def load_alias(self, filename: str) -> None:
"""
Load aliases from file.
Args:
filename (str): Filename to read aliases from
"""
with open(filename, "r") as f:
for line in f.readlines():
# Remove comments (starting with '#'), unless they are escaped
line = re.sub(r"(?<!\\)#.*", "", line)
if not line.strip():
continue # Skip empty lines
# Get key and aliases
primary, aliases = parse_aliases(line)
for name in [primary] + aliases:
# Stor a separate entry for each type of key_mod
# For each entry, also store the value itself and the
# primary name
self.vocab["none"][name] = addict.Dict(value=name,
primary=primary)
self.vocab["lower"][name.lower()] = addict.Dict(
value=name, primary=primary)
self.vocab["stem"][self.stemmer(name)] = addict.Dict(
value=name, primary=primary)
self.vocab["norm"][aliasregex.normalize(
name)] = addict.Dict(value=name, primary=primary)
def normalize_ta(name: Text, uppercase_abbr: List[Text]) -> Text:
"""Normalize TA names. Words are capitalized unless configured
to be excempt"""
res: Text = normalize(
name,
capitalize=True,
uppercase_abbr=uppercase_abbr,
)
return res
def test_aliasregex_normalization() -> None:
"Normalize test"
assert normalize("oceanlotusGroup") == "oceanlotus group"
assert normalize("APT 27") == "apt 27"
assert normalize("APT.27") == "apt 27"
assert normalize("APT-27") == "apt 27"
assert normalize("APT- 27") == "apt 27"
assert normalize("winntiGroup") == "winnti group"
def get(self,
key: str,
key_mod: str = "DEFAULT",
primary: Optional[bool] = None,
default: Optional[str] = None) -> Optional[str]:
"""
Search for entry in vocabulary
Args:
key (str): Key to search for
key_mod (str): Key modifier ("lower", "stem" or "none"). If not set,
use default value form self.config.key_mod.
primary (bool|None): If True, return primary name, if not return the
value itself from the vocabulary
default (str|None): Default value if key is not found. Default == None
"""
key_mod = self.get_key_mod(key_mod)
# If primary is not set, get from config whether we should retrieve the primary name
if primary is None:
primary = self.config.primary
# If default is None, get default value from config
if default is None:
default = self.config.default
if key_mod == "stem":
key = self.stemmer(key)
elif key_mod == "lower":
key = key.lower()
elif key_mod == "norm":
key = aliasregex.normalize(key)
value = self.vocab[key_mod].get(key)
if not value:
return default
if primary: # Return primary value
return value.primary # type: ignore
# Return value ifself
return value.value # type: ignore
def normalize_ta(name: Text) -> Text:
return normalize(
name,
capitalize=True,
uppercase_abbr=["APT", "BRONZE", "IRON", "GOLD"],
)