def create_autocomplete_job_with_data_sets():
with app.app_context():
instance = models.Instance('fr')
models.db.session.add(instance)
models.db.session.commit()
job = models.Job()
job.instance = instance
# we also create 2 datasets, one for fusio, one for autocomplete_cosmogony
for i, dset_type in enumerate(['fusio', 'cosmogony']):
dataset = models.DataSet()
dataset.type = dset_type
dataset.family_type = dataset.type
if dataset.type == 'fusio':
dataset.family_type = 'pt'
dataset.name = '/path/to/dataset_{}'.format(i)
else:
dataset.family_type = 'autocomplete_cosmogony'
dataset.name = '/path/to/dataset_cosmogony/cosmogony_europe.jsonl.gz'
models.db.session.add(dataset)
job.data_sets.append(dataset)
job.state = 'done'
models.db.session.add(job)
models.db.session.commit()
def create_and_get_dataset(ds_type, family_type, filename):
dataset = models.DataSet()
dataset.family_type = family_type
dataset.type = ds_type
# currently the name of a dataset is the path to it
dataset.name = filename
return dataset
def send_to_mimir(instance, filename, family_type):
"""
:param instance: instance to receive the data
:param filename: file to inject towards mimir
:param family_type: dataset's family type
- create a job with a data_set
- data injection towards mimir(stops2mimir, ntfs2mimir, poi2mimir)
returns action list
"""
# if mimir isn't setup do not try to import data for the autocompletion
if not current_app.config.get('MIMIR_URL'):
return []
# Bail out if the family type is not one that mimir deals with.
if family_type not in ['pt', 'poi']:
return []
# This test is to avoid creating a new job if there is no action on mimir.
if not (instance.import_ntfs_in_mimir or instance.import_stops_in_mimir):
return []
actions = []
job = models.Job()
job.instance = instance
job.state = 'running'
dataset = models.DataSet()
dataset.family_type = 'mimir'
dataset.type = 'fusio'
# currently the name of a dataset is the path to it
dataset.name = filename
models.db.session.add(dataset)
job.data_sets.append(dataset)
models.db.session.add(job)
models.db.session.commit()
if family_type == 'pt':
# Import ntfs in Mimir
if instance.import_ntfs_in_mimir:
actions.append(ntfs2mimir.si(instance.name, filename, job.id, dataset_uid=dataset.uid))
# Import stops in Mimir.
# if we are loading pt data we might want to load the stops to autocomplete
# This action is deprecated: https://github.com/CanalTP/mimirsbrunn/blob/4430eed1d81247fffa7cf32ba675a9c5ad8b1cbe/documentation/components.md#stops2mimir
if instance.import_stops_in_mimir and not instance.import_ntfs_in_mimir:
actions.append(stops2mimir.si(instance.name, filename, job.id, dataset_uid=dataset.uid))
else: # assume family_type == 'poi':
actions.append(poi2mimir.si(instance.name, filename, job.id, dataset_uid=dataset.uid))
actions.append(finish_job.si(job.id))
return actions
def create_dataset(dataset_type, dir):
dataset = models.DataSet()
dataset.type = dataset_type
dataset.family_type = '{}_family'.format(dataset_type)
dataset.name = '{}/dataset'.format(dir)
models.db.session.add(dataset)
metric = models.Metric()
metric.type = '{}2ed'.format(dataset_type)
metric.dataset = dataset
models.db.session.add(metric)
return dataset, metric
def import_autocomplete(files, autocomplete_instance, asynchronous=True, backup_file=True):
"""
Import the autocomplete'instance data files
"""
job = models.Job()
actions = []
task = {'bano': bano2mimir, 'oa': openaddresses2mimir, 'osm': osm2mimir, 'cosmogony': cosmogony2mimir}
autocomplete_dir = current_app.config['TYR_AUTOCOMPLETE_DIR']
# it's important for the admin to be loaded first, then addresses, then street, then poi
import_order = ['cosmogony', 'bano', 'oa', 'osm']
files_and_types = [(f, type_of_autocomplete_data(f)) for f in files]
files_and_types = sorted(files_and_types, key=lambda f_t: import_order.index(f_t[1]))
for f, ftype in files_and_types:
dataset = models.DataSet()
dataset.type = ftype
dataset.family_type = 'autocomplete_{}'.format(dataset.type)
if dataset.type in task:
if backup_file:
filename = move_to_backupdirectory(f, autocomplete_instance.backup_dir(autocomplete_dir))
else:
filename = f
actions.append(
task[dataset.type].si(autocomplete_instance, filename=filename, dataset_uid=dataset.uid)
)
else:
# unknown type, we skip it
current_app.logger.debug("unknown file type: {} for file {}".format(dataset.type, f))
continue
# currently the name of a dataset is the path to it
dataset.name = filename
models.db.session.add(dataset)
job.data_sets.append(dataset)
job.autocomplete_params_id = autocomplete_instance.id
if not actions:
return
models.db.session.add(job)
models.db.session.commit()
for action in actions:
action.kwargs['job_id'] = job.id
actions.append(finish_job.si(job.id))
if asynchronous:
return chain(*actions).delay(), job
else:
# all job are run in sequence and import_data will only return when all the jobs are finish
return chain(*actions).apply(), job
def send_to_mimir(instance, filename):
"""
:param instance: instance to receive the data
:param filename: file to inject towards mimir
- create a job with a data_set
- data injection towards mimir(stops2mimir, ntfs2mimir)
returns action list
"""
