def align(srcfile, tgtfile, approx_src_tgt_file):
# srcfile, tgtfile = tgtfile, srcfile
options = {
# source and target files needed by Aligner
# they can be filenames, arrays of strings or io objects.
'srcfile': srcfile,
'targetfile': tgtfile,
# translations of srcfile and targetfile, not influenced by 'factored'
# they can be filenames, arrays of strings or io objects, too.
'srctotarget': [approx_src_tgt_file],
'targettosrc': [],
# passing filenames or io object for them in respectly.
# if not passing anything or assigning None, they will use StringIO to save results.
'output-src': None,
'output-target': None,
# other options ...
'log_to': open("/dev/null", "w+")
}
a = Aligner(options)
a.mainloop()
output_src, output_target = a.results()
# output_src, output_target is StringIO because options['output-src'] is None
src = output_src.getvalue().splitlines() # StringIO member function
tgt = output_target.getvalue().splitlines() # array of string
return (src, tgt)
def _sameResultForEval(self, de, fr, s2t, t2s):
options = {
'srcfile': de,
'targetfile': fr,
'srctotarget': s2t,
'targettosrc': t2s,
'verbosity': 0,
}
a = Aligner(options)
a.mainloop()
output_src, output_target = a.results()
s2t_src = output_src.getvalue().splitlines()
s2t_trg = output_target.getvalue().splitlines()
options = {
'srcfile': fr,
'targetfile': de,
'srctotarget': t2s,
'targettosrc': s2t,
'verbosity': 0,
}
a = Aligner(options)
a.mainloop()
output_src, output_target = a.results()
t2s_src = output_src.getvalue().splitlines()
t2s_trg = output_target.getvalue().splitlines()
self.assertEqual(s2t_src, t2s_trg)
self.assertEqual(s2t_trg, t2s_src)
def main_test(self, option_function,
close_file_object = None, remove_file = None):
test_dir = os.path.dirname(os.path.abspath(__file__))
eval_dir = os.path.join(test_dir, '..', 'eval')
result_dir = os.path.join(test_dir, 'result')
refer_dir = os.path.join(test_dir, 'refer')
bleualign.log = lambda a, b:None
compare_files = []
for test_set, test_argument in [('eval1957', '-d'), ('eval1989', '-e')]:
fr_text = []
de_text = []
for filename in os.listdir(eval_dir):
if filename.startswith(test_set):
attr = filename.split('.')
if len(attr) == 3:
filepath = os.path.join(eval_dir, filename)
if attr[1] != 'clean':
if attr[2] == 'fr':
fr_text.append(filepath)
elif attr[2] == 'de':
de_text.append(filepath)
fr_text.sort()
de_text.sort()
test_files = []
test_files.append((fr_text[0:1], de_text[-3:-2]))
test_files.append((fr_text, []))
test_files.append((fr_text[1::3], de_text[-2:-1]))
test_files.append((fr_text[2:3], de_text[3:4]))
test_files.append((fr_text[0:1], []))
test_files.append((fr_text[2:], de_text[:3]))
test_files.append((fr_text, de_text))
# test_files.append(([], [])) add in another test after
# test_files.append(([], de_text))
# test_files.append(([], de_text[-1:]))
for fr_file, de_file in test_files:
srctotarget_file = fr_file
targettosrc_file = de_file
output_file = self.output_file_path(srctotarget_file, targettosrc_file)
output_path = os.path.join(result_dir , output_file)
options = getattr(self, option_function)(test_argument,
srctotarget_file, targettosrc_file, output_path)
a = Aligner(options)
a.mainloop()
output_src, output_target = a.results()
if close_file_object != None:
getattr(self, close_file_object)([output_src, output_target])
getattr(self, close_file_object)([options['targetfile']])
getattr(self, close_file_object)(options['targettosrc'])
if option_function == 'fileObjectOptions':
getattr(self, close_file_object)([options['srcfile']])
getattr(self, close_file_object)(options['srctotarget'])
refer_path = os.path.join(refer_dir , output_file)
compare_files.append((output_path + '-s', refer_path + '-s', output_src))
compare_files.append((output_path + '-t', refer_path + '-t', output_target))
# compare result with data in refer
for result_path, refer_path, output_object in compare_files:
self.cmp_files(result_path, refer_path, output_object)
if remove_file != None:
getattr(self, remove_file)(result_path)
def 對齊(self, 原來, 目標, 原來翻目標, 目標翻原來, 原來對齊=None, 目標對齊=None): 參數 = self.公家參數.copy() 參數['srcfile'] = 原來 參數['targetfile'] = 目標 參數['srctotarget'] = 原來翻目標 參數['targettosrc'] = 目標翻原來 參數['output-src'] = 原來對齊 參數['output-target'] = 目標對齊 a = Aligner(參數) a.mainloop() return a.results()
def main_test(self, option_function):
test_dir = os.path.dirname(os.path.abspath(__file__))
eval_dir = os.path.join(test_dir, '..', 'eval')
result_dir = os.path.join(test_dir, 'result')
refer_dir = os.path.join(test_dir, 'refer')
bleualign.log = lambda a, b:None
compare_files = []
for test_set, test_argument in [('eval1957', '-d'), ('eval1989', '-e')]:
fr_text = []
de_text = []
for filename in os.listdir(eval_dir):
if filename.startswith(test_set):
attr = filename.split('.')
