def test_ctor_attrs(self):
"""
Verify that a newly created Dimension object has the following
attributes:
- name (string)
- sampsize (small float value, e.g., 0.005)
- p_sum (empty dict)
- s_sum (empty dict)
- methods
> sum_total
> load
"""
dimname = 'cos'
CrawlConfig.add_config(close=True, dct=self.cfg_dict(U.my_name()))
Checkable.Checkable.ex_nihilo()
a = Dimension(name=dimname, sampsize=0.005)
for attr in [
'name',
'sampsize',
'p_sum',
's_sum',
'sum_total',
'load',
]:
self.assertTrue(
hasattr(a, attr),
"Object %s does not have expected attribute %s" % (a, attr))
def test_sum_total(self):
"""
Return the sum of all the 'count' values in either the p_sum or s_sum
dictionary.
"""
CrawlConfig.add_config(close=True, dct=self.cfg_dict(U.my_name()))
Checkable.Checkable.ex_nihilo()
a = Dimension(name='sum_total')
a.p_sum = {
'6001': {
'count': 2,
'pct': 50.0
},
'5081': {
'count': 2,
'pct': 50.0
}
}
a.s_sum = {
'6001': {
'count': 2,
'pct': 40.0
},
'5081': {
'count': 3,
'pct': 60.0
}
}
self.expected(4, a.sum_total())
self.expected(4, a.sum_total(dict=a.p_sum))
self.expected(5, a.sum_total(which='s'))
self.expected(5, a.sum_total(dict=a.s_sum))
def test_alert_email_mtcaller(self):
"""
Generate an e-mail alert and verify that it was sent (this is where we
use 'monkey patching'). For this case, caller is ''.
"""
self.dbgfunc()
fakesmtp.inbox = []
logfile = self.tmpdir('alert_email.log')
targets = "[email protected], [email protected], [email protected]"
payload = 'this is an e-mail alert'
sender = 'hpssic@' + util.hostname(long=True)
cfg = CrawlConfig.CrawlConfig()
cfg.add_section('crawler')
cfg.add_section('alerts')
cfg.set('crawler', 'logpath', logfile)
cfg.set('alerts', 'email', targets)
CrawlConfig.log(logpath=logfile, close=True)
x = Alert.Alert(caller='', msg=payload, cfg=cfg)
m = fakesmtp.inbox[0]
self.expected(targets, ', '.join(m.to_address))
self.expected(m.from_address, sender)
self.expected_in('sent mail to', util.contents(logfile))
self.expected_in(payload, m.fullmessage)
def test_ctor_attrs(self):
"""
Verify that a newly created Dimension object has the following
attributes:
- name (string)
- sampsize (small float value, e.g., 0.005)
- p_sum (empty dict)
- s_sum (empty dict)
- methods
> sum_total
> load
"""
dimname = 'cos'
CrawlConfig.add_config(close=True, dct=self.cfg_dict(U.my_name()))
Checkable.Checkable.ex_nihilo()
a = Dimension(name=dimname,
sampsize=0.005)
for attr in ['name',
'sampsize',
'p_sum',
's_sum',
'sum_total',
'load',
]:
self.assertTrue(hasattr(a, attr),
"Object %s does not have expected attribute %s" %
(a, attr))
def test_alert_shell_nospec(self):
"""
Generate a shell alert and verify that it ran. With no '%s' in the
shell alert string, no message should be offered for formatting.
"""
self.dbgfunc()
logfile = self.tmpdir('alert_shell.log')
outfile = self.tmpdir('alert_shell.out')
runfile = self.tmpdir('runme')
f = open(runfile, 'w')
f.write("#!/bin/bash\n")
f.write("echo \"ALERT: $*\" > %s\n" % outfile)
f.close()
os.chmod(
runfile, stat.S_IRUSR | stat.S_IWUSR | stat.S_IXUSR | stat.S_IRGRP
| stat.S_IWGRP | stat.S_IXGRP | stat.S_IROTH | stat.S_IXOTH)
cfg = CrawlConfig.CrawlConfig()
cfg.add_section('crawler')
cfg.add_section('AlertTest')
cfg.add_section('alert_section')
cfg.set('crawler', 'logpath', logfile)
cfg.set('AlertTest', 'alerts', 'alert_section')
cfg.set('alert_section', 'shell', runfile)
CrawlConfig.log(logpath=logfile, close=True)
x = Alert.Alert(caller='AlertTest',
msg='this is a test message',
cfg=cfg)
expected = "ran: '%s'" % runfile
self.expected_in(expected, util.contents(logfile))
self.assertPathPresent(outfile)
def test_alert_email_mtcaller(self):
"""
Generate an e-mail alert and verify that it was sent (this is where we
use 'monkey patching'). For this case, caller is ''.
"""
self.dbgfunc()
fakesmtp.inbox = []
logfile = self.tmpdir('alert_email.log')
targets = "[email protected], [email protected], [email protected]"
payload = 'this is an e-mail alert'
sender = 'hpssic@' + util.hostname(long=True)
cfg = CrawlConfig.CrawlConfig()
cfg.add_section('crawler')
cfg.add_section('alerts')
cfg.set('crawler', 'logpath', logfile)
cfg.set('alerts', 'email', targets)
CrawlConfig.log(logpath=logfile, close=True)
x = Alert.Alert(caller='', msg=payload,
cfg=cfg)
m = fakesmtp.inbox[0]
self.expected(targets, ', '.join(m.to_address))
self.expected(m.from_address, sender)
self.expected_in('sent mail to', util.contents(logfile))
self.expected_in(payload, m.fullmessage)
def test_alert_shell_nospec(self):
"""
Generate a shell alert and verify that it ran. With no '%s' in the
shell alert string, no message should be offered for formatting.