# This test is to avoid creating a new job if there is no action on mimir.
if not (instance.import_ntfs_in_mimir or instance.import_stops_in_mimir):
return []
actions = []
job = models.Job()
instance_config = load_instance_config(instance.name)
job.instance = instance
job.state = 'running'
dataset = models.DataSet()
dataset.family_type = 'mimir'
dataset.type = 'fusio'
# currently the name of a dataset is the path to it
dataset.name = filename
models.db.session.add(dataset)
job.data_sets.append(dataset)
models.db.session.add(job)
models.db.session.commit()
# Import ntfs in Mimir
if instance.import_ntfs_in_mimir:
actions.append(
ntfs2mimir.si(instance_config,
filename,
job.id,
dataset_uid=dataset.uid))
# Import stops in Mimir
# if we are loading pt data we might want to load the stops to autocomplete
if instance.import_stops_in_mimir and not instance.import_ntfs_in_mimir:
actions.append(
stops2mimir.si(instance_config,
filename,
job.id,
dataset_uid=dataset.uid))
actions.append(finish_job.si(job.id))
return actions
def create_cities_job(creation_date, path, state):
job = models.Job()
job.state = state
dataset_backup_dir = path
dataset = models.DataSet()
dataset.type = 'cities'
dataset.family_type = 'cities_family'
dataset.name = '{}'.format(dataset_backup_dir)
models.db.session.add(dataset)
job.data_sets.append(dataset)
job.created_at = creation_date
models.db.session.add(job)
def create_dataset(dataset_type):
dataset = models.DataSet()
dataset.type = dataset_type
dataset.family_type = '{}_family'.format(dataset_type)
dataset.name = '/path/to/dataset_{}'.format(dataset_type)
models.db.session.add(dataset)
metric = models.Metric()
metric.type = '{}2ed'.format(dataset_type)
metric.duration = datetime.timedelta(seconds=9.0001)
metric.dataset = dataset
models.db.session.add(metric)
return dataset, metric
def add_job_with_data_set_mimir(create_basic_job_with_data_sets):
with app.app_context():
# we also create 1 job with a dataset for mimir
instance = get_instance_from_db(name='fr')
job = models.Job()
job.instance = instance
dataset = models.DataSet()
dataset.family_type = 'mimir'
dataset.type = 'stop2mimir'
dataset.name = '/path/to/dataset_3'
models.db.session.add(dataset)
job.data_sets.append(dataset)
job.state = 'done'
models.db.session.add(job)
models.db.session.commit()
def add_job_and_data_set_with_jobstate_running(create_basic_job_with_data_sets):
with app.app_context():
# we add a new job with a dataset for mimir
instance = get_instance_from_db(name='fr')
job = models.Job()
job.instance = instance
dataset = models.DataSet()
dataset.family_type = 'osm'
dataset.type = 'osm'
dataset.name = '/path/to/dataset_osm'
models.db.session.add(dataset)
job.data_sets.append(dataset)
job.state = 'running'
models.db.session.add(job)
models.db.session.commit()
def create_autocomplete_parameter():
with app.app_context():
autocomplete_param = models.AutocompleteParameter('idf', 'OSM', 'BANO','FUSIO', 'OSM', [8, 9])
models.db.session.add(autocomplete_param)
models.db.session.commit()
# we also create 3 datasets, one for bano, 2 for osm
for i, dset_type in enumerate(['bano', 'osm', 'osm']):
job = models.Job()
dataset = models.DataSet()
dataset.type = dset_type
dataset.family_type = 'autocomplete_{}'.format(dataset.type)
dataset.name = '/path/to/dataset_{}'.format(i)
models.db.session.add(dataset)
job.autocomplete_params_id = autocomplete_param.id
job.data_sets.append(dataset)
job.state = 'done'
models.db.session.add(job)
models.db.session.commit()
def create_basic_job_with_data_sets():
with app.app_context():
instance = models.Instance('fr')
models.db.session.add(instance)
models.db.session.commit()
job = models.Job()
job.instance = instance
# we also create 2 datasets, one for fusio, one for synonym
for i, dset_type in enumerate(['fusio', 'synonym']):
dataset = models.DataSet()
dataset.type = dset_type
dataset.family_type = dataset.type
if dataset.type == 'fusio':
dataset.family_type = 'pt'
dataset.name = '/path/to/dataset_{}'.format(i)
models.db.session.add(dataset)
job.data_sets.append(dataset)
job.state = 'done'
models.db.session.add(job)
models.db.session.commit()
def import_data(files, instance, backup_file):
"""
import the data contains in the list of 'files' in the 'instance'
:param files: files to import
:param instance: instance to receive the data
:param backup_file: If True the files are moved to a backup directory, else they are not moved
run the whole data import process:
- data import in bdd (fusio2ed, gtfs2ed, poi2ed, ...)
