def user_iter(lang='en', paginate=10000000):
contrib, conn = get_contributions_table()
count_query = select([func.count(contrib.c.id)],
contrib.c.lang == lang)
s = select([contrib],
contrib.c.lang == lang).order_by(
contrib.c.id).limit(paginate)
count = conn.execute(count_query).fetchall()[0][0]
#print >>sys.stderr, 'PAGES:', count
for offset in xrange(0, count, paginate):
rs = conn.execute(s.offset(offset))
for row in rs:
## row is a RowProxy object: supports dict and list methods
## convert it to dict to use with csv.DictWriter
v = dict(row)
del v['id']
del v['lang']
v['namespace_edits'] = deserialize(decompress(b64decode(
v['namespace_edits']
))) if v['namespace_edits'] is not None else None
yield v
def user_iter(lang='en', paginate=10000000):
contrib, conn = get_contributions_table()
count_query = select([func.count(contrib.c.id)],
contrib.c.lang == lang)
s = select([contrib],
contrib.c.lang == lang).order_by(
contrib.c.id).limit(paginate)
count = conn.execute(count_query).fetchall()[0][0]
#print >>sys.stderr, 'PAGES:', count
for offset in xrange(0, count, paginate):
rs = conn.execute(s.offset(offset))
for row in rs:
## row is a RowProxy object: supports dict and list methods
## convert it to dict to use with csv.DictWriter
v = dict(row)
del v['id']
del v['lang']
v['namespace_edits'] = deserialize(decompress(b64decode(
v['namespace_edits']
))) if v['namespace_edits'] is not None else None
yield v
def __init__(self, namespaces):
global ATTR_LEN
super(ContribDict, self).__init__()
self._namespaces = namespaces
ATTR_LEN = len(namespaces)
self._d_namespaces = dict([(name.decode('utf-8'), idx) for idx, (_,
name) in enumerate(namespaces)])
self._re_welcome = re2_compile_with_fallback(r'well?come', flags=re.I)
self._re_npov = re2_compile_with_fallback(r'[ n]pov', flags=re.I)
self._re_please = re2_compile_with_fallback(r'pl(s|z|ease)',
flags=re.I)
self._re_thanks = re2_compile_with_fallback(r'th(ank|anx|x)',
flags=re.I)
self._re_revert = re2_compile_with_fallback(r'(revert| rev )',
flags=re.I)
contributions, self.connection = get_contributions_table()
self.insert = contributions.insert()
def __init__(self, namespaces):
global ATTR_LEN
super(ContribDict, self).__init__()
self._namespaces = namespaces
ATTR_LEN = len(namespaces)
self._d_namespaces = dict([(name.decode('utf-8'), idx)
for idx, (_, name) in enumerate(namespaces)
])
self._re_welcome = re2_compile_with_fallback(r'well?come', flags=re.I)
self._re_npov = re2_compile_with_fallback(r'[ n]pov', flags=re.I)
self._re_please = re2_compile_with_fallback(r'pl(s|z|ease)',
flags=re.I)
self._re_thanks = re2_compile_with_fallback(r'th(ank|anx|x)',
flags=re.I)
self._re_revert = re2_compile_with_fallback(r'(revert| rev )',
flags=re.I)
contributions, self.connection = get_contributions_table()
self.insert = contributions.insert()