"""
self.dbgfunc()
logfile = self.tmpdir('alert_shell.log')
outfile = self.tmpdir('alert_shell.out')
runfile = self.tmpdir('runme')
f = open(runfile, 'w')
f.write("#!/bin/bash\n")
f.write("echo \"ALERT: $*\" > %s\n" % outfile)
f.close()
os.chmod(runfile,
stat.S_IRUSR | stat.S_IWUSR | stat.S_IXUSR |
stat.S_IRGRP | stat.S_IWGRP | stat.S_IXGRP |
stat.S_IROTH | stat.S_IXOTH)
cfg = CrawlConfig.CrawlConfig()
cfg.add_section('crawler')
cfg.add_section('AlertTest')
cfg.add_section('alert_section')
cfg.set('crawler', 'logpath', logfile)
cfg.set('AlertTest', 'alerts', 'alert_section')
cfg.set('alert_section', 'shell', runfile)
CrawlConfig.log(logpath=logfile, close=True)
x = Alert.Alert(caller='AlertTest', msg='this is a test message',
cfg=cfg)
expected = "ran: '%s'" % runfile
self.expected_in(expected, util.contents(logfile))
self.assertPathPresent(outfile)
def test_load_new(self):
"""
With the database and checkables table in place, create a new Dimension
that is not in the table. Calling load() on it should be a no-op -- the
object should not be stored to the database and its contents should not
be changed.
"""
self.dbgfunc()
# reboot the database and call persist() to create the table without
# adding any data
U.conditional_rm(self.dbname())
CrawlConfig.add_config(close=True, dct=self.cfg_dict(U.my_name()))
Checkable.Checkable.ex_nihilo()
ignore = Dimension(name='foobar')
# get a Dimension object that is not in the table
test = Dimension(name='notindb')
# make a copy of the object for reference (not just a handle to the
# same ojbect)
ref = copy.deepcopy(test)
# call load(), which should be a no op
test.load()
# verify that the object didn't change
self.expected(ref.name, test.name)
self.expected(ref.sampsize, test.sampsize)
self.expected(ref.p_sum, test.p_sum)
self.expected(ref.s_sum, test.s_sum)
def test_load_new(self):
"""
With the database and checkables table in place, create a new Dimension
that is not in the table. Calling load() on it should be a no-op -- the
object should not be stored to the database and its contents should not
be changed.
"""
self.dbgfunc()
# reboot the database and call persist() to create the table without
# adding any data
U.conditional_rm(self.dbname())
CrawlConfig.add_config(close=True, dct=self.cfg_dict(U.my_name()))
Checkable.Checkable.ex_nihilo()
ignore = Dimension(name='foobar')
# get a Dimension object that is not in the table
test = Dimension(name='notindb')
# make a copy of the object for reference (not just a handle to the
# same ojbect)
ref = copy.deepcopy(test)
# call load(), which should be a no op
test.load()
# verify that the object didn't change
self.expected(ref.name, test.name)
self.expected(ref.sampsize, test.sampsize)
self.expected(ref.p_sum, test.p_sum)
self.expected(ref.s_sum, test.s_sum)
def test_repr(self):
"""
Method __repr__ should return <Dimension(name='foo')>.
"""
CrawlConfig.add_config(close=True, dct=self.cfg_dict(U.my_name()))
exp = "Dimension(name='foo')"
a = eval(exp)
self.expected(exp, a.__repr__())
def test_repr(self):
"""
Method __repr__ should return <Dimension(name='foo')>.
"""
CrawlConfig.add_config(close=True, dct=self.cfg_dict(U.my_name()))
exp = "Dimension(name='foo')"
a = eval(exp)
self.expected(exp, a.__repr__())
def test_load_already(self):
"""
With the database and a checkables table in place and records in the
table, calling load() on a Dimension should load the information from
the table into the object. However, it should only count records where
last_check <> 0.
