"lemmata": False,
"light": False,
"lexicon": LEXICON,
"language": "es",
"default": "NC",
"map": kwargs.get("tagset", "") != PAROLE and parole2penntreebank or None,
})
s = _en_parse(s, False, tags, chunks, relations, **kwargs)
# Use pattern.es.inflect for lemmatization:
if lemmata:
p = [find_lemmata(sentence) for sentence in s.split()]
p = "\n".join([" ".join(["/".join(token) for token in sentence]) for sentence in p])
s = TaggedString(p, tags=s.tags+["lemma"], language="es")
return s
def tag(s, tokenize=True, encoding="utf-8"):
""" Returns a list of (token, tag)-tuples from the given string.
"""
tags = []
for sentence in parse(s, tokenize, True, False, False, False, encoding).split():
for token in sentence:
tags.append((token[0], token[1]))
return tags
#### COMMAND LINE ##################################################################################
# From the folder that contains the "pattern" folder:
# python -m pattern.es.parser xml -s "A quien se hace de miel las moscas le comen." -OTCLI
if __name__ == "__main__":
commandline(parse)
s = _en_parse(s, False, tags, chunks, relations, **kwargs)
p = [[[m[token[0]]] + token[1:] for token in sentence]
for sentence in s.split()]
p = "\n".join(
[" ".join(["/".join(token) for token in sentence]) for sentence in p])
s = TaggedString(p, tags=s.tags, language="de")
# Use pattern.de.inflect for lemmatization:
if lemmata:
p = [find_lemmata(sentence) for sentence in s.split()]
s = TaggedString(p, tags=s.tags + ["lemma"], language="de")
return s
def tag(s, tokenize=True, encoding="utf-8"):
""" Returns a list of (token, tag)-tuples from the given string.
"""
tags = []
for sentence in parse(s, tokenize, True, False, False, False,
encoding).split():
for token in sentence:
tags.append((token[0], token[1]))
return tags
#### COMMAND LINE ##################################################################################
# From the folder that contains the "pattern" folder:
# python -m pattern.de.parser xml -s "Ein Unglück kommt selten allein." -OTCLI
if __name__ == "__main__":
commandline(parse)
