def from_folder(cls, path, langs=None):
from TextFolder import TextFolder
pt = TextFolder(path)
if not langs:
langs = ['pl', 'cu', 'el']
oa = pt.get_alignment(langs, 'my')
seqs = [pt.get_sentences(lang) for lang in langs]
return NewAlignment.from_old_alignment(oa, langs, seqs)
args = parser.parse_args()
print >> sys.stderr
print >> sys.stderr, ("=== Aligning %s, %s-%s ===" %
(args.folder, args.lang1, args.lang2))
set_languages(args.lang1, args.lang2)
tfolder = TextFolder(args.folder)
t1 = map(preprocess, tfolder.get_sentences(args.lang1))
t2 = map(preprocess, tfolder.get_sentences(args.lang2))
# reading hand alignment
forced_rungs = []
if args.hand:
hand_alignment = tfolder.get_alignment([args.lang1, args.lang2],
backend='hand')
forced_rungs = hand_alignment.as_ladder()
print >> sys.stderr, "%d hand-aligned pairs found." % len(forced_rungs)
# prealign
if args.prealign:
pre_alignment = list(find_matches(t1, t2, threshold=0.5, pair_count=100))
forced_rungs.extend(pre_alignment)
print >> sys.stderr, "%d sentence pairs matched." % len(pre_alignment)
forced_rungs = sorted(set(forced_rungs))
try:
a = None
a = make_composed_alignment(t1, t2, forced_rungs)
a = Alignment(a)
finally:
output_filename = '%s/%s-%s.my' % (args.folder, args.lang1, args.lang2)
x_longest = x
y_longest = y
else:
M[x][y] = 0
return (longest, len(s1) - x_longest, len(s2) - y_longest)
if __name__ == "__main__":
from Alignment import Alignment
from NewAlignment import NewAlignment
langs = ("pl", "cu")
# A - tested alignment
tf = TextFolder("texts/kanon_izr/")
aA = NewAlignment.from_old_alignment(
tf.get_alignment(langs, "my"), langs, [tf.get_sentences(lang) for lang in langs]
)
# B - correct alignment
with open("texts/kanon_izr/everything") as f:
aB = NewAlignment.read(f)
baseline = NewAlignment()
baseline.easy_append(pl=" ".join(tf.get_sentences("pl")), cu=" ".join(tf.get_sentences("cu")))
aB.pretty_print("pl", "cu")
print evaluate_alignment(aA, aB)
print evaluate_alignment(baseline, aB)