if len(attr) == 3:
filepath = os.path.join(eval_dir, filename)
if attr[1] != 'clean':
if attr[2] == 'fr':
fr_text.append(filepath)
elif attr[2] == 'de':
de_text.append(filepath)
fr_text.sort()
de_text.sort()
test_files = []
test_files.append((fr_text[0:1], de_text[-3:-2], 'articles'))
test_files.append((fr_text, [], 'sentences'))
test_files.append((fr_text, de_text, 'sentences'))
for fr_file, de_file, filter_type in test_files:
srctotarget_file = fr_file
targettosrc_file = de_file
output_file = self.output_file_path(srctotarget_file, targettosrc_file)
output_path = os.path.join(result_dir , output_file)
options = getattr(self, option_function)(test_argument, filter_type,
srctotarget_file, targettosrc_file, output_path)
a = Aligner(options)
a.mainloop()
output_src, output_target = a.results()
output_src_bad, output_target_bad = a.results_bad()
if option_function == 'fileObjectOptions':
output_src.close()
output_target.close()
output_src_bad.close()
output_target_bad.close()
refer_path = os.path.join(refer_dir , output_file)
compare_files.append((output_path + '-good-s', refer_path + '-good-s', output_src))
compare_files.append((output_path + '-good-t', refer_path + '-good-t', output_target))
compare_files.append((output_path + '-bad-s', refer_path + '-bad-s', output_src_bad))
compare_files.append((output_path + '-bad-t', refer_path + '-bad-t', output_target_bad))
for result_path, refer_path, output_object in compare_files:
self.cmp_files(result_path, refer_path, output_object)
if option_function.startswith('file'):
os.remove(result_path)
def _sameResultForEval(self,de,fr,s2t,t2s):
options = {
'srcfile':de,
'targetfile':fr,
'srctotarget':s2t,
'targettosrc':t2s,
'verbosity':0,
}
a = Aligner(options)
a.mainloop()
output_src, output_target = a.results()
s2t_src = output_src.getvalue().splitlines()
s2t_trg = output_target.getvalue().splitlines()
options = {
'srcfile':fr,
'targetfile':de,
'srctotarget':t2s,
'targettosrc':s2t,
'verbosity':0,
}
a = Aligner(options)
a.mainloop()
output_src, output_target = a.results()
t2s_src = output_src.getvalue().splitlines()
t2s_trg = output_target.getvalue().splitlines()
self.assertEqual(s2t_src, t2s_trg)
self.assertEqual(s2t_trg, t2s_src)
def test_no_translation(self):
self.assertRaises(ValueError, Aligner,
{'srcfile':self.srcfile, 'targetfile':self.targetfile})
a=Aligner(
{'srcfile':self.srcfile, 'targetfile':self.targetfile,
'no_translation_override':True})
a.close_file_streams()
a=Aligner(
{'srcfile':self.srcfile, 'targetfile':self.targetfile,
'srctotarget':[self.targetfile]})
a.close_file_streams()
a=Aligner(
{'srcfile':self.srcfile, 'targetfile':self.targetfile,
'targettosrc':[self.srcfile]})
a.close_file_streams()
def bleu_align(self, srcfile, tgtfile, hyp_src_tgt_file):
output = StringIO()
# src_out, tgt_out = StringIO(), StringIO()
options = {
'verbosity': 0,
'srcfile': srcfile,
'targetfile': tgtfile,
'srctotarget': [hyp_src_tgt_file],
'targettosrc': [],
# 'output': output,
# 'output-src': src_out, 'output-target': tgt_out,
}
a = Aligner(options)
a.mainloop()
src_out, tgt_out = a.results()
srcs = src_out.getvalue().splitlines()
tgts = tgt_out.getvalue().splitlines()
return srcs, tgts
def test_gale_church(self):
test_dir = os.path.dirname(os.path.abspath(__file__))
result_dir = os.path.join(test_dir, 'result')
refer_dir = os.path.join(test_dir, 'refer')
bleualign.log = lambda a, b:None
compare_files = []
for test_set, test_argument in [('eval1957', '-d'), ('eval1989', '-e')]:
options = load_arguments(['', test_argument, '--srctotarget', '-'])
output_file = test_set + '-galechurch'
output_path = os.path.join(result_dir , output_file)
# options['output'] = output_path
a = Aligner(options)
a.mainloop()
output_src, output_target = a.results()
