checkpoint: Manual role mapping via ID

tmp/transform.py

@@ -1,652 +0,0 @@
-"""Transform raw Unternehmensregister export (*.xml) to processed .json files for loading."""
-import dataclasses
-import glob
-import json
-import os
-import re
-import sys
-
-import xmltodict
-from tqdm import tqdm
-
-from aki_prj23_transparenzregister.models.company import (
-    Capital,
-    CapitalTypeEnum,
-    Company,
-    CompanyID,
-    CompanyRelationship,
-    CompanyRelationshipEnum,
-    CompanyToCompanyRelationship,
-    CompanyTypeEnum,
-    CurrencyEnum,
-    DistrictCourt,
-    Location,
-    PersonName,
-    PersonToCompanyRelationship,
-    RelationshipRoleEnum,
-)
-from aki_prj23_transparenzregister.utils.string_tools import (
-    remove_traling_and_leading_quotes,
-    transform_date_to_iso,
-)
-
-
-def transform_xml_to_json(source_dir: str, target_dir: str) -> None:
-    """Convert all xml files in a directory to json files.
-
-    Args:
-        source_dir (str): Directory hosting the xml files
-        target_dir (str): Target directory to move json files to
-    """
-    if not os.path.exists(target_dir):
-        os.makedirs(target_dir)
-    for source_path in [
-        os.path.normpath(i) for i in glob.glob(source_dir + "**/*.xml", recursive=True)
-    ]:
-        target_path = os.path.join(
-            target_dir, source_path.split(os.sep)[-1].replace(".xml", ".json")
-        )
-
-        with open(source_path, encoding="utf-8") as source_file:
-            # deepcode ignore HandleUnicode: Weird XML format no other solution
-            data = xmltodict.parse(source_file.read().encode())
-            with open(target_path, "w", encoding="utf-8") as json_file:
-                json_file.write(json.dumps(data))
-
-
-def parse_date_of_birth(data: dict) -> str | None:
-    """Retreives the date of birth from a stakeholder entry if possible.
-
-    Args:
-        data (dict): Stakeholder data
-
-    Returns:
-        str | None: date of birth or None if not found
-    """
-    if "tns:geburt" in (base := data["tns:beteiligter"]["tns:auswahl_beteiligter"]["tns:natuerlichePerson"]):
-        base = base["tns:geburt"]["tns:geburtsdatum"]
-        if isinstance(base, str):
-            return base
-    return None
-
-def map_role_id_to_enum(role_id: str) -> RelationshipRoleEnum:
-    match role_id:
-        case "086":
-            return RelationshipRoleEnum.GESCHAEFTSFUEHRER
-        case "285":
-            return RelationshipRoleEnum.PROKURIST
-        case _:
-            raise KeyError(f'Uknown role_id: {role_id}')
-
-
-def parse_stakeholder(data: dict) -> CompanyRelationship | None:
-    """Extract the company stakeholder/relation from a single "Beteiligung".
-
-    Args:
-        data (dict): Data export
-
-    Returns:
-        CompanyRelationship | None: Relationship if it could be processed
-    """
-    if "tns:natuerlichePerson" in data["tns:beteiligter"]["tns:auswahl_beteiligter"]:
-        # It's a Company serving as a "Kommanditist" or similar
-        # if data["Beteiligter"]["Natuerliche_Person"]["Voller_Name"]["Vorname"] is None:
-        #     return CompanyToCompanyRelationship(
-        #         **{  # type: ignore
-        #             "name": remove_traling_and_leading_quotes(
-        #                 data["Beteiligter"]["Natuerliche_Person"]["Voller_Name"][
-        #                     "Nachname"
-        #                 ]
-        #             ),
-        #             "location": Location(
-        #                 **{
-        #                     "city": data["Beteiligter"]["Natuerliche_Person"][
-        #                         "Anschrift"
-        #                     ][-1]["Ort"]
-        #                     if isinstance(
-        #                         data["Beteiligter"]["Natuerliche_Person"]["Anschrift"],
-        #                         list,
-        #                     )
-        #                     else data["Beteiligter"]["Natuerliche_Person"]["Anschrift"][
-        #                         "Ort"
-        #                     ]
-        #                 }
-        #             ),
-        #             "role": RelationshipRoleEnum(
-        #                 data["Rolle"]["Rollenbezeichnung"]["content"]
-        #             ),
-        #             "type": CompanyRelationshipEnum.COMPANY,
-        #         }
-        #     )
-        return PersonToCompanyRelationship(
-            **{  # type: ignore
-                "name": PersonName(
-                    **{
-                        "firstname": data["tns:beteiligter"]["tns:auswahl_beteiligter"]["tns:natuerlichePerson"][
-                            "tns:vollerName"
-                        ]["tns:vorname"],
-                        "lastname": data["tns:beteiligter"]["tns:auswahl_beteiligter"]["tns:natuerlichePerson"][
