def new_func(*args, **kwargs):
runval = run(func.__name__) if isinstance(run, collections.Callable) else run
if runval is False:
return
elif runval is True:
pass
elif runval is None:
if status['fail']:
return
else:
raise ValueError('run value = %s but must be True, False, or None' % runval)
logger.verbose('')
logger.verbose('-' * 60)
logger.info(' Running task: %s at %s' % (func.__name__, time.ctime()))
logger.verbose('-' * 60)
try:
func(*args, **kwargs)
pyyaks.context.store_context(status.get('context_file'))
except KeyboardInterrupt:
raise
except TaskSkip:
pass
except:
if status['fail'] is False:
logger.error('%s: %s\n\n' % (func.__name__, traceback.format_exc()))
status['fail'] = True
def _sync_stat_archive(opt, msid_files, logger, content, stat, index_tbl):
"""
Actual worker for syncing the stat archive for ``content``.
"""
# Get the last row of data from the length of the TIME.col (or archfiles?)
ft = fetch.ft
ft['content'] = content
ft['interval'] = stat
stats_dir = Path(msid_files['statsdir'].abs)
if not stats_dir.exists():
logger.debug(f'Skipping {stat} data for {content}: no directory')
return
logger.info('')
logger.info(f'Processing {stat} data for {content}')
# Get the MSIDs that are in client archive
msids = [str(fn.name)[:-3] for fn in stats_dir.glob('*.h5')]
if not msids:
logger.debug(f'Skipping {stat} data for {content}: no stats h5 files')
return
else:
logger.debug(f'Stat msids are {msids}')
last_date_id, last_date_id_file = get_last_date_id(
msid_files, msids, stat, logger)
logger.verbose(f'Got {last_date_id} as last date_id that was applied to archive')
# Get list of applicable dat objects (new data, before opt.date_stop). Also
# return ``date_id`` which is the date_id of the final data set in the list.
# This will be written as the new ``last_date_id``.
try:
dats, date_id = get_stat_data_sets(ft, index_tbl, last_date_id, logger, opt)
except urllib.error.URLError as err:
if 'timed out' in str(err):
msg = f' ERROR: timed out getting {stat} data for {content}'
logger.error(msg)
process_errors.append(msg)
return
else:
raise
if not dats:
return
dat, msids = concat_data_sets(dats, ['data'])
with DelayedKeyboardInterrupt(logger):
with timing_logger(logger, f'Applying updates to {len(msids)} h5 files'):
for msid in msids:
fetch.ft['msid'] = msid
stat_file = msid_files['stats'].abs
if os.path.exists(stat_file):
append_stat_col(dat, stat_file, msid, date_id, opt, logger)
logger.debug(f'Updating {last_date_id_file} with {date_id}')
if not opt.dry_run:
with open(last_date_id_file, 'w') as fh:
fh.write(f'{date_id}')
def sync_full_archive(opt, msid_files, logger, content, index_tbl):
"""
Sync the archive for ``content``.
:param opt:
:param msid_files:
:param logger:
:param content:
:param index_tbl: index of sync file entries
:return:
"""
# Get the last row of data from the length of the TIME.col (or archfiles?)
ft = fetch.ft
ft['content'] = content
ft['msid'] = 'TIME'
ft['interval'] = 'full'
# If no TIME.h5 file then no point in going further
time_file = Path(msid_files['msid'].abs)
if not time_file.exists():
logger.debug(f'Skipping full data for {content}: no {time_file} file')
return
logger.info('')
logger.info(f'Processing full data for {content}')
# Get the 0-based index of last available full data row
with tables.open_file(str(time_file), 'r') as h5:
last_row_idx = len(h5.root.data) - 1
# Look for index table rows that have new data => the row ends after the last existing
# data. Note: row0 and row1 correspond to the slice row0:row1, so up to but
# not including the row indexed row1 (0-based). So for 3 existing rows,
# last_row_idx=2 so to get the new row with index=3 you need row1=4, or equivalently
# row1 > n_rows. By def'n we know that row0 <= 3 at this point.
ok = index_tbl['row1'] > last_row_idx + 1
if np.count_nonzero(ok) == 0:
logger.info(f'No new sync data for {content}: no new rows in index table')
index_tbl = index_tbl[ok]
try:
dats = get_full_data_sets(ft, index_tbl, logger, opt)
except urllib.error.URLError as err:
if 'timed out' in str(err):
msg = f' ERROR: timed out getting full data for {content}'
logger.error(msg)
process_errors.append(msg)
dats = []
else:
raise
if dats:
dat, msids = concat_data_sets(dats, ['data', 'quality'])
with DelayedKeyboardInterrupt(logger):
update_full_h5_files(dat, logger, msid_files, msids, opt)
update_full_archfiles_db3(dat, logger, msid_files, opt)
def _append_h5_col_tlm(msid, epoch, times=None, values=None, apply_direct=False):
"""Append new values to an HDF5 MSID data table.
