def generate_sequential_data11(lst, lstv=None, pos="V"):
corpus = []
all_data = []
map = Mapping()
for fn in lst:
txt = read_conll2009_corpus(fn)
corpus.extend(txt)
for sen in corpus:
for w in sen:
if isinstance(w, Predicate) and getattr(w, "pos")[0] == pos:
lst = [map.add_value(w.form + "_" + w.lemma + "." + w.sense)]
arglst = []
for arg in w.arguments:
hn = get_represented_form(sen, arg)
if hn is not None:
arglst.append(
map.add_value(hn + "_" + w.arguments[arg]))
arglst.append(map.add_value("EOS"))
lst.extend(arglst)
all_data.append(lst)
X = [[x[i] for i in range(len(x) - 1)] for x in all_data]
Y = [[x[i + 1] for i in range(len(x) - 1)] for x in all_data]
Xv = None
Yv = None
if lstv is not None:
corpusv = []
all_datav = []
for fn in lstv:
txt = read_conll2009_corpus(fn)
corpusv.extend(txt)
for sen in corpusv:
for w in sen:
if isinstance(w, Predicate) and getattr(w, "pos")[0] == pos:
lst = [
map.add_value(w.form + "_" + w.lemma + "." + w.sense)
]
arglst = []
for arg in w.arguments:
hn = get_represented_form(sen, arg)
if hn is not None:
arglst.append(
map.add_value(hn + "_" + w.arguments[arg]))
arglst.append(map.add_value("EOS"))
lst.extend(arglst)
all_datav.append(lst)
Xv = [[x[i] for i in range(len(x) - 1)] for x in all_datav]
Yv = [[x[i + 1] for i in range(len(x) - 1)] for x in all_datav]
return X, Y, Xv, Yv, map
def generate_sequential_data21_getmap(lst, lstv=None, pos="V", data="form"):
corpus = []
mapX1 = Mapping()
mapX2 = Mapping()
mapY1 = Mapping()
for fn in lst:
txt = read_conll2009_corpus(fn)
corpus.extend(txt)
for sen in corpus:
for w in sen:
if isinstance(w, Predicate) and getattr(w, "pos")[0] == pos:
mapX1.add_value(w.form)
mapX2.add_value("PRED")
mapY1.add_value(w.form + "_PRED")
for arg in w.arguments:
if data == "origin":
hn = sen[arg].form
else:
hn = get_represented_form(sen, arg)
if hn is not None:
mapX1.add_value(hn)
mapX2.add_value(w.arguments[arg])
mapY1.add_value(hn + "_" + w.arguments[arg])
mapX1.add_value("EOS")
mapX2.add_value("EOS")
mapY1.add_value("EOS_EOS")
return mapX1, mapX2, mapY1
def generate_sequential_data11(lst,
lstv=None,
pos="V",
data="form",
count=100000):
corpus = []
all_data = []
map = Mapping()
for i in range(1, len(lst)):
txt = read_conll2009_corpus(lst[i])
corpus.extend(txt)
if count is not None:
c = 0
corpus_select = []
for s in corpus:
c += 1
if c < count:
corpus_select.append(s)
else:
break
corpus = corpus_select
corpus.extend(read_conll2009_corpus(lst[0]))
for sen in corpus:
for w in sen:
if isinstance(w, Predicate) and getattr(w, "pos")[0] == pos:
lst = [map.add_value(w.form + "_PRED")]
arglst = []
for arg in w.arguments:
if data == "origin":
hn = sen[arg].form
else:
hn = get_represented_form(sen, arg)
if hn is not None:
arglst.append(
map.add_value(hn + "_" + w.arguments[arg]))
arglst.append(map.add_value("EOS"))
lst.extend(arglst)
all_data.append(lst)
X = [[x[i] for i in range(len(x) - 1)] for x in all_data]
Y = [[x[i + 1] for i in range(len(x) - 1)] for x in all_data]
Xv = None
Yv = None
if lstv is not None:
corpusv = []
all_datav = []
for fn in lstv:
txt = read_conll2009_corpus(fn)
corpusv.extend(txt)
for sen in corpusv:
for w in sen:
if isinstance(w, Predicate) and getattr(w, "pos")[0] == pos:
lst = [map.add_value(w.form + "_PRED")]
arglst = []
for arg in w.arguments:
if data == "origin":
hn = sen[arg].form
else:
hn = get_represented_form(sen, arg)
if hn is not None:
arglst.append(
map.add_value(hn + "_" + w.arguments[arg]))
arglst.append(map.add_value("EOS"))
lst.extend(arglst)
all_datav.append(lst)
Xv = [[x[i] for i in range(len(x) - 1)] for x in all_datav]
Yv = [[x[i + 1] for i in range(len(x) - 1)] for x in all_datav]
return X, Y, Xv, Yv, map
def generate_sequential_data21(corpus,
lstv=None,
pos="V",
data="form",
mapX1=None,
mapX2=None,
mapY1=None):
all_data = []
if mapX1 is not None:
for sen in corpus:
for w in sen:
if isinstance(w, Predicate) and getattr(w, "pos")[0] == pos:
lst = [(mapX1.get_index(w.form), mapX2.get_index("PRED"),
mapY1.get_index(w.form + "_PRED"))]
arglst = []
for arg in w.arguments:
if data == "origin":
