def _get_edges_rids_opa(event):
"""
Return a list of edges (resource_id_parent, resource_id_child),
a set of rejected resource_ids and a dict of
{resource_id: (obj, parent, attr)}.
"""
# {status: [obj, ]}
edges = [] # list of tuples (parent, child)
# {resource_id, }
rejected_rid = set()
# {resource_id: [obj, parent, attr]}
rejected_opa = {}
# first make a list objects with eval status as well as all other
# objects they refer to, using resource_ids
opa_iter = yield_obj_parent_attr(event, has_attr="evaluation_status")
for obj, parent, attr in opa_iter:
rid = str(obj.resource_id)
is_rejected = obj.evaluation_status == "rejected"
if is_rejected:
rejected_rid.add(rid)
rejected_opa[rid] = (obj, parent, attr)
# now recurse object and find all objects this one refers to
opa_iter_2 = yield_obj_parent_attr(obj, cls=ev.ResourceIdentifier)
for obj2, _, _ in opa_iter_2:
edges.append((rid, str(obj2)))
return edges, rejected_rid, rejected_opa
def _events_to_tables(
event_iterable: Union[obspy.Catalog, Sequence[ev.Event]]
) -> Dict[str, pd.DataFrame]:
""" Create tables from an event iterable """
obj_dict = defaultdict(list)
seen_ids = set()
def _obj_to_dict(obj, parent_id=None, event_id=None):
""" convert objects to flat dicts for conversion to pandas tables. """
# dont process anything twice, only obj with resource_ids go in tables
if id(obj) in seen_ids or not hasattr(obj, "resource_id"):
return
# get the expected datatypes and fields, then extract values into dict
name = getattr(obj, "__name__", type(obj).__name__)
obj_dict[name].append(_get_values(obj, parent_id, event_id))
for event in event_iterable:
event_id = str(event.resource_id)
for obj, parent, attr in yield_obj_parent_attr(event):
parent_id = getattr(parent, "resource_id", None)
_obj_to_dict(obj, parent_id=parent_id, event_id=event_id)
obj_dict.pop("ResourceIdentifier", None)
tables = {
i: pd.DataFrame(v).set_index("resource_id")
for i, v in obj_dict.items()
}
tables["ID"] = _create_resource_id_tables(tables)
return tables
def attach_all_resource_ids(event: Event):
""" recurse all objects in a events and set referred objects """
rid_to_object = {}
# first pass, bind all resource ids to parent
for rid, parent, attr in yield_obj_parent_attr(event, ResourceIdentifier):
if attr == "resource_id":
# if the object has already been set and is not unique, raise
rid.set_referred_object(parent)
if rid.id in rid_to_object:
assert rid.get_referred_object() is rid_to_object[rid.id]
# else set referred object
rid_to_object[rid.id] = parent
# second pass, bind all other resource ids to correct resource ids
for rid, parent, attr in yield_obj_parent_attr(event, ResourceIdentifier):
if attr != "resource_id" and rid.id in rid_to_object:
rid.set_referred_object(rid_to_object[rid.id])
def prune_catalog(self, catalog):
""" recurse a events and set all attrs that eval to False to None.
This is needed to overcome some Catalog oddities to fairly compare two
catalogs. """
skips = (obspy.UTCDateTime, ev.ResourceIdentifier)
cat = catalog.copy()
for obj, parent, attr in yield_obj_parent_attr(cat):
if isinstance(obj, skips):
continue
for item, val in obj.__dict__.items():
setattr(obj, item, val or None)
return cat
def simple_catalog_to_merge(self, bingham_catalog):
"""
Create a simple catalog to merge into bingham_cat using only one event.
"""
cat = obspy.Catalog(events=bingham_catalog[:2]).copy()
# drop first pick
cat[0].picks = cat[0].picks[1:]
# modify the picks to whole seconds, reset pick IDS
for pick, _, _ in yield_obj_parent_attr(cat, ev.Pick):
pick.time -= (pick.time.timestamp) % 1
pick.id = ev.ResourceIdentifier(referred_object=pick)
return cat
def _remove_empty_quantity_errors(catalog):
"""
Copy the catalog and set all empty QunatityErrors to None.
This is needed to evaluate equality of catalogs that may have
None or empty QuantityErrors.
Fixed in https://github.com/obspy/obspy/pull/2185
"""
cat = catalog.copy()
for obj, parent, attr in yield_obj_parent_attr(cat, cls=ev.QuantityError):
if not obj:
setattr(parent, attr, None)
return cat
def simple_catalog_to_merge(self, bingham_catalog):
"""
Create a simple catalog to merge into bingham_cat using only two event.
"""
events = sorted(bingham_catalog, key=get_reference_time)
cat = obspy.Catalog(events=events[:2]).copy()
# drop first pick
cat[0].picks = cat[0].picks[1:]
# modify the picks to whole seconds, reset pick IDS
for pick, _, _ in yield_obj_parent_attr(cat, ev.Pick):
nearest_second = np.round(pick.time.timestamp)
pick.time = obspy.UTCDateTime(nearest_second)
# pick.resource_id = ev.ResourceIdentifier(referred_object=pick)
return cat
def duplicate_events(
event: catalog_or_event, resource_generator: Optional[Callable] = None
) -> catalog_or_event:
"""
Duplicate an event.
Creates a copy of the event, and assigns new resource IDs to all
internal objects (ie picks, station_magnitudes, etc.). Will not
assign new resource_id to those representing external resources
(eg analyst, velocity models, etc.).
Parameters
----------
event
The event to duplicate.
resource_generator:
A callable that can be used to generate resource IDs.
"""
new = copy.deepcopy(event)
gen_func = resource_generator or ResourceIdentifier
id_map = {} # map old id to new
# first pass through bind resource id to parent if attr is resource_id
for rid, parent, attr in yield_obj_parent_attr(new, ResourceIdentifier):
if attr == "resource_id":
try:
prefix = rid.prefix
except AttributeError:
prefix = "smi:local"
new_rid = gen_func(referred_object=parent, prefix=prefix)
id_map[rid.id] = new_rid
setattr(parent, attr, new_rid)
# second pass, swap out resource ids with the same id as those in id_map
for rid, parent, attr in yield_obj_parent_attr(new, ResourceIdentifier):
if rid.id in id_map and attr != "resource_id":
setattr(parent, attr, id_map[rid.id])
return new
def test_catalog(self, null_catalog):
""" ensure all nullish catalog chars are replaced """
cat = obsplus.utils.replace_null_nlsc_codes(null_catalog.copy())
for pick, _, _ in yield_obj_parent_attr(cat, cls=ev.Pick):
wid = pick.waveform_id
assert wid.location_code == ""
def test_nullish_codes_replaced(self, cat_nullish_nslc_codes):
""" Nullish location codes should be replace with empty strings. """
kwargs = dict(obj=cat_nullish_nslc_codes, cls=WaveformStreamID)
for obj, _, _ in yield_obj_parent_attr(**kwargs):
assert obj.location_code == ""
def _decompose_generic(events):
""" Decompose an event or a catalog. """
out = defaultdict(list)
for obj, parent, attr in yield_obj_parent_attr(events):
out[type(obj)].append(obj)
return out