- export bdd to nav file
- update the jormungandr db with the new data for the instance
- reload the krakens
"""
actions = []
job = models.Job()
instance_config = load_instance_config(instance.name)
job.instance = instance
job.state = 'pending'
task = {
'gtfs': gtfs2ed,
'fusio': fusio2ed,
'osm': osm2ed,
'geopal': geopal2ed,
'fare': fare2ed,
'poi': poi2ed,
'synonym': synonym2ed,
}
for _file in files:
filename = None
dataset = models.DataSet()
dataset.type = type_of_data(_file)
if dataset.type in task:
if backup_file:
filename = move_to_backupdirectory(
_file, instance_config.backup_directory)
else:
filename = _file
actions.append(task[dataset.type].si(instance_config, filename))
else:
#unknown type, we skip it
current_app.logger.debug("unknwn file type: {} for file {}".format(
dataset.type, _file))
continue
#currently the name of a dataset is the path to it
dataset.name = filename
models.db.session.add(dataset)
job.data_sets.append(dataset)
if actions:
models.db.session.add(job)
models.db.session.commit()
for action in actions:
action.kwargs['job_id'] = job.id
#We pass the job id to each tasks, but job need to be commited for
#having an id
binarisation = [
ed2nav.si(instance_config, job.id),
nav2rt.si(instance_config, job.id)
]
aggregate = aggregate_places.si(instance_config, job.id)
#We pass the job id to each tasks, but job need to be commited for
#having an id
actions.append(group(chain(*binarisation), aggregate))
actions.append(reload_data.si(instance_config, job.id))
actions.append(finish_job.si(job.id))
chain(*actions).delay()
job.state = 'pending'
task = {
'gtfs': gtfs2ed,
'fusio': fusio2ed,
'osm': osm2ed,
'geopal': geopal2ed,
'fare': fare2ed,
'poi': poi2ed,
'synonym': synonym2ed,
'shape': shape2ed,
}
for _file in files:
filename = None
dataset = models.DataSet()
# NOTE: for the moment we do not use the path to load the data here
# but we'll need to refactor this to take it into account
dataset.type, _ = utils.type_of_data(_file)
dataset.family_type = utils.family_of_data(dataset.type)
if dataset.type in task:
if backup_file:
filename = move_to_backupdirectory(_file,
instance_config.backup_directory)
else:
filename = _file
actions.append(task[dataset.type].si(instance_config, filename, dataset_uid=dataset.uid))
else:
#unknown type, we skip it
current_app.logger.debug("unknwn file type: {} for file {}"
.format(dataset.type, _file))
def import_data(
files, instance, backup_file, asynchronous=True, reload=True, custom_output_dir=None, skip_mimir=False
):
"""
import the data contains in the list of 'files' in the 'instance'
:param files: files to import
:param instance: instance to receive the data
:param backup_file: If True the files are moved to a backup directory, else they are not moved
:param asynchronous: If True all jobs are run in background, else the jobs are run in sequence the function
will only return when all of them are finish
:param reload: If True kraken would be reload at the end of the treatment
:param custom_output_dir: subdirectory for the nav file created. If not given, the instance default one is taken
:param skip_mimir: skip importing data into mimir
run the whole data import process:
- data import in bdd (fusio2ed, gtfs2ed, poi2ed, ...)
