def translate_struct(ref_lang,
target_lang,
ref_struct,
target_struct,
db,
token,
parent_key=None):
for k, v in ref_struct.items():
kp = k.split("|")
kr = kp[0]
full_key = kr if parent_key is None else f'{parent_key}.{kr}'
if isinstance(v, str):
kh = hash(v)
rh = db.get(target_lang, full_key)
if rh == kh:
continue # translation is still up-to-date
print(
f"Translating {full_key} from {ref_lang} to {target_lang}...")
try:
if target_lang == ref_lang:
translation = deserialize_text(serialize_text(v))
else:
translation = deserialize_text(
translate(serialize_text(v), ref_lang, target_lang,
token))
except:
print("Cannot translate, skipping...")
continue
target_struct[k] = translation
db.set(target_lang, full_key, kh)
else:
if not isinstance(target_struct.get(k), dict):
target_struct[k] = {}
translate_struct(ref_lang,
target_lang,
v,
target_struct[k],
db,
token,
parent_key=full_key)
return target_struct
def deserialize_struct(struct, translated=False):
for k, v in struct.items():
if isinstance(v, str):
if not translated:
# we make a roundtrip to remove translation markup...
struct[k] = deserialize_text(serialize_text(v))
# if there are <tr-snip> tags, we only return the text within them.
if re.match(r".*<tr-snip>", v):
snippets = []
for m in re.finditer(r"<tr-snip>([^<]+?)</tr-snip>", v):
snippets.append(m.group(1).strip())
struct[k] = " ".join(snippets)
else:
deserialize_struct(v, translated=translated)
return process_filters(struct)
def translate_struct(ref_lang, target_lang, translations, db, token):
if not target_lang in translations:
translations[target_lang] = {}
for k, v in translations.get('$' + ref_lang, {}).items():
kp = k.split("|")
kr = kp[0]
kh = hash(v)
rh = db.get(target_lang, kr)
if rh == kh:
continue #translation is still good
print(f"Translating {kr} from {ref_lang} to {target_lang}...")
try:
if target_lang == ref_lang:
translation = deserialize_text(serialize_text(v))
else:
translation = deserialize_text(
translate(serialize_text(v), ref_lang, target_lang, token))
except KeyboardInterrupt:
raise
except:
continue
translations[target_lang][kr] = translation
db.set(target_lang, kr, kh)
return translations
