def region(self, ctr_lat, ctr_lon, mcc, cells):
region = None
regions = [cell.region for cell in cells]
unique_regions = set(regions)
if len(unique_regions) == 1:
region = regions[0]
else:
# Choose the area region based on the majority of cells
# inside each region.
grouped_regions = defaultdict(int)
for reg in regions:
grouped_regions[reg] += 1
max_count = max(grouped_regions.values())
max_regions = sorted([k for k, v in grouped_regions.items()
if v == max_count])
# If we get a tie here, randomly choose the first.
region = max_regions[0]
if len(max_regions) > 1:
# Try to break the tie based on the center of the area,
# but keep the randomly chosen region if this fails.
area_region = GEOCODER.region_for_cell(
ctr_lat, ctr_lon, mcc)
if area_region is not None:
region = area_region
return region
def validate(cls, entry, _raise_invalid=False, **kw):
validated = super(CellShard, cls).validate(
entry, _raise_invalid=_raise_invalid, **kw
)
if validated is not None:
if "cellid" not in validated:
validated["cellid"] = (
validated["radio"],
validated["mcc"],
validated["mnc"],
validated["lac"],
validated["cid"],
)
if (
("region" not in validated or not validated["region"])
and validated["lat"] is not None
and validated["lon"] is not None
):
validated["region"] = GEOCODER.region_for_cell(
validated["lat"], validated["lon"], validated["mcc"]
)
return validated
def upgrade():
bind = op.get_bind()
from ichnaea.geocode import GEOCODER
log.info('Update cell_area regions.')
stmt = '''\
UPDATE cell_area
SET `region` = "{code}"
WHERE `radio` IN (0, 1, 2, 3) AND `mcc` = {mcc} AND `region` IS NULL
'''
length = len(MCC_TO_REGION)
for i, (mcc, code) in enumerate(MCC_TO_REGION.items()):
op.execute(sa.text(stmt.format(code=code, mcc=mcc)))
if (i > 0 and i % 10 == 0):
log.info('Updated %s of %s regions.', i, length)
log.info('Updated %s of %s regions.', length, length)
stmt = 'SELECT COUNT(*) FROM cell_area WHERE region IS NULL'
todo = bind.execute(stmt).fetchone()[0]
log.info('Updating remaining %s areas.', todo)
stmt = '''\
SELECT HEX(`areaid`), `mcc`, `lat`, `lon`
FROM cell_area
WHERE `region` IS NULL
'''
rows = bind.execute(stmt).fetchall()
areas = {}
i = 0
for row in rows:
if (i > 0 and i % 5000 == 0):
log.info('Geocoded %s of %s areas.', i, todo)
code = GEOCODER.region_for_cell(row.lat, row.lon, row.mcc)
if code not in areas:
areas[code] = []
areas[code].append(row[0])
i += 1
log.info('Geocoded %s of %s areas.', todo, todo)
stmt = '''\
UPDATE cell_area
SET `region` = "{code}"
WHERE `areaid` in ({ids})
'''
for code, areaids in areas.items():
if not code:
continue
ids = 'UNHEX("' + '"), UNHEX("'.join(areaids) + '")'
op.execute(sa.text(stmt.format(code=code, ids=ids)))
log.info('Updated %s region.', code)
def upgrade():
bind = op.get_bind()
from ichnaea.geocode import GEOCODER
log.info('Update cell_area regions.')
stmt = '''\
UPDATE cell_area
SET `region` = "{code}"
WHERE `radio` IN (0, 1, 2, 3) AND `mcc` = {mcc} AND `region` IS NULL
'''
length = len(MCC_TO_REGION)
for i, (mcc, code) in enumerate(MCC_TO_REGION.items()):
op.execute(sa.text(stmt.format(code=code, mcc=mcc)))
if (i > 0 and i % 10 == 0):
log.info('Updated %s of %s regions.', i, length)
log.info('Updated %s of %s regions.', length, length)
stmt = 'SELECT COUNT(*) FROM cell_area WHERE region IS NULL'
todo = bind.execute(stmt).fetchone()[0]
log.info('Updating remaining %s areas.', todo)
stmt = '''\
SELECT HEX(`areaid`), `mcc`, `lat`, `lon`
FROM cell_area
WHERE `region` IS NULL
'''
rows = bind.execute(stmt).fetchall()
areas = {}
i = 0
for row in rows:
if (i > 0 and i % 5000 == 0):
log.info('Geocoded %s of %s areas.', i, todo)
code = GEOCODER.region_for_cell(row.lat, row.lon, row.mcc)
if code not in areas:
areas[code] = []
areas[code].append(row[0])
i += 1
log.info('Geocoded %s of %s areas.', todo, todo)
stmt = '''\
UPDATE cell_area
SET `region` = "{code}"
WHERE `areaid` in ({ids})
'''
for code, areaids in areas.items():
if not code:
continue
ids = 'UNHEX("' + '"), UNHEX("'.join(areaids) + '")'
op.execute(sa.text(stmt.format(code=code, ids=ids)))
log.info('Updated %s region.', code)
def validate(cls, entry, _raise_invalid=False, **kw):
validated = super(CellAreaMixin, cls).validate(
entry, _raise_invalid=_raise_invalid, **kw)
if validated is not None and 'areaid' not in validated:
validated['areaid'] = (
validated['radio'],
validated['mcc'],
validated['mnc'],
validated['lac'],
)
if (('region' not in validated or not validated['region']) and
validated['lat'] is not None and
validated['lon'] is not None):
validated['region'] = GEOCODER.region_for_cell(
validated['lat'], validated['lon'], validated['mcc'])
return validated
def validate(cls, entry, _raise_invalid=False, **kw):
validated = super(CellAreaMixin, cls).validate(
entry, _raise_invalid=_raise_invalid, **kw)
if validated is not None and 'areaid' not in validated:
validated['areaid'] = (
validated['radio'],
validated['mcc'],
validated['mnc'],
validated['lac'],
)
if (('region' not in validated or not validated['region']) and
validated['lat'] is not None and
validated['lon'] is not None):
validated['region'] = GEOCODER.region_for_cell(
validated['lat'], validated['lon'], validated['mcc'])
return validated
