def get_langlinks(self):
"""
Retrieve the list of hyperlinks to translation of the current page
Returns
-------
langlinks : list
List of language codes (e.g "en", "fr", "es", "ru", etc)
todo: put a link to a page with the list of languages
"""
api = API(self.lang)
langlinks = []
params = {
"format": "json",
"action": "query",
"titles": self.title,
"prop": "langlinks",
"lllimit": 500
}
r = api.get(params)
# print r
page = r["query"]["pages"][ list(r["query"]["pages"].keys())[0] ]
if ("langlinks" in page):
langlinks += page["langlinks"]
return langlinks
def fetch_info(self, title, opt_params={ "prop": "info", "inprop": "url" }, lang="en"):
api = API(lang)
params = {
"format": "json",
"action": "query",
"titles": u""+title
# "rvprop": "content",
# "redirects": ""
}
params.update(opt_params)
r = api.get(params)
# print r.json()
pages = r["query"]["pages"]
self.page_id = list(pages.keys())[0]
self.title = pages[ self.page_id ]["title"]
self.lang = lang
self.url = pages[ self.page_id ]["fullurl"]
self.data.update(pages[ self.page_id ])
# print r.url
# print r.text
return r
def get_diff_full(self, rev_id=""):
"""
Retrieve the full json response from a request for diff.
Parameters
----------
rev_id : string, optional
If no revision id is supplied, the method retrieve the diff from the
current version of the page and compare it to its predecessor.
"""
api = API(self.lang)
q = {
"format": "json",
"action": "query",
"titles": self.title,
"redirects":"true",
#"rvparse" : "true",
"prop": "info|revisions",
"inprop": "url",
# "rvlimit": 1,
# "rvprop": "content",
"rvdiffto" : "prev"
}
if rev_id != "":
q.update({ "rvlimit":1, "rvstartid": rev_id })
r = api.get(q)
return r
def fetch_contribs(self):
""" get all contributions from a user """
api = API()
contribs = []
params = {
"action":"query",
"format": "json",
"list":"usercontribs",
"ucuser": self.name,
"uclimit": "500",
"continue": ""
}
while True:
r = api.get(params)
contribs += r["query"]["usercontribs"]
if "continue" in r:
# print r["continue"]
params.update(r["continue"])
else:
break
return contribs
def test_request_too_long_success():
api = API()