"""
self.dbgfunc()
U.conditional_rm(self.dbname())
CrawlConfig.add_config(close=True, dct=self.cfg_dict(U.my_name()))
Checkable.Checkable.ex_nihilo()
chk = Checkable.Checkable
testdata = [
chk(rowid=1, path="/abc/001", type='f', cos='6001', checksum=0,
last_check=0),
chk(rowid=2, path="/abc/002", type='f', cos='6002', checksum=0,
last_check=5),
chk(rowid=3, path="/abc/003", type='f', cos='6003', checksum=1,
last_check=0),
chk(rowid=4, path="/abc/004", type='f', cos='6001', checksum=1,
last_check=17),
chk(rowid=5, path="/abc/005", type='f', cos='6002', checksum=0,
last_check=0),
chk(rowid=6, path="/abc/006", type='f', cos='6003', checksum=0,
last_check=8),
chk(rowid=7, path="/abc/007", type='f', cos='6001', checksum=0,
last_check=0),
chk(rowid=8, path="/abc/008", type='f', cos='6002', checksum=0,
last_check=19),
chk(rowid=9, path="/abc/009", type='f', cos='6003', checksum=0,
last_check=0),
]
# insert some test data into the table
for t in testdata:
t.persist()
# get a default Dimension with the same name as the data in the table
q = Dimension(name='cos')
# this should load the data from the table into the object
q.load()
# verify the loaded data in the object
self.expected('cos', q.name)
self.assertTrue('6001' in q.p_sum.keys(),
"Expected '6001' in p_sum.keys()")
self.assertTrue('6002' in q.p_sum.keys(),
"Expected '6001' in p_sum.keys()")
self.assertTrue('6003' in q.p_sum.keys(),
"Expected '6003' in p_sum.keys()")
self.assertTrue('6001' in q.s_sum.keys(),
"Expected '6001' in s_sum.keys()")
self.assertTrue('6002' in q.s_sum.keys(),
"Expected '6002' in s_sum.keys()")
self.assertTrue('6003' in q.s_sum.keys(),
"Expected '6003' in s_sum.keys()")
def main(cfg):
"""
Plugin example
"""
try:
msg = cfg.get('example', 'message')
except ConfigParser.NoOptionError:
msg = 'No message in configuration'
CrawlConfig.log('EXAMPLE: This is plugin EXAMPLE saying "%s"' % msg)
def main(cfg):
"""
Migration Purge Record Ager (mpra) reads the database tables BFMIGRREC and
BFPURGEREC and reports migration and purge records that are older than the
age specified in the configuration.
"""
if cfg is None:
cfg = CrawlConfig.get_config()
age = cfg.get_time('mpra', 'age')
end = time.time() - age
start = mpra_lib.mpra_fetch_recent("migr")
#
# If the configured age has been moved back in time, so that end is before
# start, we need to reset and start scanning from the beginning of time.
#
if end < start:
start = 0
CrawlConfig.log("migr recs after %d (%s) before %d (%s)" %
(start, util.ymdhms(start), end, util.ymdhms(end)))
result = mpra_lib.age("migr", start=start, end=end, mark=True)
CrawlConfig.log("found %d migration records in the range" % result)
rval = result
start = mpra_lib.mpra_fetch_recent("purge")
CrawlConfig.log("Looking for expired purge locks")
result = mpra_lib.xplocks(mark=True)
CrawlConfig.log("found %d expired purge locks" % result)
rval += result
return rval
def test_ctor_defaults(self):
"""
A new Dimension with only the name specified should have the right
defaults.
"""
dimname = 'cos'
CrawlConfig.add_config(close=True, dct=self.cfg_dict(U.my_name()))
a = Dimension(name=dimname)
self.expected(dimname, a.name)
self.expected(0.01, a.sampsize)
self.expected({}, a.p_sum)
self.expected({}, a.s_sum)
def test_ctor_defaults(self):
"""
A new Dimension with only the name specified should have the right
defaults.
"""
dimname = 'cos'
CrawlConfig.add_config(close=True, dct=self.cfg_dict(U.my_name()))
a = Dimension(name=dimname)
self.expected(dimname, a.name)
self.expected(0.01, a.sampsize)
self.expected({}, a.p_sum)
self.expected({}, a.s_sum)
def test_ctor_bad_attr(self):
"""
Attempting to create a Dimension with attrs that are not in the
settable list should get an exception.
"""
dimname = 'bad_attr'
CrawlConfig.add_config(close=True, dct=self.cfg_dict(U.my_name()))
got_exception = False
self.assertRaisesMsg(StandardError,
"Attribute 'catl' is not valid",
Dimension, name=dimname, catl=[1, 2, 3])
self.assertRaisesMsg(StandardError,
"Attribute 'aardvark' is not valid",
Dimension, name=dimname, aardvark='Fanny Brice')
def test_html_report(self):
"""
Try running 'html report > filename' and verify that 1) no traceback
occurs and 2) something is actually written to the output file.
"""
self.dbgfunc()
cfpath = self.tmpdir("crawl.cfg")
cfg = CrawlConfig.add_config()
cfg.crawl_write(open(cfpath, 'w'))
cmd = "html report --config %s" % cfpath
CrawlConfig.log(cmd, close=True)
result = pexpect.run(cmd)
if "HPSS Unavailable" in result:
pytest.skip("HPSS Unavailable")
self.validate_report(result)
def test_html_report(self):
"""
Try running 'html report > filename' and verify that 1) no traceback
occurs and 2) something is actually written to the output file.
