Source code for openquake.calculators.extract
# -*- coding: utf-8 -*-
# vim: tabstop=4 shiftwidth=4 softtabstop=4
#
# Copyright (C) 2017-2018 GEM Foundation
#
# OpenQuake is free software: you can redistribute it and/or modify it
# under the terms of the GNU Affero General Public License as published
# by the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# OpenQuake is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU Affero General Public License for more details.
#
# You should have received a copy of the GNU Affero General Public License
# along with OpenQuake. If not, see <http://www.gnu.org/licenses/>.
import collections
import operator
import logging
from h5py._hl.dataset import Dataset
from h5py._hl.group import Group
import numpy
try:
from functools import lru_cache
except ImportError:
from openquake.risklib.utils import memoized
else:
memoized = lru_cache(100)
from openquake.baselib.hdf5 import ArrayWrapper, vstr
from openquake.baselib.general import group_array, deprecated
from openquake.baselib.python3compat import encode, decode
from openquake.calculators import getters
from openquake.calculators.export.loss_curves import get_loss_builder
from openquake.commonlib import calc, util
U32 = numpy.uint32
F32 = numpy.float32
F64 = numpy.float64
[docs]def barray(iterlines):
"""
Array of bytes
"""
lst = [line.encode('utf-8') for line in iterlines]
arr = numpy.array(lst)
return arr
[docs]def extract_(dstore, dspath):
"""
Extracts an HDF5 path object from the datastore, for instance
extract('sitecol', dstore). It is also possibly to extract the
attributes, for instance with extract('sitecol.attrs', dstore).
"""
obj = dstore[dspath]
if isinstance(obj, Dataset):
return ArrayWrapper(obj.value, obj.attrs)
elif isinstance(obj, Group):
return ArrayWrapper(numpy.array(list(obj)), obj.attrs)
else:
return obj
[docs]class Extract(dict):
"""
A callable dictionary of functions with a single instance called
`extract`. Then `extract(dstore, fullkey)` dispatches to the function
determined by the first part of `fullkey` (a slash-separated
string) by passing as argument the second part of `fullkey`.
For instance extract(dstore, 'sitecol), extract(dstore, 'asset_values/0')
etc.
"""
[docs] def add(self, key, cache=False):
def decorator(func):
self[key] = memoized(func) if cache else func
return func
return decorator
def __call__(self, dstore, key):
try:
k, v = key.split('/', 1)
except ValueError: # no slashes
k, v = key, ''
if k in self:
return self[k](dstore, v)
else:
return extract_(dstore, key)
extract = Extract()
# used by the QGIS plugin
[docs]@extract.add('realizations')
def extract_realizations(dstore, dummy):
"""
Extract an array of realizations. Use it as /extract/realizations
"""
return dstore['csm_info'].rlzs
[docs]@extract.add('asset_values', cache=True)
def extract_asset_values(dstore, sid):
"""
Extract an array of asset values for the given sid. Use it as
/extract/asset_values/0
:returns:
(aid, loss_type1, ..., loss_typeN) composite array
"""
if sid:
return extract(dstore, 'asset_values')[int(sid)]
assetcol = extract(dstore, 'assetcol')
asset_refs = assetcol.asset_refs
assets_by_site = assetcol.assets_by_site()
lts = assetcol.loss_types
time_event = assetcol.time_event
dt = numpy.dtype([('aref', asset_refs.dtype), ('aid', numpy.uint32)] +
[(str(lt), numpy.float32) for lt in lts])
data = []
for assets in assets_by_site:
vals = numpy.zeros(len(assets), dt)
for a, asset in enumerate(assets):
vals[a]['aref'] = asset_refs[a]
vals[a]['aid'] = asset.ordinal
for lt in lts:
vals[a][lt] = asset.value(lt, time_event)
data.append(vals)
return data
[docs]@extract.add('asset_tags')
def extract_asset_tags(dstore, tagname):
"""
Extract an array of asset tags for the given tagname. Use it as
/extract/asset_tags or /extract/asset_tags/taxonomy
"""
tagcol = dstore['assetcol/tagcol']
if tagname:
yield tagname, barray(tagcol.gen_tags(tagname))
for tagname in tagcol.tagnames:
yield tagname, barray(tagcol.gen_tags(tagname))
[docs]@extract.add('hazard')
def extract_hazard(dstore, what):
"""
Extracts hazard curves and possibly hazard maps and/or uniform hazard
spectra. Use it as /extract/hazard/mean or /extract/hazard/rlz-0, etc
"""
oq = dstore['oqparam']
sitecol = dstore['sitecol']
rlzs_assoc = dstore['csm_info'].get_rlzs_assoc()
yield 'sitecol', sitecol
yield 'oqparam', oq
yield 'imtls', oq.imtls
yield 'realizations', dstore['csm_info'].rlzs
yield 'checksum32', dstore['/'].attrs['checksum32']
nsites = len(sitecol)
M = len(oq.imtls)
P = len(oq.poes)
for statname, pmap in getters.PmapGetter(dstore, rlzs_assoc).items(what):
for imt in oq.imtls:
key = 'hcurves/%s/%s' % (imt, statname)
arr = numpy.zeros((nsites, len(oq.imtls[imt])))
for sid in pmap:
arr[sid] = pmap[sid].array[oq.imtls(imt), 0]
logging.info('extracting %s', key)
yield key, arr
try:
hmap = dstore['hmaps/' + statname]
except KeyError: # for statname=rlz-XXX
hmap = calc.make_hmap(pmap, oq.imtls, oq.poes)
for p, poe in enumerate(oq.poes):
key = 'hmaps/poe-%s/%s' % (poe, statname)
arr = numpy.zeros((nsites, M))
idx = [m * P + p for m in range(M)]
for sid in pmap:
arr[sid] = hmap[sid].array[idx, 0]
logging.info('extracting %s', key)
yield key, arr
[docs]def get_mesh(sitecol, complete=True):
"""
:returns:
a lon-lat or lon-lat-depth array depending if the site collection
is at sea level or not
"""
sc = sitecol.complete if complete else sitecol
if sc.at_sea_level():
mesh = numpy.zeros(len(sc), [('lon', F64), ('lat', F64)])
mesh['lon'] = sc.lons
mesh['lat'] = sc.lats
else:
mesh = numpy.zeros(len(sc), [('lon', F64), ('lat', F64),
('depth', F64)])
mesh['lon'] = sc.lons
mesh['lat'] = sc.lats
mesh['depth'] = sc.depths
return mesh
[docs]def hazard_items(dic, mesh, *extras, **kw):
"""
:param dic: dictionary of arrays of the same shape
:param mesh: a mesh array with lon, lat fields of the same length
:param extras: optional triples (field, dtype, values)
:param kw: dictionary of parameters (like investigation_time)
:returns: a list of pairs (key, value) suitable for storage in .npz format
"""
for item in kw.items():
yield item
arr = dic[next(iter(dic))]
dtlist = [(str(field), arr.dtype) for field in sorted(dic)]
for field, dtype, values in extras:
dtlist.append((str(field), dtype))
array = numpy.zeros(arr.shape, dtlist)
for field in dic:
array[field] = dic[field]
for field, dtype, values in extras:
array[field] = values
yield 'all', util.compose_arrays(mesh, array)
def _get_dict(dstore, name, imts, imls):
dic = {}
dtlist = []
for imt, imls in zip(imts, imls):
dt = numpy.dtype([(str(iml), F32) for iml in imls])
dtlist.append((imt, dt))
for statname, curves in dstore[name].items():
dic[statname] = curves.value.view(dtlist).flatten()
return dic
[docs]@extract.add('hcurves')
def extract_hcurves(dstore, what):
"""
Extracts hazard curves. Use it as /extract/hcurves/mean or
/extract/hcurves/rlz-0, /extract/hcurves/stats, /extract/hcurves/rlzs etc
"""
if 'hcurves' not in dstore:
return []
oq = dstore['oqparam']
sitecol = dstore['sitecol']
mesh = get_mesh(sitecol, complete=False)
dic = _get_dict(dstore, 'hcurves', oq.imtls, oq.imtls.values())
