def test_create_simulation(self):
papi = postgresdb(self.session)
model_code = 'swat'
sim_name = 'TonySwatSimulation'
sim_description = 'My SWAT Simulation'
#sim_start = dt.datetime(2014,01,01,0,0,0)
#sim_end = dt.datetime(2014,02,01,0,0,0)
timestepvalue = 1
timestepunitid = None # query this
inputdatasetid = None # query this
#startoffset=-6
#endoffset=-6
# build exchange items from file
config = os.path.realpath('../../tests/test_data/configuration.ini')
params = gui.parse_config(config)
eitems = mdl.build_exchange_items_from_config(params)
# add some data to simulate 'output' exchange items
vals= [(dt.datetime(2014,1,1,0,0,0) + dt.timedelta(days=i), i) for i in range(0,100)]
output_item1 = eitems['output'][0]
output_item1.geometries()[0].datavalues().set_timeseries(vals)
vals= [(dt.datetime(2014,1,1,0,0,0) + dt.timedelta(days=i), 2*i) for i in range(0,100)]
output_item1.geometries()[1].datavalues().set_timeseries(vals)
vals= [(dt.datetime(2014,1,1,0,0,0) + dt.timedelta(days=i), 2**i) for i in range(0,100)]
output_item2 = eitems['output'][1]
output_item2.geometries()[0].datavalues().set_timeseries(vals)
outputs = [output_item1,output_item2]
sim = papi.create_simulation(preferences_path='../../data/preferences',
config_params=params,
output_exchange_items= outputs)
print 'Successfully inserted Simulation: %d'%sim.SimulationID
def test_create_dataset(self):
papi = postgresdb(self.session)
def test_insert_ts_results(self):
papi = postgresdb(self.session)
papi.insert_result_ts()
pass
def test_set_user_preferences(self):
papi = postgresdb(self.session)
papi.set_user_preferences(self.prefs)
def run_time_step(obj, ds=None):
# store db sessions
db_sessions = {}
# ThreadManager
dispatcher = ThreadManager().get_dispatcher()
sim_st = time.time()
# todo: move this into function
# determine execution order
msg = "> Determining execution order... "
dispatcher.putOutput(msg)
exec_order = obj.determine_execution_order()
msg = "done"
dispatcher.putOutput(msg)
for i in range(0, len(exec_order)):
msg = "> %d.) %s" % (i + 1, obj.get_model_by_id(exec_order[i]).name())
dispatcher.putOutput(msg)
links = {}
spatial_maps = {}
simulation_status = {}
# todo: move this into function
msg = "> [PRE-RUN] Performing spatial mapping... "
dispatcher.putOutput(msg)
for modelid in exec_order:
# get links
l = obj.get_from_links_by_model(modelid)
links[modelid] = l
for linkid, link in l.iteritems():
# build spatial maps
source = link.source_exchange_item()
target = link.target_exchange_item()
key = generate_link_key(link)
# spatial_maps[key] = spatial.transform(source.get_all_datasets().keys(), target.get_all_datasets().keys())
spatial_interp = link.spatial_interpolation()
if spatial_interp:
spatial_maps[key] = spatial_interp.transform(
source.get_all_datasets().keys(), target.get_all_datasets().keys()
)
# set model status to RUNNING
obj.get_model_by_id(modelid).instance().status(Status.Running)
# initialize the simulation_status dictionary
simulation_status[modelid] = obj.get_model_by_id(modelid).instance().status()
# store model db sessions
session = obj.get_model_by_id(modelid).instance().session()
if session is None:
try: # this is necessary if no db connection exists
session = obj.get_default_db()["session"]
except:
pass
# todo: need to consider other databases too!
db_sessions[modelid] = postgresdb(session)
msg = "done"
dispatcher.putOutput(msg)
# todo: move this into a time-horizon checking function.
# this should check that the time-horizon is valid.
# determine minimum overlapping timespan to set start and end times
msg = "> [PRE-RUN] Validating simulation time-horizon... "
dispatcher.putOutput(msg)
global_simulation_start = datetime.datetime(3000, 1, 1)
global_simulation_end = datetime.datetime(1800, 1, 1)
for modelid in exec_order:
inst = obj.get_model_by_id(modelid).instance()
global_simulation_start = (
inst.simulation_start() if inst.simulation_start() < global_simulation_start else global_simulation_start
)
global_simulation_end = (
inst.simulation_end() if inst.simulation_end() > global_simulation_end else global_simulation_end
)
if global_simulation_start >= global_simulation_end:
raise Exception(
"Invalid start and end times!\nstart:%s\nend:%s"
% (str(global_simulation_start), str(global_simulation_end))
)
msg = "done"
dispatcher.putOutput(msg)
# todo: move this into function
# prepare all models
for modelid in exec_order:
model_obj = obj.get_model_by_id(modelid)
model_inst = model_obj.instance()
if model_inst.status() != Status.Ready:
model_inst.prepare()
iter_count = 1
# run simulation until all models reach a FINISHED state
while not all(stat == Status.Finished for stat in simulation_status.values()):
# TODO: This needs to be modified to operate under loop control! For instance if the simulation reaches a point where the current model doesnt reach or exceed the target model, the loop should be broken and restart from the beginning. This will also allow multithreading of multiple loops during a composition that might be helpful for calibrations.