-                            "tns:vollerName"
-                        ]["tns:nachname"],
-                    }
-                ),
-                "date_of_birth": parse_date_of_birth(data),
-                "location": Location(
-                    **{
-                        "city": data["tns:beteiligter"]["tns:auswahl_beteiligter"]["tns:natuerlichePerson"]["tns:anschrift"][
-                            -1
-                        ]["tns:ort"]
-                        if isinstance(
-                            data["tns:beteiligter"]["tns:auswahl_beteiligter"]["tns:natuerlichePerson"]["tns:anschrift"], list
-                        )
-                        else data["tns:beteiligter"]["tns:auswahl_beteiligter"]["tns:natuerlichePerson"]["tns:anschrift"][
-                            "tns:ort"
-                        ]
-                    }
-                ),
-                # TODO get role via ID
-                "role": map_role_id_to_enum(
-                    data["tns:rolle"]["tns:rollenbezeichnung"]["code"]
-                ),
-                "type": CompanyRelationshipEnum.PERSON,
-            }
-        )
-    if "Organisation" in data["Beteiligter"]:
-        return CompanyToCompanyRelationship(
-            **{  # type: ignore
-                "role": RelationshipRoleEnum(
-                    data["Rolle"]["Rollenbezeichnung"]["content"]
-                ),
-                "name": remove_traling_and_leading_quotes(
-                    data["Beteiligter"]["Organisation"]["Bezeichnung"][
-                        "Bezeichnung_Aktuell"
-                    ]
-                ),
-                "location": Location(
-                    **{
-                        "city": data["Beteiligter"]["Organisation"]["Anschrift"]["Ort"],
-                        "street": data["Beteiligter"]["Organisation"]["Anschrift"][
-                            "Strasse"
-                        ]
-                        if "Strasse" in data["Beteiligter"]["Organisation"]["Anschrift"]
-                        else None,
-                        "house_number": data["Beteiligter"]["Organisation"][
-                            "Anschrift"
-                        ]["Hausnummer"]
-                        if "Hausnummer"
-                        in data["Beteiligter"]["Organisation"]["Anschrift"]
-                        else None,
-                        "zip_code": data["Beteiligter"]["Organisation"]["Anschrift"][
-                            "Postleitzahl"
-                        ]
-                        if "Postleitzahl"
-                        in data["Beteiligter"]["Organisation"]["Anschrift"]
-                        else None,
-                    }
-                ),
-                "type": CompanyRelationshipEnum.COMPANY,
-            }
-        )
-    return None
-
-
-def normalize_street(street: str) -> str:
-    """Normalize street names by extending them to `Straße` or `straße`.
-
-    Args:
-        street (str): Name of street
-
-    Returns:
-        str: Normalized street name
-    """
-    if street is None:
-        return None
-    regex = r"(Str\.|Strasse)"
-    street = re.sub(regex, "Straße", street)
-    regex = r"(str\.|strasse)"
-    street = re.sub(regex, "straße", street)
-    return street.strip()
-
-
-def loc_from_beteiligung(data: dict) -> Location:
-    """Extract the company location from the first relationship in the export.
-
-    Args:
-        data (dict): Data export
-
-    Returns:
-        Location: location
-    """
-    base_path = [
-        "tns:grunddaten",
-        "tns:verfahrensdaten",
-        "tns:beteiligung",
-        0,
-        "tns:beteiligter",
-        "tns:auswahl_beteiligter",
-        "tns:organisation",
-        "tns:anschrift"
-    ]
-    base = traversal(data, base_path)
-
-    house_number = None
-    street = None
-    if "tns:strasse" in base:
-        regex = r".(\d+)$"
-        hits = re.findall(regex, base["tns:strasse"])
-        if len(hits) == 1:
-            house_number = hits[0]
-            street = base["tns:strasse"][: (-1 * len(house_number))]
-            if "tns:hausnummer" in base:
-                house_number = house_number + base["tns:hausnummer"]
-        else:
-            if "tns:hausnummer" in base:
-                house_number = base["tns:hausnummer"]
-            street = base["tns:strasse"]
-    return Location(
-        **{
-            "city": base["tns:ort"],
-            "zip_code": base["tns:postleitzahl"],
-            "street": normalize_street(street),  # type: ignore
-            "house_number": house_number,
-        }
-    )
-
-
-def name_from_beteiligung(data: dict) -> str:
-    """Extract the Company name from an Unternehmensregister export by using the first relationship found.
