def test_hashdelete_ok_str(self):
"""
If hashdelete gets a string argument, it should work
"""
try:
paths = self.paths + " %s/hashnot" % self.hdir
h = hpss.HSI(verbose=("verbose" in testhelp.testargs()))
# make sure the hashables all have a checksum stored
x = h.hashlist(self.plist)
for path in self.plist:
if util.rgxin("\(?none\)? %s" % path, x):
h.hashcreate(path)
# run hashdelete on the string
result = h.hashdelete(paths)
h.quit()
# verify the results
self.expected_in("hashdelete", result)
for path in self.paths.split():
exp = "hash deleted: \(?md5\)? %s" % path
self.expected_in(exp, result)
exp = "hash deleted: \(?md5\)? %s/hashnot" % self.hdir
self.assertFalse(util.rgxin(exp, result),
"'%s' not expected in %s" %
(exp, util.line_quote(result)))
except hpss.HSIerror as e:
if MSG.hpss_unavailable in str(e):
pytest.skip(str(e))
def test_hashlist_ok_str(self):
"""
If hashlist gets a string argument, it should work
"""
try:
paths = self.paths + " %s/hashnot" % self.hdir
h = hpss.HSI(verbose=("verbose" in testhelp.testargs()))
# make sure the hashables all have a checksum stored
x = h.hashlist(self.plist)
for path in self.plist:
if util.rgxin("\(?none\)? %s" % path, x):
h.hashcreate(path)
# run the test payload
result = h.hashlist(paths)
h.quit()
self.expected_in("hashlist", result)
for path in self.paths.split():
exp = "\(?md5\)? %s" % path
self.expected_in(exp, result)
exp = "\(?none\)? %s/hashnot" % self.hdir
self.expected_in(exp, result)
except hpss.HSIerror as e:
if MSG.hpss_unavailable in str(e):
pytest.skip(str(e))
def test_hashdelete_ok_list(self):
"""
If hashdelete get a list argument, it should work
"""
try:
plist = self.plist + [self.hdir + "/hashnot"]
h = hpss.HSI(verbose=("verbose" in testhelp.testargs()))
# make sure the hashables all have a checksum stored
x = h.hashlist(self.plist)
for path in self.plist:
if util.rgxin("\(?none\)? %s" % path, x):
h.hashcreate(path)
# run hashdelete on the list
result = h.hashdelete(plist)
h.quit()
# verify the results
self.expected_in("hashdelete", result)
for path in self.plist:
self.expected_in("hash deleted: md5 %s" % path, result)
exp = "\(?none\)? %s/hashnot" % self.hdir
self.assertTrue(exp not in result,
"'%s' not expected in %s" %
(exp,
util.line_quote(result)))
except hpss.HSIerror as e:
if MSG.hpss_unavailable in str(e):
pytest.skip(str(e))
def test_rgxin(self):
"""
Routine rgxin(needle, haystack) is analogous to the Python expression
"needle in haystack" with needle being a regexp.
"""
self.dbgfunc()
rgx = "a\(?b\)?c"
rgx2 = "(dog|fox|over)"
fstring = "The quick brown fox jumps over the lazy dog"
tstring1 = "Now we know our abc's"
tstring2 = "With parens: a(b)c"
self.assertTrue(U.rgxin(rgx, tstring1),
"'%s' should match '%s'" % (rgx, tstring1))
self.assertTrue(U.rgxin(rgx, tstring2),
"'%s' should match '%s'" % (rgx, tstring2))
self.assertFalse(U.rgxin(rgx, fstring),
"'%s' should NOT match '%s'" % (rgx, fstring))
self.expected('abc', U.rgxin(rgx, tstring1))
self.expected('a(b)c', U.rgxin(rgx, tstring2))
self.expected('fox', U.rgxin(rgx2, fstring))
def test_rgxin(self):
"""
Routine rgxin(needle, haystack) is analogous to the Python expression
"needle in haystack" with needle being a regexp.
"""
self.dbgfunc()
rgx = "a\(?b\)?c"
rgx2 = "(dog|fox|over)"
fstring = "The quick brown fox jumps over the lazy dog"
tstring1 = "Now we know our abc's"
tstring2 = "With parens: a(b)c"
self.assertTrue(U.rgxin(rgx, tstring1),
"'%s' should match '%s'" % (rgx, tstring1))
self.assertTrue(U.rgxin(rgx, tstring2),
"'%s' should match '%s'" % (rgx, tstring2))
self.assertFalse(U.rgxin(rgx, fstring),
"'%s' should NOT match '%s'" % (rgx, fstring))
self.expected('abc', U.rgxin(rgx, tstring1))
self.expected('a(b)c', U.rgxin(rgx, tstring2))
self.expected('fox', U.rgxin(rgx2, fstring))
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
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