msg = "Executing Loop %d" % iter_count
dispatcher.putOutput(msg)
# print '\nExecuting Loop %d' % iter_count
# loop through models and execute run
for modelid in exec_order:
# get the current model instance
model_obj = obj.get_model_by_id(modelid)
model_inst = model_obj.instance()
# get the target simulation times from the model links (including its own endtime)
target_times = []
if len(links[modelid]) > 0:
# add the target current time
for linkid, link in links[modelid].iteritems():
# target_model = target[0]
target_model = link.target_component()
target_times.append(target_model.instance().current_time())
else:
# add the current time for the source model (this will force the model to step only once)
target_times.append(model_inst.current_time())
# time the model until it reaches or surpasses all target times (requested time)
current_time = model_inst.current_time()
while current_time <= max(target_times):
# update simulation status
simulation_status[modelid] = model_inst.status()
if model_inst.status() != Status.Running and model_inst.status() != Status.Ready:
# exit without calling run_timestep
msg = "> %s " % (datetime.datetime.strftime(current_time, "%m-%d-%Y %H:%M:%S"))
dispatcher.putOutput(msg)
msg = "> %s | %s \n" % (model_inst.name(), model_inst.status())
dispatcher.putOutput(msg)
break
# get model input data
input_data = model_inst.inputs()
msg = "> %s " % (datetime.datetime.strftime(current_time, "%m-%d-%Y %H:%M:%S"))
dispatcher.putOutput(msg)
msg = "> %s | %s \n" % (model_inst.name(), model_inst.status())
dispatcher.putOutput(msg)
# run model timestep
model_inst.run_timestep(input_data, current_time)
# get the new current time
current_time = model_inst.current_time()
sys.stdout.write("\n")
# get all outputs
output_exchange_items = model_inst.outputs()
# update the outgoing links for this component
update.update_links(obj, links[modelid], output_exchange_items, spatial_maps)
iter_count += 1
for modelid in exec_order:
# get the current model instance
model_obj = obj.get_model_by_id(modelid)
model_inst = model_obj.instance()
# save results
items = model_inst.save()
# todo: save outputs to database!
if len(items) > 0:
simulation_dbapi = db_sessions[modelid]
# set these input data as exchange items in stdlib or wrapper class
simulation = simulation_dbapi.create_simulation(
preferences_path=obj.preferences,
config_params=model_obj.get_config_params(),
output_exchange_items=items,
)
msg = (
"> \n"
+ "> ------------------------------------------\n"
+ "> Simulation Summary \n"
+ "> ------------------------------------------\n"
+ "> Completed without error :)\n"
+ "> Simulation duration: %3.2f seconds\n" % (time.time() - sim_st)
+ "> ------------------------------------------"
)
dispatcher.putOutput(msg)
def run_time_step(obj, ds=None):
# store db sessions
db_sessions = {}
# ThreadManager
dispatcher = ThreadManager().get_dispatcher()
sim_st = time.time()
# todo: move this into function
# determine execution order
msg = '> Determining execution order... '
dispatcher.putOutput(msg)
exec_order = obj.determine_execution_order()
msg = 'done'
dispatcher.putOutput(msg)
for i in range(0, len(exec_order)):
msg = '> %d.) %s' % (i + 1, obj.get_model_by_id(exec_order[i]).name())
dispatcher.putOutput(msg)
links = {}
spatial_maps = {}
simulation_status = {}
# todo: move this into function
msg = '> [PRE-RUN] Performing spatial mapping... '
dispatcher.putOutput(msg)
for modelid in exec_order:
# get links
l = obj.get_from_links_by_model(modelid)
links[modelid] = l
for linkid, link in l.iteritems():
# build spatial maps
source = link.source_exchange_item()
target = link.target_exchange_item()
key = generate_link_key(link)
#spatial_maps[key] = spatial.transform(source.get_all_datasets().keys(), target.get_all_datasets().keys())
spatial_interp = link.spatial_interpolation()
if spatial_interp:
spatial_maps[key] = spatial_interp.transform(source.get_all_datasets().keys(),
target.get_all_datasets().keys())
# set model status to RUNNING
obj.get_model_by_id(modelid).instance().status(Status.Running)
# initialize the simulation_status dictionary
simulation_status[modelid] = obj.get_model_by_id(modelid).instance().status()
# store model db sessions
session = obj.get_model_by_id(modelid).instance().session()
if session is None:
try: # this is necessary if no db connection exists
session = obj.get_default_db()['session']
except:
pass
# todo: need to consider other databases too!
db_sessions[modelid] = postgresdb(session)
msg = 'done'
dispatcher.putOutput(msg)
