User:Edoderoobot/nl-street2en-description
Jump to navigation
Jump to search
import pywikibot
from pywikibot import pagegenerators
import pywikibot.data.wikidataquery as wdquery
import codecs #used in logfiles, unicoded strings
import sys
import datetime
from datetime import datetime, date, time
debugedo=True
debugedo=False
townlanguages=['nl','en','de','fr']
default_query='claim[31:79007] and claim[17:55]' #a street in the Netherlands
default_language = 'nl'
items2do = 0
def log_premature(itemno):
with codecs.open("nl-street2en-name-description.prelog.csv","a", encoding="utf-8") as logfile:
logfile.write('%s\n' % (itemno))
logfile.close
def logme(verbose, formatstring, *parameters):
with codecs.open("nl-street2en-name-description.log.csv", "a", encoding="utf-8") as logfile:
formattedstring = u'%s%s' % (formatstring, '\n')
try:
logfile.write(formattedstring % (parameters) )
except :
exctype, value = sys.exc_info()[:2]
print("1) Error writing to logfile on: [%s] [%s]" % (exctype, value))
verbose = True #now I want to see what!
logfile.close()
if verbose:
print(formatstring % (parameters))
class WDBot():
"""
A bot to process items on Wikidata
"""
def __init__(self, generator):
"""
Arguments:
* generator - A generator that yields itempage objects.
"""
self.generator = generator
self.repo = pywikibot.Site().data_repository()
def run(self,lng):
"""
Starts the robot.
"""
site = pywikibot.getSite('sv')
repo = site.data_repository()
items_found=0
for WDIquery in self.generator:
if items_found> 0 :
break
if WDIquery.exists() :
log_premature(WDIquery.title()) #log which item we process ... in case of an error, I know which item it is
WDIquery.get(get_redirect=True)
items_found += action_one_item(repo,WDIquery,lng)
def WikidataQueryItemPageGenerator(query, site=None):
"""Generate pages that result from the given WikidataQuery.
@param query: the WikidataQuery query string.
"""
global items2do
if site is None:
site = pywikibot.Site()
repo = site.data_repository()
wd_queryset = wdquery.QuerySet(query)
wd_query = wdquery.WikidataQuery(cacheMaxAge=0)
data = wd_query.query(wd_queryset)
items2do = data[u'status'][u'items']
pywikibot.output(u'retrieved %d items' % data[u'status'][u'items'])
for item in data[u'items']:
yield pywikibot.ItemPage(repo, u'Q' + str(item))
def get_town(repo,wditem,property):
town = u''
if (town=='') and (property in wditem.claims):
LNKtown = wditem.claims.get(property)[0].getTarget()
if not (LNKtown is None):
wdi_town = pywikibot.ItemPage(repo,LNKtown.title())
wdi_town.get(get_redirect=True)
for trylng in townlanguages:
if town=='':
if trylng in wdi_town.labels: town=wdi_town.labels[trylng]
return(town)
def action_one_item(repo, wditem, lng):
global items2do
items2do -= 1
str1 = '{:>10d}'.format(items2do)
str2 = '{:>10}'.format(wditem.title())
sys.stdout.write("\r%s%s" % (str1, str2)) #print how many items we still have to do ...
try:
lng = 'en'
data = {}
town = ''
if not(lng in wditem.descriptions): #not already filled
if town=='': town = get_town(repo,wditem,'P276')
if town=='': town = get_town(repo,wditem,'P131')
if town=='':
my_description = "street in the Netherlands"
else :
my_description = "street in %s, the Netherlands" % town
if (lng not in wditem.descriptions):
data.update({'descriptions': {lng:my_description}} )
if not (lng in wditem.labels): #name of town is still blank?
data.update({'labels': {lng:wditem.labels['nl']}})
if len(data)>0:
#print(data)
wditem.editEntity(data,summary=u'en-dutch-street, python code on https://goo .gl/EGL2ZE, logfile on https://goo .gl/BezTim')
else :
return(0)
logme(False,"%s|%s",datetime.now().strftime("%Y-%b-%d/%H:%M:%S"),wditem.title())
except ValueError:
return(0)
return(1)
def main():
lng = default_language
pigenerator = pagegenerators.PreloadingItemGenerator(pagegenerators.WikidataItemGenerator(WikidataQueryItemPageGenerator(query)))
wikidataBot = WDBot(pigenerator)
wikidataBot.run(lng)
if __name__ == "__main__":
if debugedo:
print("debug is on")
else:
print("Klaar voor de start")
main()