"""
self.dbgfunc()
cfpath = self.tmpdir("crawl.cfg")
cfg = CrawlConfig.add_config()
cfg.crawl_write(open(cfpath, 'w'))
cmd = "html report --config %s" % cfpath
CrawlConfig.log(cmd, close=True)
result = pexpect.run(cmd)
if "HPSS Unavailable" in result:
pytest.skip("HPSS Unavailable")
self.validate_report(result)
def test_env_set_pre(self):
"""
TEST: set predefined environment variable from [env] entry
unconditionally
EXP: the old value gets overwritten
"""
self.dbgfunc()
sname = 'env'
evname = 'UTIL_TEST'
pre_val = "one:two:three"
add = "four:five:six"
exp = add
# make sure the target env variable is set to a known value
with util.tmpenv(evname, pre_val):
# create a config object with an 'env' section and a non-'+' option
cfg = CrawlConfig.CrawlConfig()
cfg.add_section(sname)
cfg.set(sname, evname, add)
# pass the config object to util.env_update()
util.env_update(cfg)
# verify that the target env variable now contains the new value
# and the old value is gone
self.expected(exp, os.environ[evname])
self.assertTrue(
pre_val not in os.environ[evname],
"The old value should be gone but still seems " +
"to be hanging around")
def test_env_add_folded_pre(self):
"""
TEST: add to a preset environment variable from a folded [env]
entry
EXP: the value gets set to the payload with the whitespace squeezed out
"""
self.dbgfunc()
sname = 'env'
evname = 'UTIL_TEST'
pre_val = "one:two:three"
add = "four:\n five:\n six"
exp = ":".join([pre_val, re.sub("\n\s*", "", add)])
# make sure the target env variable has the expected value
with util.tmpenv(evname, pre_val):
# create a config object with an 'env' section and a folded '+'
# option
cfg = CrawlConfig.CrawlConfig()
cfg.add_section(sname)
cfg.set(sname, evname, '+' + add)
# pass the config object to util.env_update()
util.env_update(cfg)
# verify that the variable was set to the expected value
self.expected(exp, os.environ[evname])
def test_env_add_pre(self):
"""
TEST: add to a predefined environment variable from [env] entry
EXP: payload is appended to the old value
"""
self.dbgfunc()
sname = 'env'
evname = 'UTIL_TEST'
pre_val = "one:two:three"
add = "four:five:six"
exp = ":".join([pre_val, add])
# make sure the target env variable is set to a known value
with util.tmpenv(evname, pre_val):
# create a config object with an 'env' section and a '+' option
cfg = CrawlConfig.CrawlConfig()
cfg.add_section(sname)
cfg.set(sname, evname, "+" + add)
# pass the config object to util.env_update()
util.env_update(cfg)
# verify that the target env variable now contains both old and
# added values
self.expected(exp, os.environ[evname])
def test_env_add_folded_none(self):
"""
TEST: add to an undefined environment variable from a folded [env]
entry
EXP: the value gets set to the payload with the whitespace squeezed out
"""
self.dbgfunc()
sname = 'env'
evname = 'UTIL_TEST'
add = "four:\n five:\n six"
exp = re.sub("\n\s*", "", add)
# make sure the target env variable is not defined
with util.tmpenv(evname, None):
# create a config object with an 'env' section and a '+' option
cfg = CrawlConfig.CrawlConfig()
cfg.add_section(sname)
cfg.set(sname, evname, '+' + add)
# pass the config object to util.env_update()
util.env_update(cfg)
# verify that the variable was set to the expected value
self.expected(exp, os.environ[evname])
def test_env_set_folded_none(self):
"""
TEST: set undefined environment variable from a folded [env] entry
unconditionally
EXP: the value gets set
"""
self.dbgfunc()
sname = 'env'
evname = 'UTIL_TEST'
newval = "one:\n two:\n three"
exp = re.sub("\n\s*", "", newval)
# make sure the target env variable is not defined
with util.tmpenv(evname, None):
# create a config object with an 'env' section and a non-'+' option
cfg = CrawlConfig.CrawlConfig()
cfg.add_section(sname)
cfg.set(sname, evname, newval)
# pass the config object to util.env_update()
util.env_update(cfg)
# verify that the variable was set to the expected value
self.expected(exp, os.environ[evname])
def test_ctor_reset_atime_cfg_true(self):
"""
If reset_atime is specified in the config as True, it should be True
"""
cf_name = self.tmpdir(util.my_name() + ".cfg")
# write out a config file with no reset_atime spec
self.write_cfg_file(cf_name, self.cfg_d)
# make the test config the default
CrawlConfig.get_config(cfname=cf_name, reset=True)
# get an hpss.HSI object and check its reset_atime attribute
h = hpss.HSI(connect=False)
self.expected(True, h.reset_atime)
CrawlConfig.get_config(reset=True, soft=True)
def main(cfg):
"""
Generate an html-formatted report and store it at the designated location
"""
CrawlConfig.log("html_plugin starting")
fpath = cfg.get('html', 'output_path')
rpt = html_lib.get_html_report(cfg=cfg)
npath = fpath + '.new'
opath = fpath + '.old'
with open(npath, 'w') as out:
out.write(rpt)
if os.path.exists(fpath):
os.rename(fpath, opath)
os.rename(npath, fpath)
CrawlConfig.log("html_plugin finished")
def main(cfg):
"""
HSI demo
"""
CrawlConfig.log("hsi-demo: sending output to hsi.out")
hsi_prompt = "]:"
S = pexpect.spawn("/opt/public/bin/hsi")
S.logfile = f = open("hsi.out", 'a')
S.expect(hsi_prompt)
S.sendline("ls")
S.expect(hsi_prompt)
S.sendline("quit")
S.expect(pexpect.EOF)
S.logfile.close()
S.close()
def test_alert_log(self):
"""
Generate a log alert and verify that the message was written to the
correct log file.
