def test_latin(self):
self.assertEqual(
u"Jean Michel Jarre", util.translate_from_sortname(u"Jean Michel Jarre", u"Jarre, Jean Michel")
)
self.assertNotEqual(
u"Jarre, Jean Michel", util.translate_from_sortname(u"Jean Michel Jarre", u"Jarre, Jean Michel")
)
def test_kanji(self):
self.failUnlessEqual(
u"Tetsuya Komuro",
util.translate_from_sortname(u"小室哲哉", u"Komuro, Tetsuya"))
self.failIfEqual(
u"Komuro, Tetsuya",
util.translate_from_sortname(u"小室哲哉", u"Komuro, Tetsuya"))
self.failIfEqual(
u"小室哲哉", util.translate_from_sortname(u"小室哲哉", u"Komuro, Tetsuya"))
def test_kanji(self):
self.assertEqual(
"Tetsuya Komuro",
util.translate_from_sortname("小室哲哉", "Komuro, Tetsuya"))
# see _reverse_sortname(), cases with 3 or 4 chunks
self.assertEqual("c b a",
util.translate_from_sortname("小室哲哉", "a, b, c"))
self.assertEqual("b a, d c",
util.translate_from_sortname("小室哲哉", "a, b, c, d"))
def test_latin(self):
self.assertEqual(
u"Jean Michel Jarre",
util.translate_from_sortname(u"Jean Michel Jarre",
u"Jarre, Jean Michel"))
self.assertNotEqual(
u"Jarre, Jean Michel",
util.translate_from_sortname(u"Jean Michel Jarre",
u"Jarre, Jean Michel"))
def test_kanji(self):
self.assertEqual(
u"Tetsuya Komuro",
util.translate_from_sortname(u"小室哲哉", u"Komuro, Tetsuya"))
self.assertNotEqual(
u"Komuro, Tetsuya",
util.translate_from_sortname(u"小室哲哉", u"Komuro, Tetsuya"))
self.assertNotEqual(
u"小室哲哉", util.translate_from_sortname(u"小室哲哉", u"Komuro, Tetsuya"))
def test_latin(self):
self.failUnlessEqual(
u"Jean Michel Jarre",
util.translate_from_sortname(u"Jean Michel Jarre",
u"Jarre, Jean Michel"))
self.failIfEqual(
u"Jarre, Jean Michel",
util.translate_from_sortname(u"Jean Michel Jarre",
u"Jarre, Jean Michel"))
def test_kanji2(self):
self.assertEqual(
u"Ayumi Hamasaki & Keiko", util.translate_from_sortname(u"浜崎あゆみ & KEIKO", u"Hamasaki, Ayumi & Keiko")
)
self.assertNotEqual(
u"浜崎あゆみ & KEIKO", util.translate_from_sortname(u"浜崎あゆみ & KEIKO", u"Hamasaki, Ayumi & Keiko")
)
self.assertNotEqual(
u"Hamasaki, Ayumi & Keiko", util.translate_from_sortname(u"浜崎あゆみ & KEIKO", u"Hamasaki, Ayumi & Keiko")
)
def test_kanji2(self):
self.failUnlessEqual(
u"Ayumi Hamasaki & Keiko",
util.translate_from_sortname(u"浜崎あゆみ & KEIKO",
u"Hamasaki, Ayumi & Keiko"))
self.failIfEqual(
u"浜崎あゆみ & KEIKO",
util.translate_from_sortname(u"浜崎あゆみ & KEIKO",
u"Hamasaki, Ayumi & Keiko"))
self.failIfEqual(
u"Hamasaki, Ayumi & Keiko",
util.translate_from_sortname(u"浜崎あゆみ & KEIKO",
u"Hamasaki, Ayumi & Keiko"))
def test_kanji2(self):
self.assertEqual(
u"Ayumi Hamasaki & Keiko",
util.translate_from_sortname(u"浜崎あゆみ & KEIKO",
u"Hamasaki, Ayumi & Keiko"))
self.assertNotEqual(
u"浜崎あゆみ & KEIKO",
util.translate_from_sortname(u"浜崎あゆみ & KEIKO",
u"Hamasaki, Ayumi & Keiko"))
self.assertNotEqual(
u"Hamasaki, Ayumi & Keiko",
util.translate_from_sortname(u"浜崎あゆみ & KEIKO",
u"Hamasaki, Ayumi & Keiko"))
def test_cyrillic(self):
self.assertEqual(
U"Pyotr Ilyich Tchaikovsky",
util.translate_from_sortname(u"Пётр Ильич Чайковский",
u"Tchaikovsky, Pyotr Ilyich"))
self.assertNotEqual(
u"Tchaikovsky, Pyotr Ilyich",
util.translate_from_sortname(u"Пётр Ильич Чайковский",
u"Tchaikovsky, Pyotr Ilyich"))
self.assertNotEqual(
u"Пётр Ильич Чайковский",
util.translate_from_sortname(u"Пётр Ильич Чайковский",
u"Tchaikovsky, Pyotr Ilyich"))
def test_cyrillic(self):
self.assertEqual(
u"Pyotr Ilyich Tchaikovsky",
util.translate_from_sortname(u"Пётр Ильич Чайковский", u"Tchaikovsky, Pyotr Ilyich"),
)
self.assertNotEqual(
u"Tchaikovsky, Pyotr Ilyich",
util.translate_from_sortname(u"Пётр Ильич Чайковский", u"Tchaikovsky, Pyotr Ilyich"),
