mirror of
https://github.com/moparisthebest/SickRage
synced 2024-11-05 17:05:03 -05:00
214 lines
7.5 KiB
Python
214 lines
7.5 KiB
Python
# -*- coding: utf-8 -*-
|
|
|
|
# ########################## Copyrights and license ############################
|
|
# #
|
|
# Copyright 2012 Vincent Jacques <vincent@vincent-jacques.net> #
|
|
# Copyright 2012 Zearin <zearin@gonk.net> #
|
|
# Copyright 2013 AKFish <akfish@gmail.com> #
|
|
# Copyright 2013 Bill Mill <bill.mill@gmail.com> #
|
|
# Copyright 2013 Vincent Jacques <vincent@vincent-jacques.net> #
|
|
# Copyright 2013 davidbrai <davidbrai@gmail.com> #
|
|
# #
|
|
# This file is part of PyGithub. http://jacquev6.github.com/PyGithub/ #
|
|
# #
|
|
# PyGithub is free software: you can redistribute it and/or modify it under #
|
|
# the terms of the GNU Lesser General Public License as published by the Free #
|
|
# Software Foundation, either version 3 of the License, or (at your option) #
|
|
# any later version. #
|
|
# #
|
|
# PyGithub is distributed in the hope that it will be useful, but WITHOUT ANY #
|
|
# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS #
|
|
# FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more #
|
|
# details. #
|
|
# #
|
|
# You should have received a copy of the GNU Lesser General Public License #
|
|
# along with PyGithub. If not, see <http://www.gnu.org/licenses/>. #
|
|
# #
|
|
# ##############################################################################
|
|
|
|
import github.GithubObject
|
|
|
|
|
|
class PaginatedListBase:
|
|
def __init__(self):
|
|
self.__elements = list()
|
|
|
|
def __getitem__(self, index):
|
|
assert isinstance(index, (int, slice))
|
|
if isinstance(index, (int, long)):
|
|
self.__fetchToIndex(index)
|
|
return self.__elements[index]
|
|
else:
|
|
return self._Slice(self, index)
|
|
|
|
def __iter__(self):
|
|
for element in self.__elements:
|
|
yield element
|
|
while self._couldGrow():
|
|
newElements = self._grow()
|
|
for element in newElements:
|
|
yield element
|
|
|
|
def _isBiggerThan(self, index):
|
|
return len(self.__elements) > index or self._couldGrow()
|
|
|
|
def __fetchToIndex(self, index):
|
|
while len(self.__elements) <= index and self._couldGrow():
|
|
self._grow()
|
|
|
|
def _grow(self):
|
|
newElements = self._fetchNextPage()
|
|
self.__elements += newElements
|
|
return newElements
|
|
|
|
class _Slice:
|
|
def __init__(self, theList, theSlice):
|
|
self.__list = theList
|
|
self.__start = theSlice.start or 0
|
|
self.__stop = theSlice.stop
|
|
self.__step = theSlice.step or 1
|
|
|
|
def __iter__(self):
|
|
index = self.__start
|
|
while not self.__finished(index):
|
|
if self.__list._isBiggerThan(index):
|
|
yield self.__list[index]
|
|
index += self.__step
|
|
else:
|
|
return
|
|
|
|
def __finished(self, index):
|
|
return self.__stop is not None and index >= self.__stop
|
|
|
|
|
|
class PaginatedList(PaginatedListBase):
|
|
"""
|
|
This class abstracts the `pagination of the API <http://developer.github.com/v3/#pagination>`_.