"""
self.dbgfunc()
logfile = self.tmpdir('alert_log.log')
cfg = CrawlConfig.CrawlConfig()
cfg.add_section('crawler')
cfg.add_section('AlertTest')
cfg.add_section('alert_section')
cfg.set('crawler', 'logpath', logfile)
cfg.set('AlertTest', 'alerts', 'alert_section')
cfg.set('alert_section', 'log', "%s")
CrawlConfig.log(logpath=logfile, close=True)
x = Alert.Alert(caller='AlertTest', msg='this is a test message',
cfg=cfg)
self.expected_in('this is a test message', util.contents(logfile))
def test_sum_total(self):
"""
Return the sum of all the 'count' values in either the p_sum or s_sum
dictionary.
"""
CrawlConfig.add_config(close=True, dct=self.cfg_dict(U.my_name()))
Checkable.Checkable.ex_nihilo()
a = Dimension(name='sum_total')
a.p_sum = {'6001': {'count': 2, 'pct': 50.0},
'5081': {'count': 2, 'pct': 50.0}
}
a.s_sum = {'6001': {'count': 2, 'pct': 40.0},
'5081': {'count': 3, 'pct': 60.0}
}
self.expected(4, a.sum_total())
self.expected(4, a.sum_total(dict=a.p_sum))
self.expected(5, a.sum_total(which='s'))
self.expected(5, a.sum_total(dict=a.s_sum))
def test_maybe_update_hsi_cant(muh_prep, tmpdir):
"""
If we don't have write permission on the target, then even if we should
update, we can't. In this case, should log a message.
"""
pytest.dbgfunc()
lp = tmpdir.join('crawl.test.log')
rf = test_maybe_update_hsi_cant
path = ":".join([rf.bin.strpath, rf.hsihome])
with U.tmpenv('PATH', path):
CrawlConfig.log(logpath=lp.strpath, close=True)
hpss.maybe_update_hsi()
c = rf.file.read()
assert 'not changed' in c
assert os.path.exists(lp.strpath)
c = lp.read()
assert MSG.hsi_wrap_ood in c
CrawlConfig.log(close=True)
def test_alert_log(self):
"""
Generate a log alert and verify that the message was written to the
correct log file.
"""
self.dbgfunc()
logfile = self.tmpdir('alert_log.log')
cfg = CrawlConfig.CrawlConfig()
cfg.add_section('crawler')
cfg.add_section('AlertTest')
cfg.add_section('alert_section')
cfg.set('crawler', 'logpath', logfile)
cfg.set('AlertTest', 'alerts', 'alert_section')
cfg.set('alert_section', 'log', "%s")
CrawlConfig.log(logpath=logfile, close=True)
x = Alert.Alert(caller='AlertTest',
msg='this is a test message',
cfg=cfg)
self.expected_in('this is a test message', util.contents(logfile))
def test_ctor_bad_attr(self):
"""
Attempting to create a Dimension with attrs that are not in the
settable list should get an exception.
"""
dimname = 'bad_attr'
CrawlConfig.add_config(close=True, dct=self.cfg_dict(U.my_name()))
got_exception = False
self.assertRaisesMsg(StandardError,
"Attribute 'catl' is not valid",
Dimension,
name=dimname,
catl=[1, 2, 3])
self.assertRaisesMsg(StandardError,
"Attribute 'aardvark' is not valid",
Dimension,
name=dimname,
aardvark='Fanny Brice')
def test_ctor_reset_atime_call_true(self):
"""
If reset_atime is specified in the call as True, it should be True,
even if it's specified as False in the config
"""
cf_name = self.tmpdir(util.my_name() + ".cfg")
# write out a config file with no reset_atime spec
cfg = copy.deepcopy(self.cfg_d)
cfg['cv']['reset_atime'] = 'no'
self.write_cfg_file(cf_name, cfg)
# make the test config the default
CrawlConfig.get_config(cfname=cf_name, reset=True)
# get an hpss.HSI object and check its reset_atime attribute
h = hpss.HSI(connect=False, reset_atime=True)
self.expected(True, h.reset_atime)
CrawlConfig.get_config(reset=True, soft=True)
def test_ctor_reset_atime_default(self):
"""
If reset_atime is not specified in the config or argument list, it
should default to False
"""
cf_name = self.tmpdir(util.my_name() + ".cfg")
# write out a config file with no reset_atime spec
cd = copy.deepcopy(self.cfg_d)
del cd['cv']['reset_atime']
self.write_cfg_file(cf_name, cd)
# make the test config the default
CrawlConfig.get_config(cfname=cf_name, reset=True)
# get an hpss.HSI object and check its reset_atime attribute
h = hpss.HSI(connect=False)
self.expected(False, h.reset_atime)
CrawlConfig.get_config(reset=True, soft=True)
def test_dlog(self):
"""
Test method dlog on daemon object
"""
lfname = self.tmpdir('daemon.dlog.log')
lf = CrawlConfig.log(logpath=lfname)
a = daemon.Daemon(self.tmpdir("daemon_pid"), logger=lf)
logmsg = "testing the dlog method of %s" % a
a.dlog(logmsg)
self.assertTrue(
logmsg in util.contents(lfname), "Expected '%s' in '%s'" %
(logmsg, util.line_quote(util.contents(lfname))))
def test_alert_use_other(self):
"""
A use directive sends us to another config section where we generate a
log alert and verify that the message was written to the correct log
file.