)
self.assertNotEqual(
u"Пётр Ильич Чайковский",
util.translate_from_sortname(u"Пётр Ильич Чайковский", u"Tchaikovsky, Pyotr Ilyich"),
)
def _translate_artist_node(node, config=None):
transl, translsort = None, None
if config and config.setting['translate_artist_names']:
locale = config.setting["artist_locale"]
lang = locale.split("_")[0]
if "alias_list" in node.children:
found_primary = found_locale = False
for alias in node.alias_list[0].alias:
if alias.attribs.get(
"type") != "Search hint" and "locale" in alias.attribs:
if alias.locale == locale:
transl, translsort = alias.text, alias.attribs[
"sort_name"]
if alias.attribs.get("primary") == "primary":
return (transl, translsort)
found_locale = True
elif alias.locale == lang and not (found_locale
or found_primary):
transl, translsort = alias.text, alias.attribs[
"sort_name"]
if alias.attribs.get("primary") == "primary":
found_primary = True
if lang == "en" and not transl:
transl = translate_from_sortname(node.name[0].text,
node.sort_name[0].text)
return (transl, translsort)
def _translate_artist_node(node, config=None):
transl = None
if config and config.setting['translate_artist_names']:
locale = config.setting["artist_locale"]
lang = locale.split("_")[0]
if "alias_list" in node.children:
for alias in node.alias_list[0].alias:
if "locale" in alias.attribs:
if alias.locale == locale:
return alias.text
elif alias.locale == lang:
transl = alias.text
if lang == "en" and not transl:
transl = translate_from_sortname(node.name[0].text, node.sort_name[0].text)
return transl
def _translate_artist_node(node, config=None):
transl = None
if config and config.setting['translate_artist_names']:
locale = config.setting["artist_locale"]
lang = locale.split("_")[0]
if "alias_list" in node.children:
for alias in node.alias_list[0].alias:
if "locale" in alias.attribs:
if alias.locale == locale:
return alias.text
elif alias.locale == lang:
transl = alias.text
if lang == "en" and not transl:
transl = translate_from_sortname(node.name[0].text,
node.sort_name[0].text)
return transl
def _translate_artist_node(node):
config = get_config()
transl, translsort = None, None
if config.setting['translate_artist_names']:
locale = config.setting["artist_locale"]
lang = locale.split("_")[0]
if "aliases" in node:
result = (-1, (None, None))
for alias in node['aliases']:
if not alias["primary"]:
continue
if "locale" not in alias:
continue
parts = []
if alias['locale'] == locale:
score = 0.8
elif alias['locale'] == lang:
score = 0.6
elif alias['locale'].split("_")[0] == lang:
score = 0.4
else:
continue
parts.append((score, 5))
if alias["type"] == "Artist name":
score = 0.8
elif alias["type"] == "Legal Name":
score = 0.5
else:
# as 2014/09/19, only Artist or Legal names should have the
# Primary flag
score = 0.0
parts.append((score, 5))
comb = linear_combination_of_weights(parts)
if comb > result[0]:
result = (comb, (alias['name'], alias["sort-name"]))
transl, translsort = result[1]
if not transl:
translsort = node['sort-name']
transl = translate_from_sortname(node['name'] or "", translsort)
else:
transl, translsort = node['name'], node['sort-name']
return (transl, translsort)
def _translate_artist_node(node):
transl, translsort = None, None
if config.setting['translate_artist_names']:
locale = config.setting["artist_locale"]
lang = locale.split("_")[0]
if "alias_list" in node.children:
result = (-1, (None, None))
for alias in node.alias_list[0].alias:
if alias.attribs.get("primary") != "primary":
continue
if "locale" not in alias.attribs:
