def load(self, filename):
"""Load classifier from FILENAME"""
debug.trace_fmtd(4, "tc.load({f})", f=filename)
try:
(self.keys, self.classifier) = system.load_object(filename)
except (TypeError, ValueError):
system.print_stderr("Problem loading classifier from {f}: {exc}".
format(f=filename, exc=sys.exc_info()))
return
def main(args):
"""Supporting code for command-line processing"""
debug.trace_fmtd(6, "main({a})", a=args)
if (len(args) != 2):
system.print_stderr("Usage: {p} model".format(p=args[0]))
return
model = args[1]
start_web_controller(model)
return
def change_directory(dir):
"""Changes into file system DIR (ignoring errors)"""
debug_print("change_directory(%s)" % dir, 4)
try:
os.chdir(dir)
except:
print_stderr("Unable to change into directory '%s': %s" %
(filename, sys.exc_info()))
return
def make_directory(dir):
"""Creates file system DIR (ignoring errors)"""
debug_print("make_directory(%s)" % dir, 4)
try:
os.mkdir(dir)
except:
print_stderr("Unable to create directory '%s': %s" %
(dir, sys.exc_info()))
return
def write_file(filename, text):
"""Outpus TEXT to FILENAME"""
debug_print("write_file(%s, _)" % filename, 5)
try:
with open(filename, 'w') as f:
f.write(text)
except:
print_stderr("Unable to write file '%s': %s" %
(filename, sys.exc_info()))
return
def run_command(command_line, level=5):
"""Runs COMMAND_LINE and returns the output (as a string), with debug tracing at specified LEVEL"""
# Issue command
debug_print("Running command: %s" % command_line, level=level)
(status, output) = getstatusoutput(command_line)
if (status != 0):
print_stderr("Warning: problem running command (status=%d): %s" %
(status, command_line))
# Return output
debug_print("Command output: %s\n" % output, level=level + 1)
return (output)
def usage():
"""Show command-line usage"""
# TODO: remove path from script filename
script = (__file__ or "n/a")
system.print_stderr(
"Usage: {scr} training-file model-file [testing]".format(scr=script))
system.print_stderr("")
system.print_stderr("Notes:")
system.print_stderr(
"- Use - to indicate the file is not needed (e.g., existing training model)."
)
system.print_stderr(
"- You need to supply either training file or model file.")
system.print_stderr("- The testing file is optional when training.")
return
def main():
"""
Main routine: parse arguments and perform main processing
TODO: revise comments
Note: Used to avoid conflicts with globals (e.g., if this were done at end of script).
"""
#
debug_print("start %s: %s" % (__file__, debug.timestamp()), 3)
# Parse command-line, showing usage statement if no arguments given (or --help)
args = sys.argv
debug_print("argv = %s" % sys.argv, 3)
num_args = len(args)
if ((num_args == 1) or ((num_args > 1) and (args[1] == "--help"))):
print_stderr("Usage: %s [--help] URL" % args[0])
print_stderr(
"Example: %s 'http://en.wikipedia.org/wiki/Category:Major_League_Baseball_players'"
% args[0])
print_stderr("Notes:")
print_stderr(
"- MAKE_SUBDIRS recreates category hierarchy in file system")
print_stderr("- SKIP_SUBCATS disables subcat traversals")
print_stderr("- SKIP_PAGES disables page downloading")
sys.exit()
arg_pos = 1
while (arg_pos < num_args) and (args[arg_pos][0] == "-"):
debug_print("args[%d]: %s" % (arg_pos, args[arg_pos]), 3)
if (args[arg_pos] == "-"):
# note: - is used to avoid usage statement with file input from stdin
pass
else:
print_stderr("Error: unexpected argument '%s'" % args[arg_pos])
sys.exit()
arg_pos += 1
url = args[arg_pos]
# Optionally create and change into output directory for wiki pages
if OUTPUT_DIR != ".":
make_directory(OUTPUT_DIR)
change_directory(OUTPUT_DIR)
# Process wiki categorization file
# TODO: accept file input as well (e.g., '2013 Pittsburgh Pirates season - Wikipedia, the free encyclopedia.html')
download_category_articles(url)
debug_print("stop %s: %s" % (__file__, debug.timestamp()), 3)