hn = sen[arg].form
else:
hn = get_represented_form(sen, arg)
if hn is not None:
arglst.append(
(mapX1.get_index(hn),
mapX2.get_index(w.arguments[arg]),
mapY1.get_index(hn + "_" + w.arguments[arg])))
arglst.append(
(mapX1.get_index("EOS"), mapX2.get_index("EOS"),
mapY1.get_index("EOS_EOS")))
lst.extend(arglst)
all_data.append(lst)
X1 = [[x[i][0] for i in range(len(x) - 1)] for x in all_data]
X2 = [[x[i][1] for i in range(len(x) - 1)] for x in all_data]
Y = [[x[i + 1][2] for i in range(len(x) - 1)] for x in all_data]
Xv1 = None
Xv2 = None
Yv = None
if lstv is not None:
corpusv = []
all_datav = []
for fn in lstv:
txt = read_conll2009_corpus(fn)
corpusv.extend(txt)
for sen in corpusv:
for w in sen:
if isinstance(w, Predicate) and getattr(w,
"pos")[0] == pos:
lst = [(mapX1.get_index(w.form),
mapX2.get_index("PRED"),
mapY1.get_index(w.form + "_PRED"))]
arglst = []
for arg in w.arguments:
if data == "origin":
hn = sen[arg].form
else:
hn = get_represented_form(sen, arg)
if hn is not None:
arglst.append(
(mapX1.get_index(hn),
mapX2.get_index(w.arguments[arg]),
mapY1.get_index(hn + "_" +
w.arguments[arg])))
arglst.append(
(mapX1.get_index("EOS"), mapX2.get_index("EOS"),
mapY1.get_index("EOS_EOS")))
lst.extend(arglst)
all_datav.append(lst)
Xv1 = [[x[i][0] for i in range(len(x) - 1)] for x in all_datav]
Xv2 = [[x[i][1] for i in range(len(x) - 1)] for x in all_datav]
Yv = [[x[i + 1][2] for i in range(len(x) - 1)] for x in all_datav]
return X1, X2, Y, Xv1, Xv2, Yv, mapX1, mapX2, mapY1
else:
mapX1 = Mapping()
mapX2 = Mapping()
mapY1 = Mapping()
for txttrain in corpus:
all_data.extend()
for sen in corpus:
for w in sen:
if isinstance(w, Predicate) and getattr(w, "pos")[0] == pos:
lst = [(mapX1.add_value(w.form), mapX2.add_value("PRED"),
mapY1.add_value(w.form + "_PRED"))]
arglst = []
for arg in w.arguments:
if data == "origin":
hn = sen[arg].form
else:
hn = get_represented_form(sen, arg)
if hn is not None:
arglst.append(
(mapX1.add_value(hn),
mapX2.add_value(w.arguments[arg]),
mapY1.add_value(hn + "_" + w.arguments[arg])))
arglst.append(
(mapX1.add_value("EOS"), mapX2.add_value("EOS"),
mapY1.add_value("EOS_EOS")))
lst.extend(arglst)
all_data.append(lst)
X1 = [[x[i][0] for i in range(len(x) - 1)] for x in all_data]
X2 = [[x[i][1] for i in range(len(x) - 1)] for x in all_data]
Y = [[x[i + 1][2] for i in range(len(x) - 1)] for x in all_data]
Xv1 = None
Xv2 = None
Yv = None
if lstv is not None:
corpusv = []
all_datav = []
for fn in lstv:
txt = read_conll2009_corpus(fn)
corpusv.extend(txt)
for sen in corpusv:
for w in sen:
if isinstance(w, Predicate) and getattr(w,
"pos")[0] == pos:
lst = [(mapX1.add_value(w.form),
mapX2.add_value("PRED"),
mapY1.add_value(w.form + "_PRED"))]
arglst = []
for arg in w.arguments:
if data == "origin":
hn = sen[arg].form
else:
hn = get_represented_form(sen, arg)
if hn is not None:
arglst.append(
(mapX1.add_value(hn),
mapX2.add_value(w.arguments[arg]),
mapY1.add_value(hn + "_" +
w.arguments[arg])))
arglst.append(
(mapX1.add_value("EOS"), mapX2.add_value("EOS"),
mapY1.add_value("EOS_EOS")))
lst.extend(arglst)
all_datav.append(lst)
Xv1 = [[x[i][0] for i in range(len(x) - 1)] for x in all_datav]
Xv2 = [[x[i][1] for i in range(len(x) - 1)] for x in all_datav]
Yv = [[x[i + 1][2] for i in range(len(x) - 1)] for x in all_datav]
return X1, X2, Y, Xv1, Xv2, Yv, mapX1, mapX2, mapY1
def readAll(self):
txt = []
for f in self.input_file:
txt=read_conll2009_corpus(f)
return txt
words.append(l.strip())
if len(words) != 0:
if end is None:
if idx >= start:
sens.append(words)
else:
if idx >= start:
if idx < end:
sens.append(words)
for sen in sens:
conll2009sen = read_conll2009_sentence(sen, read_label, use_gold=use_gold)
txt.append(conll2009sen)
return txt
if __name__ == "__main__":
lst = ["/home/quynh/working/Data/conll2009/train.conll2009.pp.txt"]
reader = Conll2009BatchReader(1000, lst)
count = 0
while True:
txt = reader.next()
if len(txt) == 0:
break
count+=len(txt)
print (count)
txt = read_conll2009_corpus("/home/quynh/working/Data/conll2009/train.conll2009.pp.txt")
print (len(txt))
