gnd json

import simplejson
import pywikibot
import sys

filename = unicode(sys.argv[1])
instanceof = unicode(sys.argv[2])
replace = None
if len(sys.argv) == 4:
    replace = unicode(sys.argv[3])

print 'loading', filename
with open(filename) as f:
        a = simplejson.load(f)

site=pywikibot.Site('de', 'wikipedia')
repo=site.data_repository()
for i,j in a.items():
        s = pywikibot.Page(site, i)
        item = pywikibot.ItemPage.fromPage(s)
        item.get()
        skip = False
        if u'P227' in item.claims:
                for c in item.claims[u'P227']:
                    if c.target != j:
                        print '* found gnd', c.target, 'instead of', j, 'for', i, item.getID()
                skip = True
        if not skip:
                c = pywikibot.Claim(repo, u'P227')
                c.setTarget(j)
                item.addClaim(c)
        skip = False
        if u'P31' in item.claims:
                deletes = []
                for c in item.claims[u'P31']:
                    if replace and c.target.getID() == replace:
                        print '* deleting type', c.target, 'of', i, item.getID()
                        deletes.append(c)
                    elif c.target.getID() != instanceof:
                        print '* found type', c.target, 'of', i, item.getID()
                        skip = True
                    else:
                        skip = True
                if deletes:
                    item.removeClaims(deletes)
        if not skip:
                c2= pywikibot.Claim(repo, u'P31')
                b = pywikibot.ItemPage(repo, instanceof)
                c2.setTarget(b)
                item.addClaim(c2)