return hazard_items(dic, mesh, investigation_time=oq.investigation_time)
[docs]@extract.add('hmaps')
def extract_hmaps(dstore, what):
"""
Extracts hazard maps. Use it as /extract/hmaps/mean or
/extract/hmaps/rlz-0, etc
"""
oq = dstore['oqparam']
sitecol = dstore['sitecol']
mesh = get_mesh(sitecol)
dic = _get_dict(dstore, 'hmaps', oq.imtls, [oq.poes] * len(oq.imtls))
return hazard_items(dic, mesh, investigation_time=oq.investigation_time)
[docs]@extract.add('uhs')
def extract_uhs(dstore, what):
"""
Extracts uniform hazard spectra. Use it as /extract/uhs/mean or
/extract/uhs/rlz-0, etc
"""
oq = dstore['oqparam']
imts = oq.imt_periods()
mesh = get_mesh(dstore['sitecol'])
rlzs_assoc = dstore['csm_info'].get_rlzs_assoc()
dic = {}
imts_dt = numpy.dtype([(str(imt), F32) for imt in imts])
uhs_dt = numpy.dtype([(str(poe), imts_dt) for poe in oq.poes])
for name, hcurves in getters.PmapGetter(dstore, rlzs_assoc).items(what):
hmap = calc.make_hmap_array(hcurves, oq.imtls, oq.poes, len(mesh))
uhs = numpy.zeros(len(hmap), uhs_dt)
for field in hmap.dtype.names:
imt, poe = field.split('-')
if imt in imts_dt.names:
uhs[poe][imt] = hmap[field]
dic[name] = uhs
return hazard_items(dic, mesh, investigation_time=oq.investigation_time)
def _agg(losses, idxs):
shp = losses.shape[1:]
if not idxs:
# no intersection, return a 0-dim matrix
return numpy.zeros((0,) + shp, losses.dtype)
# numpy.array wants lists, not sets, hence the sorted below
return losses[numpy.array(sorted(idxs))].sum(axis=0)
def _filter_agg(assetcol, losses, selected, stats=''):
# losses is an array of shape (A, ..., R) with A=#assets, R=#realizations
aids_by_tag = assetcol.get_aids_by_tag()
idxs = set(range(len(assetcol)))
tagnames = []
for tag in selected:
tagname, tagvalue = tag.split('=', 1)
if tagvalue == '*':
tagnames.append(tagname)
else:
idxs &= aids_by_tag[tag]
if len(tagnames) > 1:
raise ValueError('Too many * as tag values in %s' % tagnames)
elif not tagnames: # return an array of shape (..., R)
return ArrayWrapper(
_agg(losses, idxs), dict(selected=encode(selected), stats=stats))
else: # return an array of shape (T, ..., R)
[tagname] = tagnames
_tags = list(assetcol.tagcol.gen_tags(tagname))
all_idxs = [idxs & aids_by_tag[t] for t in _tags]
# NB: using a generator expression for all_idxs caused issues (?)
data, tags = [], []
for idxs, tag in zip(all_idxs, _tags):
agglosses = _agg(losses, idxs)
if len(agglosses):
data.append(agglosses)
tags.append(tag)
return ArrayWrapper(
numpy.array(data),
dict(selected=encode(selected), tags=encode(tags), stats=stats))
[docs]def get_loss_type_tags(what):
try:
loss_type, query_string = what.rsplit('?', 1)
except ValueError: # no question mark
loss_type, query_string = what, ''
tags = query_string.split('&') if query_string else []
return loss_type, tags
[docs]@extract.add('agg_losses')
def extract_agg_losses(dstore, what):
"""
Aggregate losses of the given loss type and tags. Use it as
/extract/agg_losses/structural?taxonomy=RC&zipcode=20126
/extract/agg_losses/structural?taxonomy=RC&zipcode=*
:returns:
an array of shape (T, R) if one of the tag names has a `*` value
an array of shape (R,), being R the number of realizations
an array of length 0 if there is no data for the given tags
"""
loss_type, tags = get_loss_type_tags(what)
if not loss_type:
raise ValueError('loss_type not passed in agg_losses/<loss_type>')
l = dstore['oqparam'].lti[loss_type]
if 'losses_by_asset' in dstore: # scenario_risk
stats = None
losses = dstore['losses_by_asset'][:, :, l]['mean']
elif 'avg_losses-stats' in dstore: # event_based_risk, classical_risk