refer_path = os.path.join(refer_dir , output_file)
compare_files.append((output_path + '-s', refer_path + '-s', output_src))
compare_files.append((output_path + '-t', refer_path + '-t', output_target))
# compare result with data in refer
for result_path, refer_path, output_object in compare_files:
self.cmp_files(result_path, refer_path, output_object)
def bleu_align(self, srcfile, tgtfile, hyp_src_tgt_file=None):
output = StringIO()
options = {
'srcfile': srcfile,
'targetfile': tgtfile,
'galechurch': True if hyp_src_tgt_file is None else False,
'no_translation_override':
True if hyp_src_tgt_file is None else False,
'srctotarget': [hyp_src_tgt_file] if hyp_src_tgt_file else [],
'targettosrc': [],
'verbosity': 0,
}
a = Aligner(options)
a.mainloop()
src_out, tgt_out = a.results()
srcs = src_out.getvalue().splitlines()
tgts = tgt_out.getvalue().splitlines()
return srcs, tgts
def test_gale_church(self):
test_dir = os.path.dirname(os.path.abspath(__file__))
result_dir = os.path.join(test_dir, 'result')
refer_dir = os.path.join(test_dir, 'refer')
bleualign.log = lambda a, b: None
compare_files = []
for test_set, test_argument in [('eval1957', '-d'),
('eval1989', '-e')]:
options = load_arguments(['', test_argument, '--srctotarget', '-'])
output_file = test_set + '-galechurch'
output_path = os.path.join(result_dir, output_file)
# options['output'] = output_path
a = Aligner(options)
a.mainloop()
output_src, output_target = a.results()
refer_path = os.path.join(refer_dir, output_file)
compare_files.append(
(output_path + '-s', refer_path + '-s', output_src))
compare_files.append(
(output_path + '-t', refer_path + '-t', output_target))
# compare result with data in refer
for result_path, refer_path, output_object in compare_files:
self.cmp_files(result_path, refer_path, output_object)
options['verbosity'] = 1
options['printempty'] = False
options['output'] = None
jobs = []
for source_document in [d for d in os.listdir(directory) if d.endswith('.' + source_suffix)]:
source_document = os.path.join(directory, source_document)
target_document = source_document[:-len(source_suffix)] + target_suffix
translation_document = source_document[:-len(source_suffix)] + translation_suffix
# Sanity checks
for f in source_document, target_document, translation_document:
if not os.path.isfile(f):
sys.stderr.write('ERROR: File {0} expected, but not found\n'.format(f))
exit()
jobs.append((source_document, target_document, translation_document))
for (source_document,target_document,translation_document) in jobs:
options['srcfile'] = source_document
options['targetfile'] = target_document
options['srctotarget'] = [translation_document]
options['output-src'] = source_document + '.aligned'
options['output-target'] = target_document + '.aligned'
a = Aligner(options)
a.mainloop()
jobs = []
for source_document in [
d for d in os.listdir(directory) if d.endswith('.' + source_suffix)
]:
source_document = os.path.join(directory, source_document)
target_document = source_document[:-len(source_suffix)] + target_suffix
translation_document = source_document[:-len(source_suffix
)] + translation_suffix
# Sanity checks
for f in source_document, target_document, translation_document:
if not os.path.isfile(f):
sys.stderr.write(
'ERROR: File {0} expected, but not found\n'.format(f))
exit()
jobs.append((source_document, target_document, translation_document))
for (source_document, target_document, translation_document) in jobs:
options['srcfile'] = source_document
options['targetfile'] = target_document
options['srctotarget'] = [translation_document]
options['output-src'] = source_document + '.aligned'
options['output-target'] = target_document + '.aligned'
a = Aligner(options)
a.mainloop()