# todo: move this into a time-horizon checking function.
# this should check that the time-horizon is valid.
# determine minimum overlapping timespan to set start and end times
msg = '> [PRE-RUN] Validating simulation time-horizon... '
dispatcher.putOutput(msg)
global_simulation_start = datetime.datetime(3000, 1, 1)
global_simulation_end = datetime.datetime(1800, 1, 1)
for modelid in exec_order:
inst = obj.get_model_by_id(modelid).instance()
global_simulation_start = inst.simulation_start() if inst.simulation_start() < global_simulation_start else global_simulation_start
global_simulation_end = inst.simulation_end() if inst.simulation_end() > global_simulation_end else global_simulation_end
if global_simulation_start >= global_simulation_end:
raise Exception("Invalid start and end times!\nstart:%s\nend:%s" % (
str(global_simulation_start), str(global_simulation_end)))
msg = 'done'
dispatcher.putOutput(msg)
# todo: move this into function
# prepare all models
for modelid in exec_order:
model_obj = obj.get_model_by_id(modelid)
model_inst = model_obj.instance()
if model_inst.status() != Status.Ready:
model_inst.prepare()
iter_count = 1
# run simulation until all models reach a FINISHED state
while not all(stat == Status.Finished for stat in simulation_status.values()):
# TODO: This needs to be modified to operate under loop control! For instance if the simulation reaches a point where the current model doesnt reach or exceed the target model, the loop should be broken and restart from the beginning. This will also allow multithreading of multiple loops during a composition that might be helpful for calibrations.
msg = 'Executing Loop %d' % iter_count
dispatcher.putOutput(msg)
# print '\nExecuting Loop %d' % iter_count
# loop through models and execute run
for modelid in exec_order:
# get the current model instance
model_obj = obj.get_model_by_id(modelid)
model_inst = model_obj.instance()
# get the target simulation times from the model links (including its own endtime)
target_times = []
if len(links[modelid]) > 0:
# add the target current time
for linkid, link in links[modelid].iteritems():
#target_model = target[0]
target_model = link.target_component()
target_times.append(target_model.instance().current_time())
else:
# add the current time for the source model (this will force the model to step only once)
target_times.append(model_inst.current_time())
# time the model until it reaches or surpasses all target times (requested time)
current_time = model_inst.current_time()
while current_time <= max(target_times):
# update simulation status
simulation_status[modelid] = model_inst.status()
if model_inst.status() != Status.Running and \
model_inst.status() != Status.Ready:
# exit without calling run_timestep
msg = '> %s ' % (datetime.datetime.strftime(current_time, "%m-%d-%Y %H:%M:%S"))
dispatcher.putOutput(msg)
msg = '> %s | %s \n' % (model_inst.name(), model_inst.status())
dispatcher.putOutput(msg)
break
# get model input data
input_data = model_inst.inputs()
msg = '> %s ' % (datetime.datetime.strftime(current_time, "%m-%d-%Y %H:%M:%S"))
dispatcher.putOutput(msg)
msg = '> %s | %s \n' % (model_inst.name(), model_inst.status())
dispatcher.putOutput(msg)
# run model timestep
model_inst.run_timestep(input_data, current_time)
# get the new current time
current_time = model_inst.current_time()
sys.stdout.write('\n')
# get all outputs
output_exchange_items = model_inst.outputs()
# update the outgoing links for this component
update.update_links(obj, links[modelid], output_exchange_items, spatial_maps)
iter_count += 1
for modelid in exec_order:
# get the current model instance
model_obj = obj.get_model_by_id(modelid)
model_inst = model_obj.instance()
# save results
items = model_inst.save()
# todo: save outputs to database!
if len(items) > 0:
simulation_dbapi = db_sessions[modelid]
# set these input data as exchange items in stdlib or wrapper class
simulation = simulation_dbapi.create_simulation(preferences_path=obj.preferences,
config_params=model_obj.get_config_params(),
output_exchange_items=items,
)
msg = '> \n' + \
'> ------------------------------------------\n' + \
'> Simulation Summary \n' + \
'> ------------------------------------------\n' + \
'> Completed without error :)\n' + \
'> Simulation duration: %3.2f seconds\n' % (time.time()-sim_st) + \
'> ------------------------------------------'
dispatcher.putOutput(msg)