-
-    Args:
-        data (dict): Data export
-
-    Returns:
-        str: Company name
-    """
-    path = [
-        "tns:grunddaten",
-        "tns:verfahrensdaten",
-        "tns:beteiligung",
-        0,
-        "tns:beteiligter",
-        "tns:auswahl_beteiligter",
-        "tns:organisation",
-        "tns:bezeichnung",
-        "tns:bezeichnung.aktuell"
-    ]
-    name = traversal(data, path)
-    return remove_traling_and_leading_quotes(name)
-
-
-def map_rechtsform(company_name: str, data: dict) -> CompanyTypeEnum | None:
-    """Extracts the company type from a given Unternehmensregister export.
-
-    Args:
-        company_name (str): Name of the company as a fallback solution
-        data (dict): Data export
-
-    Returns:
-        CompanyTypeEnum | None: Company type if found
-    """
-    try:
-        path = [
-            "tns:fachdatenRegister",
-            "tns:basisdatenRegister",
-            "tns:rechtstraeger",
-            "tns:angabenZurRechtsform",
-            "tns:rechtsform",
-            "code"
-        ]
-        return CompanyTypeEnum(
-            traversal(data, path)
-        )
-    except Exception:
-        if (
-            company_name.endswith("GmbH")
-            or company_name.endswith("UG")
-            or company_name.endswith("UG (haftungsbeschränkt)")
-        ):
-            return CompanyTypeEnum("Gesellschaft mit beschränkter Haftung")
-        if company_name.endswith("SE"):
-            return CompanyTypeEnum("Europäische Aktiengesellschaft (SE)")
-        if company_name.endswith("KG"):
-            return CompanyTypeEnum("Kommanditgesellschaft")
-        return None
-
-
-def map_capital(data: dict, company_type: CompanyTypeEnum) -> Capital | None:
-    """Extracts the company capital from the given Unternehmensregister export.
-
-    Args:
-        data (dict): Data export
-        company_type (CompanyTypeEnum): Type of company (e.g., 'Gesellschaft mit beschränkter Haftung')
-
-    Returns:
-        Capital | None: Company Capital if found
-    """
-    # Early return
-    if "tns:auswahl_zusatzangaben" not in data["tns:fachdatenRegister"]:
-        return None
-    capital: dict = {"Zahl": 0.0, "Waehrung": ""}
-    if company_type == CompanyTypeEnum.KG:
-        capital_type = "Hafteinlage"
-        base = data["tns:fachdatenRegister"]["tns:auswahl_zusatzangaben"][
-            "tns:personengesellschaft"
-        ]["tns:zusatzKG"]["tns:datenKommanditist"]
-        if isinstance(base, list):
-            for entry in base:
-                # TODO link to persons using Ref_Rollennummer then extract ["Hafteinlage"] as below
-                capital["Zahl"] = capital["Zahl"] + float(entry["Hafteinlage"]["Zahl"])
-                capital["Waehrung"] = entry["Hafteinlage"]["Waehrung"]
-        elif isinstance(base, dict):
-            capital = base["Hafteinlage"]
-    elif company_type in [
-        CompanyTypeEnum.GMBH,
-        CompanyTypeEnum.SE,
-        CompanyTypeEnum.AG,
-        CompanyTypeEnum.KGaA,
-        CompanyTypeEnum.AUSLAENDISCHE_RECHTSFORM,
-        CompanyTypeEnum.OHG,
-    ]:
-        if (
-            "tns:kapitalgesellschaft"
-            not in data["tns:fachdatenRegister"]["tns:auswahl_zusatzangaben"]
-        ):
-            base = data["tns:fachdatenRegister"]["tns:auswahl_zusatzangaben"][
-                "tns:personengesellschaft"
-            ]
-        else:
-            base = data["tns:fachdatenRegister"]["tns:auswahl_zusatzangaben"][
-                "tns:kapitalgesellschaft"
-            ]
-        if "tns:zusatzGmbH" in base:
-            capital_type = "Stammkapital"
-            capital = base["tns:zusatzGmbH"]["tns:stammkapital"]
-        elif "tns:zusatzAktiengesellschaft" in base:
-            capital_type = "Grundkapital"
-            capital = base["tns:zusatzAktiengesellschaft"]["tns:grundkapital"]["tns:zahl"]
-    elif company_type in [
-        CompanyTypeEnum.EINZELKAUFMANN,
-        CompanyTypeEnum.EG,
-        CompanyTypeEnum.PARTNERSCHAFT,
-        CompanyTypeEnum.PARTNERGESELLSCHAFT,
-        CompanyTypeEnum.PARTNERSCHAFTSGESELLSCHAFT,
-        None,
-    ]:
-        return None
-    # Catch entries having the dict but with null values
-    if not all(capital.values()):
-        return None
-    return Capital(
-        **{  # type: ignore
-            "value": float(capital["tns:zahl"]),
-            "currency": CurrencyEnum(capital["tns:waehrung"]["code"]),
-            "type": CapitalTypeEnum(capital_type),
-        }
-    )
-
-
-def map_business_purpose(data: dict) -> str | None:
-    """Extracts the "Geschäftszweck" from a given Unternehmensregister export.
