Translation#

For convenience, we’ll use a dummy fetcher used for testing to generates translation data instead of fetching real translations. Generated data is on the form <placeholder-name>-of-<id>, eg 'placeholder0-of-id0'. Since we use the '{x}, {y}' translation format, all translated values will take form form 'x-of-<id>, y-of-<id>'.

[1]:

from rics.translation import Translator

NAMES_TO_TRANSLATE = "name"
translate = Translator(fmt="{x}, {y}").translate

/home/dev/git/rics/src/rics/translation/_translator.py:156: UserWarning: No fetcher given. Translation data will be automatically generated.
  warnings.warn("No fetcher given. Translation data will be automatically generated.", UserWarning, stacklevel=-1)

Translating `dict` keys#

When translating using inplace=False (the default), the Translator will always try to return an object of the same type. The Translator doesn’t know what to do with the dict_keys class, so we wrap the values we want to translate using a known type.

[2]:

a_dict = {f"k{i}": i for i in range(4)}
print(a_dict)

translated_keys = translate(list(a_dict), names=NAMES_TO_TRANSLATE)
{tk: a_dict[k] for k, tk in zip(a_dict, translated_keys)}  # tk is translated

{'k0': 0, 'k1': 1, 'k2': 2, 'k3': 3}

[2]:

{'x-of-k0, y-of-k0': 0,
 'x-of-k1, y-of-k1': 1,
 'x-of-k2, y-of-k2': 2,
 'x-of-k3, y-of-k3': 3}

Translating Pandas `index` and `columns`#

Recipies for translating involving pandas types.

The attribute argument may be used to translate a member variable instead.

[3]:

import pandas as pd

[4]:

df = pd.DataFrame(
    data=[["00", "01", "02", "03"], ["10", "11", "12", "13"], ["20", "21", "22", "23"]],
    columns=[f"col{i}" for i in range(4)],
    index=[f"idx{i}" for i in range(3)],
)
df.index.name = NAMES_TO_TRANSLATE
s = df.col0
df

[4]:

	col0	col1	col2	col3
name
idx0	00	01	02	03
idx1	10	11	12	13
idx2	20	21	22	23

[5]:

translate(df, inplace=True)
translate(df, attribute="columns")  # Returns df after overwriting df.columns
translate(df, attribute="index")  # Returns df after overwriting df.index

[5]:

	x-of-col0, y-of-col0	x-of-col1, y-of-col1	x-of-col2, y-of-col2	x-of-col3, y-of-col3
name
x-of-idx0, y-of-idx0	x-of-00, y-of-00	x-of-01, y-of-01	x-of-02, y-of-02	x-of-03, y-of-03
x-of-idx1, y-of-idx1	x-of-10, y-of-10	x-of-11, y-of-11	x-of-12, y-of-12	x-of-13, y-of-13
x-of-idx2, y-of-idx2	x-of-20, y-of-20	x-of-21, y-of-21	x-of-22, y-of-22	x-of-23, y-of-23

[6]:

translate(s, inplace=True)
translate(s, attribute="index")

[6]:

name
x-of-idx0, y-of-idx0    x-of-00, y-of-00
x-of-idx1, y-of-idx1    x-of-10, y-of-10
x-of-idx2, y-of-idx2    x-of-20, y-of-20
Name: col0, dtype: object

Function signature#

[7]:

?translate

Signature:
translate(
    translatable: ~Translatable,
    names: Union[~NameType, Iterable[~NameType]] = None,
    ignore_names: Union[~NameType, Iterable[~NameType], Callable[[~NameType], bool]] = None,
    inplace: bool = False,
    override_function: Callable[[~NameType, Set[~SourceType], List[~IdType]], Union[~SourceType, Dict[~SourceType, List[~IdType]], NoneType]] = None,
    maximal_untranslated_fraction: float = 1.0,
    reverse: bool = False,
    attribute: str = None,
) -> Optional[~Translatable]
Docstring:
Translate IDs to human-readable strings.

Args:
    translatable: A data structure to translate.
    names: Explicit names to translate. Derive from `translatable` if ``None``.
    ignore_names: Names **not** to translate, or a predicate ``(str) -> bool``.
    inplace: If ``True``, translate in-place and return ``None``.
    override_function: A callable ``(value, candidates, ids)`` returning one of

        * ``None`` (use regular mapping logic)
        * a ``source`` to use, or
        * a split mapping ``{source: [ids_for_source..]}``. This forces IDs to be fetched from
          different sources in spite of being labelled with the same name.

    maximal_untranslated_fraction: The maximum fraction of IDs for which translation may fail before an error is
        raised. 1=disabled. Ignored in `reverse` mode.
    reverse: If ``True``, perform translations back to IDs. Offline mode only.
    attribute: If given, translate ``translatable.attribute`` instead. If ``inplace=False``, the translated
        attribute will be assigned to `translatable` using
        ``setattr(translatable, attribute, <translated-attribute>)``.

Returns:
    A copy of translated copy of `translatable` if ``inplace=False``, otherwise ``None``.

Raises:
    UntranslatableTypeError: If ``type(translatable)`` cannot be translated.
    AttributeError: If `names` are not given and cannot be derived from `translatable`.
    MappingError: If required (explicitly given) names fail to map to a source.
    ValueError: If `maximal_untranslated_fraction` is not a valid fraction.
    TooManyFailedTranslationsError: If translation fails for more than `maximal_untranslated_fraction` of IDs.
    ConnectionStatusError: If ``reverse=True`` while the ``Translator`` is online.
    UnknownSourceError: If `override_function` returns a source which is not known.

See Also:
    The :meth:`.Mapper.apply` function, which performs both placeholder and name-to-source mapping.
File:      ~/git/rics/src/rics/translation/_translator.py
Type:      method

[ ]: