Vultr is planning on disabling support for TLSv1 within the coming months. Your API connections would need to start using v1.2 in order to remain compatible.

As a maintainer I want to publish the test coverage to give developers confidence in using the library.

Test coverage should be measured by travis and published to coveralls.
see: https://www.jeffknupp.com/blog/2013/08/16/open-sourcing-a-python-project-the-right-way/
heading: Continuous Integration with TravisCI

I cannot create new vps from my snapshot.
vultr.server.create(dcid, planid, osid, {'SNAPSHOTID':snapshot_id})
Everything is OK except this new vps is not from my snapshot.
sample snapshot_id = '3c35c279a5ba8'
Thanks!

As a maintainer I want to automate publication to pypi when pull requests are merged.

@see: http://docs.travis-ci.com/user/deployment/pypi/

At least in method server_start (as well as in another methods, for example, server_destroy) there is no subid argument which causes NameError.

Also I believe subid argument in server_list method should be optional.

Is this still under active support?

Hi,

Vultr api is rather unified, why not to try universalize it? I've written following code, which is working perfectly for me. And you don't need anything except this code, which is pretty convenient, as all new features will be automatically supported by the module.

#!/usr/bin/python
import json
import requests
from urlparse import urljoin
from os import path


class Vultr(object):
    def __init__(self, api):
        self.apikey = api
        self.url = "https://api.vultr.com/v1/"

    def __getattr__(self, key):
        return VultrMethod(self, key)


class VultrMethod(object):
    def __init__(self, client, method_name):
        self.client = client
        self.method_name = method_name

    def __getattr__(self, key):
        return VultrMethod(self.client, '.'.join((self.method_name, key)))

    def __call__(self, **kwargs):
        method = self.method_name.split('.')
        headers = {'API-Key': self.client.apikey}
        return getattr(requests, method[0])(urljoin(
            self.client.url, path.join(method[1], method[2])
            ), headers=headers, data=kwargs)


def main():
    vultr = Vultr(api="APIKEY")
    create_server = vultr.post.server.create(DCID=6, VPSPLANID=201, OSID=164, SNAPSHOTID="910594a93622c")
    print create_server.text


if __name__ == "__main__":
    main()

When making a request via POST, all options are passed as POST data. In the POST-dependent functions in Vultr's API, the API key should be passed as a URL parameter and the rest of the options as POST data. For example, see server_create: https://www.vultr.com/api/#server_create

Example Request:
POST https://api.vultr.com/v1/server/create?api_key=APIKEY
DCID=1
VPSPLANID=1
OSID=127

The API key is passed as the 'api_key' URL parameter, while DCID is passed as POST data.

The request function for python-vultr (which is a great module, by the way, thank you for putting it together) does not assemble the request this way; instead, it places all options in the params{} dict, which is passed as either GET or POST. This is the current implementation:

    def request(self, path, params={}, method='GET'):
        if not path.startswith('/'):
            path = '/' + path
        url = self.api_endpoint + path
        params['api_key'] = self.api_key

        try:
            if method == 'POST':
                resp = requests.post(url, data=params, timeout=60)
            elif method == 'GET':
                resp = requests.get(url, params=params, timeout=60)
            else:
                raise VultrError('Unsupported method %s' % method)

This means that all requests using method='POST' will not correctly pass the API key. Instead, the API key should be in params, while the rest of the options should be in data: something like requests.post(url, params=api_key, data=params, timeout=60)

I used a one-change workaround in lieu of a more elegant solution:

if method == 'POST':
                resp = requests.post(url, params={'api_key':params.get('api_key', None)}, data=params, timeout=60)

Not a tasteful fix, but I would rather change as little as possible pending a fix in the master branch.

Thanks again for your work on this module! It's been really helpful.

Hello. Got the problem

ERROR 403: Invalid or missing API key. Check that your API key is present and matches your assigned key.

changed

API_KEY = environ.get('VULTR_KEY')

to

API_KEY = 'VULTR_KEY'

now working.

Small problem, only applies to server_create(). The SSHKEYID is stored in the params{} dict in lowercase:

if sshkeyid is not None:
            params['sshkeyid'] = sshkeyid

Vultr ignores the value of the key called sshkeyid, as it's only looking for one called SSHKEYID. More a limitation on their end than in python-vultr, but an easy fix:

if sshkeyid is not None:
            params['SSHKEYID'] = sshkeyid

Was wondering how I'm able to pass label/extra parameters to snapshot.

The following method of adding params doesn't work. Was wondering if it had anything to do with how snapshot_create was written.

params = dict()
params['label] = 'new snapshot'
vultr.snapshot_create(subid,params=params)

Thanks!

We refactored all the optional request parameters into a params dictionary, but didn't update the README.rst at the same time. We probably also consider completing the documentation of the API parameters in the README.rst.

supercedes #30.

As a maintainer I want to automate documentation publishing to reduce the overhead of supporting the user community.

see: https://www.jeffknupp.com/blog/2013/08/16/open-sourcing-a-python-project-the-right-way/
Heading: Documentation with Sphinx

As a maintainer I want to minimize the work of publishing documentation to support the user community.

see: https://www.jeffknupp.com/blog/2013/08/16/open-sourcing-a-python-project-the-right-way/
heading: ReadTheDocs for Continuous Documentation Integration

Hi,

I've tried the example on the page along with the sample base_list.py. Neither works for me. Python 2.7 or 3.4 on FreeBSD.

2016-02-21 19:33:37,906 INFO [main():77] Vultr API Client Python Library
2016-02-21 19:33:37,907 INFO [main():78] URL: https://www.vultr.com/api/
Traceback (most recent call last):
  File "basic_list.py", line 81, in <module>
    main()
  File "basic_list.py", line 79, in main
    dump_info()
  File "basic_list.py", line 25, in dump_info
    vultr.account.info(), indent=2
AttributeError: 'Vultr' object has no attribute 'account'

Basically the same error on both. I'm setting my API key as an environment variable and it should be passed properly.

Is the example out of date or am I doing something wrong?

Thank you,
Teran

The print statements in vultr.py cause import errors, due to their syntax. Everything seems to work fine if those are reformatted:

if __name__ == '__main__':
    print("Vultr API Python Libary")
    print("http://vultr.com")

    api_key = ''
    if len(sys.argv) > 1:
        api_key = sys.argv[1]

    vultr = Vultr(api_key)

    print(vultr.iso_list())
    print(vultr.plans_list())
    print(vultr.regions_list())
    print(vultr.os_list())
    print(vultr.app_list())

    if api_key:
        print(vultr.account_info())

As a maintainer I want to continuously test against multiple versions of python, and make testing easier on contributing developers.

Tox seems to be the current standard in the python community.
see: https://www.jeffknupp.com/blog/2013/08/16/open-sourcing-a-python-project-the-right-way/
heading: Standardized Testing with Tox

Taking a look at the README.rst and the following calls are incomplete;

• dns.create_record
• server.create