- export bdd to nav file
- update the jormungandr db with the new data for the instance
- reload the krakens
"""
actions = []
job = models.Job()
instance_config = load_instance_config(instance.name)
job.instance = instance
job.state = 'running'
task = {
'gtfs': gtfs2ed,
'fusio': fusio2ed,
'osm': osm2ed,
'geopal': geopal2ed,
'fare': fare2ed,
'poi': poi2ed,
'synonym': synonym2ed,
'shape': shape2ed,
}
for _file in files:
filename = None
dataset = models.DataSet()
# NOTE: for the moment we do not use the path to load the data here
# but we'll need to refactor this to take it into account
try:
dataset.type, _ = utils.type_of_data(_file)
dataset.family_type = utils.family_of_data(dataset.type)
except Exception:
if backup_file:
move_to_backupdirectory(_file, instance_config.backup_directory)
current_app.logger.debug(
"Corrupted source file : {} moved to {}".format(_file, instance_config.backup_directory)
)
continue
if dataset.type in task:
if backup_file:
filename = move_to_backupdirectory(_file, instance_config.backup_directory)
else:
filename = _file
actions.append(task[dataset.type].si(instance_config, filename, dataset_uid=dataset.uid))
else:
# unknown type, we skip it
current_app.logger.debug("unknown file type: {} for file {}".format(dataset.type, _file))
continue
# currently the name of a dataset is the path to it
dataset.name = filename
models.db.session.add(dataset)
job.data_sets.append(dataset)
if actions:
models.db.session.add(job)
models.db.session.commit()
# We pass the job id to each tasks, but job need to be commited for having an id
for action in actions:
action.kwargs['job_id'] = job.id
# Create binary file (New .nav.lz4)
binarisation = [ed2nav.si(instance_config, job.id, custom_output_dir)]
actions.append(chain(*binarisation))
# Reload kraken with new data after binarisation (New .nav.lz4)
if reload:
actions.append(reload_data.si(instance_config, job.id))
if not skip_mimir:
for dataset in job.data_sets:
actions.extend(send_to_mimir(instance, dataset.name, dataset.family_type))
else:
current_app.logger.info("skipping mimir import")
actions.append(finish_job.si(job.id))
if asynchronous:
return chain(*actions).delay()
else:
# all job are run in sequence and import_data will only return when all the jobs are finish
return chain(*actions).apply()
def update_data():
for instance in models.Instance.query.all():
current_app.logger.debug("Update data of : %s" % instance.name)
instance_config = load_instance_config(instance.name)
files = glob.glob(instance_config.source_directory + "/*")
actions = []
job = models.Job()
job.instance = instance
job.state = 'pending'
for _file in files:
dataset = models.DataSet()
filename = None
dataset.type = type_of_data(_file)
if dataset.type == 'gtfs':
filename = move_to_backupdirectory(
_file, instance_config.backup_directory)
actions.append(gtfs2ed.si(instance_config, filename))
elif dataset.type == 'fusio':
filename = move_to_backupdirectory(
_file, instance_config.backup_directory)
actions.append(fusio2ed.si(instance_config, filename))
elif dataset.type == 'osm':
filename = move_to_backupdirectory(
_file, instance_config.backup_directory)
actions.append(osm2ed.si(instance_config, filename))
elif dataset.type == 'geopal':
filename = move_to_backupdirectory(
_file, instance_config.backup_directory)
actions.append(geopal2ed.si(instance_config, filename))
elif dataset.type == 'fare':
filename = move_to_backupdirectory(
_file, instance_config.backup_directory)
actions.append(fare2ed.si(instance_config, filename))
elif dataset.type == 'poi':
filename = move_to_backupdirectory(
_file, instance_config.backup_directory)
actions.append(poi2ed.si(instance_config, filename))
elif dataset.type == 'synonym':
filename = move_to_backupdirectory(
_file, instance_config.backup_directory)
actions.append(synonym2ed.si(instance_config, filename))
else:
#unknown type, we skip it
continue
#currently the name of a dataset is the path to it
dataset.name = filename
models.db.session.add(dataset)
job.data_sets.append(dataset)
if actions:
models.db.session.add(job)
models.db.session.commit()
for action in actions:
action.kwargs['job_id'] = job.id
#We pass the job id to each tasks, but job need to be commited for
#having an id
binarisation = [
ed2nav.si(instance_config, job.id),
nav2rt.si(instance_config, job.id)
]
aggregate = aggregate_places.si(instance_config, job.id)
#We pass the job id to each tasks, but job need to be commited for
#having an id
actions.append(group(chain(*binarisation), aggregate))
actions.append(reload_data.si(instance_config, job.id))
actions.append(finish_job.si(job.id))
chain(*actions).delay()
def import_data(
files,
instance,
backup_file,
asynchronous=True,
reload=True,
custom_output_dir=None,
skip_mimir=False,
skip_2ed=False,
):
"""
import the data contains in the list of 'files' in the 'instance'
:param files: files to import
:param instance: instance to receive the data
:param backup_file: If True the files are moved to a backup directory, else they are not moved
:param asynchronous: If True all jobs are run in background, else the jobs are run in sequence the function
will only return when all of them are finish
:param reload: If True kraken would be reload at the end of the treatment
:param custom_output_dir: subdirectory for the nav file created. If not given, the instance default one is taken
:param skip_mimir: skip importing data into mimir
:param skip_2ed: skip inserting last_load_dataset files into ed database
run the whole data import process:
- data import in bdd (fusio2ed, gtfs2ed, poi2ed, ...)