-
-    Args:
-        data (dict): Data export
-
-    Returns:
-        str | None: Business purpose if found
-    """
-    try:
-        path = [
-            "tns:fachdatenRegister",
-            "tns:basisdatenRegister",
-            "tns:gegenstand"
-        ]
-        return traversal(data, path)
-    except KeyError:
-        return None
-
-
-def extract_date_from_string(value: str) -> str | None:
-    """Extract a date in ISO format from the given string if possible.
-
-    Args:
-        value (str): Input text
-
-    Returns:
-        str | None: Date in ISO format, None if not found
-    """
-    date_regex = [  # type: ignore
-        {"regex": r"\d{1,2}\.\d{1,2}\.\d{4}", "mapper": transform_date_to_iso},
-        {"regex": r"\d{4}-\d{1,2}-\d{1,2}", "mapper": None},
-    ]
-    results = []
-    for regex in date_regex:
-        result = re.findall(regex["regex"], value)  # type: ignore
-        if len(result) == 1:
-            relevant_data = result[0]
-            if regex["mapper"] is not None:  # type: ignore
-                results.append(regex["mapper"](relevant_data))  # type: ignore
-            else:
-                results.append(relevant_data)
-    if len(results) != 1:
-        return None
-    return results[0]
-
-
-def map_founding_date(data: dict) -> str | None:
-    """Extracts the founding date from a given Unternehmensregister export.
-
-    Args:
-        data (dict): Data export
-
-    Returns:
-        str | None: Founding date if found
-    """
-    text = str(data)
-    entry_date = re.findall(
-        r".Tag der ersten Eintragung:(\\n| )?(\d{1,2}\.\d{1,2}\.\d{2,4})", text
-    )
-    if len(entry_date) == 1:
-        return transform_date_to_iso(entry_date[0][1])
-
-    entry_date = re.findall(
-        r".Gesellschaftsvertrag vom (\d{1,2}\.\d{1,2}\.\d{2,4})", text
-    )
-    if len(entry_date) == 1:
-        return transform_date_to_iso(entry_date[0])
-    if (
-        "tns:satzungsdatum"
-        in data["tns:fachdatenRegister"]["tns:basisdatenRegister"]
-    ):
-        path = [
-            "tns:fachdatenRegister",
-            "tns:basisdatenRegister",
-            "tns:satzungsdatum",
-            "tns:aktuellesSatzungsdatum"
-        ]
-        return traversal(data, path)
-    # No reliable answer
-    return None
-
-def traversal(data: dict, path: list[str | int]) -> any:
-    current = data
-    for key in path:
-        try:
-            current = current[key]
-        except:
-            raise KeyError(f"Key {key} not found")
-    return current
-
-
-def map_hr_number(data: dict) -> str:
-    hr_prefix = data["tns:grunddaten"]["tns:verfahrensdaten"]["tns:instanzdaten"][
-        "tns:aktenzeichen"
-    ]["tns:auswahl_aktenzeichen"]["tns:aktenzeichen.strukturiert"]["tns:register"][
-        "code"
-    ]
-    hr_number = data["tns:grunddaten"]["tns:verfahrensdaten"]["tns:instanzdaten"][
-        "tns:aktenzeichen"
-    ]["tns:auswahl_aktenzeichen"]["tns:aktenzeichen.strukturiert"]["tns:laufendeNummer"]
-    hr_full = f"{hr_prefix} {hr_number}"
-    return hr_full
-
-def map_district_court(data: dict) -> DistrictCourt:
-    base_path = [
-        "tns:grunddaten",
-        "tns:verfahrensdaten",
-        "tns:beteiligung",
-        1,
-        "tns:beteiligter",
-        "tns:auswahl_beteiligter",
-        "tns:organisation"
-    ]
-    path = [*base_path,
-        "tns:bezeichnung",
-        "tns:bezeichnung.aktuell"
-    ]
-    name = traversal(data, path)
-    path = [*base_path,
-        "tns:sitz",
-        "tns:ort"
-    ]
-    city = traversal(data, path)
-    return DistrictCourt(name=name, city=city)
-
-
-def map_company_id(data: dict) -> CompanyID:
-    """Retrieve Company ID from export.
