CLI should reflect the full Javascript API

[ire-and-curses]

The command line interface to the API is a convenient way to access common functions (individual card retrieval, set retrieval). For non-Javascript developers, it also provides a way to cleanly access the API. It would be great if the full API (formats, types, sets) was supported. I personally would make use of at least the sets function if it were available to me. Also, if I add searching by type to my project (I may do), then the types function would be very useful.

[robdennis]

I was about to take a look at this myself since I too could use the {{sets}} function, but I see David created a branch for it already, presumably to take it on: https://github.com/davidchambers/tutor/tree/cli-feature-parity

[davidchambers]

I just turned this issue into a pull request. Feel free to pull from this branch and give the new functionality a spin. I'd appreciate a code review before I pull this change into master. At the very least, I'll take another look myself one day soon.

[robdennis]

One of the cli tests failed one time, then passed:

 $ tutor types prints types:
 Error: timeout of 10000ms exceeded
  at Object.<anonymous> (/Users/rad3/development/tutor/node_modules/mocha/lib/runnable.js:167:14)
  at Timer.list.ontimeout (timers.js:101:19)

not really about this change, but I think this help text for "set" is out of date: set [options] output one page of cards from the named set

since it's actually the entire set now

other than that, looks good to me.

[robdennis]

as an example of how easy this makes things for external consumers, I will now download all of gatherer to an 8.1 MB file using python 2.7 (maybe this is a decent README example?):

import os
import itertools
import logging
import json
import subprocess

logging.basicConfig(level=logging.DEBUG)
log = logging.getLogger()

TUTOR_DIR = os.path.realpath(os.path.dirname(__file__))
TUTOR_PATH = os.path.join(TUTOR_DIR, 'tutor')

def _query_tutor(command_parameters):
    """

    :param command_parameters:
    :return: the json-parsed output of the tutor command
    """

    command = [TUTOR_PATH, '--format', 'json'] + command_parameters

    try:
        return json.loads(subprocess.check_output(command).strip())
    except (subprocess.CalledProcessError, ValueError) as e:
        raise subprocess.CalledProcessError(u'problem with command',
                                            u' '.join(command))

def _query_tutor_for_cards_by_set(set_name):
    """
    Depends on: installation of the tutor CLI, that path in settings,
    the path to node in your $PATH environment variable

    :param set_name: the set's name, which will be passed to
        tutor's command line
    :return: a list of cards represented as dictionaries
    """

    log.debug('searching for set name: %r', set_name)
    return _query_tutor(['set', set_name])

def _query_tutor_for_all_sets():
    log.debug('searching for sets')
    return _query_tutor(['sets'])

def write_all_cards_from_gatherer_to_file():
    all_sets = _query_tutor_for_all_sets()
    cards = {
        card['name']: card
        for card in itertools.chain(*[_query_tutor_for_cards_by_set(set_name)
                                      for set_name in all_sets])
    }

    json.dump(cards, open('cards.json', 'wb'))

if __name__ == '__main__':

    write_all_cards_from_gatherer_to_file()

[davidchambers]

Thanks for taking a look at this pull request, @rdennis463. :)

not really about this change, but I think this help text for "set" is out of date: set [options] output one page of cards from the named set

since it's actually the entire set now

Good catch. I'll update this description.

The Python code above is very cool. I think the wiki (which I've just enabled) is the best place for it. Go ahead and create a page for it. It might be a good idea to link to it from the CLI section of the README, too.