[MRG+1] Included new optional parameter in startproject command line

[feliperuhland]

I don't know if anyone creates project like me.
I have a habit to create git repository and then I create scrapy project.

You can see something similar at django startproject command.

Unfortunately, I had to rewrite copytree function, to make sure that the startproject won't fail if it project_dir already exists.

I am available to any feedback,

Thanks.

[eliasdorneles]

Hello, @feliperuhland.

I like the feature idea, it will be nice to be able to do scrapy startproject myproj . inside a directory.

The build is failing for Python 3, can you have a look?

[feliperuhland]

Thanks, @eliasdorneles . I am working on it.

[eliasdorneles]

The template should also be updated to print a message with the correct project dir.

I just tried it with:

$ scrapy startproject myproj customdir

And it printed:

You can start your first spider with:
    cd myproj
    scrapy genspider example example.com

It should point me to cd customdir

[feliperuhland]

@eliasdorneles you're right.
I will fix this.

[eliasdorneles]

Looks great, thanks @feliperuhland !

[eliasdorneles]

hey @feliperuhland, I'm looking at this again, and I'm thinking this check if scrapy.cfg exists is needed in both cases (args == 1 or args == 2), right?
looks like it should be outside of this if.

[feliperuhland]

Not really, because in the case of len(args) == 1 the directory will be created and can't have the scrapy.cfg. But I can't see any problem in this change. What do you think? Thanks.

[eliasdorneles]

Since we're using a custom copytree that now works in existing directories, I think we shouldn't assume the directory will be created.
Users can now do mkdir myproj; scrapy startproject myproj and scrapy will work without complaining, therefore the extra check is a good idea.

In fact, I just tried running scrapy startproject myproj twice and it breaks the second time.

[eliasdorneles]

Also, do you think we could get rid of this function, if we use the ignore callable argument for shutil.copytree ?

[feliperuhland]

Hi, @eliasdorneles. I would like to remove that function, but the original implementation always creates the destination directory and that is not useful to us. I'm not happy of rewrite some internal library behavior. Do you have any other ideias? Thanks again.

[eliasdorneles]

Ah yeah, you're right.
The problem is that os.makedirs call that is always called and requires the directory not to exist.
Can you perhaps add a docstring explaining this, then?

[feliperuhland]

@eliasdorneles I done what you suggested and I think it's good. Can you read the docstring to see if it's good enough?

Thanks again!

[eliasdorneles]

Yup, looks great now. Thank you, @feliperuhland !

@redapple this looks ready. How do you feel about merging it for 1.2?

[redapple]

Thanks @feliperuhland !