def main(argv=None):
"""ThaiNLP command line."""
if not argv:
argv = sys.argv
parser = argparse.ArgumentParser(
prog="thainlp",
description="Thai natural language processing.",
usage=("thainlp <command> [options]\n\n"
"Example:\n\n"
"thainlp data catalog\n\n"
"--"),
)
parser.add_argument(
"command",
type=str,
choices=cli.COMMANDS,
help="text processing action",
)
args = parser.parse_args(argv[1:2])
cli.exit_if_empty(args.command, parser)
if hasattr(cli, args.command):
command = getattr(cli, args.command)
command.App(argv)
def main(args=None):
"""The main routine of PyThaiNLP command line."""
if args is None:
args = sys.argv[1:]
parser = argparse.ArgumentParser(
usage="pythainlp command [subcommand] [options]")
parser.add_argument("command",
type=str,
default="",
nargs="?",
help="[%s]" % "|".join(cli.COMMANDS))
args = parser.parse_args(sys.argv[1:2])
cli.exit_if_empty(args.command, parser)
if hasattr(cli, args.command):
command = getattr(cli, args.command)
command.App(sys.argv)
else:
print(
f"Command not available: {args.command}\nPlease run with --help for alternatives"
)
def __init__(self, argv):
parser = argparse.ArgumentParser(**cli.make_usage("tokenize"))
parser.add_argument("subcommand",
type=str,
nargs="?",
help="[subword|syllable|word|sent]")
args = parser.parse_args(argv[2:3])
cli.exit_if_empty(args.subcommand, parser)
subcommand = str.lower(args.subcommand)
argv = argv[3:]
if subcommand.startswith("word"):
WordTokenizationApp("word", argv)
elif subcommand.startswith("syl"):
SyllableTokenizationApp("syllable", argv)
elif subcommand.startswith("subw"):
SubwordTokenizationApp("subword", argv)
elif subcommand.startswith("sent"):
SubwordTokenizationApp("sent", argv)
else:
raise NotImplementedError(
f"Subcommand not available: {subcommand}")
def __init__(self, argv):
parser = argparse.ArgumentParser(**cli.make_usage("corpus"))
parser.add_argument(
"subcommand",
type=str,
default="",
nargs="?",
help="[download|remove]" # there should be a "list" subcommand
)
parser.add_argument(
"--name",
type=str,
help="corpus's name",
)
args = parser.parse_args(argv[2:])
cli.exit_if_empty(args.subcommand, parser)
subcommand = str.lower(args.subcommand)
if hasattr(App, subcommand):
getattr(App, subcommand)(args)
else:
raise NotImplementedError(
f"Subcommand not available: {subcommand}")
def __init__(self, name, argv):
parser = argparse.ArgumentParser(**cli.make_usage("tokenize " + name))
parser.add_argument(
"-t",
"--text",
type=str,
help="input text",
)
parser.add_argument("-s",
"--sep",
type=bool,
help=f"default: {self.separator}",
default=self.separator)
parser.add_argument("-e",
"--engine",
type=str,
help=f"default: {self.engine}",
default=self.engine)
parser.add_argument("-w",
"--keep-whitespace",
type=bool,
help=f"default: {self.keep_whitespace}",
default=self.keep_whitespace)
args = parser.parse_args(argv)
self.args = args
cli.exit_if_empty(args.text, parser)
result = self.run(args.text, engine=args.engine)
print(self.separator.join(result))
def __init__(self, name, argv):
parser = argparse.ArgumentParser(
**cli.make_usage("tokenization " + name)
)
parser.add_argument(
"--text",
type=str,
help="input text",
)
parser.add_argument(
"--engine",
type=str,
help="default: %s" % self.default_engine,
default=self.default_engine
)
args = parser.parse_args(argv)
self.args = args
cli.exit_if_empty(args.text, parser)
print(f"Using engine={args.engine}")
result = self.run(args.text, engine=args.engine)
print(self.separator.join(result))
def __init__(self, name, argv):
parser = argparse.ArgumentParser(**cli.make_usage("tokenize " + name))
parser.add_argument(
"text",
type=str,
nargs="?",
help="input text",
)
parser.add_argument(
"-s",
"--sep",
dest="separator",
type=str,
help=f"default: {self.separator}",
default=self.separator,
)
parser.add_argument(
"-a",
"--algo",
dest="algorithm",
type=str,
help=f"default: {self.algorithm}",
default=self.algorithm,
)
parser.add_argument(
"-w",
"--keep-whitespace",
dest="keep_whitespace",
action="store_true",
)
parser.add_argument(
"-nw",
"--no-whitespace",
dest="keep_whitespace",
action="store_false",
)
parser.set_defaults(keep_whitespace=True)
args = parser.parse_args(argv)
self.args = args
cli.exit_if_empty(args.text, parser)
result = self.run(
args.text,
engine=args.algorithm,
keep_whitespace=args.keep_whitespace,
)
print(args.separator.join(result) + args.separator)
def __init__(self, argv):
parser = argparse.ArgumentParser(**cli.make_usage("tag"))