continue
parts = []
if alias.locale == locale:
score = 0.8
elif alias.locale == lang:
score = 0.6
elif alias.locale.split("_")[0] == lang:
score = 0.4
else:
continue
parts.append((score, 5))
if alias.attribs.get("type") == "Artist name":
score = 0.8
elif alias.attribs.get("type") == "Legal Name":
score = 0.5
else:
# as 2014/09/19, only Artist or Legal names should have the
# Primary flag
score = 0.0
parts.append((score, 5))
comb = linear_combination_of_weights(parts)
if comb > result[0]:
result = (comb, (alias.text, alias.attribs["sort_name"]))
transl, translsort = result[1]
if not transl:
translsort = node.sort_name[0].text
transl = translate_from_sortname(node.name[0].text, translsort)
else:
transl, translsort = node.name[0].text, node.sort_name[0].text
return (transl, translsort)
def _translate_artist_node(node):
transl, translsort = None, None
if config.setting['translate_artist_names']:
locale = config.setting["artist_locale"]
lang = locale.split("_")[0]
if "aliases" in node:
result = (-1, (None, None))
for alias in node['aliases']:
if not alias["primary"]:
continue
if "locale" not in alias:
continue
parts = []
if alias['locale'] == locale:
score = 0.8
elif alias['locale'] == lang:
score = 0.6
elif alias['locale'].split("_")[0] == lang:
score = 0.4
else:
continue
parts.append((score, 5))
if alias["type"] == "Artist name":
score = 0.8
elif alias["type"] == "Legal Name":
score = 0.5
else:
# as 2014/09/19, only Artist or Legal names should have the
# Primary flag
score = 0.0
parts.append((score, 5))
comb = linear_combination_of_weights(parts)
if comb > result[0]:
result = (comb, (alias['name'], alias["sort-name"]))
transl, translsort = result[1]
if not transl:
translsort = node['sort-name']
transl = translate_from_sortname(node['name'] or "", translsort)
else:
transl, translsort = node['name'], node['sort-name']
return (transl, translsort)
def _translate_artist_node(node):
transl, translsort = None, None
if config.setting['translate_artist_names']:
locale = config.setting["artist_locale"]
lang = locale.split("_")[0]
if "alias_list" in node.children:
result = (-1, (None, None))
for alias in node.alias_list[0].alias:
if alias.attribs.get("primary") != "primary":
continue
if "locale" not in alias.attribs:
continue
parts = []
if alias.locale == locale:
score = 0.8
elif alias.locale == lang:
score = 0.6
elif alias.locale.split("_")[0] == lang:
score = 0.4
else:
continue
parts.append((score, 5))
if alias.attribs.get("type") == u"Artist name":
score = 0.8
elif alias.attribs.get("type") == u"Legal Name":
score = 0.5
else:
# as 2014/09/19, only Artist or Legal names should have the
# Primary flag
score = 0.0
parts.append((score, 5))
comb = linear_combination_of_weights(parts)
if comb > result[0]:
result = (comb, (alias.text, alias.attribs["sort_name"]))
transl, translsort = result[1]
if not transl:
translsort = node.sort_name[0].text
transl = translate_from_sortname(node.name[0].text, translsort)
else:
transl, translsort = node.name[0].text, node.sort_name[0].text
return (transl, translsort)
def _translate_artist_node(node):
transl, translsort = None, None
if config.setting['translate_artist_names']:
locale = config.setting["artist_locale"]
lang = locale.split("_")[0]
if "alias_list" in node.children:
found_primary = found_locale = False
for alias in node.alias_list[0].alias:
if alias.attribs.get("type") != "Search hint" and "locale" in alias.attribs:
if alias.locale == locale:
transl, translsort = alias.text, alias.attribs["sort_name"]
if alias.attribs.get("primary") == "primary":