stats = dstore['avg_losses-stats'].attrs['stats']
losses = dstore['avg_losses-stats'][:, :, l]
elif 'avg_losses-rlzs' in dstore: # event_based_risk, classical_risk
stats = [b'mean']
losses = dstore['avg_losses-rlzs'][:, :, l]
else:
raise KeyError('No losses found in %s' % dstore)
return _filter_agg(dstore['assetcol'], losses, tags, stats)
[docs]@extract.add('agg_damages')
def extract_agg_damages(dstore, what):
"""
Aggregate damages of the given loss type and tags. Use it as
/extract/agg_damages/structural?taxonomy=RC&zipcode=20126
:returns:
array of shape (R, D), being R the number of realizations and D
the number of damage states or array of length 0 if there is no
data for the given tags
"""
loss_type, tags = get_loss_type_tags(what)
if 'dmg_by_asset' in dstore: # scenario_damage
losses = dstore['dmg_by_asset'][loss_type]['mean']
else:
raise KeyError('No damages found in %s' % dstore)
return _filter_agg(dstore['assetcol'], losses, tags)
def _get_curves(curves, li):
shp = curves.shape + curves.dtype.shape
return curves.value.view(F32).reshape(shp)[:, :, :, li]
[docs]@extract.add('agg_curves')
def extract_agg_curves(dstore, what):
"""
Aggregate loss curves of the given loss type and tags for
event based risk calculations. Use it as
/extract/agg_curves/structural?taxonomy=RC&zipcode=20126
:returns:
array of shape (S, P), being P the number of return periods
and S the number of statistics
"""
oq = dstore['oqparam']
loss_type, tags = get_loss_type_tags(what)
if 'curves-stats' in dstore: # event_based_risk
losses = _get_curves(dstore['curves-stats'], oq.lti[loss_type])
stats = dstore['curves-stats'].attrs['stats']
elif 'curves-rlzs' in dstore: # event_based_risk, 1 rlz
losses = _get_curves(dstore['curves-rlzs'], oq.lti[loss_type])
assert losses.shape[1] == 1, 'There must be a single realization'
stats = [b'mean'] # suitable to be stored as hdf5 attribute
else:
raise KeyError('No curves found in %s' % dstore)
res = _filter_agg(dstore['assetcol'], losses, tags, stats)
cc = dstore['assetcol/cost_calculator']
res.units = cc.get_units(loss_types=[loss_type])
res.return_periods = get_loss_builder(dstore).return_periods
return res
[docs]@extract.add('aggregate_by')
def extract_aggregate_by(dstore, what):
"""
/extract/aggregate_by/taxonomy,occupancy/curves/structural
yield pairs (<stat>, <array of shape (T, O, S, P)>)
/extract/aggregate_by/taxonomy,occupancy/avg_losses/structural
yield pairs (<stat>, <array of shape (T, O, S)>)
"""
try:
tagnames, name, loss_type = what.split('/')
except ValueError: # missing '/' at the end
tagnames, name = what.split('/')
loss_type = ''
assert name in ('avg_losses', 'curves'), name
tagnames = tagnames.split(',')
assetcol = dstore['assetcol']
oq = dstore['oqparam']
dset, stats = _get(dstore, name)
for s, stat in enumerate(stats):
if loss_type:
array = dset[:, s, oq.lti[loss_type]]
else:
array = dset[:, s]
aw = ArrayWrapper(assetcol.aggregate_by(tagnames, array), {})
for tagname in tagnames:
setattr(aw, tagname, getattr(assetcol.tagcol, tagname))
if not loss_type:
aw.extra = ('loss_type',) + oq.loss_dt().names
if name == 'curves':
aw.return_period = dset.attrs['return_periods']
aw.tagnames = encode(tagnames + ['return_period'])
else:
aw.tagnames = encode(tagnames)
yield decode(stat), aw
[docs]@extract.add('losses_by_asset')
def extract_losses_by_asset(dstore, what):
loss_dt = dstore['oqparam'].loss_dt()
rlzs = dstore['csm_info'].get_rlzs_assoc().realizations
assets = util.get_assets(dstore)
if 'losses_by_asset' in dstore:
losses_by_asset = dstore['losses_by_asset'].value
