def get_closed_projects(projs, pj_con, seconds):
"""Takes list of project and gives project list that are closed
more than given time(as seconds)
:param list projs: list of projects to check
:param obj pj_con: connection object to project database
:param int seconds: Days/hours converted as seconds to check
"""
closed_projs = []
for proj in projs:
if proj not in pj_con.name_view.keys():
logger.warn(
"Project {} is not in database, so SKIPPING it..".format(proj))
continue
proj_db_obj = pj_con.get_entry(proj)
try:
proj_close_date = proj_db_obj['close_date']
except KeyError:
logger.warn(
"Project {} is either open or too old, so SKIPPING it..".
format(proj))
continue
if misc.to_seconds(days=misc.days_old(
proj_close_date, date_format='%Y-%m-%d')) > seconds:
closed_projs.append(proj)
return closed_projs
def get_closed_proj_info(prj, pdoc):
"""check and return a dict if project is closed"""
pdict = None
if not pdoc:
logger.warn(
"Seems like project {} dont have a proper statudb document, skipping it"
.format(prj))
elif "close_date" in pdoc:
closed_date = pdoc['close_date']
closed_days = misc.days_old(closed_date, "%Y-%m-%d")
if closed_days is not None and isinstance(closed_days, int):
pdict = {
'name':
pdoc.get('project_name'),
'pid':
pdoc.get('project_id'),
'closed_date':
closed_date,
'closed_days':
closed_days,
'bioinfo_responsible':
pdoc.get('project_summary',
{}).get('bioinfo_responsible',
'').encode('ascii', 'ignore')
}
else:
logger.warn(
"Problem calculating closed days for project {} with close data {}. Skipping it"
.format(pdoc.get('project_name'), closed_date))
return pdict
def cleanup_uppmax(site, days, dry_run=False):
"""Remove project/run that have been closed more than 'days'
from the given 'site' on uppmax
:param str site: site where the cleanup should be performed
:param int days: number of days to check for closed projects
"""
days = check_days(site, days, config)
if not days:
return
root_dir = CONFIG.get("cleanup").get(site).get("root")
deleted_log = CONFIG.get("cleanup").get("deleted_log")
assert os.path.exists(os.path.join(root_dir, deleted_log)), "Log directory {} doesn't exist in {}".format(
deleted_log, root_dir
)
log_file = os.path.join(root_dir, "{fl}/{fl}.log".format(fl=deleted_log))
# make a connection for project db #
pcon = statusdb.ProjectSummaryConnection()
assert pcon, "Could not connect to project database in StatusDB"
if site != "archive":
## work flow for cleaning up illumina/analysis ##
projects = [p for p in os.listdir(root_dir) if re.match(filesystem.PROJECT_RE, p)]
list_to_delete = get_closed_projects(projects, pcon, days)
else:
##work flow for cleaning archive ##
list_to_delete = []
archived_in_swestore = filesystem.list_runs_in_swestore(
path=CONFIG.get("cleanup").get("swestore").get("root"), no_ext=True
)
runs = [r for r in os.listdir(root_dir) if re.match(filesystem.RUN_RE, r)]
with filesystem.chdir(root_dir):
for run in runs:
fc_date = run.split("_")[0]
if misc.days_old(fc_date) > days:
if run in archived_in_swestore:
list_to_delete.append(run)
else:
logger.warn(
"Run {} is older than {} days but not in " "swestore, so SKIPPING".format(run, days)
)
## delete and log
for item in list_to_delete:
if dry_run:
logger.info("Will remove {} from {}".format(item, root_dir))
continue
try:
shutil.rmtree(os.path.join(root_dir, item))
logger.info("Removed project {} from {}".format(item, root_dir))
with open(log_file, "a") as to_log:
to_log.write("{}\t{}\n".format(item, datetime.strftime(datetime.now(), "%Y-%m-%d %H:%M")))
except OSError:
logger.warn("Could not remove path {} from {}".format(item, root_dir))
continue
def cleanup_uppmax(site, days, dry_run=False):
"""Remove project/run that have been closed more than 'days'
from the given 'site' on uppmax
:param str site: site where the cleanup should be performed
:param int days: number of days to check for closed projects
"""
days = check_days(site, days, config)
if not days:
return
root_dir = CONFIG.get('cleanup').get(site).get('root')
deleted_log = CONFIG.get('cleanup').get('deleted_log')
assert os.path.exists(os.path.join(root_dir,deleted_log)), "Log directory {} doesn't exist in {}".format(deleted_log,root_dir)
log_file = os.path.join(root_dir,"{fl}/{fl}.log".format(fl=deleted_log))
# make a connection for project db #
pcon = statusdb.ProjectSummaryConnection()
assert pcon, "Could not connect to project database in StatusDB"
if site != "archive":
## work flow for cleaning up illumina/analysis ##
projects = [ p for p in os.listdir(root_dir) if re.match(filesystem.PROJECT_RE,p) ]
list_to_delete = get_closed_projects(projects, pcon, days)
else:
##work flow for cleaning archive ##
list_to_delete = []
archived_in_swestore = filesystem.list_runs_in_swestore(path=CONFIG.get('cleanup').get('swestore').get('root'), no_ext=True)
runs = [ r for r in os.listdir(root_dir) if re.match(filesystem.RUN_RE,r) ]
with filesystem.chdir(root_dir):
for run in runs:
fc_date = run.split('_')[0]
if misc.days_old(fc_date) > days:
if run in archived_in_swestore:
list_to_delete.append(run)
else:
logger.warn("Run {} is older than {} days but not in "
"swestore, so SKIPPING".format(run, days))
