if fst_count % 10 == 0:
str_fst.minimize()
fst_count += 1
stderr.write("\n")
ustr_fsts.intersect(ustr_fst)
all_ustr_fsts.append(ustr_fsts)
str_fsts.intersect(str_fst)
all_str_fsts.append(str_fsts)
ustr_model = libhfst.regex(base)
str_model = libhfst.regex(base)
for i, fst in enumerate(all_ustr_fsts):
print "USTR: %u of %u" % (i, len(all_ustr_fsts))
ustr_model.intersect(fst)
ustr_model.minimize()
for i, fst in enumerate(all_str_fsts):
print "STR: %u of %u" % (i, len(all_str_fsts))
str_model.intersect(fst)
str_model.minimize()
out = libhfst.create_hfst_output_stream(argv[1] + ".ustr", libhfst.TROPICAL_OPENFST_TYPE, 1)
out.write(ustr_model)
out = libhfst.create_hfst_output_stream(argv[1] + ".str", libhfst.TROPICAL_OPENFST_TYPE, 1)
out.write(str_model)
rule_str = ' '.join([center, left_context, center_context])
return regex(rule_str), rule_str
if __name__ == '__main__':
is_structured = 0
unstructured_model = {}
structured_rules = regex('?*')
structured_model = regex('?*')
oustr = open(argv[1] + '.ustr', 'wb')
ostr = create_hfst_output_stream(argv[1] + '.str', TROPICAL_OPENFST_TYPE,
1)
seen_struct_feats = set()
for i, line in enumerate(map(lambda x: x.strip(), stdin)):
if line == '':
continue
if line == STRUCTID:
stderr.write("Structured features.\n")
is_structured = 1
elif line == UNSTRUCTID:
stderr.write("Unstructured features.\n")
is_structured = 0
else:
if is_structured:
fields = line.split(' ')
for i in iocounts:
odistr = sorted([(iocounts[i][o], o) for o in iocounts[i]])
odistr.reverse()
tot = 0
for m, o in odistr:
outputs[i].append(o)
tot += m
if m >= TH:
break
outputs['#'].append('#')
outputs['_#_'].append('_#_')
out = libhfst.create_hfst_output_stream("",
libhfst.TROPICAL_OPENFST_TYPE, 1)
ustr_model = libhfst.HfstInputStream(argv[2]).read()
str_model = libhfst.HfstInputStream(argv[3]).read()
for i, line in enumerate(imap(lambda x: x.strip(), stdin)):
stderr.write("LINE: %u\r" % i)
expr = ''
if line == '':
continue
chars = ('_#_ _#_ # ' + line.replace('0','"0"') + ' # _#_ _#_').split(' ')
for char in chars:
expr += ('%s [%s] £ ' % (escape(char),
'|'.join([escape(c) for c in outputs[char]])))