-
-    Args:
-        data (dict): Data export
-
-    Returns:
-        CompanyID: ID of the company
-    """
-    return CompanyID(
-        **{
-            "hr_number": map_hr_number(data),
-            "district_court": map_district_court(data)
-        }
-    )
-
-
-def map_last_update(data: dict) -> str:
-    """Extract last update date from export.
-
-    Args:
-        data (dict): Unternehmensregister export
-
-    Returns:
-        str: Last update date
-    """
-    path = [
-        "tns:fachdatenRegister",
-        "tns:auszug",
-        "tns:letzteEintragung"
-    ]
-    return traversal(data, path)
-
-
-def map_co_relation(data: dict) -> dict:
-    """Search for and map the c/o relation from location.street if possible.
-
-    Args:
-        data (dict): Company dict
-
-    Returns:
-        dict: Modified Company dict
-    """
-    street = data["location"].street
-    if street is None:
-        return data
-    parts = street.split(",")
-    co_company = None
-    co_company_index = None
-    for index, part in enumerate(parts):
-        trimmed_part = part.strip()
-        result = re.findall(r"^c\/o(.*)$", trimmed_part)
-        if len(result) == 1:
-            co_company = result[0].strip()
-            co_company_index = index
-    if co_company_index is not None:
-        del parts[co_company_index]
-        street = "".join(parts).strip()
-        data["location"].street = street
-
-        if co_company is not None and co_company != "":
-            relation = CompanyToCompanyRelationship(
-                RelationshipRoleEnum.CARE_OF,  # type: ignore
-                Location(
-                    data["location"].city,
-                    street,
-                    data["location"].house_number,
-                    data["location"].zip_code,
-                ),
-                CompanyRelationshipEnum.COMPANY,  # type: ignore
-                co_company,
-            )
-            data["relationships"].append(relation)
-    return data
-
-
-def map_unternehmensregister_json(data: dict) -> Company:
-    """Processes the Unternehmensregister structured export to a Company by using several helper methods.
-
-    Args:
-        data (dict): Data export
-
-    Returns:
-        Company: Transformed data
-    """
-    root_key = list(data.keys())[0]
-    data = data[root_key]
-    result: dict = {"relationships": []}
-
-    result["id"] = map_company_id(data)
-    result["name"] = name_from_beteiligung(data)
-
-    result["location"] = loc_from_beteiligung(data)
-    result["last_update"] = map_last_update(data)
-
-    result["company_type"] = map_rechtsform(result["name"], data)
-    result["capital"] = map_capital(data, result["company_type"])
-    result["business_purpose"] = map_business_purpose(data)
-    result["founding_date"] = map_founding_date(data)
-
-    # TODO adapt...
