Verified Commit d63945e3 authored by Jakob Moser's avatar Jakob Moser
Browse files

Try implementing better heuristic

parent 82acdbb9

poolpay/fints/to_compatibility_regex.py

0 → 100644
+33 −0
Original line number Diff line number Diff line 
from functools import reduce



REPLACEMENTS = {

    "ä": "(ä|ae|a)",

    "ö": "(ö|oe|o)",

    "ü": "(ü|ue|u)",

    "ß": "(ß|ss)",

    "Ä": "(Ä|Ae|A)",

    "Ö": "(Ö|Oe|O)",

    "Ü": "(Ü|Ue|U)",

    "ẞ": "(ẞ|SS)",

}





def to_compatibility_regex(evil_unicode_string: str) -> str:

    """

    > Schon unsere Großväter wussten, was gut ist: Ein selbstbereiteter String mit irgendwelchen Zeichen drin.

    > Doch sie wussten auch ein Lied davon zu singen, wie schwer es ist, so einen String zu verarbeiten.

    > Ja, so war es damals.

    >

    > Und heute?



    Converts a normal string, which might contain evil characters, such as ä, ö, ü, or ß, to a regular

    expression matching variants of those characters which you can use if you are a German bank and have

    never heard of Unicode before.



    :see: https://www.youtube.com/watch?v=P56pk5mDewU

    """

    return reduce(

        lambda string, replacement: string.replace(*replacement),

        REPLACEMENTS.items(),

        evil_unicode_string,

    )

poolpay/model/Person.py

+22 −3
Original line number Diff line number Diff line
@@ -4,9 +4,10 @@ from typing import TYPE_CHECKING, Literal, Self 


from babel.numbers import format_currency

from sqlalchemy import select

from sqlalchemy.orm import Mapped, mapped_column, relationship

from sqlalchemy.orm import Mapped, QueryableAttribute, mapped_column, relationship



from poolpay import db

from poolpay.fints.to_compatibility_regex import to_compatibility_regex

from poolpay.model.Base import Base



if TYPE_CHECKING:
@@ -133,9 +134,19 @@ class Person(Base): 


        If no persons can be found or there are too many plausible matches, None is returned.

        """

        # TODO Improve heuristics



        # We currently ignore middle names

        first_name, *_, last_name = full_name.split(" ")



        # We try to deal with umlauts etc.

        first_name_regex = to_compatibility_regex(first_name)

        last_name_regex = to_compatibility_regex(last_name)



        def _try_get_single_person(

            field: QueryableAttribute, regex: str

        ) -> Self | None:

            persons = (

            db.session.execute(select(cls).where(cls.name == full_name.split(" ")[0]))

                db.session.execute(select(cls).where(field.regexp_match(regex)))

                .scalars()

                .all()

            )
@@ -144,3 +155,11 @@ class Person(Base): 
                return None



            return persons[0]



        potential_last_name_match = _try_get_single_person(

            cls.cl_account_name, last_name_regex.lower()

        )

        if potential_last_name_match:

            return potential_last_name_match



        return _try_get_single_person(cls.name, first_name_regex)