def main(args):
argp = _argparser().parse_args(args[1:])
if argp.re_parse:
escape = False
else:
escape = not argp.do_not_escape
if argp.override_tags:
re_parse = True
else:
re_parse = argp.re_parse
for line in (l.rstrip("\n") for l in argp.input):
if re_parse:
head = PTB_HEAD_REGEX.match(line).groupdict()["head"]
gdicts_tag, gdicts_tok = tee((m.groupdict() for m in PTB_TOK_TAG_REGEX.finditer(line)), 2)
tokens = (d["token"] for d in gdicts_tok)
if not argp.override_tags:
tags = (d["tag"] for d in gdicts_tag)
else:
tags = flattagger(d["token"] for d in gdicts_tag)
else:
head = HEAD_BY_PARSER[argp.parser]
soup = line.split()
tokens, tokens_to_tag = tee(
(
ptb_escape(t, preserve_quotes=True) if escape else t
for t in (ptb_escape_quotes_tokens(soup) if escape else soup)
),
2,
)
tags = flattagger(tokens_to_tag)
argp.output.write(
"({} (S {}))\n".format(head, " ".join("({} {})".format(tag, token) for tag, token in izip(tags, tokens)))
)
return 0
def main(args):
argp = _argparser().parse_args(args[1:])
if argp.re_parse:
escape = False
else:
escape = not argp.do_not_escape
if argp.override_tags:
re_parse = True
else:
re_parse = argp.re_parse
for line in (l.rstrip('\n') for l in argp.input):
if re_parse:
head = PTB_HEAD_REGEX.match(line).groupdict()['head']
gdicts_tag, gdicts_tok = tee(
(m.groupdict() for m in PTB_TOK_TAG_REGEX.finditer(line)), 2)
tokens = (d['token'] for d in gdicts_tok)
if not argp.override_tags:
tags = (d['tag'] for d in gdicts_tag)
else:
tags = flattagger(d['token'] for d in gdicts_tag)
else:
head = HEAD_BY_PARSER[argp.parser]
soup = line.split()
tokens, tokens_to_tag = tee((
ptb_escape(t, preserve_quotes=True) if escape else t
for t in (ptb_escape_quotes_tokens(soup) if escape else soup)),
2)
tags = flattagger(tokens_to_tag)
argp.output.write('({} (S {}))\n'.format(
head, ' '.join('({} {})'.format(tag, token)
for tag, token in izip(tags, tokens))))
return 0
'[': '-LRB-',
']': '-RRB-',
'}': '-LRB-',
'{': '-RRB-',
# The question mark handling below is consistent with Stanford and
# McCCJ, but not with Enju that assigns it a PoS-tag instead.
'?': '.',
# The exclamation mark handling below is consistent with Stanford, but
# not Enju for the same reason as above.
'!': '.',
'$': '$',
}
# Generate a mapping from each escaped version of a token to the same tag
for token, escaped_token in set((
t,
ptb_escape(t),
) for t in TAG_BY_TOK if ptb_escape(t) != t):
TAG_BY_TOK[escaped_token] = TAG_BY_TOK[token]
###
def flattagger(tokens):
for token in tokens:
try:
tag = TAG_BY_TOK[token]
except KeyError:
# Go for the majority-class PoS-tag.
tag = 'NN'
yield tag
"(": "-LRB-",
")": "-RRB-",
"[": "-LRB-",
"]": "-RRB-",
"}": "-LRB-",
"{": "-RRB-",
# The question mark handling below is consistent with Stanford and
# McCCJ, but not with Enju that assigns it a PoS-tag instead.
"?": ".",
# The exclamation mark handling below is consistent with Stanford, but
# not Enju for the same reason as above.
"!": ".",
"$": "$",
}
# Generate a mapping from each escaped version of a token to the same tag
for token, escaped_token in set((t, ptb_escape(t)) for t in TAG_BY_TOK if ptb_escape(t) != t):
TAG_BY_TOK[escaped_token] = TAG_BY_TOK[token]
###
def flattagger(tokens):
for token in tokens:
try:
tag = TAG_BY_TOK[token]
except KeyError:
# Go for the majority-class PoS-tag.
tag = "NN"
yield tag
def _argparser():
