def fetch_alignment(basename, langs, backend='hunalign'):
assert langs
real_langs = list(set(lang[:2]
for lang in langs))
if len(real_langs) == 1:
text_len = len(fetch_sentences(basename, real_langs[0]))
return Alignment.create_straight(text_len, len(langs))
elif len(real_langs) == 2:
try:
a = Alignment.from_file("%s/%s-%s.%s" %
(basename, real_langs[0], real_langs[1], backend))
except IOError:
real_langs.reverse()
a = Alignment.from_file("%s/%s-%s.%s" %
(basename, real_langs[0], real_langs[1], backend))
else: # len(real_langs) == 3 :(
a1 = Alignment.from_file('%s/pl-cu.%s' % (basename, backend)).as_ladder()
a2 = Alignment.from_file('%s/cu-el.%s' % (basename, backend)).as_ladder()
a3 = Alignment.from_file('%s/pl-el.%s' % (basename, backend)).as_ladder()
a3 = [(b, a) for (a, b) in a3] # reversed
a = merge_3_alignments(a1, a2, a3)
real_langs = ['pl', 'cu', 'el'] # needed later
columns = _transpose(a.data)
columns_map = { real_langs[i] : columns[i]
for i in range(len(real_langs)) }
# common part for 2 and 3
chosen_columns = [columns_map[lang[:2]] for lang in langs]
chosen_columns.append(columns[2])
return Alignment(_transpose(chosen_columns))
def extract_bisents(file1, lang1, file2, lang2, alignment_file):
assert isinstance(lang1, (unicode, str)) and len(lang1)==2
assert isinstance(lang2, (unicode, str)) and len(lang2)==2
t1 = Text.from_file(file1, lang1)
t2 = Text.from_file(file2, lang2)
alignment = Alignment.from_file(alignment_file)
bisents = alignment.as_pairs(t1.as_sentences_flat(),
t2.as_sentences_flat())
return bisents
def get_alignment(self, langs, backend=None):
"""like fetcher"""
assert len(langs) >= 2
assert not backend or backend in possible_backends
real_langs = list(set(lang[:2]
for lang in langs))
if len(real_langs) == 1:
text_len = len(fetch_sentences(basename, real_langs[0]))
return Alignment.create_straight(text_len, len(langs))
elif len(real_langs) == 2:
a = None
for i in range(2):
for b in ([backend] if backend else possible_backends):
try:
langs_string = '-'.join(str(l) for l in real_langs)
a = Alignment.from_file(self._p(langs_string + '.' + b))
break
except IOError:
continue
if a:
break
real_langs.reverse()
if not a:
raise IOError
else: # len(real_langs) == 3 :(
a1 = self.get_alignment(['pl', 'cu'], backend).as_ladder()
a2 = self.get_alignment(['cu', 'el'], backend).as_ladder()
a3 = self.get_alignment(['pl', 'el'], backend).as_ladder()
# a3 = [(b, a) for (a, b) in a3] # reversed
a = merge_3_alignments(a1, a2, a3)
real_langs = ['pl', 'cu', 'el'] # needed later
columns = _transpose(a.data)
columns_map = { real_langs[i] : columns[i]
for i in range(len(real_langs)) }
# common part for 2 and 3
chosen_columns = [columns_map[lang[:2]] for lang in langs]
chosen_columns.append(columns[2])
return Alignment(_transpose(chosen_columns))
def gen():
yield (0, 0, 0)
prev_i2 = 0
for (i1, i2) in al12:
for _i2 in range(prev_i2+1, i2+1):
try:
i3s = map23[_i2]
for i3 in i3s:
# if map31[i3] == i1:
yield (i1, _i2, i3)
except KeyError:
pass
prev_i2 = i2
return Alignment(list(gen()), no_costs=True)
if __name__ == '__main__':
import sys
name = sys.argv[1]
a1 = Alignment.from_file(name + '/pl-cu.my').as_ladder()
a2 = Alignment.from_file(name + '/cu-el.my').as_ladder()
a3 = Alignment.from_file(name + '/pl-el.my').as_ladder()
a3 = [(b, a) for (a, b) in a3]
ma = merge_3_alignments(a1, a2, a3)
ma.pretty_print(Text.from_file(name + '/pl.txt', lang='pl').as_sentences_flat(),
Text.from_file(name + '/cu.txt', lang='cu').as_sentences_flat(),
Text.from_file(name + '/el.txt', lang='el').as_sentences_flat())