data = {'action': 'query', 'redirects': '', 'titles': u'\u0baa\u0bbf\u0baf\u0bcb\u0ba4\u0bcd\u0ba4\u0bb0\u0bcd \u0b87\u0bb2\u0bc0\u0b9a\u0bcd \u0b9a\u0bbe\u0baf\u0bcd\u0b95\u0bcd\u0b95\u0bcb\u0bb5\u0bcd\u0b9a\u0bc1\u0b95\u0bcd\u0b95\u0bbf|\u0b9a\u0bc1\u0bb1\u0bcd\u0bb1\u0bc1\u0bb2\u0bbe|\u0bb5\u0bcb\u0bb2\u0bcd\u0b95\u0bbe \u0b86\u0bb1\u0bc1|\u0baa\u0bc0\u0bb0\u0b99\u0bcd\u0b95\u0bbf \u0bb5\u0ba3\u0bcd\u0b9f\u0bbf|\u0bb0\u0baf\u0bbe\u0b9a\u0ba9\u0bcd \u0b92\u0baa\u0bcd\u0bb2\u0bbe\u0bb8\u0bcd\u0ba4\u0bc1|\u0b92\u0ba9\u0bcd\u0bb1\u0bc1\u0b95\u0bcd\u0b95\u0bc1 \u0bae\u0bc7\u0bb1\u0bcd\u0baa\u0b9f\u0bcd\u0b9f \u0b95\u0ba3\u0bcd\u0b9f\u0b99\u0bcd\u0b95\u0bb3\u0bbf\u0bb2\u0bcd \u0b85\u0bae\u0bc8\u0ba8\u0bcd\u0ba4\u0bc1\u0bb3\u0bcd\u0bb3 \u0ba8\u0bbe\u0b9f\u0bc1\u0b95\u0bb3\u0bbf\u0ba9\u0bcd \u0baa\u0b9f\u0bcd\u0b9f\u0bbf\u0baf\u0bb2\u0bcd|\u0ba8\u0bc6\u0baa\u0bcd\u0baa\u0bcb\u0bb2\u0bbf\u0baf\u0baa\u0bcd \u0baa\u0bcb\u0bb0\u0bcd\u0b95\u0bb3\u0bcd|\u0ba8\u0bc0\u0baa\u0bcd\u0baa\u0bcb \u0b86\u0bb1\u0bc1|\u0baa\u0bc6\u0bb0\u0bc1 \u0bb5\u0bc6\u0b9f\u0bbf\u0baa\u0bcd\u0baa\u0bc1\u0b95\u0bcd \u0b95\u0bcb\u0b9f\u0bcd\u0baa\u0bbe\u0b9f\u0bc1|\u0bae\u0bc8\u0b95\u0bcd\u0b95\u0bb2\u0bcd \u0b83\u0baa\u0bc6\u0bb2\u0bcd\u0baa\u0bcd\u0bb8\u0bcd|\u0baa\u0bbe\u0bb0\u0bcd\u0bae\u0bc1\u0bb2\u0bbe 1|\u0ba4\u0bbf\u0b9f\u0bcd\u0b9f\u0bae\u0bbf\u0b9f\u0bcd\u0b9f \u0baa\u0bca\u0bb0\u0bc1\u0bb3\u0bbe\u0ba4\u0bbe\u0bb0\u0bae\u0bcd|\u0b9c\u0baa\u0bcd\u0baa\u0bbe\u0ba9\u0bcd|\u0b9a\u0bbe\u0bb0\u0bcd\u0baa\u0bbe\u0ba3\u0bcd\u0bae\u0bc8 \u0bae\u0b95\u0bcd\u0b95\u0bb3\u0bbe\u0b9f\u0bcd\u0b9a\u0bbf|\u0baa\u0bc1\u0ba4\u0bbf\u0baf \u0bb5\u0bb3\u0bb0\u0bcd\u0b9a\u0bcd\u0b9a\u0bbf \u0bb5\u0b99\u0bcd\u0b95\u0bbf|\u0b86\u0bb0\u0bcd\u0bae\u0bc0\u0ba9\u0bbf\u0baf\u0bbe|\u0baa\u0bbf\u0bb1\u0bbf\u0bae\u0bcb\u0bb0\u0bcd\u0b9a\u0bc1\u0b95\u0bcd\u0b95\u0bbf \u0ba8\u0bbf\u0bb2\u0baa\u0bcd\u0baa\u0bb0\u0baa\u0bcd\u0baa\u0bc1|\u0b9f\u0bbf\u0bb2\u0bcd\u0bae\u0bbe \u0bb0\u0bc2\u0b9a\u0bc6\u0b83\u0baa\u0bcd|\u0b85\u0ba9\u0bc8\u0ba4\u0bcd\u0ba4\u0bc1 \u0b87\u0bb1\u0bc8\u0b95\u0bcd \u0b95\u0bca\u0bb3\u0bcd\u0b95\u0bc8|\u0baa\u0bc6\u0b9f\u0bcd\u0bb0\u0bcb\u0bb2\u0bbf\u0baf \u0b89\u0bb1\u0bcd\u0baa\u0ba4\u0bcd\u0ba4\u0bbf \u0b85\u0b9f\u0bbf\u0baa\u0bcd\u0baa\u0b9f\u0bc8\u0baf\u0bbf\u0bb2\u0bcd \u0ba8\u0bbe\u0b9f\u0bc1\u0b95\u0bb3\u0bbf\u0ba9\u0bcd \u0baa\u0b9f\u0bcd\u0b9f\u0bbf\u0baf\u0bb2\u0bcd|2014 \u0baa\u0bbf\u0bb0\u0bbf\u0b95\u0bcd\u0bb8\u0bcd \u0bae\u0bbe\u0ba8\u0bbe\u0b9f\u0bc1|\u0baa\u0bc6\u0bb2\u0bcd\u0b95\u0bcb\u0bb0\u0ba4\u0bcd \u0b92\u0baa\u0bcd\u0bb2\u0bbe\u0bb8\u0bcd\u0ba4\u0bc1|\u0baa\u0bbf\u0bb2\u0bbf\u0baa\u0bcd\u0baa\u0bc0\u0ba9\u0bcd\u0b9a\u0bc1|\u0baa\u0bc1\u0ba4\u0bc1\u0baa\u0bcd\u0baa\u0bbf\u0b95\u0bcd\u0b95\u0ba4\u0bcd\u0ba4\u0b95\u0bcd\u0b95 \u0ba8\u0bc0\u0bb0\u0bcd\u0bb5\u0bb3\u0bae\u0bcd \u0b95\u0bca\u0ba3\u0bcd\u0b9f \u0ba8\u0bbe\u0b9f\u0bc1\u0b95\u0bb3\u0bbf\u0ba9\u0bcd \u0baa\u0b9f\u0bcd\u0b9f\u0bbf\u0baf\u0bb2\u0bcd|\u0b85\u0ba3\u0bc1 \u0b86\u0baf\u0bc1\u0ba4 \u0b9a\u0b95\u0bcd\u0ba4\u0bbf\u0baf\u0bc1\u0b9f\u0bc8\u0baf \u0ba8\u0bbe\u0b9f\u0bc1\u0b95\u0bb3\u0bcd|\u0bb2\u0bbf\u0baf\u0bcb \u0b9f\u0bbe\u0bb2\u0bcd\u0bb8\u0bcd\u0b9f\u0bbe\u0baf\u0bcd|\u0b85\u0bae\u0bc6\u0bb0\u0bbf\u0b95\u0bcd\u0b95\u0b95\u0bcd \u0b95\u0bbe\u0b99\u0bcd\u0b95\u0bbf\u0bb0\u0b9a\u0bc1 \u0ba8\u0bc2\u0bb2\u0b95\u0bae\u0bcd|\u0b86\u0bb0\u0bcd\u0b95\u0bcd\u0b95\u0bbe\u0b99\u0bcd\u0b95\u0bc6\u0bb2\u0bcd\u0b9a\u0bbf\u0b95\u0bcd \u0b93\u0baa\u0bb2\u0bbe\u0b9a\u0bc1\u0ba4\u0bcd\u0ba4\u0bc1|\u0bae\u0ba9\u0bbf\u0ba4 \u0bb5\u0bb3\u0bb0\u0bcd\u0b9a\u0bcd\u0b9a\u0bbf\u0b9a\u0bcd \u0b9a\u0bc1\u0b9f\u0bcd\u0b9f\u0bc6\u0ba3\u0bcd|\u0bb2\u0bbf\u0ba4\u0bcd\u0ba4\u0bc1\u0bb5\u0bc7\u0ba9\u0bbf\u0baf\u0bbe|\u0baf\u0bc7\u0bb0\u0bcd\u0b9a\u0bbf|\u0ba8\u0bbe\u0b9f\u0bc1\u0b95\u0bb3\u0bbf\u0ba9\u0bcd \u0bb5\u0bc6\u0bb3\u0bbf\u0b95\u0bcd\u0b95\u0b9f\u0ba9\u0bcd \u0baa\u0b9f\u0bcd\u0b9f\u0bbf\u0baf\u0bb2\u0bcd|\u0bb5\u0bbe\u0b9a\u0bbf\u0b99\u0bcd\u0b9f\u0ba9\u0bcd, \u0b9f\u0bbf. \u0b9a\u0bbf.|\u0baa\u0bc6\u0bb0\u0bc1|\u0bae\u0bb1\u0bc1\u0bae\u0bb2\u0bb0\u0bcd\u0b9a\u0bcd\u0b9a\u0bbf (\u0b90\u0bb0\u0bcb\u0baa\u0bcd\u0baa\u0bbe)|\u0b95\u0bbf\u0baf\u0bc2\u0baa\u0bbe \u0b8f\u0bb5\u0bc1\u0b95\u0ba3\u0bc8 \u0ba8\u0bc6\u0bb0\u0bc1\u0b95\u0bcd\u0b95\u0b9f\u0bbf|\u0bae\u0bbe\u0bb8\u0bcd\u0b95\u0bcb \u0b85\u0bb0\u0b9a\u0bc1\u0baa\u0bcd \u0baa\u0bb2\u0bcd\u0b95\u0bb2\u0bc8\u0b95\u0bcd\u0b95\u0bb4\u0b95\u0bae\u0bcd|\u0ba4\u0bae\u0ba4\u0bc7\u0ba4\u0bb5\u0bcb \u0bb5\u0bbe\u0ba9\u0bc2\u0bb0\u0bcd\u0ba4\u0bbf \u0ba8\u0bbf\u0bb2\u0bc8\u0baf\u0bae\u0bcd|\u0bae\u0ba9\u0bbf\u0ba4 \u0bb5\u0bb3\u0bb0\u0bcd\u0b9a\u0bcd\u0b9a\u0bbf \u0b9a\u0bc1\u0b9f\u0bcd\u0b9f\u0bc6\u0ba3\u0bcd\u0ba3\u0bbf\u0ba9\u0bcd \u0baa\u0b9f\u0bbf \u0ba8\u0bbe\u0b9f\u0bc1\u0b95\u0bb3\u0bbf\u0ba9\u0bcd \u0baa\u0b9f\u0bcd\u0b9f\u0bbf\u0baf\u0bb2\u0bcd|\u0b9a\u0bc8\u0baa\u0bcd\u0baa\u0bbf\u0bb0\u0b9a\u0bc1|\u0bb2\u0bc6\u0ba9\u0bbf\u0ba9\u0bcd\u0b95\u0bbf\u0bb0\u0bbe\u0b9f\u0bcd \u0bae\u0bc1\u0bb1\u0bcd\u0bb1\u0bc1\u0b95\u0bc8|\u0baa\u0bbf\u0bb0\u0bc7\u0b9a\u0bbf\u0bb2\u0bcd|\u0baa\u0ba9\u0bcd\u0ba9\u0bbe\u0b9f\u0bcd\u0b9f\u0bc1 \u0bae\u0ba9\u0bcd\u0ba9\u0bbf\u0baa\u0bcd\u0baa\u0bc1 \u0b85\u0bb5\u0bc8|\u0b86\u0bb3\u0bcd\u0bb5\u0bc0\u0ba4 \u0bae\u0bca\u0ba4\u0bcd\u0ba4 \u0ba4\u0bc7\u0b9a\u0bbf\u0baf \u0b89\u0bb1\u0bcd\u0baa\u0ba4\u0bcd\u0ba4\u0bbf \u0b85\u0b9f\u0bbf\u0baa\u0bcd\u0baa\u0b9f\u0bc8\u0baf\u0bbf\u0bb2\u0bcd \u0ba8\u0bbe\u0b9f\u0bc1\u0b95\u0bb3\u0bcd \u0baa\u0b9f\u0bcd\u0b9f\u0bbf\u0baf\u0bb2\u0bcd|\u0bb2\u0bbf\u0baf\u0bcb\u0ba9\u0bbf\u0b9f\u0bcd \u0baa\u0bbf\u0bb0\u0bc6\u0bb7\u0bcd\u0ba9\u0bc6\u0bb5\u0bcd|\u0bae\u0bca\u0ba4\u0bcd\u0ba4 \u0ba4\u0bc7\u0b9a\u0bbf\u0baf \u0b89\u0bb1\u0bcd\u0baa\u0ba4\u0bcd\u0ba4\u0bbf \u0b85\u0b9f\u0bbf\u0baa\u0bcd\u0baa\u0b9f\u0bc8\u0baf\u0bbf\u0bb2\u0bcd \u0ba8\u0bbe\u0b9f\u0bc1\u0b95\u0bb3\u0bcd \u0baa\u0b9f\u0bcd\u0b9f\u0bbf\u0baf\u0bb2\u0bcd|\u0b9a\u0bae\u0bcd\u0baa\u0bcb|\u0b90\u0b95\u0bcd\u0b95\u0bbf\u0baf \u0ba8\u0bbe\u0b9f\u0bc1\u0b95\u0bb3\u0bcd \u0bae\u0ba9\u0bbf\u0ba4 \u0b89\u0bb0\u0bbf\u0bae\u0bc8\u0b95\u0bb3\u0bcd \u0b86\u0ba3\u0bc8\u0baf\u0bae\u0bcd|\u0b9c\u0bbf8', 'format': 'json'}
r = api.post(data)
assert "pages" in r["query"]
def get_links(self, extra_params={}):
"""
Retrieve links contained by a wikipedia page according to the API
Parameters
----------
extra_params : dict, optional
By default, the method will only retrieve links from the namespace 0
(usual pages) and skipped everything like templates, etc.
You can still get `the other namespaces
<http://en.wikipedia.org/wiki/Wikipedia:Namespace>`_ by updating the query
with an extra parameters.
>>> p.get_links({ plnamspace: 12 })
Returns
-------
links : list
See Also
--------
"""
links = []
api = API(self.lang)
params = {
"format": "json",
"action": "query",
