def fetch_events(cls, headers, fill=True):
"""
Get Events from given run(s).
Parameters
----------
headers : RunHeader or iterable of RunHeader
The headers to fetch the events for
fill : bool, optional
If non-scalar data should be filled in, Defaults to True
Yields
------
event : Event
The event, optionally with non-scalar data filled in
"""
try:
headers.items()
except AttributeError:
pass
else:
headers = [headers]
for header in headers:
descriptors = find_event_descriptors(
run_start=header.run_start_uid)
for descriptor in descriptors:
for event in find_events(descriptor=descriptor):
if fill:
fill_event(event)
yield event
def test_document_funcs_for_smoke():
global run_start_uid, descriptor_uid
# todo this next line will break once NSLS-II/metadatastore#142 is merged
run_start, = find_run_starts(uid=run_start_uid)
descriptors = [desc for desc in find_event_descriptors(uid=descriptor_uid)]
run_stop, = find_run_stops(uid=run_stop_uid)
documents = [run_start, run_stop]
documents.extend(descriptors)
attrs = ['__repr__', '__str__', '_repr_html_', ]
for doc, attr in product(documents, attrs):
getattr(doc, attr)()
def update(self):
"""Obtain a fresh list of the relevant Events."""
# like fetch_events, but we don't fill in the data right away
events = []
for header in self.headers:
descriptors = find_event_descriptors(run_start=header.run_start_uid)
for descriptor in descriptors:
events.extend(list(find_events(descriptor=descriptor)))
if not events:
return
new_events = []
for event in events:
if event.uid not in self._known_uids:
new_events.append(event)
self._known_uids.add(event.uid)
# The major performance savings is here: only fill the new events.
[fill_event(event) for event in new_events]
self._queue.append(new_events) # the entry can be an empty list
def from_run_start(cls, run_start, verify_integrity=True):
"""
Build a Header from a RunStart Document.
Parameters
----------
run_start : metadatastore.document.Document
Returns
-------
header : dataportal.broker.Header
"""
header = Header()
header._name = "Header"
header.event_descriptors = list(
find_event_descriptors(run_start=run_start))
run_stops = list(find_run_stops(run_start=run_start.uid))
try:
run_stop, = run_stops
except ValueError:
num = len(run_stops)
run_stop = None
if num == 0:
error_msg = ("A RunStop record could not be found for the "
"run with run_start_uid {0}".format(run_start.uid))
warnings.warn(error_msg)
else:
error_msg = (
"{0} RunStop records (uids {1}) were found for the run "
"with run_start_uid {2}".format(
num, [rs.uid for rs in run_stops], run_start.uid))
if verify_integrity:
raise IntegrityError(error_msg)
else:
warnings.warn(error_msg)
# Map keys from RunStart and RunStop onto Header.
run_start_renames = {'start_time': 'time',
'start_datetime': 'time_as_datetime',
'scan_id': 'scan_id',
'beamline_id': 'beamline_id',
'owner': 'owner',
'group': 'group',
'project': 'project',
'run_start_id': 'id',
'run_start_uid': 'uid'}
run_start_renames_back = {v: k for k, v
in run_start_renames.items()}
for k in run_start:
new_key = run_start_renames_back.get(k, k)
header[new_key] = run_start[k]
if run_stop is not None:
run_stop_renames = {'stop_time': 'time',
'stop_datetime': 'time_as_datetime',
'exit_reason': 'reason',
'exit_status': 'exit_status',
'run_stop_id': 'id',
'run_stop_uid': 'uid'}
run_stop_renames_back = {v: k for k, v
in run_stop_renames.items()}
for k in run_stop:
new_key = run_stop_renames_back.get(k, k)
header[new_key] = run_stop[k]
del header['run_start']
run_start._name = 'Header'
# Strip unuseful recursion. We don't mess with underlying Documents,
# but the header is a special case, and the repr should reflect
# its structure accurately.
for ev_desc in header.event_descriptors:
ev_desc.run_start = ev_desc.run_start.uid
return header
def find_headers(cls, **kwargs):
"""Given search criteria, find Headers describing runs.
This function returns a list of dictionary-like objects encapsulating
the metadata for a run -- start time, instruments uses, and so on.
In addition to the Parameters below, advanced users can specifiy
arbitrary queries that are passed through to mongodb.
Parameters
----------
start_time : time-like, optional
Include Headers for runs started after this time. Valid
"time-like" representations are:
- float timestamps (seconds since 1970), such as time.time()
- '2015'
- '2015-01'
- '2015-01-30'
- '2015-03-30 03:00:00'
- Python datetime objects, such as datetime.datetime.now()
stop_time: time-like, optional
Include Headers for runs started before this time. See
`start_time` above for examples.
beamline_id : str, optional
String identifier for a specific beamline
project : str, optional
Project name
owner : str, optional
The username of the logged-in user when the scan was performed
scan_id : int, optional
Integer scan identifier
uid : str, optional
Globally unique id string provided to metadatastore
_id : str or ObjectId, optional
The unique id generated by mongo
data_key : str, optional
The alias (e.g., 'motor1') or PV identifier of data source
Returns
-------
data : list
Header objects
Examples
--------
>>> find_headers(start_time='2015-03-05', stop_time='2015-03-10')
>>> find_headers(data_key='motor1')
>>> find_headers(data_key='motor1', start_time='2015-03-05')
"""
data_key = kwargs.pop('data_key', None)
run_start = find_run_starts(**kwargs)
if data_key is not None:
node_name = 'data_keys.{0}'.format(data_key)
query = {node_name: {'$exists': True}}
descriptors = []
for rs in run_start:
descriptor = find_event_descriptors(run_start=rs, **query)
for d in descriptor:
descriptors.append(d)
# query = {node_name: {'$exists': True},
# 'run_start_id': {'$in': [ObjectId(rs.id) for rs in run_start]}}
# descriptors = find_event_descriptors(**query)
result = []
known_uids = deque()
for descriptor in descriptors:
if descriptor.run_start.uid not in known_uids:
rs = descriptor.run_start
known_uids.append(rs.uid)
result.append(rs)
run_start = result
result = []
for rs in run_start:
result.append(Header.from_run_start(rs))
return Headers(result)