"""
self.dbgfunc()
logfile = self.tmpdir('alert_use.log')
cfg = CrawlConfig.CrawlConfig()
cfg.add_section('crawler')
cfg.add_section('AlertTest')
cfg.add_section('alert_section')
cfg.add_section('other_section')
cfg.set('crawler', 'logpath', logfile)
cfg.set('AlertTest', 'alerts', 'alert_section')
cfg.set('alert_section', 'use', "other_section")
cfg.set('other_section', 'log', "%s")
CrawlConfig.log(logpath=logfile, close=True)
payload = 'this is a test message from %s' % util.my_name()
x = Alert.Alert(caller='AlertTest', msg=payload, cfg=cfg)
self.expected_in(payload, util.contents(logfile))
def test_ctor_no_cv_section(self):
"""
If there is no cv section in the config, reset_atime and hash_algorithm
should take on their default values.
"""
self.dbgfunc()
cfg = copy.deepcopy(self.cfg_d)
del cfg['cv']
zcfg = CrawlConfig.add_config(close=True, dct=cfg)
self.assertFalse(zcfg.has_section('cv'))
h = hpss.HSI(connect=False)
self.expected(False, h.reset_atime)
self.expected(None, h.hash_algorithm)
def test_alert_use_other(self):
"""
A use directive sends us to another config section where we generate a
log alert and verify that the message was written to the correct log
file.
"""
self.dbgfunc()
logfile = self.tmpdir('alert_use.log')
cfg = CrawlConfig.CrawlConfig()
cfg.add_section('crawler')
cfg.add_section('AlertTest')
cfg.add_section('alert_section')
cfg.add_section('other_section')
cfg.set('crawler', 'logpath', logfile)
cfg.set('AlertTest', 'alerts', 'alert_section')
cfg.set('alert_section', 'use', "other_section")
cfg.set('other_section', 'log', "%s")
CrawlConfig.log(logpath=logfile, close=True)
payload = 'this is a test message from %s' % util.my_name()
x = Alert.Alert(caller='AlertTest', msg=payload,
cfg=cfg)
self.expected_in(payload, util.contents(logfile))
def test_alert_email_defcfg(self):
"""
Generate an e-mail alert using the default config and verify that it
was sent (this is where we use 'monkey patching').
"""
self.dbgfunc()
fakesmtp.inbox = []
CrawlConfig.add_config(close=True)
# with U.tmpenv('CRAWL_CONF', 'hpssic_test.cfg'):
with U.tmpenv('CRAWL_CONF', None):
logfile = self.tmpdir('alert_email.log')
targets = "[email protected], [email protected]"
payload = 'this is an e-mail alert'
sender = 'hpssic@' + util.hostname(long=True)
CrawlConfig.log(logpath=logfile, close=True)
x = Alert.Alert(caller='cv', msg=payload)
m = fakesmtp.inbox[0]
self.expected(', '.join(m.to_address), targets)
self.expected(m.from_address, sender)
self.expected_in('sent mail to', util.contents(logfile))
self.expected_in(payload, m.fullmessage)
def tcc_priority(globspec, cosinfo):
"""
Handle any files matching globspec. Return the number of files processed.
"""
rval = 0
cfg = CrawlConfig.get_config()
pri_compdir = cfg.get_d(tcc_lib.sectname(), 'completed', '/tmp')
for filepath in glob.glob(globspec):
tcc_lib.check_file(filepath, verbose=False, plugin=True)
cpath = U.pathjoin(pri_compdir, U.basename(filepath))
os.rename(filepath, cpath)
return rval
def test_dlog(self):
"""
Test method dlog on daemon object
"""
lfname = self.tmpdir('daemon.dlog.log')
lf = CrawlConfig.log(logpath=lfname)
a = daemon.Daemon(self.tmpdir("daemon_pid"), logger=lf)
logmsg = "testing the dlog method of %s" % a
a.dlog(logmsg)
self.assertTrue(logmsg in util.contents(lfname),
"Expected '%s' in '%s'" %
(logmsg,
util.line_quote(util.contents(lfname))))
def test_alert_email_defcfg(self):
"""
Generate an e-mail alert using the default config and verify that it
was sent (this is where we use 'monkey patching').
"""
self.dbgfunc()
fakesmtp.inbox = []
CrawlConfig.add_config(close=True)
# with U.tmpenv('CRAWL_CONF', 'hpssic_test.cfg'):
with U.tmpenv('CRAWL_CONF', None):
logfile = self.tmpdir('alert_email.log')
targets = "[email protected], [email protected]"
payload = 'this is an e-mail alert'
sender = 'hpssic@' + util.hostname(long=True)
CrawlConfig.log(logpath=logfile, close=True)
x = Alert.Alert(caller='cv', msg=payload)
m = fakesmtp.inbox[0]
self.expected(', '.join(m.to_address), targets)
self.expected(m.from_address, sender)
self.expected_in('sent mail to', util.contents(logfile))
self.expected_in(payload, m.fullmessage)
def tcc_priority(globspec, cosinfo):
"""
Handle any files matching globspec. Return the number of files processed.
"""
rval = 0
cfg = CrawlConfig.get_config()
pri_compdir = cfg.get_d(tcc_lib.sectname(), 'completed', '/tmp')
for filepath in glob.glob(globspec):
tcc_lib.check_file(filepath, verbose=False, plugin=True)
cpath = U.pathjoin(pri_compdir, U.basename(filepath))
os.rename(filepath, cpath)
return rval
def main(cfg):
"""
This plugin will generate a report and send it to the designated e-mail
address(es).
