi package: Should I insert my name and email beside yours, or list myself as the sole author?

Thanks

P.S.

Here is a link to the fork: https://github.com/George-lewis/superpipe

The readme is still in the process of being rewritten, however I believe the tests may be of interest

It'd be great if you could apply the @pipes transformation at the level of the class/module instead of having to annotate each function within.

Hey! Just tried to install the package from pip, and it's failing with:

 src/Bot-new   mainline  pip3 install pipeop
Collecting pipeop
  Downloading https://files.pythonhosted.org/packages/f3/71/5449fe7a4ca9ecef67bca9ceb13a163d7326fc0c325b6642694e7da5c0fc/pipeop-0.2.2.tar.gz
    Complete output from command python setup.py egg_info:
    Traceback (most recent call last):
      File "<string>", line 1, in <module>
      File "/private/var/folders/bw/539xc4113rd4lcc56p27lq1jghmd7n/T/pip-install-ujbf0qn9/pipeop/setup.py", line 5, in <module>
        with open('README.rst', 'r') as f:
    FileNotFoundError: [Errno 2] No such file or directory: 'README.rst'

    ----------------------------------------
Command "python setup.py egg_info" failed with error code 1 in /private/var/folders/bw/539xc4113rd4lcc56p27lq1jghmd7n/T/pip-install-ujbf0qn9/pipeop/

Hey, thank you for creating this, just what I'm looking for 😃

It'd be great if you could add Python3 support!
Currently, if I try to use it in Python3, I'm getting the following error:

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-6-28ed7b59ff38> in <module>()
      5     return a * b
      6 
----> 7 @pipes
      8 def calc():
      9     print(1 >> add(2) >> times(3))  # prints 9

/usr/local/anaconda3/lib/python3.6/site-packages/pipeop/__init__.py in pipes(func)
     17 def pipes(func):
     18     # name of our replacement function
---> 19     pipe_func_name = '__pipes_{}'.format(func.func_code.co_name)
     20 
     21     # variable context where decorator added

AttributeError: 'function' object has no attribute 'func_code'

It appears that functions cannot be included conditionally in a series of pipes. For instance:

def noop(arg):
    return arg

def sum(a, b):
    return a + b

@pipes
def main():
    print(
        5
        >> sum(2)
        >> (sum(sys.argv[1]) if len(sys.argv) > 1 else noop)
    )

Produces:

TypeError: sum() missing 1 required positional argument: 'b'

This would be a nice feature to have when constructing a complex pipeline which requires some sort of conditional logic.

Thank you so much for this brilliant little thing. I've just copy&pasted your main code as a tooling decorator into https://github.com/amogorkon/justuse and it does its job without complaining even once.
For example
ordered = releases >> _filter_by_platform(tags=get_supported()) >> _sort_releases

last_version = project.releases >> reversed >> iter >> next

meta = (names 
<< filter(DIST_PKG_INFO_REGEX.search) 
<< map(functions.read_entry) 
>> dict
)

Pipes make this so much nicer and less cluttered. Readability counts! I hope you will continue maintaining this beautiful thing and maybe even make a PEP for inclusion into the core language 👍

This probably is also the only way to fix a little issue that happens when you write a small chain of pipes like

a = (foo
>> bar
<< baz
)

which would look ugly on a single line a = foo >> bar << baz with those arrows piping in different directions - alas, black thinks so otherwise, so it turns the multi-line pipe into a single line.

PS: You really should update your examples on your README to python 3 print-as-a-function ;-)

def calc():
    print 1 >> add(2) >> times(3)  # prints 9

should be

def calc():
    1 >> add(2) >> times(3)  >> print # prints 9