Let --data-path be specified when running download.py scripts

[ExplodingCabbage]

Description

Adds a --data-path (or -d) argument to the download.py scripts that lets the path to download to be specified.

Motivation and Context

Makes downloading to a custom location easier; resolves #637

How Has This Been Tested?

Manually. After pip install -e ., I did a git clean -xdf to clean my local repo, then, in sequence:

• Checked that

  import spacy
  spacy.load('en')('Hello, world')[0].pos_
  

  gives the empty string

• Ran python3 -m spacy.en.download all -d ~/some_dir

• Checked that spacy.load('en')('Hello, world')[0].pos_ still gave the empty string but that spacy.util.set_data_path('/home/mark/some_dir'); spacy.load('en')('Hello, world')[0].pos_ didn't.

• Ran python3 -m spacy.en.download all

• Checked that spacy.load('en')('Hello, world')[0].pos_ no longer gave the empty string.

• Tried running python3 -m spacy.en.download all and python3 -m spacy.en.download all -d ~/some_dir again and confirmed that both exited early saying that the model was already installed.

Types of changes

• Bug fix (non-breaking change fixing an issue)
• New feature (non-breaking change adding functionality to spaCy)
• Breaking change (fix or feature causing change to spaCy's existing functionality)
• Documentation (Addition to documentation of spaCy)

Checklist:

• My code follows spaCy's code style. (*as far as I can tell - there's no actual style guide in CONTRIBUTING.md yet, so I just tried to imitate what was nearby)
• My change requires a change to spaCy's documentation. (ideally, anyway)
• I have updated the documentation accordingly.
• I have added tests to cover my changes. (no tests for the downloader currently exist; any that we'd want to add would probably want to be CI-only, since they'll be extremely slow for anyone on a slow internet connection)
• All new and existing tests passed.

[ExplodingCabbage]

I'm gonna quickly add in docs, then this should be ready to merge.

[ExplodingCabbage]

Docs added, and looked at (both on the site on the local server and in the README on GitHub) to make sure the formatting came out right. Still not quite ready for merge - in the process of documenting ways to specify the path at load time, I realised that passing a path to the Language constructor is broken (fixed by b0a07c2) and I want to:

• Make sure it's properly fixed, and
• Make sure the tests still pass

Once that's done, I'll hand this over to y'all.

[ExplodingCabbage]

This is ready for review, I think.

[honnibal]

This is great — especially appreciate the docs!

[ines]

Agree – looks great and thanks for adding the sections to the docs and readme! 👍
Feel free to add yourself to the CONTRIBUTORS.md, btw.

[ExplodingCabbage]

Feel free to add yourself to the CONTRIBUTORS.md, btw.

I'll leave it to you to do so or not as you see fit - I feel like I probably shouldn't be making the decision about whether I get added or not. :)

[ines]

@ExplodingCabbage Sure, we'd be happy to add you! 😃

Just saw that your GitHub profile only shows your user name – which name would you like me to use for the CONTRIBUTORS.md? We usually list contributors as "Full name, username", but an alias instead of the full name would be fine, too.

[ExplodingCabbage]

@ines My real name is "Mark Amery" (and I've just updated my profile to show it). :)