"""
rval = 0
try:
if cfg is None:
cfg = CrawlConfig.get_config()
subject = "%s %s" % (cfg.get(
'rpt',
'subject'), time.strftime("%Y.%m%d %H:%M:%S", time.localtime()))
CrawlMail.send(sender=cfg.get('rpt', 'sender'),
to='rpt.recipients',
subj=subject,
msg=rpt_lib.get_report())
except Exception as e:
rval = 1
CrawlConfig.log("Failure in rpt_lib: '%s'" % str(e))
return rval
def main(cfg):
"""
This plugin will generate a report and send it to the designated e-mail
address(es).
"""
rval = 0
try:
if cfg is None:
cfg = CrawlConfig.get_config()
subject = "%s %s" % (cfg.get('rpt', 'subject'),
time.strftime("%Y.%m%d %H:%M:%S",
time.localtime()))
CrawlMail.send(sender=cfg.get('rpt', 'sender'),
to='rpt.recipients',
subj=subject,
msg=rpt_lib.get_report())
except Exception as e:
rval = 1
CrawlConfig.log("Failure in rpt_lib: '%s'" % str(e))
return rval
def check_hash_algorithm(self, cf_stem, alg, checkfor=None):
"""
With hash_algorithm = *alg* in config
"""
if checkfor is None:
checkfor = alg
# generate a config file and make it the default config
cf_name = self.tmpdir(cf_stem + ".cfg")
cd = copy.deepcopy(self.cfg_d)
if alg == '(none)':
del cd['cv']['hash_algorithm']
else:
cd['cv']['hash_algorithm'] = alg
self.write_cfg_file(cf_name, cd)
CrawlConfig.get_config(cfname=cf_name, reset=True)
# Get an hsi object
testfile = self.plist[1]
try:
h = hpss.HSI()
except hpss.HSIerror as e:
if MSG.hpss_unavailable in str(e):
pytest.skip(str(e))
# if necessary, delete any hash on the test file
result = h.hashlist(testfile)
if "(none)" not in result:
h.hashdelete(testfile)
# generate a hash on the test file
h.hashcreate(testfile)
# verify that the hash created is of the proper type
result = h.hashlist(testfile)
self.expected_in(checkfor, result)
def test_get_html_report(self):
"""
Call html_lib.get_html_report() directly
"""
self.dbgfunc()
c = CrawlConfig.add_config()
db = CrawlDBI.DBI(dbtype="crawler")
dbschem.drop_table(table="lscos")
self.expected(False, db.table_exists(table="lscos"))
try:
result = html_lib.get_html_report('')
except hpss.HSIerror as e:
if MSG.hpss_unavailable in str(e):
pytest.skip(str(e))
self.expected(True, db.table_exists(table="lscos"))
db.close()
self.validate_report(result)
def main(cfg):
"""
Main entry point for the cv plugin
"""
# Get stuff we need -- the logger object, dataroot, etc.
CrawlConfig.log("firing up")
plugdir = cfg.get('crawler', 'plugin-dir')
dataroot = util.csv_list(cfg.get(plugin_name, 'dataroot'))
odds = cfg.getfloat(plugin_name, 'odds')
n_ops = int(cfg.get(plugin_name, 'operations'))
# Initialize our statistics
(t_checksums, t_matches, t_failures) = get_stats()
(checksums, matches, failures) = (0, 0, 0)
# Fetch the list of HPSS objects that we're looking at from the
# database
try:
clist = Checkable.Checkable.get_list(prob=odds, rootlist=dataroot)
except CrawlDBI.DBIerror as e:
if any([util.rgxin(msg, str(e))
for msg in ["no such table: checkables",
"Table '.*' doesn't exist"]]):
CrawlConfig.log("calling ex_nihilo")
Checkable.Checkable.ex_nihilo(dataroot=dataroot)
clist = Checkable.Checkable.get_list(prob=odds)
else:
raise
except StandardError as e:
if 'Please call .ex_nihilo()' in str(e):
CrawlConfig.log("calling ex_nihilo")
Checkable.Checkable.ex_nihilo(dataroot=dataroot)
clist = Checkable.Checkable.get_list(prob=odds)
else:
raise
# We're going to process n_ops things in the HPSS namespace
for op in range(n_ops):
# if the list from the database is empty, there's nothing to do
if 0 < len(clist):
# but it's not, so grab the first item and check it
item = clist.pop(0)
CrawlConfig.log("[%d] checking %s" % (item.rowid, item))
ilist = item.check()
# Expected outcomes that check can return:
# list of Checkables: read dir or checksummed files (may be empty)
# Alert: checksum verify failed
# 'access denied': unaccessible directory
# 'matched': a checksum was verified
# 'checksummed': file was checksummed
# 'skipped': file was skipped
# 'unavailable': HPSS is temporarily unavailable
# StandardError: invalid Checkable type (not 'f' or 'd')
#
if type(ilist) == str:
if ilist == "access denied":
CrawlConfig.log("dir %s not accessible" % item.path)
# clist.remove(item)
elif ilist == "matched":
matches += 1
CrawlConfig.log("%s checksums matched" % item.path)
elif ilist == "checksummed":
# checksums += 1
CrawlConfig.log("%s checksummed" % item.path)
elif ilist == "skipped":
CrawlConfig.log("%s skipped" % item.path)