## delete and log
for item in list_to_delete:
if dry_run:
logger.info('Will remove {} from {}'.format(item,root_dir))
continue
try:
shutil.rmtree(os.path.join(root_dir,item))
logger.info('Removed project {} from {}'.format(item,root_dir))
with open(log_file,'a') as to_log:
to_log.write("{}\t{}\n".format(item,datetime.strftime(datetime.now(),'%Y-%m-%d %H:%M')))
except OSError:
logger.warn("Could not remove path {} from {}"
.format(item,root_dir))
continue
def cleanup_swestore(days, dry_run=False):
"""Remove archived runs from swestore
:param int days: Threshold days to check and remove
"""
days = check_days('swestore', days, config)
if not days:
return
runs = filesystem.list_runs_in_swestore(path=CONFIG.get('cleanup').get('swestore').get('root'))
for run in runs:
date = run.split('_')[0]
if misc.days_old(date) > days:
if dry_run:
logger.info('Will remove file {} from swestore'.format(run))
continue
misc.call_external_command('irm -f {}'.format(run))
logger.info('Removed file {} from swestore'.format(run))
def cleanup_swestore(seconds, dry_run=False):
"""Remove archived runs from swestore
:param int seconds: Days/hours converted as seconds to check
"""
seconds = check_default(site, seconds, CONFIG)
if not seconds:
return
runs = filesystem.list_runs_in_swestore(path=CONFIG.get('cleanup').get('swestore').get('root'))
for run in runs:
date = run.split('_')[0]
if misc.to_seconds(misc.days_old(date)) > seconds:
if dry_run:
logger.info('Will remove file {} from swestore'.format(run))
continue
misc.call_external_command('irm -f {}'.format(run))
logger.info('Removed file {} from swestore'.format(run))
def cleanup_swestore(days, dry_run=False):
"""Remove archived runs from swestore
:param int days: Threshold days to check and remove
"""
days = check_days('swestore', days, config)
if not days:
return
runs = filesystem.list_runs_in_swestore(path=CONFIG.get('cleanup').get('swestore').get('root'))
for run in runs:
date = run.split('_')[0]
if misc.days_old(date) > days:
if dry_run:
logger.info('Will remove file {} from swestore'.format(run))
continue
misc.call_external_command('irm -f {}'.format(run))
logger.info('Removed file {} from swestore'.format(run))
def cleanup_swestore(seconds, dry_run=False):
"""Remove archived runs from swestore
:param int seconds: Days/hours converted as seconds to check
"""
seconds = check_default(site, seconds, CONFIG)
if not seconds:
return
runs = filesystem.list_runs_in_swestore(
path=CONFIG.get('cleanup').get('swestore').get('root'))
for run in runs:
date = run.split('_')[0]
if misc.to_seconds(misc.days_old(date)) > seconds:
if dry_run:
logger.info('Will remove file {} from swestore'.format(run))
continue
misc.call_external_command('irm -f {}'.format(run))
logger.info('Removed file {} from swestore'.format(run))
def get_closed_proj_info(prj, pdoc):
"""check and return a dict if project is closed"""
pdict = None
if not pdoc:
logger.warn("Seems like project {} dont have a proper statudb document, skipping it".format(prj))
elif "close_date" in pdoc:
closed_date = pdoc['close_date']
closed_days = misc.days_old(closed_date, "%Y-%m-%d")
if closed_days is not None and isinstance(closed_days, int):
pdict = {'name' : pdoc.get('project_name'),
'pid' : pdoc.get('project_id'),
'closed_date' : closed_date,
'closed_days' : closed_days,
'bioinfo_responsible' : pdoc.get('project_summary',{}).get('bioinfo_responsible','')}
else:
logger.warn("Problem calculating closed days for project {} with close data {}. Skipping it".format(
pdoc.get('project_name'), closed_date))
return pdict
def get_closed_projects(projs, pj_con, days):
"""Takes list of project and gives project list that are closed
more than given check 'days'
:param list projs: list of projects to check
:param obj pj_con: connection object to project database
:param int days: number of days to check
"""
closed_projs = []
for proj in projs:
if proj not in pj_con.name_view.keys():
logger.warn("Project {} is not in database, so SKIPPING it..".format(proj))
continue
proj_db_obj = pj_con.get_entry(proj)
try:
proj_close_date = proj_db_obj["close_date"]
except KeyError:
logger.warn("Project {} is either open or too old, so SKIPPING it..".format(proj))
continue
if misc.days_old(proj_close_date, date_format="%Y-%m-%d") > days:
closed_projs.append(proj)
return closed_projs
def get_closed_projects(projs, pj_con, seconds):
"""Takes list of project and gives project list that are closed
more than given time(as seconds)
:param list projs: list of projects to check
:param obj pj_con: connection object to project database
:param int seconds: Days/hours converted as seconds to check
"""
closed_projs = []
for proj in projs:
if proj not in pj_con.name_view.keys():
logger.warn("Project {} is not in database, so SKIPPING it.."
.format(proj))
continue
proj_db_obj = pj_con.get_entry(proj)
try:
proj_close_date = proj_db_obj['close_date']
except KeyError:
logger.warn("Project {} is either open or too old, so SKIPPING it..".format(proj))
continue
if misc.to_seconds(days=misc.days_old(proj_close_date,date_format='%Y-%m-%d')) > seconds:
closed_projs.append(proj)
return closed_projs