# source and target files needed by Aligner
# they can be filenames, arrays of strings or io objects.
'srcfile':
os.path.join(current_path, '..', 'eval', 'eval1989.de'),
'targetfile':
os.path.join(current_path, '..', 'eval', 'eval1989.fr'),
# translations of srcfile and targetfile, not influenced by 'factored'
# they can be filenames, arrays of strings or io objects, too.
'srctotarget':
[os.path.join(current_path, '..', 'eval', 'eval1957.europarlfull.fr')],
'targettosrc': [],
# passing filenames or io object for them in respectly.
# if not passing anything or assigning None, they will use StringIO to save results.
'output-src':
None,
'output-target':
None,
# other options ...
}
a = Aligner(options)
a.mainloop()
output_src, output_target = a.results()
# output_src, output_target is StringIO because options['output-src'] is None
src = output_src.getvalue() # StringIO member function
trg = output_target.getvalue().splitlines() # array of string
print('output_src.getvalue()')
print(src[:30])
print()
print('output_target.getvalue().splitlines()')
print(trg[:3])
import os
from bleualign.align import Aligner
if __name__ == '__main__':
current_path = os.path.dirname(os.path.abspath(__file__))
options = {
# source and target files needed by Aligner
# they can be filenames, arrays of strings or io objects.
'srcfile':os.path.join(current_path, '..', 'eval', 'eval1989.de'),
'targetfile': os.path.join(current_path, '..', 'eval', 'eval1989.fr'),
# translations of srcfile and targetfile, not influenced by 'factored'
# they can be filenames, arrays of strings or io objects, too.
'srctotarget': [os.path.join(current_path, '..', 'eval', 'eval1957.europarlfull.fr')],
'targettosrc': [],
# passing filenames or io object for them in respectly.
# if not passing anything or assigning None, they will use StringIO to save results.
'output-src': None, 'output-target': None,
# other options ...
}
a = Aligner(options)
a.mainloop()
output_src, output_target = a.results()
# output_src, output_target is StringIO because options['output-src'] is None
src = output_src.getvalue() # StringIO member function
trg = output_target.getvalue().splitlines() # array of string
print('output_src.getvalue()')
print(src[:30])
print()
print('output_target.getvalue().splitlines()')
print(trg[:3])
def main_test(self, option_function):
test_dir = os.path.dirname(os.path.abspath(__file__))
eval_dir = os.path.join(test_dir, '..', 'eval')
result_dir = os.path.join(test_dir, 'result')
refer_dir = os.path.join(test_dir, 'refer')
bleualign.log = lambda a, b: None
compare_files = []
for test_set, test_argument in [('eval1957', '-d'),
('eval1989', '-e')]:
fr_text = []
de_text = []
for filename in os.listdir(eval_dir):
if filename.startswith(test_set):
attr = filename.split('.')