Parameters
----------
msids : <class 'list'> of msids with data buffered for appending to the
archive.
"""
global _times
global _values
ft['msid'] = msid
values_h5 = tables.open_file(
f"{TELEMETRY_ARCHIVE}/data/tlm/{msid}/values.h5",
mode='a'
)
if values_h5.__contains__('/values') is False:
raise IOError(f"Archive file for {msid} value data does not contain root '/values'")
times_h5 = tables.open_file(
f"{TELEMETRY_ARCHIVE}/data/tlm/{msid}/times.h5",
mode='a'
)
if times_h5.__contains__('/times') is False:
raise IOError(f"Archive file for {msid} time data does not contain root '/times'")
# Index should point to current number of rows
index = values_h5.root.values.nrows
# TODO: Filter interval before append
# if index != 0:
# last_time = times_h5.root.times[index - 1]
# tstart = np.atleast_1d(_times[msid])[0]
# tstop = np.atleast_1d(_times[msid])[-1]
try:
if apply_direct == True:
values_h5.root.values.append(np.atleast_1d(values))
times_h5.root.times.append(np.atleast_1d(times))
else:
values_h5.root.values.append(np.atleast_1d(_values[msid]))
times_h5.root.times.append(np.atleast_1d(_times[msid]))
except Exception as err:
logger.error(f'{msid} couldnt append the normal way {type(_values[msid])} | {[_values[msid]]} | {_values[msid]}')
_update_index_file(msid, epoch, index)
values_h5.close()
times_h5.close()
return 0
def new_func(*args, **kwargs):
try:
self.setup()
func(*args, **kwargs)
except (KeyboardInterrupt, TaskSkip):
raise
except:
if status['fail'] is False:
logger.error('%s: %s\n\n' % (func.__name__, traceback.format_exc()))
status['fail'] = True
raise
finally:
self.teardown()
def process_obsids(start, stop):
obsids = events.obsids.filter(start, stop)
for obsid_event in obsids:
obsid = obsid_event.obsid
filename = os.path.join('data', str(obsid) + '.pkl')
if os.path.exists(filename):
logger.info('Skipping obsid {}, file exists'.format(obsid))
continue
else:
logger.info('**********************************')
logger.info('Processing obsid {}'.format(obsid))
logger.info('**********************************')
try:
dat = get_obsid(obsid)
except Exception as err:
logger.error('ERROR in obsid {}: {}\n'.format(obsid, err))
open(filename + '.ERR', 'w')
else:
pickle.dump(dat, open(filename, 'w'), protocol=-1)
logger.info('Success for {}\n'.format(obsid))
def main(args=None):
global fetch # fetch module, see below
# Setup for updating the sync repository
opt = get_options(args)
# Set up logging
loglevel = int(opt.log_level)
logger = pyyaks.logger.get_logger(name='cheta_update_client_archive', level=loglevel,
format="%(asctime)s %(message)s")
# If --data-root is supplied then set the fetch msid_files basedir via ENG_ARCHIVE
# prior to importing fetch. This ensures that ``content`` is consistent with
# the destination archive.
if opt.data_root is not None:
if not Path(opt.data_root).exists():
raise FileNotFoundError(
f'local cheta archive directory {Path(opt.data_root).absolute()} not found')
os.environ['ENG_ARCHIVE'] = opt.data_root
fetch = importlib.import_module('.fetch', __package__)
# Turn things around and define data_root based on fetch, relying on it to
# find the archive if --data-root is not specified.
opt.data_root = fetch.msid_files.basedir
logger.info(f'Running cheta_update_client_archive version {__version__}')
logger.info(f' {__file__}')
logger.info('')
logger.info(f'Updating client archive at {fetch.msid_files.basedir}')
if opt.add_msids:
add_msids(opt, logger)
return
if opt.content:
contents = opt.content
else:
# fetch.content is a dict of {MSID: content_type} values
contents = set(fetch.content.values())
# Global list of timeout errors
process_errors.clear()
for content in sorted(contents):
fetch.ft['content'] = content
index_tbl = get_index_tbl(content, logger, opt)
if index_tbl is not None:
sync_full_archive(opt, fetch.msid_files, logger, content, index_tbl)
for stat in STATS_DT:
sync_stat_archive(opt, fetch.msid_files, logger, content, stat, index_tbl)
if process_errors:
logger.error('')
logger.error('PROCESS ERRORS or WARNINGS:')
for process_error in process_errors:
logger.error(process_error)
def update_index_file(index_file, opt, logger):
"""Update the top-level index file of data available in the sync archive
:param index_file: Path of index ECSV file
:param opt: options
:param logger: output logger
:return: index table (astropy Table)
"""
if index_file.exists():