parser.add_argument("subcommand", type=str, nargs="?", help="[pos]")
args = parser.parse_args(argv[2:3])
cli.exit_if_empty(args.subcommand, parser)
subcommand = str.lower(args.subcommand)
argv = argv[3:]
if subcommand == "pos":
POSTaggingApp("Part-of-Speech tagging", argv)
else:
raise NotImplementedError(
f"Subcommand not available: {subcommand}")
def __init__(self, argv):
parser = argparse.ArgumentParser(
prog="tokenize",
description="Break a text into small units (tokens).",
usage=(
'thainlp tokenize <token_type> [options] "<text>"\n\n'
"token_type:\n\n"
"subword subword (may not be a linguistic unit)\n"
"syllable syllable\n"
"word word\n"
"sent sentence\n\n"
"options:\n\n"
"--sep or -s <separator> specify custom separator\n"
" (default is a space)\n"
"--algo or -a <algorithm> tokenization algorithm\n"
" (see API doc for more info)\n"
"--keep-whitespace or -w keep whitespaces in output\n"
" (default)\n\n"
"<separator> and <text> should be inside double quotes.\n\n"
"Example:\n\n"
'thainlp tokenize word -s "|" "ใต้แสงนีออนเปลี่ยวเหงา"\n\n'
"--"),
)
parser.add_argument(
"token_type",
type=str,
help="[subword|syllable|word|sent]",
)
args = parser.parse_args(argv[2:3])
cli.exit_if_empty(args.token_type, parser)
token_type = str.lower(args.token_type)
argv = argv[3:]
if token_type.startswith("w"):
WordTokenizationApp("word", argv)
elif token_type.startswith("sy"):
SyllableTokenizationApp("syllable", argv)
elif token_type.startswith("su"):
SubwordTokenizationApp("subword", argv)
elif token_type.startswith("se"):
SentenceTokenizationApp("sent", argv)
else:
print(f"Token type not available: {token_type}")
def __init__(self, argv):
parser = argparse.ArgumentParser("soundex")
parser.add_argument(
"--text",
type=str,
help="text",
)
parser.add_argument("--engine",
type=str,
help="[udom83|lk82|metasound] (default: udom83)",
default="udom83")
args = parser.parse_args(argv[2:])
cli.exit_if_empty(args.text, parser)
sx = soundex(args.text, engine=args.engine)
print(sx)
def __init__(self, argv):
parser = argparse.ArgumentParser(
prog="benchmark",
description=("Benchmark for various tasks;\n"
"currently, we have only for word tokenization."),
usage=("thainlp benchmark [task] [task-options]\n\n"
"tasks:\n\n"
"word-tokenization benchmark word tokenization\n\n"
"--"),
)
parser.add_argument("task", type=str, help="[word-tokenization]")
args = parser.parse_args(argv[2:3])
cli.exit_if_empty(args.task, parser)
task = str.lower(args.task)
task_argv = argv[3:]
if task == "word-tokenization":
WordTokenizationBenchmark(task, task_argv)
def __init__(self, argv):
parser = argparse.ArgumentParser(**cli.make_usage("tagging"))
parser.add_argument(
"command",
type=str,
nargs="?",
help="[pos]"
)
args = parser.parse_args(argv[2:3])
command = args.command
cli.exit_if_empty(args.command, parser)
argv = argv[3:]
if command == "pos":
POSTaggingApp("Part-of-Speech tagging", argv)
else:
raise ValueError(f"no command:{subcommand}")
def __init__(self, argv):
parser = argparse.ArgumentParser(**cli.make_usage("tokenization"))
parser.add_argument(
"command",
type=str,
nargs="?",
help="[word|syllable]"
)
args = parser.parse_args(argv[2:3])
command = args.command
cli.exit_if_empty(command, parser)
argv = argv[3:]
if command == "word":
WordTokenizationApp("word", argv)
elif command == "syllable":
SyllableTokenizationApp("syllable", argv)
def __init__(self, argv):
parser = argparse.ArgumentParser("soundex")
parser.add_argument(
"subcommand",
type=str,
nargs="?",
help="[udom83|lk82|metasound]"
)
parser.add_argument(
"-t",
"--text",
type=str,
help="input text",
)
args = parser.parse_args(argv[2:3])
cli.exit_if_empty(args.subcommand, parser)
subcommand = str.lower(args.subcommand)
cli.exit_if_empty(args.text, parser)
sdx = ""
if subcommand.startswith("udom"):
sdx = soundex(args.text, engine="udom83")
elif subcommand.startswith("lk"):
sdx = soundex(args.text, engine="lk82")
elif subcommand.startswith("meta"):
sdx = soundex(args.text, engine="metasound")
else:
raise NotImplementedError(
f"Subcommand not available: {subcommand}")
print(sdx)
def __init__(self, argv):
parser = argparse.ArgumentParser(**cli.make_usage("corpus"))
parser.add_argument(
"--name",
type=str,
help="corpus's name",
)
parser.add_argument("command",
type=str,
default="",
nargs="?",
help="[download|remove]")
args = parser.parse_args(argv[2:])
cli.exit_if_empty(args.command, parser)
command = args.command
if hasattr(App, command):
getattr(App, command)(args)
else:
print("No command available: %s" % command)
