def extract_entries(rows):
for cols in rows:
labels = {}
for (i,lab) in enumerate(header):
if lab != None and (cols[i] == '1' or cols[i] == '0'):
labels[lab] = bool(int(cols[i]))
ids = []
if options.idfields != None:
ids = [myutils.int_if(cols[i]) for i in options.idfields]
else:
ids = [idbase]
text = None
ent = {'entry': {'id': dict(zip([header[i] for i in options.idfields], ids))}}
if options.labels != None and labels != {}:
ent.update({'labels': labels})
ent['entry'].update({'content': {'added': [],
'removed': []},
'comment': ''})
if options.textfield != None:
ent['entry']['content']['added'] = [cols[options.textfield].decode('UTF-8')]
yield (cols,ent)
db = collection['talkpage_diffs_raw']
if options.idfields != None:
digits = re.compile('\d+')
table = filter(lambda x: reduce(lambda s,y: s and y, [digits.match(x[i]) for i in options.idfields]), table)
if options.overwrite and options.idfields != None:
# get existing entries
existings = {}
query = {}
for i in options.idfields:
query['entry.id.' + header[i]] = {'$exists': True}
query['entry.content'] = {'$exists': True}
for x in db.find(query, {'entry.id': 1, 'entry.content': 1}):
existings[tuple([x['entry']['id'][header[i]] for i in options.idfields])] = True
table = filter(lambda x: not existings.has_key(tuple([myutils.int_if(x[i]) for i in options.idfields])), table)
if options.overwrite and options.idfields == None:
print >>sys.stderr, 'overwrite requires idfields'
def extract_entries(rows):
for cols in rows:
labels = {}
for (i,lab) in enumerate(header):
if lab != None and (cols[i] == '1' or cols[i] == '0'):
labels[lab] = bool(int(cols[i]))
ids = []
if options.idfields != None:
ids = [myutils.int_if(cols[i]) for i in options.idfields]
else:
ids = [idbase]