for rlz in rlzs:
# I am exporting the 'mean' and ignoring the 'stddev'
losses = cast(losses_by_asset[:, rlz.ordinal]['mean'], loss_dt)
data = util.compose_arrays(assets, losses)
yield 'rlz-%03d' % rlz.ordinal, data
elif 'avg_losses-stats' in dstore:
avg_losses = dstore['avg_losses-stats'].value
stats = dstore['avg_losses-stats'].attrs['stats']
for s, stat in enumerate(stats):
losses = cast(avg_losses[:, s], loss_dt)
data = util.compose_arrays(assets, losses)
yield stat, data
elif 'avg_losses-rlzs' in dstore: # there is only one realization
avg_losses = dstore['avg_losses-rlzs'].value
losses = cast(avg_losses, loss_dt)
data = util.compose_arrays(assets, losses)
yield 'rlz-000', data
[docs]@extract.add('losses_by_event')
def extract_losses_by_event(dstore, what):
dic = group_array(dstore['losses_by_event'].value, 'rlzi')
for rlzi in dic:
yield 'rlz-%03d' % rlzi, dic[rlzi]
def _gmf_scenario(data, num_sites, imts):
# convert data into the composite array expected by QGIS
eids = sorted(numpy.unique(data['eid']))
eid2idx = {eid: idx for idx, eid in enumerate(eids)}
E = len(eid2idx)
gmf_dt = numpy.dtype([(imt, (F32, (E,))) for imt in imts])
gmfa = numpy.zeros(num_sites, gmf_dt)
for rec in data:
arr = gmfa[rec['sid']]
for imt, gmv in zip(imts, rec['gmv']):
arr[imt][eid2idx[rec['eid']]] = gmv
return gmfa, E
[docs]@extract.add('gmf_data')
def extract_gmf_scenario_npz(dstore, what):
oq = dstore['oqparam']
mesh = get_mesh(dstore['sitecol'])
n = len(mesh)
data_by_rlzi = group_array(dstore['gmf_data/data'].value, 'rlzi')
for rlzi in data_by_rlzi:
gmfa, e = _gmf_scenario(data_by_rlzi[rlzi], n, oq.imtls)
logging.info('Exporting array of shape %s for rlz %d',
(n, e), rlzi)
yield 'rlz-%03d' % rlzi, util.compose_arrays(mesh, gmfa)
[docs]def build_damage_dt(dstore, mean_std=True):
"""
:param dstore: a datastore instance
:param mean_std: a flag (default True)
:returns:
a composite dtype loss_type -> (mean_ds1, stdv_ds1, ...) or
loss_type -> (ds1, ds2, ...) depending on the flag mean_std
"""
oq = dstore['oqparam']
damage_states = ['no_damage'] + list(
dstore.get_attr(oq.risk_model, 'limit_states'))
dt_list = []
for ds in damage_states:
ds = str(ds)
if mean_std:
dt_list.append(('%s_mean' % ds, F32))
dt_list.append(('%s_stdv' % ds, F32))
else:
dt_list.append((ds, F32))
damage_dt = numpy.dtype(dt_list)
loss_types = dstore.get_attr(oq.risk_model, 'loss_types')
return numpy.dtype([(str(lt), damage_dt) for lt in loss_types])
[docs]def build_damage_array(data, damage_dt):
"""
:param data: an array of length N with fields 'mean' and 'stddev'
:param damage_dt: a damage composite data type loss_type -> states
:returns: a composite array of length N and dtype damage_dt
"""
L = len(data) if data.shape else 1
dmg = numpy.zeros(L, damage_dt)
for lt in damage_dt.names:
for i, ms in numpy.ndenumerate(data[lt]):
if damage_dt[lt].names[0].endswith('_mean'):
lst = []
for m, s in zip(ms['mean'], ms['stddev']):
lst.append(m)
lst.append(s)
dmg[lt][i] = tuple(lst)
else:
dmg[lt][i] = ms['mean']
return dmg
[docs]@extract.add('dmg_by_asset')
def extract_dmg_by_asset_npz(dstore, what):
damage_dt = build_damage_dt(dstore)
rlzs = dstore['csm_info'].get_rlzs_assoc().realizations
data = dstore['dmg_by_asset']
assets = util.get_assets(dstore)
for rlz in rlzs:
dmg_by_asset = build_damage_array(data[:, rlz.ordinal], damage_dt)
yield 'rlz-%03d' % rlz.ordinal, util.compose_arrays(
assets, dmg_by_asset)
[docs]@extract.add('event_based_mfd')
def extract_mfd(dstore, what):
"""
Display num_ruptures by magnitude for event based calculations.