# Start time of last update contained in the sync repo (if it exists), but do not look
# back more than max_lookback days. This is relevant for rarely sampled
# content like cpe1eng.
filetime0 = (DateTime(opt.date_stop) - opt.max_lookback).secs
index_tbl = Table.read(index_file)
if len(index_tbl) == 0:
# Need to start with a fresh index_tbl since the string column will end up
# with a length=1 string (date_id) and add_row later will give the wrong result.
index_tbl = None
else:
filetime0 = max(filetime0, index_tbl['filetime1'][-1])
else:
# For initial index file creation use the --date-start option
index_tbl = None
filetime0 = DateTime(opt.date_start).secs
max_secs = int(opt.max_days * 86400)
time_stop = DateTime(opt.date_stop).secs
# Step through the archfile files entries and collect them into groups of up
# to --max-days based on file time stamp (which is an integer in CXC secs).
rows = []
filename = fetch.msid_files['archfiles'].abs
logger.verbose(f'Opening archfiles {filename}')
with DBI(dbi='sqlite', server=filename) as dbi:
while True:
filetime1 = min(filetime0 + max_secs, time_stop)
logger.verbose(
f'select from archfiles '
f'filetime > {DateTime(filetime0).fits[:-4]} {filetime0} '
f'filetime <= {DateTime(filetime1).fits[:-4]} {filetime1} ')
archfiles = dbi.fetchall(f'select * from archfiles '
f'where filetime > {filetime0} '
f'and filetime <= {filetime1} '
f'order by filetime ')
# Found new archfiles? If so get a new index table row for them.
if len(archfiles) > 0:
rows.append(get_row_from_archfiles(archfiles))
filedates = DateTime(archfiles['filetime']).fits
logger.verbose(f'Got {len(archfiles)} archfiles rows from '
f'{filedates[0]} to {filedates[-1]}')
filetime0 = filetime1
# Stop if already queried out to the end of desired time range
if filetime1 >= time_stop:
break
if not rows:
logger.info(f'No updates available for content {fetch.ft["content"]}')
return index_tbl
# Create table from scratch or add new rows. In normal processing there
# will just be one row per run.
if index_tbl is None:
index_tbl = Table(rows)
else:
for row in rows:
index_tbl.add_row(row)
if not index_file.parent.exists():
logger.info(f'Making directory {index_file.parent}')
index_file.parent.mkdir(exist_ok=True, parents=True)
msg = check_index_tbl_consistency(index_tbl)
if msg:
msg += '\n'
msg += '\n'.join(index_tbl.pformat(max_lines=-1, max_width=-1))
logger.error(f'Index table inconsistency: {msg}')
return None
logger.info(f'Writing {len(rows)} row(s) to index file {index_file}')
index_tbl.write(index_file, format='ascii.ecsv')
return index_tbl
def main():
global logger
opt = get_opt()
logger = pyyaks.logger.get_logger(name='kadi',
level=opt.log_level,
format="%(asctime)s %(message)s")
log_run_info(logger.info, opt)
# Set the global root data directory. This gets used in the django
# setup to find the sqlite3 database file.
os.environ['KADI'] = os.path.abspath(opt.data_root)
from .paths import EVENTS_DB_PATH
logger.info('Event database : {}'.format(EVENTS_DB_PATH()))
logger.info('')
from .events import models
# Allow for a cmd line option --start. If supplied then loop the
# effective value of opt.stop from start to the cmd line
# --stop in steps of --max-lookback-time
if opt.start is None:
date_stops = [opt.stop]
else:
t_starts = np.arange(
DateTime(opt.start).secs,
DateTime(opt.stop).secs, opt.loop_days * 86400.)
date_stops = [DateTime(t).date for t in t_starts]
date_stops.append(opt.stop)
# Get the event classes in models module
EventModels = [
Model for name, Model in vars(models).items()
if (isinstance(Model, type) # is a class
and issubclass(Model, models.BaseEvent) # is a BaseEvent subclass
and 'Meta' not in Model.__dict__ # is not a base class
and hasattr(Model, 'get_events') # can get events
)
]
# Filter on ---model command line arg(s)
if opt.models:
EventModels = [
x for x in EventModels if any(
re.match(y, x.__name__) for y in opt.models)
]
# Update priority (higher priority value means earlier in processing)
EventModels = sorted(EventModels,
key=lambda x: x.update_priority,
reverse=True)
for EventModel in EventModels:
try:
if opt.delete_from_start and opt.start is not None:
delete_from_date(EventModel, opt.start)
for date_stop in date_stops:
update(EventModel, date_stop)
except Exception:
# Something went wrong, but press on with processing other EventModels
import traceback
logger.error(f'ERROR in processing {EventModel}')
logger.error(f'Traceback:\n{traceback.format_exc()}')