if len(attr) == 3:
filepath = os.path.join(eval_dir, filename)
if attr[1] != 'clean':
if attr[2] == 'fr':
fr_text.append(filepath)
elif attr[2] == 'de':
de_text.append(filepath)
fr_text.sort()
de_text.sort()
test_files = []
test_files.append((fr_text[0:1], de_text[-3:-2], 'articles'))
test_files.append((fr_text, [], 'sentences'))
test_files.append((fr_text, de_text, 'sentences'))
for fr_file, de_file, filter_type in test_files:
srctotarget_file = fr_file
targettosrc_file = de_file
output_file = self.output_file_path(srctotarget_file,
targettosrc_file)
output_path = os.path.join(result_dir, output_file)
options = getattr(self,
option_function)(test_argument, filter_type,
srctotarget_file,
targettosrc_file,
output_path)
a = Aligner(options)
a.mainloop()
output_src, output_target = a.results()
output_src_bad, output_target_bad = a.results_bad()
if option_function == 'fileObjectOptions':
output_src.close()
output_target.close()
output_src_bad.close()
output_target_bad.close()
refer_path = os.path.join(refer_dir, output_file)
compare_files.append((output_path + '-good-s',
refer_path + '-good-s', output_src))
compare_files.append((output_path + '-good-t',
refer_path + '-good-t', output_target))
compare_files.append((output_path + '-bad-s',
refer_path + '-bad-s', output_src_bad))
compare_files.append(
(output_path + '-bad-t', refer_path + '-bad-t',
output_target_bad))
for result_path, refer_path, output_object in compare_files:
self.cmp_files(result_path, refer_path, output_object)
if option_function.startswith('file'):
os.remove(result_path)
def test_no_translation(self):
self.assertRaises(ValueError, Aligner, {
'srcfile': self.srcfile,
'targetfile': self.targetfile
})
a = Aligner({
'srcfile': self.srcfile,
'targetfile': self.targetfile,
'no_translation_override': True
})
a.close_file_streams()
a = Aligner({
'srcfile': self.srcfile,
'targetfile': self.targetfile,
'srctotarget': [self.targetfile]
})
a.close_file_streams()
a = Aligner({
'srcfile': self.srcfile,
'targetfile': self.targetfile,
'targettosrc': [self.srcfile]
})
a.close_file_streams()
'targettosrc': [],
# passing filenames or io object for them in respectly.
# if not passing anything or assigning None, they will use StringIO to save results.
'output-src':
None,
'output-target':
None,
'output-src-bad':
None,
'output-target-bad':
None,
# other options ...
}
options['filter'] = 'sentences'
options['filterthreshold'] = 66
a = Aligner(options)
a.mainloop()
output_src, output_target = a.results()
output_src_bad, output_target_bad = a.results_bad(
) # if you set options['filter']
# output_src is StringIO because options['output-src'] is None
src = output_src.getvalue() # StringIO member function
trg = output_target.getvalue().splitlines() # array of string
print('output_src.getvalue()')
print(src[:30])
print()
print('output_target.getvalue().splitlines()')
print(trg[:3])
print()
print('filterthreshold for choice good part of alignment: {0}%'.format(
options['filterthreshold']))
# they can be filenames, arrays of strings or io objects.
'srcfile':os.path.join(current_path, '..', 'eval', 'eval1989.de'),
'targetfile': os.path.join(current_path, '..', 'eval', 'eval1989.fr'),
# translations of srcfile and targetfile, not influenced by 'factored'
# they can be filenames, arrays of strings or io objects, too.
'srctotarget': [os.path.join(current_path, '..', 'eval', 'eval1957.europarlfull.fr')],
'targettosrc': [],
# passing filenames or io object for them in respectly.
# if not passing anything or assigning None, they will use StringIO to save results.
'output-src': None, 'output-target': None,
'output-src-bad': None, 'output-target-bad': None,
# other options ...
}
options['filter'] = 'sentences'
options['filterthreshold'] = 66
a = Aligner(options)
a.mainloop()
output_src, output_target = a.results()
output_src_bad, output_target_bad = a.results_bad() # if you set options['filter']
# output_src is StringIO because options['output-src'] is None
src = output_src.getvalue() # StringIO member function
trg = output_target.getvalue().splitlines() # array of string
print('output_src.getvalue()')
print(src[:30])
print()
print('output_target.getvalue().splitlines()')
print(trg[:3])
print()
print('filterthreshold for choice good part of alignment: {0}%'.format(options['filterthreshold']))
print('number of good/bad alignmemts: {0}'.format(len(output_src.getvalue().splitlines()), len(output_src_bad.getvalue().splitlines())))