-    for i in range(
-        2, len(data["tns:grunddaten"]["tns:verfahrensdaten"]["tns:beteiligung"])
-    ):
-        people = parse_stakeholder(
-            data["tns:grunddaten"]["tns:verfahrensdaten"]["tns:beteiligung"][i]
-        )
-        result["relationships"].append(people)
-    result = map_co_relation(result)
-    return Company(**result)
-
-
-if __name__ == "__main__":
-    from loguru import logger
-
-    base_path = "./Jupyter/API-tests/Unternehmensregister/data/Unternehmensregister"
-    for file in tqdm(glob.glob1(f"{base_path}/export", "*.json")):
-        path = os.path.join(f"{base_path}/export", file)
-        with open(path, encoding="utf-8") as file_object:
-            try:
-                company: Company = map_unternehmensregister_json(
-                    json.loads(file_object.read())
-                )
-
-                name = "".join(e for e in company.name if e.isalnum())[:50]
-
-                with open(
-                    f"{base_path}/transformed/{name}.json",
-                    "w+",
-                    encoding="utf-8",
-                ) as export_file:
-                    json.dump(
-                        dataclasses.asdict(company), export_file, ensure_ascii=False
-                    )
-            except Exception as e:
-                logger.error(e)
-                logger.error(f"Error in processing {path}")
-                sys.exit(1)

tmp/transformation.ipynb

@@ -2,14 +2,14 @@
  "cells": [
   {
    "cell_type": "code",
-   "execution_count": 8,
+   "execution_count": 1,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Company(id=CompanyID(district_court=DistrictCourt(name='Amtsgericht Hamm', city='Hamm'), hr_number='HRB 5363'), location=Location(city='Bönen', street='Siemensstraße', house_number='25-27', zip_code='59199'), name='GEA Farm Technologies GmbH', last_update='2023-10-27', relationships=[PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Oelde', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Reinhard', lastname='Gebing'), date_of_birth='1964-04-26'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Wetter', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Markus', lastname='Kreft'), date_of_birth='1966-04-03'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Holzminden', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Kai', lastname='Luntz'), date_of_birth='1970-12-04'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Rheda-Wiedenbrück', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Thomas', lastname='Mader'), date_of_birth='1972-05-24'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.GESCHAEFTSFUEHRER: 'Geschäftsführer(in)'>, location=Location(city='Düsseldorf', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Peter', lastname='Lauwers'), date_of_birth='1970-03-26'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.GESCHAEFTSFUEHRER: 'Geschäftsführer(in)'>, location=Location(city='Erkrath', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Erkul', lastname='Basaran'), date_of_birth='1977-05-06'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Bochum', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Henrik', lastname='Böttner'), date_of_birth='1982-11-07'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Oelde', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Ulrich', lastname='Raßenhövel'), date_of_birth='1969-04-16'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Herdecke', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Andreas', lastname='Naroska'), date_of_birth='1967-03-23'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Witten', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Mark', lastname='Kramps'), date_of_birth='1967-09-04'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Dortmund', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Ralf', lastname='Barkmeyer'), date_of_birth='1974-02-28'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Tönnisvorst', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Holger', lastname='Siegwarth'), date_of_birth='1967-05-13'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Herne', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Oliver', lastname='Liß'), date_of_birth='1981-04-13'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.GESCHAEFTSFUEHRER: 'Geschäftsführer(in)'>, location=Location(city='Göppingen', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Liang', lastname='Cheng'), date_of_birth='1980-12-29'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Beckum', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Astrid', lastname='Dörner-Rodeheger'), date_of_birth='1968-12-24'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Dortmund', streeLine truncated