Example: http://127.0.0.1:8800/v1/calc/30/extract/event_based_mfd
"""
dd = collections.defaultdict(int)
for rup in dstore['ruptures'].value:
dd[rup['mag']] += 1
dt = numpy.dtype([('mag', float), ('freq', int)])
magfreq = numpy.array(sorted(dd.items(), key=operator.itemgetter(0)), dt)
return magfreq
[docs]@extract.add('src_loss_table')
def extract_src_loss_table(dstore, loss_type):
"""
Extract the source loss table for a give loss type, ordered in decreasing
order. Example:
http://127.0.0.1:8800/v1/calc/30/extract/src_loss_table/structural
"""
oq = dstore['oqparam']
li = oq.lti[loss_type]
source_ids = dstore['source_info']['source_id']
idxs = dstore['ruptures'].value[['srcidx', 'grp_id']]
losses = dstore['rup_loss_table'][:, li]
slt = numpy.zeros(len(source_ids), [('grp_id', U32), (loss_type, F32)])
for loss, (srcidx, grp_id) in zip(losses, idxs):
slt[srcidx][loss_type] += loss
slt[srcidx]['grp_id'] = grp_id
slt = util.compose_arrays(source_ids, slt, 'source_id')
slt.sort(order=loss_type)
return slt[::-1]
[docs]@extract.add('mean_std_curves')
def extract_mean_std_curves(dstore, what):
"""
Yield imls/IMT and poes/IMT containg mean and stddev for all sites
"""
getter = getters.PmapGetter(dstore)
arr = getter.get_mean().array
for imt in getter.imtls:
yield 'imls/' + imt, getter.imtls[imt]
yield 'poes/' + imt, arr[:, getter.imtls(imt)]
[docs]@extract.add('composite_risk_model.attrs')
def crm_attrs(dstore, what):
"""
:returns:
the attributes of the risk model, i.e. limit_states, loss_types,
min_iml and covs, needed by the risk exporters.
"""
name = dstore['oqparam'].risk_model
return ArrayWrapper(0, dstore.get_attrs(name))
def _get(dstore, name):
try:
dset = dstore[name + '-stats']
return dset, [b.decode('utf8') for b in dset.attrs['stats']]
except KeyError: # single realization
return dstore[name + '-rlzs'], ['mean']
[docs]@deprecated('This feature will be removed soon')
@extract.add('losses_by_tag')
def losses_by_tag(dstore, tag):
"""
Statistical average losses by tag. For instance call
$ oq extract losses_by_tag/occupancy
"""
dt = [(tag, vstr)] + dstore['oqparam'].loss_dt_list()
aids = dstore['assetcol/array'][tag]
dset, stats = _get(dstore, 'avg_losses')
arr = dset.value
tagvalues = dstore['assetcol/tagcol/' + tag][1:] # except tagvalue="?"
for s, stat in enumerate(stats):
out = numpy.zeros(len(tagvalues), dt)
for li, (lt, lt_dt) in enumerate(dt[1:]):
for i, tagvalue in enumerate(tagvalues):
out[i][tag] = tagvalue
counts = arr[aids == i + 1, s, li].sum()
if counts:
out[i][lt] = counts
yield stat, out
[docs]@deprecated('This feature will be removed soon')
@extract.add('curves_by_tag')
def curves_by_tag(dstore, tag):
"""
Statistical loss curves by tag. For instance call
$ oq extract curves_by_tag/occupancy
"""
dt = ([(tag, vstr), ('return_period', U32)] +
dstore['oqparam'].loss_dt_list())
aids = dstore['assetcol/array'][tag]
dset, stats = _get(dstore, 'curves')
periods = dset.attrs['return_periods']
arr = dset.value
P = arr.shape[2] # shape (A, S, P, LI)
tagvalues = dstore['assetcol/tagcol/' + tag][1:] # except tagvalue="?"
for s, stat in enumerate(stats):
out = numpy.zeros(len(tagvalues) * P, dt)
for li, (lt, lt_dt) in enumerate(dt[2:]):
n = 0
for i, tagvalue in enumerate(tagvalues):
for p, period in enumerate(periods):
out[n][tag] = tagvalue
out[n]['return_period'] = period
counts = arr[aids == i + 1, s, p, li].sum()
if counts:
out[n][lt] = counts
n += 1
yield stat, out