def _process_long_opt(self, rargs, values):
opt = rargs[0]
try:
OptionParser._process_long_opt(self, rargs, values)
except BadOptionError:
self.ignored.append(opt)
self.eat_args(rargs)
def _process_long_opt(self, rargs, values):
arg = rargs[0]
if "=" not in arg:
opt = self._match_long_opt(arg)
option = self._long_opt[opt]
if option.takes_value():
self.error("The '%s' option requires an argument with '%s='." %
(opt, opt))
return OptionParser._process_long_opt(self, rargs, values)
def _process_args2(self, largs, rargs, values):
"""_process_args(largs : [string],
rargs : [string],
values : Values)
Process command-line arguments and populate 'values', consuming
options and arguments from 'rargs'. If 'allow_interspersed_args' is
false, stop at the first non-option argument. If true, accumulate any
interspersed non-option arguments in 'largs'.
"""
while rargs:
arg = rargs[0]
p = re.match('^-[0-9.]+$', arg) # Identify a numeric argument
if p:
del rargs[0]
raise BadOptionError(arg)
#self.error(_("%s unrecognized number in arguments") % arg)
# We handle bare "--" explicitly, and bare "-" is handled by the
# standard arg handler since the short arg case ensures that the
# len of the opt string is greater than 1.
if arg == "--":
del rargs[0]
return
elif arg[0:2] == "--":
# process a single long option (possibly with value(s))
OptionParser._process_long_opt(self, rargs, values)
elif arg[:1] == "-" and len(arg) > 1:
# process a cluster of short options (possibly with
# value(s) for the last one only)
OptionParser._process_short_opts(self, rargs, values)
elif self.allow_interspersed_args:
largs.append(arg)
del rargs[0]
else:
return # stop now, leave this arg in rargs
def _process_args2(self, largs, rargs, values):
"""_process_args(largs : [string],
rargs : [string],
values : Values)
Process command-line arguments and populate 'values', consuming
options and arguments from 'rargs'. If 'allow_interspersed_args' is
false, stop at the first non-option argument. If true, accumulate any
interspersed non-option arguments in 'largs'.
"""
while rargs:
arg = rargs[0]
p = re.match('^-[0-9.]+$', arg) # Identify a numeric argument
if p:
del rargs[0]
raise BadOptionError(arg)
#self.error(_("%s unrecognized number in arguments") % arg)
# We handle bare "--" explicitly, and bare "-" is handled by the
# standard arg handler since the short arg case ensures that the
# len of the opt string is greater than 1.
if arg == "--":
del rargs[0]
return
elif arg[0:2] == "--":
# process a single long option (possibly with value(s))
OptionParser._process_long_opt(self, rargs, values)
elif arg[:1] == "-" and len(arg) > 1:
# process a cluster of short options (possibly with
# value(s) for the last one only)
OptionParser._process_short_opts(self, rargs, values)
elif self.allow_interspersed_args:
largs.append(arg)
del rargs[0]
else:
return # stop now, leave this arg in rargs
def _process_long_opt(self, rargs, values):
try:
OptionParser._process_long_opt(self, rargs, values)
except BadOptionError as err:
self.largs.append(err.opt_str)
def _process_long_opt(self, rargs, values):
try:
OptionParser._process_long_opt(self, rargs, values)
except BadOptionError, err:
self.largs.append(err.opt_str)
def _process_long_opt(self, rargs, values):
arg = rargs[0]
try:
OptionParser._process_long_opt(self, rargs, values)
except BadOptionError, err:
self.largs.append(arg)
dest="bills_list",
default=None)
parser.add_option("--dictionary",
type="string",
dest="dictionary",
default=None)
process_long_opt = parser._process_long_opt
def _process_long_opt(rargs, values):
try:
process_long_opt(rargs, values)
except BadOptionError, m:
# print "Ignoring bad option error: %s" % m
pass
parser._process_long_opt = _process_long_opt
OPT, args = parser.parse_args()
NGRAM_SIZE = 5
class Mapper(mapreduce.Mapper):
"""Counts the number of words in each line.
"""
def Start(self):
self.ignore_chars = ".,:?'`\"$+-/;%\t<>()0123456789=*\\|{}^_~&[]#@!\x98\x99\xb4\x86\x84\xc3"
self.stemmer = tokenizer.POSStemmer()
self.bills_list = None
if OPT.bills_list:
print "Reading bills."
self.bills_list = {}