return (transl, translsort)
found_locale = True
elif alias.locale == lang and not (found_locale or found_primary):
transl, translsort = alias.text, alias.attribs["sort_name"]
if alias.attribs.get("primary") == "primary":
found_primary = True
if lang == "en" and not transl:
transl = translate_from_sortname(node.name[0].text, node.sort_name[0].text)
return (transl, translsort)
def test_cyrillic(self):
self.failUnlessEqual(U"Pyotr Ilyich Tchaikovsky", util.translate_from_sortname(u"Пётр Ильич Чайковский", u"Tchaikovsky, Pyotr Ilyich"))
self.failIfEqual(u"Tchaikovsky, Pyotr Ilyich", util.translate_from_sortname(u"Пётр Ильич Чайковский", u"Tchaikovsky, Pyotr Ilyich"))
self.failIfEqual(u"Пётр Ильич Чайковский", util.translate_from_sortname(u"Пётр Ильич Чайковский", u"Tchaikovsky, Pyotr Ilyich"))
def test_kanji(self):
self.failUnlessEqual(u"Tetsuya Komuro", util.translate_from_sortname(u"小室哲哉", u"Komuro, Tetsuya"))
self.failIfEqual(u"Komuro, Tetsuya", util.translate_from_sortname(u"小室哲哉", u"Komuro, Tetsuya"))
self.failIfEqual(u"小室哲哉", util.translate_from_sortname(u"小室哲哉", u"Komuro, Tetsuya"))
def test_kanji2(self):
self.failUnlessEqual(u"Ayumi Hamasaki & Keiko", util.translate_from_sortname(u"浜崎あゆみ & KEIKO", u"Hamasaki, Ayumi & Keiko"))
self.failIfEqual(u"浜崎あゆみ & KEIKO", util.translate_from_sortname(u"浜崎あゆみ & KEIKO", u"Hamasaki, Ayumi & Keiko"))
self.failIfEqual(u"Hamasaki, Ayumi & Keiko", util.translate_from_sortname(u"浜崎あゆみ & KEIKO", u"Hamasaki, Ayumi & Keiko"))
def _translate_artist_node(node):
config = get_config()
transl, translsort = None, None
if config.setting['translate_artist_names']:
if config.setting['translate_artist_names_script_exception']:
threshhold = config.setting["artist_script_exception_weighting"] / 100
detected_scripts = list_script_weighted(node["name"], threshhold)
for script_id in config.setting["artist_script_exceptions"]:
if script_id in detected_scripts:
return node['name'], node['sort-name']
def check_higher_score(locale_dict, locale, score):
return locale not in locale_dict or score > locale_dict[locale][0]
# Prepare dictionaries of available locale aliases
full_locales = {}
root_locales = {}
if "aliases" in node:
for alias in node['aliases']:
if not alias["primary"]:
continue
if "locale" not in alias:
continue
full_locale = alias['locale']
root_locale = full_locale.split("_")[0]
full_parts = []
root_parts = []
score = 0.8
full_parts.append((score, 5))
if '_' in full_locale:
score = 0.4
root_parts.append((score, 5))
if alias["type"] == "Artist name":
score = 0.8
elif alias["type"] == "Legal Name":
score = 0.5
else:
# as 2014/09/19, only Artist or Legal names should have the
# Primary flag
score = 0.0
full_parts.append((score, 5))
root_parts.append((score, 5))
comb = linear_combination_of_weights(full_parts)
if check_higher_score(full_locales, full_locale, comb):
full_locales[full_locale] = (comb, (alias['name'], alias["sort-name"]))
comb = linear_combination_of_weights(root_parts)
if check_higher_score(root_locales, root_locale, comb):
root_locales[root_locale] = (comb, (alias['name'], alias["sort-name"]))
# First pass to match full locale if available
for locale in config.setting["artist_locales"]:
if locale in full_locales:
return full_locales[locale][1]
# Second pass to match root locale if available
for locale in config.setting["artist_locales"]:
lang = locale.split("_")[0]
if lang in root_locales:
return root_locales[lang][1]
# No matches found in available alias locales
translsort = node['sort-name']
transl = translate_from_sortname(node['name'] or "", translsort)
