def run(self, filters_to_run, files):
'''Performs a processing run, given a list of filter names to run, and a list of file specifiers.'''
filters = []
for filter_name, args in filters_to_run:
# For a no-args switch, optparse passes in None; we substitute an empty tuple for
# consistency
if not args: args = ()
try:
filter_class = self.available_filters_dict[filter_name]
actual, expected = len(args), get_argcount_for_method(filter_class.__init__)
if actual != expected:
warn("Skipping filter %s; %d arguments given, %d expected.", filter_name, actual, expected)
continue
filters.append(filter_class(*args))
except KeyError:
err("No filter with name `%s' found.", filter_name)
# convert short notation in file specifiers to proper paths
def expand_short_notation(fn):
# short notation is
# corpus:ss,dd,deriv -> corpus/chtb_ssdd.fid:deriv
m = re.match(r'([^:]+):(\d+),(\d+),(\d+)', fn)
if m:
corpus_dir, sec, doc, deriv = m.groups()
return os.path.join(corpus_dir, 'chtb_%02d%02d.fid:%d' % (int(sec), int(doc), int(deriv)))
return fn
files = [expand_short_notation(file) for file in files]
self.run_filters(filters, files)
def do_run(self, args):
'''Runs the given filter with the given arguments on the working set.'''
args = args.split()
if not args: return
filter_name = args.pop(0)
if not filter_name: # no filter with the requested switch was found
return
if filter_name.startswith('--'): # a long option name was given
filter_name = self.get_filter_by_switch(filter_name)
elif filter_name.startswith('-'): # a short option name was given
filter_name = self.get_filter_by_switch(filter_name)
# Special case: for a one-arg filter any number of arguments are treated
# as a single argument.
if (filter_name in self.tracer and
get_argcount_for_method(self.tracer[filter_name].__init__) == 1):
filter_args = (' '.join(args), )
elif args is not None:
filter_args = args
else:
filter_args = None
def action():
self.tracer.run( [(filter_name, filter_args)], self.files )
print
self.redirecting_stdout(action, filter_name, filter_args)
def add_filter_to_optparser(parser, filter):
'''Given a filter, this registers it with the option parser, allowing it to be selected on the
command line.'''
argcount = get_argcount_for_method(filter.__init__)
opt_dict = {
'help': filter.__doc__, # Help string is the filter docstring
'dest': filter.long_opt, # Destination variable is the same as the long option name
'metavar': filter.arg_names,# Metavar names are supplied by the filter
'action': 'callback',
'callback': run_builtin_filter,
'callback_args': (filter.__name__, ) # Pass in the filter's class name
}
# If the filter expects arguments, set the correct number of arguments and assume that it takes
# string arguments (for consistency).
if argcount > 0:
opt_dict['nargs'] = argcount
opt_dict['type'] = 'string'
parser.add_option("-" + filter.opt, "--" + filter.long_opt, **opt_dict)
def run(self, filters_to_run, files):
'''Performs a processing run, given a list of filter names to run, and a list of file specifiers.'''
filters = []
for filter_name, args in filters_to_run:
# For a no-args switch, optparse passes in None; we substitute an empty tuple for
# consistency
if not args: args = ()
try:
filter_class = self.available_filters_dict[filter_name]
actual, expected = len(args), get_argcount_for_method(
filter_class.__init__)
if actual != expected:
warn(
"Skipping filter %s; %d arguments given, %d expected.",
filter_name, actual, expected)
continue
filters.append(filter_class(*args))
except KeyError:
err("No filter with name `%s' found.", filter_name)
# convert short notation in file specifiers to proper paths
def expand_short_notation(fn):
# short notation is
# corpus:ss,dd,deriv -> corpus/chtb_ssdd.fid:deriv
m = re.match(r'([^:]+):(\d+),(\d+),(\d+)', fn)
if m:
corpus_dir, sec, doc, deriv = m.groups()
return os.path.join(
corpus_dir,
'chtb_%02d%02d.fid:%d' % (int(sec), int(doc), int(deriv)))
return fn
files = [expand_short_notation(file) for file in files]
self.run_filters(filters, files)
def add_filter_to_optparser(parser, filter):
'''Given a filter, this registers it with the option parser, allowing it to be selected on the
command line.'''
argcount = get_argcount_for_method(filter.__init__)
opt_dict = {
'help': filter.__doc__, # Help string is the filter docstring
'dest': filter.
long_opt, # Destination variable is the same as the long option name
'metavar':
filter.arg_names, # Metavar names are supplied by the filter
'action': 'callback',
'callback': run_builtin_filter,
'callback_args': (filter.__name__, ) # Pass in the filter's class name
}
# If the filter expects arguments, set the correct number of arguments and assume that it takes
# string arguments (for consistency).
if argcount > 0:
opt_dict['nargs'] = argcount
opt_dict['type'] = 'string'
parser.add_option("-" + filter.opt, "--" + filter.long_opt, **opt_dict)
def LongTemplate(filter_name, filter):
return ("\t%s (%s)\n\t\t(%d args, -%s, --%s%s)" %
(filter_name, filter.__module__,
get_argcount_for_method(
filter.__init__), filter.opt, filter.long_opt,
(' ' + filter.arg_names) if filter.arg_names else ''))
def LongTemplate(filter_name, filter):
return ("\t%s (%s)\n\t\t(%d args, -%s, --%s%s)" %
(filter_name, filter.__module__,
get_argcount_for_method(filter.__init__),
filter.opt, filter.long_opt,
(' '+filter.arg_names) if filter.arg_names else ''))