"titles": self.title,
"prop": "links",
"pllimit": 500,
"plnamespace": 0,
"continue":""
}
params.update(extra_params)
while True:
r = api.get(params)
if "links" in r["query"]["pages"][ self.page_id ]:
l = r["query"]["pages"][ self.page_id ]["links"]
links.extend(l)
if "continue" in r:
params.update(r["continue"])
else:
break
return links
def get_revisions(self, extra_params={}):
"""
Parameters
----------
extra_params : dictionary
Returns
-------
revisions : list
todo: document revisions@get_revisions
"""
api = API()
params = {
"format": "json",
"action": "query",
"titles": self.title,
"prop": "revisions",
"rvprop": "user|userid|timestamp|size|ids|sha1|comment|content",
"rvlimit": "max",
"redirects": "",
"continue": ""
}
params.update(extra_params)
# print params
revisions = []
while True:
r = api.get(params)
# print r
pages = r["query"]["pages"]
page = pages[ list(pages.keys())[0] ]
revisions += page["revisions"]
if "continue" in r:
params.update(r["continue"])
else:
break
return revisions
def get_categories(self, extra_params={}):
"""
Retrieve a list of all categories used on the provided pages
Parameters
----------
extra_params : dict, optional
- http://www.mediawiki.org/wiki/API:Property/Categories
Returns
-------
links : list
See Also
--------
"""
categories = []
api = API(self.lang)
params = {
"format": "json",
"action": "query",
"titles": self.title,
"prop": "categories",
"cllimit": 500,
"continue":""
}
while True:
r = api.get(params)
if "categories" in r["query"]["pages"][ self.page_id ]:
c = r["query"]["pages"][ self.page_id ]["categories"]
categories.extend(c)
if "continue" in r:
params.update(r["continue"])
else:
break
return categories
def get_revisions_list(self, extra_params={}):
"""
Retrieve all the revisions and their info
Return
------
revisions : list
"""
api = API()
revisions = []
params = {
"format": "json",
"action": "query",
"titles": self.title,
"prop": "revisions",
"rvprop": "user|userid|timestamp|size|ids|sha1|comment",
"rvlimit": "max",
"redirects": "",
"continue": ""
}
while True:
r = api.get(params)
pages = r["query"]["pages"]
page = pages[ list(pages.keys())[0] ]
revisions += page["revisions"]
if "continue" in r:
params.update(r["continue"])
else:
break
return revisions