|
|
|
|
You can simply enumerate through instances of this class::
|
|
|
|
for repo in user.get_repos():
|
|
print repo.name
|
|
|
|
You can also index them or take slices::
|
|
|
|
second_repo = user.get_repos()[1]
|
|
first_repos = user.get_repos()[:10]
|
|
|
|
If you want to iterate in reversed order, just do::
|
|
|
|
for repo in user.get_repos().reversed:
|
|
print repo.name
|
|
|
|
And if you really need it, you can explicitely access a specific page::
|
|
|
|
some_repos = user.get_repos().get_page(0)
|
|
some_other_repos = user.get_repos().get_page(3)
|
|
"""
|
|
|
|
def __init__(self, contentClass, requester, firstUrl, firstParams):
|
|
PaginatedListBase.__init__(self)
|
|
self.__requester = requester
|
|
self.__contentClass = contentClass
|
|
self.__firstUrl = firstUrl
|
|
self.__firstParams = firstParams or ()
|
|
self.__nextUrl = firstUrl
|
|
self.__nextParams = firstParams or {}
|
|
if self.__requester.per_page != 30:
|
|
self.__nextParams["per_page"] = self.__requester.per_page
|
|
self._reversed = False
|
|
self.__totalCount = None
|
|
|
|
@property
|
|
def totalCount(self):
|
|
if not self.__totalCount:
|
|
self._grow()
|
|
|
|
return self.__totalCount
|
|
|
|
def _getLastPageUrl(self):
|
|
headers, data = self.__requester.requestJsonAndCheck(
|
|
"GET",
|
|
self.__firstUrl,
|
|
parameters=self.__nextParams
|
|
)
|
|
links = self.__parseLinkHeader(headers)
|
|
lastUrl = links.get("last")
|
|
return lastUrl
|
|
|
|
@property
|
|
def reversed(self):
|
|
r = PaginatedList(self.__contentClass, self.__requester, self.__firstUrl, self.__firstParams)
|
|
r.__reverse()
|
|
return r
|
|
|
|
def __reverse(self):
|
|
self._reversed = True
|
|
lastUrl = self._getLastPageUrl()
|
|
if lastUrl:
|
|
self.__nextUrl = lastUrl
|
|
|
|
def _couldGrow(self):
|
|
return self.__nextUrl is not None
|
|
|
|
def _fetchNextPage(self):
|
|
headers, data = self.__requester.requestJsonAndCheck(
|
|
"GET",
|
|
self.__nextUrl,
|
|
parameters=self.__nextParams
|
|
)
|
|
|
|
self.__nextUrl = None
|
|
if len(data) > 0:
|
|
links = self.__parseLinkHeader(headers)
|
|
if self._reversed:
|
|
if "prev" in links:
|
|
self.__nextUrl = links["prev"]
|
|
elif "next" in links:
|
|
self.__nextUrl = links["next"]
|
|
self.__nextParams = None
|
|
|
|
if 'items' in data:
|
|
self.__totalCount = data['total_count']
|
|
data = data["items"]
|
|
|
|
content = [
|
|
self.__contentClass(self.__requester, headers, element, completed=False)
|
|
for element in data if element is not None
|
|
]
|
|
if self._reversed:
|
|
return content[::-1]
|
|
return content
|
|
|
|
def __parseLinkHeader(self, headers):
|
|
links = {}
|
|
if "link" in headers:
|
|
linkHeaders = headers["link"].split(", ")
|
|
for linkHeader in linkHeaders:
|
|
(url, rel) = linkHeader.split("; ")
|
|
url = url[1:-1]
|
|
rel = rel[5:-1]
|
|
links[rel] = url
|
|
return links
|
|
|
|
def get_page(self, page):
|
|
params = dict(self.__firstParams)
|
|
if page != 0:
|
|
params["page"] = page + 1
|
|
if self.__requester.per_page != 30:
|
|
params["per_page"] = self.__requester.per_page
|
|
headers, data = self.__requester.requestJsonAndCheck(
|
|
"GET",
|
|
self.__firstUrl,
|
|
parameters=params
|
|
)
|
|
|
|
if 'items' in data:
|
|
self.__totalCount = data['total_count']
|
|
data = data["items"]
|
|
|
|
return [
|
|
self.__contentClass(self.__requester, headers, element, completed=False)
|
|
for element in data
|
|
]
|