+      "Company(id=CompanyID(district_court=DistrictCourt(name='Amtsgericht Hamm', city='Hamm'), hr_number='HRB 5363'), location=Location(city='Bönen', street='Siemensstraße', house_number='25-27', zip_code='59199'), name='GEA Farm Technologies GmbH', last_update='2023-10-27', relationships=[PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Oelde', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Reinhard', lastname='Gebing'), date_of_birth='1964-04-26'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Wetter', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Markus', lastname='Kreft'), date_of_birth='1966-04-03'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Holzminden', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Kai', lastname='Luntz'), date_of_birth='1970-12-04'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Rheda-Wiedenbrück', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Thomas', lastname='Mader'), date_of_birth='1972-05-24'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.GESCHAEFTSFUEHRER: 'Geschäftsführer(in)'>, location=Location(city='Düsseldorf', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Peter', lastname='Lauwers'), date_of_birth='1970-03-26'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.GESCHAEFTSFUEHRER: 'Geschäftsführer(in)'>, location=Location(city='Erkrath', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Erkul', lastname='Basaran'), date_of_birth='1977-05-06'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Bochum', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Henrik', lastname='Böttner'), date_of_birth='1982-11-07'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Oelde', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Ulrich', lastname='Raßenhövel'), date_of_birth='1969-04-16'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Herdecke', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Andreas', lastname='Naroska'), date_of_birth='1967-03-23'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Witten', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Mark', lastname='Kramps'), date_of_birth='1967-09-04'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Dortmund', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Ralf', lastname='Barkmeyer'), date_of_birth='1974-02-28'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Tönnisvorst', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Holger', lastname='Siegwarth'), date_of_birth='1967-05-13'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Herne', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Oliver', lastname='Liß'), date_of_birth='1981-04-13'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.GESCHAEFTSFUEHRER: 'Geschäftsführer(in)'>, location=Location(city='Göppingen', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Liang', lastname='Cheng'), date_of_birth='1980-12-29'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Beckum', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Astrid', lastname='Dörner-Rodeheger'), date_of_birth='1968-12-24'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Dortmund', streeLine truncated
      ]
     }
    ],
@@ -26,40 +26,63 @@
     "        json.dump(dataclasses.asdict(company_data), file, ensure_ascii=False)"
    ]
   },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "# Test"
-   ]
-  },
   {
    "cell_type": "code",
-   "execution_count": 9,
+   "execution_count": 2,
    "metadata": {},
    "outputs": [
     {
-     "ename": "AssertionError",
-     "evalue": "",
-     "output_type": "error",
-     "traceback": [
-      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
-      "\u001b[1;31mAssertionError\u001b[0m                            Traceback (most recent call last)",
-      "\u001b[1;32mc:\\Users\\trist\\Documents\\Code\\M.Sc\\aki_prj23_transparenzregister\\tmp\\transformation.ipynb Cell 3\u001b[0m line \u001b[0;36m9\n\u001b[0;32m      <a href='vscode-notebook-cell:/c%3A/Users/trist/Documents/Code/M.Sc/aki_prj23_transparenzregister/tmp/transformation.ipynb#W2sZmlsZQ%3D%3D?line=6'>7</a>\u001b[0m \u001b[39mwith\u001b[39;00m \u001b[39mopen\u001b[39m(\u001b[39m'\u001b[39m\u001b[39m../tmp/tests/GEAFarmTechnologiesGmbH.json\u001b[39m\u001b[39m'\u001b[39m, \u001b[39m\"\u001b[39m\u001b[39mr\u001b[39m\u001b[39m\"\u001b[39m) \u001b[39mas\u001b[39;00m file:\n\u001b[0;32m      <a href='vscode-notebook-cell:/c%3A/Users/trist/Documents/Code/M.Sc/aki_prj23_transparenzregister/tmp/transformation.ipynb#W2sZmlsZQ%3D%3D?line=7'>8</a>\u001b[0m     expected_result \u001b[39m=\u001b[39m json\u001b[39m.\u001b[39mload(file)\n\u001b[1;32m----> <a href='vscode-notebook-cell:/c%3A/Users/trist/Documents/Code/M.Sc/aki_prj23_transparenzregister/tmp/transformation.ipynb#W2sZmlsZQ%3D%3D?line=8'>9</a>\u001b[0m     \u001b[39massert\u001b[39;00m result \u001b[39m==\u001b[39m expected_result\n",
-      "\u001b[1;31mAssertionError\u001b[0m: "
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Company(id=CompanyID(district_court=DistrictCourt(name='Amtsgericht Charlottenburg', city='Berlin'), hr_number='HRB 153385 B'), location=Location(city='Berlin', street='Valeska-Gert-Straße', house_number='5', zip_code='10243'), name='Zalando Lounge Service GmbH', last_update='2022-10-05', relationships=[PersonToCompanyRelationship(role=<RelationshipRoleEnum.GESCHAEFTSFUEHRER: 'Geschäftsführer(in)'>, location=Location(city='Berlin', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Martin', lastname='Rost'), date_of_birth='1982-09-24'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.GESCHAEFTSFUEHRER: 'Geschäftsführer(in)'>, location=Location(city='Berlin', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Karen', lastname='Kennes'), date_of_birth='1979-06-22'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Berlin', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Andre', lastname='Hörschel'), date_of_birth='1973-06-15')], company_type=<CompanyTypeEnum.GMBH: 'Gesellschaft mit beschränkter Haftung'>, capital=Capital(value=25000.0, currency=<CurrencyEnum.EURO: 'EUR'>, type=<CapitalTypeEnum.STAMMKAPITAL: 'Stammkapital'>), business_purpose='Die Erbringung von Dienstleistungen für e-Commerce Unternehmen im Bereich Kundenservice und Logistik.', founding_date='2014-02-05')\n"