- export bdd to nav file
- update the jormungandr db with the new data for the instance
- reload the krakens
"""
actions = []
job = models.Job()
instance_config = load_instance_config(instance.name)
job.instance = instance
job.state = 'running'
task = {
'gtfs': gtfs2ed,
'fusio': fusio2ed,
'osm': osm2ed,
'geopal': geopal2ed,
'fare': fare2ed,
'poi': poi2ed,
'synonym': synonym2ed,
'shape': shape2ed,
}
def process_ed2nav():
models.db.session.add(job)
models.db.session.commit()
# We pass the job id to each tasks, but job need to be commited for having an id
for action in actions:
action.kwargs['job_id'] = job.id
# Create binary file (New .nav.lz4)
binarisation = [ed2nav.si(instance_config, job.id, custom_output_dir)]
actions.append(chain(*binarisation))
# Reload kraken with new data after binarisation (New .nav.lz4)
if reload:
actions.append(reload_data.si(instance_config, job.id))
if not skip_mimir:
for dataset in job.data_sets:
actions.extend(
send_to_mimir(instance, dataset.name, dataset.family_type))
else:
current_app.logger.info("skipping mimir import")
actions.append(finish_job.si(job.id))
# We should delete old backup directories related to this instance
actions.append(
purge_instance.si(
instance.id,
current_app.config['DATASET_MAX_BACKUPS_TO_KEEP']))
if asynchronous:
return chain(*actions).delay()
else:
# all job are run in sequence and import_data will only return when all the jobs are finish
return chain(*actions).apply()
if skip_2ed:
# For skip_2ed, skip inserting last_load_dataset files into ed database
return process_ed2nav()
for _file in files:
filename = None
dataset = models.DataSet()
# NOTE: for the moment we do not use the path to load the data here
# but we'll need to refactor this to take it into account
try:
dataset.type, _ = utils.type_of_data(_file)
dataset.family_type = utils.family_of_data(dataset.type)
except Exception:
if backup_file:
move_to_backupdirectory(_file,
instance_config.backup_directory)
current_app.logger.debug(
"Corrupted source file : {} moved to {}".format(
_file, instance_config.backup_directory))
continue
if dataset.type in task:
if backup_file:
filename = move_to_backupdirectory(
_file,
instance_config.backup_directory,
manage_sp_char=True)
else:
filename = _file
has_pt_planner_loki = (
hasattr(instance, 'pt_planners_configurations')
and "loki" in instance.pt_planners_configurations)
if has_pt_planner_loki:
loki_data_source = instance.pt_planners_configurations.get(
'loki', {}).get('data_source')
if loki_data_source is not None:
if loki_data_source == "minio":
if dataset.type == "fusio":
actions.append(
fusio2s3.si(instance_config,
filename,
dataset_uid=dataset.uid))
if dataset.type == "gtfs":
actions.append(
gtfs2s3.si(instance_config,
filename,
dataset_uid=dataset.uid))
elif loki_data_source == "local" and dataset.type in [
"fusio", "gtfs"
]:
zip_file = zip_if_needed(filename)
dest = os.path.join(
os.path.dirname(instance_config.target_file),
"ntfs.zip")
shutil.copy(zip_file, dest)
else:
current_app.logger.debug(
"unknown loki data_source '{}' for coverage '{}'".
format(loki_data_source, instance.name))
actions.append(task[dataset.type].si(instance_config,
filename,
dataset_uid=dataset.uid))
else:
# unknown type, we skip it
current_app.logger.debug(
"unknown file type: {} for file {}".format(
dataset.type, _file))
continue
# currently the name of a dataset is the path to it
dataset.name = filename
dataset.state = "pending"
models.db.session.add(dataset)
job.data_sets.append(dataset)
if actions:
return process_ed2nav()