def get_verse_alignment_mp(self, verse_nums, edition_pairs):
res = []
ps_lang, pt_lang, index_t, alignments = None, None, None, None # if we have multiple edition pairs of the same languages, we use prev loaded files!
for edition_1, edition_2 in edition_pairs:
aligns = {}
if self.get_lang_from_edition(
edition_1) == self.get_lang_from_edition(edition_2):
res.append((edition_1, edition_2, aligns))
continue
if edition_1 in self.bert_files and edition_2 in self.bert_files:
LOG.info("going to super aglingment for: {}, {}".format(
edition_1, edition_2))
res.append((edition_1, edition_2, super().get_verse_alignment(
verse_nums, self.lang_prf_map[edition_1],
self.lang_prf_map[edition_2])))
continue
LOG.info("getting eflomal aglingment for: {} , {}".format(
edition_1, edition_2))
s_lang, t_lang, s_edition, t_edition = self.get_ordered_editions(
edition_1, edition_2)
s_lang_file = self.edition_file_mapping[s_edition]
t_lang_file = self.edition_file_mapping[t_edition]
revert = False
if s_edition == edition_2:
revert = True
if s_lang != ps_lang or t_lang != pt_lang:
alignments = self.get_alignment(s_lang, t_lang)
index_t = self.get_index(s_lang, t_lang)
ps_lang, pt_lang = s_lang, t_lang
index = None
if s_lang_file in index_t:
if t_lang_file in index_t[s_lang_file]:
index = index_t[s_lang_file][t_lang_file]
if index is not None:
LOG.info(
"getting verse, {}, {}, {}, {}, {}, {}, {}, {}".format(
edition_1, edition_2, s_lang, t_lang, ps_lang, pt_lang,
len(index_t), len(index)))
for verse in verse_nums:
if verse in index:
aligns[verse] = self.create_ordered_alignment(
alignments, index[verse], revert)
LOG.info("verses got")
else:
LOG.warning("couldn't find index for: " + s_edition + ", " +
t_edition)
res.append((edition_1, edition_2, aligns))
return res
def read_langs_order_file(self):
res = []
try:
with open(self.lang_order_file_path, 'r') as inf:
for l in inf:
res.append(l.strip())
except FileNotFoundError as e:
LOG.warning("Langs order file not found")
return res
def read_dict_file(self, file_path, do_lower=False):
res = {}
try:
with open(file_path, "r") as mapping_list:
for l in mapping_list:
if l.startswith('#'):
continue
if do_lower:
l.lower()
pair = l.strip().split('\t')
res[pair[0].strip()] = pair[1].strip()
except FileNotFoundError:
LOG.warning(f"file {file_path} not found")
return res