      ]
     }
    ],
    "source": [
     "import json\n",
+    "import dataclasses\n",
+    "from transform import map_unternehmensregister_json\n",
     "\n",
-    "result = None\n",
-    "expected_result = None\n",
-    "with open('../tmp/transformed/GEAFarmTechnologiesGmbH.json', 'r') as file_a:\n",
-    "    result = json.load(file_a)\n",
-    "with open('../tmp/tests/GEAFarmTechnologiesGmbH.json', \"r\") as file:\n",
-    "    expected_result = json.load(file)\n",
-    "    assert result == expected_result"
+    "with open('../tmp/json/ZalandoLoungeServiceGmbH.json', \"r\") as file:\n",
+    "    content = json.load(file)\n",
+    "    company_data = map_unternehmensregister_json(content)\n",
+    "    print(company_data)\n",
+    "    with open('../tmp/transformed/ZalandoLoungeServiceGmbH.json', \"w+\", encoding=\"utf-8\") as file:\n",
+    "        json.dump(dataclasses.asdict(company_data), file, ensure_ascii=False)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Company(id=CompanyID(district_court=DistrictCourt(name='Amtsgericht Charlottenburg', city='Berlin'), hr_number='HRB 158855 B'), location=Location(city='Berlin', street='Valeska-Gert-Straße', house_number='5', zip_code='10243'), name='Zalando SE', last_update='2023-07-04', relationships=[PersonToCompanyRelationship(role=<RelationshipRoleEnum.VORSTAND: 'Vorstand'>, location=Location(city='Berlin', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Robert', lastname='Gentz'), date_of_birth='1983-09-24'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.VORSTAND: 'Vorstand'>, location=Location(city='Berlin', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='David', lastname='Schneider'), date_of_birth='1982-07-29'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.VORSTAND: 'Vorstand'>, location=Location(city='Berlin', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='David', lastname='Schröder'), date_of_birth='1982-11-27'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.VORSTAND: 'Vorstand'>, location=Location(city='Berlin', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Astrid', lastname='Arndt'), date_of_birth='1971-11-05'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.VORSTAND: 'Vorstand'>, location=Location(city='Passau', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Sandra', lastname='Dembeck'), date_of_birth='1974-03-06'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Berlin', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Martin', lastname='Rost'), date_of_birth='1982-09-24'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Berlin', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Jan-Hendrik', lastname='Bartels'), date_of_birth='1980-11-05'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Berlin', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Ulrich', lastname='Kalk'), date_of_birth='1978-05-27'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Berlin', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Anne', lastname='Pascual'), date_of_birth='1976-01-03'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Knokke/Belgien', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Bruno', lastname='Vanhoorickx'), date_of_birth='1981-08-25'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Berlin', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Andreas', lastname='Antrup'), date_of_birth='1983-06-27'), PersonToCompanyRelationship(role=<RelationshipRoleEnum.PROKURIST: 'Prokurist(in)'>, location=Location(city='Berlin', street=None, house_number=None, zip_code=None), type=<CompanyRelationshipEnum.PERSON: 'Person'>, name=PersonName(firstname='Lena', lastname='Wallenhorst'), date_of_birth='1978-07-20')], company_type=<CompanyTypeEnum.SE: 'Europäische Aktiengesellschaft (SE)'>, capital=Capital(value=263531672.0, currency=<CurrencyEnum.EURO: 'EUR'>, type=<CapitalTypeEnum.GRUNDKAPITAL: 'Grundkapital'>), business_purpose='Die Entwicklung, Vermarktung und Erbringung von Internetdienstleistungen (E-Commerce-Handel mit Waren verschiedener Art, insbesondere Bekleidung und Schuhe), die Entwicklung, Herstellung, Vermarktung und der Handel mit solchen Waren, insbesondere Bekleidung und Schuhe, die Erbringung von Logistikdienstleistungen, digitalen Dienstleistungen und alle mit dem vorgenannten Unternehmensgegenstand zusammenhängenden Dienstleistungen.', founding_date='2023-05-24')\n"
+     ]
+    }
+   ],
+   "source": [
+    "import json\n",
+    "import dataclasses\n",
+    "from transform import map_unternehmensregister_json\n",
+    "\n",
+    "with open('../tmp/json/ZalandoSE.json', \"r\") as file:\n",
+    "    content = json.load(file)\n",
+    "    company_data = map_unternehmensregister_json(content)\n",
+    "    print(company_data)\n",
+    "    with open('../tmp/transformed/ZalandoSE.json', \"w+\", encoding=\"utf-8\") as file:\n",
+    "        json.dump(dataclasses.asdict(company_data), file, ensure_ascii=False)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Test"
    ]
   }
  ],