def get_globalusage(self, site, image, shared = False):
self.connect_http()
if type(site) is str:
hostname = site
apipath = '/w/api.php'
else:
hostname = site.hostname()
apipath = site.apipath()
kwargs = {'action': 'query', 'titles': u'File:' + image,
'prop': 'globalusage|imageinfo',
'iiprop': '', 'guprop': 'namespace', 'gulimit': '500'}
while True:
res = self.http.query_api(hostname, apipath, **kwargs)
if not res or not res['query'] or not res['query']['pages']:
return
if res['query']['pages'].values()[0].get('imagerepository') == 'local' and shared:
return
usages = res['query']['pages'].values()[0].get('globalusage', ())
for usage in usages:
title = usage['title'].replace(' ', '_')
namespace = int(usage['ns'])
site = family(usage['wiki'])
if namespace != 0:
yield site, (namespace, strip_ns(title), title)
else:
yield site, (namespace, title, title)
if 'globalusage' in res.get('query-continue', ()):
kwargs.update(res['query-continue']['globalusage'])
else:
return
def __init__(self, limit = 100,
mysql_default_server = 2, mysql_host_prefix = 'sql-s', mysql_kwargs = {},
no_db = False, use_autoconn = False,
http_retry_timeout = 30, http_max_retries = -1,
http_callback = lambda *args: None,
mysql_retry_timeout = 60,
mysql_max_retries = -1, mysql_callback = lambda *args: None):
self.http = None
self.http_retry_timeout = http_retry_timeout
self.http_max_retries = http_max_retries
self.http_callback = http_callback
if no_db: return
self.mysql_host_prefix = mysql_host_prefix
if 'host' in mysql_kwargs: del mysql_kwargs['host']
self.mysql_kwargs = mysql_kwargs
self.use_autoconn = use_autoconn
self.mysql_retry_timeout = mysql_retry_timeout
self.mysql_max_retries = mysql_max_retries
self.mysql_callback = mysql_callback
self.connections = []
# Mapping database name -> mysql connection
self.databases = {}
# Mapping server id -> mysql connection
self.servers = {}
# Mapping database name -> (lang, family)
self.sites = {}
self.unknown_families = []
# Mapping family name -> family object
self.known_families = {}
database, cursor = self.connect_mysql(mysql_host_prefix + str(mysql_default_server))
self.servers[mysql_default_server] = (database, cursor)
# Find where the databases are located
cursor.execute('SELECT dbname, domain, server FROM toolserver.wiki ORDER BY size DESC LIMIT %s', (limit, ))
for dbname, domain, server in cursor.fetchall():
if server not in self.servers:
self.servers[server] = self.connect_mysql(mysql_host_prefix + str(server))
# FIXME: wikimediafoundation!
try:
lang, fam = family(domain)
if fam not in self.known_families:
self.known_families[fam] = wikipedia.Family(fam, fatal = False)
except (RuntimeError, ValueError):
self.unknown_families.append(domain)
else:
self.sites[dbname] = (lang, fam)
self.databases[dbname] = self.servers[server]
def startElement(self, name, attr):
if name == "Family":
self.family = family()
self.family.name = attr.getValue("Name")
if name == "Product":
attributes = attr.getNames()
prod = product()
prod.name = attr.getValue("Name")
if "DisplayMode" in attributes:
prod.displayMode = attr.getValue("DisplayMode")
else:
prod.displayMode = "Visible"
self.family.products.append(prod)
def return_family_list(n=1000, m=5, seed=0):
# generate fam_list
G = nx.random_graphs.barabasi_albert_graph(n, m, seed)
fam_list = list()
adj = list(G.edges())
for g in G:
fam = family(label=g)
fam_list.append(fam)
for g_1, g_2 in adj:
f_1 = fam_list[g_1]
f_2 = fam_list[g_2]
f_1.relate(f_2)
return fam_list, G
def __init__(self, limit = 100,
mysql_default_server = 3, mysql_host_prefix = 'sql-s', mysql_host_suffix = '',
mysql_kwargs = {}, no_db = False, use_autoconn = False,
http_retry_timeout = 30, http_max_retries = -1,
http_callback = lambda *args: None,
mysql_retry_timeout = 60,
mysql_max_retries = -1, mysql_callback = lambda *args: None):
self.http = None
self.http_retry_timeout = http_retry_timeout
self.http_max_retries = http_max_retries
self.http_callback = http_callback
if no_db: return
self.mysql_host_prefix = mysql_host_prefix
self.mysql_kwargs = mysql_kwargs.copy() # To be safe
if 'host' in self.mysql_kwargs: del self.mysql_kwargs['host']
self.use_autoconn = use_autoconn
self.mysql_retry_timeout = mysql_retry_timeout
self.mysql_max_retries = mysql_max_retries
self.mysql_callback = mysql_callback
self.connections = []
# Mapping database name -> mysql connection
self.databases = {}
# Mapping server id -> mysql connection
self.servers = {}
# Mapping database name -> (lang, family)
self.sites = {}
self.domains = {}
self.unknown_families = []
# Mapping family name -> family object
self.known_families = {}
database, cursor = self.connect_mysql(mysql_host_prefix + str(mysql_default_server))
self.servers[mysql_default_server] = (database, cursor)
# Find where the databases are located
cursor.execute('SELECT dbname, domain, server FROM toolserver.wiki ORDER BY size DESC LIMIT %s', (limit, ))
for dbname, domain, server in cursor.fetchall():
if server not in self.servers:
self.servers[server] = self.connect_mysql(mysql_host_prefix + str(server) + mysql_host_suffix)
# FIXME: wikimediafoundation!
# TODO: This is one big mess
try:
lang, fam = family(domain)
if fam not in self.known_families:
self.known_families[fam] = wikipedia.Family(fam, fatal = False)
except (RuntimeError, ValueError, SyntaxError):
self.unknown_families.append(domain)
else:
self.sites[dbname] = (lang, fam)
self.databases[dbname] = self.servers[server]
self.domains[dbname] = domain