else:
transl, translsort = node['name'], node['sort-name']
return (transl, translsort)
def _translate_artist_node(node):
config = get_config()
transl, translsort = None, None
if config.setting['translate_artist_names']:
if config.setting['translate_artist_names_script_exception']:
log_text = 'Script alpha characters found in "{0}": '.format(
node["name"], )
detected_scripts = detect_script_weighted(node["name"])
if detected_scripts:
log_text += "; ".join(
list("{0} ({1:.1f}%)".format(
scr_id, detected_scripts[scr_id] * 100)
for scr_id in detected_scripts))
else:
log_text += "None"
log.debug(log_text)
if detected_scripts:
if config.setting["script_exceptions"]:
log_text = " found in selected scripts: " + "; ".join(
list("{0} ({1}%)".format(scr[0], scr[1])
for scr in config.setting["script_exceptions"]))
for script_id, script_weighting in config.setting[
"script_exceptions"]:
if script_id in detected_scripts and detected_scripts[
script_id] >= script_weighting / 100:
log.debug("Match" + log_text)
return node['name'], node['sort-name']
log.debug("No match" + log_text)
else:
log.warning(
"No scripts selected for translation exception match check."
)
def check_higher_score(locale_dict, locale, score):
return locale not in locale_dict or score > locale_dict[locale][0]
# Prepare dictionaries of available locale aliases
full_locales = {}
root_locales = {}
if "aliases" in node:
for alias in node['aliases']:
if not alias["primary"]:
continue
if "locale" not in alias:
continue
full_locale = alias['locale']
root_locale = full_locale.split("_")[0]
full_parts = []
root_parts = []
score = 0.8
full_parts.append((score, 5))
if '_' in full_locale:
score = 0.4
root_parts.append((score, 5))
if alias["type"] == "Artist name":
score = 0.8
elif alias["type"] == "Legal Name":
score = 0.5
else:
# as 2014/09/19, only Artist or Legal names should have the
# Primary flag
score = 0.0
full_parts.append((score, 5))
root_parts.append((score, 5))
comb = linear_combination_of_weights(full_parts)
if check_higher_score(full_locales, full_locale, comb):
full_locales[full_locale] = (comb, (alias['name'],
alias["sort-name"]))
comb = linear_combination_of_weights(root_parts)
if check_higher_score(root_locales, root_locale, comb):
root_locales[root_locale] = (comb, (alias['name'],
alias["sort-name"]))
# First pass to match full locale if available
for locale in config.setting["artist_locales"]:
if locale in full_locales:
return full_locales[locale][1]
# Second pass to match root locale if available
for locale in config.setting["artist_locales"]:
lang = locale.split("_")[0]
if lang in root_locales:
return root_locales[lang][1]
# No matches found in available alias locales
translsort = node['sort-name']
transl = translate_from_sortname(node['name'] or "", translsort)
else:
transl, translsort = node['name'], node['sort-name']
return (transl, translsort)
def test_kanji(self):
self.assertEqual(u"Tetsuya Komuro", util.translate_from_sortname(u"小室哲哉", u"Komuro, Tetsuya"))
self.assertNotEqual(u"Komuro, Tetsuya", util.translate_from_sortname(u"小室哲哉", u"Komuro, Tetsuya"))
self.assertNotEqual(u"小室哲哉", util.translate_from_sortname(u"小室哲哉", u"Komuro, Tetsuya"))
def test_latin(self):
self.assertEqual("thename",
util.translate_from_sortname("thename", "sort, name"))
def test_latin(self):
self.failUnlessEqual(u"Jean Michel Jarre", util.translate_from_sortname(u"Jean Michel Jarre", u"Jarre, Jean Michel"))
self.failIfEqual(u"Jarre, Jean Michel", util.translate_from_sortname(u"Jean Michel Jarre", u"Jarre, Jean Michel"))