elif ilist == "unavailable":
CrawlConfig.log("HPSS is not available")
break
else:
CrawlConfig.log("unexpected string returned " +
"from Checkable: '%s'" % ilist)
elif type(ilist) == list:
CrawlConfig.log("in %s, found:" % item)
for n in ilist:
CrawlConfig.log(">>> %s" % str(n))
if 'f' == n.type and n.checksum != 0:
CrawlConfig.log(".. previously checksummed")
# checksums += 1
elif isinstance(ilist, Checkable.Checkable):
CrawlConfig.log("Checkable returned - file checksummed" +
" - %s, %s" % (ilist.path, ilist.checksum))
# checksums += 1
elif isinstance(ilist, Alert.Alert):
CrawlConfig.log("Alert generated: '%s'" %
ilist.msg())
failures += 1
else:
CrawlConfig.log("unexpected return val from " +
"Checkable.check: %s: %r" %
(type(ilist), ilist))
# Report the statistics in the log
# ** For checksums, we report the current total minus the previous
# ** For matches and failures, we counted them up during the iteration
# ** See the description of get_stats for why we don't store total
# checksums
p_checksums = t_checksums
t_matches += matches
t_failures += failures
cv_lib.update_stats((t_matches, t_failures))
(t_checksums, t_matches, t_failures) = get_stats()
CrawlConfig.log("files checksummed: %d; " % (t_checksums - p_checksums) +
"checksums matched: %d; " % matches +
"failures: %d" % failures)
CrawlConfig.log("totals checksummed: %d; " % t_checksums +
"matches: %d; " % t_matches +
"failures: %d" % t_failures)
# Report the dimension data in the log
d = Dimension.Dimension(name='cos')
t = Dimension.Dimension(name='cart')
CrawlConfig.log(d.report())
CrawlConfig.log(t.report())
return failures
last_obj_id,
last_obj_id,
correct,
error)
errcount += error
CrawlConfig.log("last nsobject in range: %d" % last_obj_id)
return errcount
# -----------------------------------------------------------------------------
def tcc_priority(globspec, cosinfo):
"""
Handle any files matching globspec. Return the number of files processed.
"""
rval = 0
cfg = CrawlConfig.get_config()
pri_compdir = cfg.get_d(tcc_lib.sectname(), 'completed', '/tmp')
for filepath in glob.glob(globspec):
tcc_lib.check_file(filepath, verbose=False, plugin=True)
cpath = U.pathjoin(pri_compdir, U.basename(filepath))
os.rename(filepath, cpath)
return rval
# -----------------------------------------------------------------------------
if __name__ == '__main__':
main(CrawlConfig.get_config())
def main(cfg):
"""
Tape Copy Checker retrieves the necessary information from the DB2 database
to find files where the number of copies stored may not match the number
called for by the COS.
"""
# retrieve configuration items as needed
how_many = int(cfg.get_d(tcc_lib.sectname(), 'operations', 10))
CrawlConfig.log("tape-copy-checker: firing up for %d items" % how_many)
# retrieve COS info
cosinfo = tcc_lib.get_cos_info()
# check for priority file(s)
pri_glob = cfg.get_d(tcc_lib.sectname(), 'priority', '')
if pri_glob != '':
if 0 < tcc_priority(pri_glob, cosinfo):
return
# get the nsobject_id of the next bitfile to process from mysql
next_nsobj_id = tcc_lib.get_next_nsobj_id(cfg)
CrawlConfig.log("next nsobject id = %d" % next_nsobj_id)
# fetch the next N bitfiles from DB2
CrawlConfig.log("looking for nsobject ids between %d and %d"
% (next_nsobj_id, next_nsobj_id+how_many-1))
try:
bfl = tcc_lib.get_bitfile_set(int(next_nsobj_id),
how_many)
except U.HpssicError as e:
bfl = []
pass
CrawlConfig.log("got %d bitfiles" % len(bfl))
errcount = 0
if len(bfl) == 0:
for oid in range(next_nsobj_id, next_nsobj_id+how_many):
tcc_lib.record_checked_ids(cfg, oid, oid, 1, 0)
if cfg.getboolean(tcc_lib.sectname(), 'verbose'):
CrawlConfig.log("Object %d is not complete" % oid)
errcount += 1
else:
# for each bitfile, if it does not have the right number of copies,
# report it
for bf in bfl:
correct = 1
error = 0
if bf['SC_COUNT'] != cosinfo[bf['BFATTR_COS_ID']]:
tcc_lib.tcc_report(bf, cosinfo)
correct = 0
error = 1
CrawlConfig.log("%s %s %d != %d" %
(bf['OBJECT_ID'],
tcc_lib.hexstr(bf['BFID']),
bf['SC_COUNT'],
cosinfo[bf['BFATTR_COS_ID']]))
elif cfg.getboolean(tcc_lib.sectname(), 'verbose'):
CrawlConfig.log("%s %s %d == %d" %
(bf['OBJECT_ID'],
tcc_lib.hexstr(bf['BFID']),
bf['SC_COUNT'],
cosinfo[bf['BFATTR_COS_ID']]))
last_obj_id = int(bf['OBJECT_ID'])
tcc_lib.record_checked_ids(cfg,
last_obj_id,
last_obj_id,
correct,
error)
errcount += error
CrawlConfig.log("last nsobject in range: %d" % last_obj_id)
return errcount
