EINService/EINService.py

import requests as rq
from bs4 import BeautifulSoup as bsp
import re
from abc import ABC, abstractmethod
from dataclasses import dataclass
from datetime import datetime as dt
from pprint import pprint as prt
import pandas as pd


SAMPLE_EIN = "59-1571026"

@dataclass
class EINData:

    def __init__(self, ein: str, buinessName: str, address1: str, city:str, state:str, zip: str, phone: str) -> None:
        if re.search("\d{2}(-|)\d{7}", str(ein)) == None:
            raise Exception(f"Invalid EIN: {ein}")
        self.ein = ein.strip().replace('-','')
        self.buinessName = buinessName.lower().strip()
        self.address1 = address1.lower().strip()
        self.city = city.lower().strip()
        self.state = state.lower().strip()
        self.phone = phone.strip().replace('-','').replace('(',"").replace(')','').replace('+','')
        self.zip = int(zip.replace('-','').strip())

    
    def __str__(self) -> str:
        return f"""EIN: {self.ein}\t | Name: {self.buinessName}\t\t| Address: {self.address1}\t\t| City: {self.city}\t| State: {self.state}\t| Phone: {self.phone}"""      

    def get_ein(self) -> str:
        return f"{self.ein[0:2]}-{self.ein[2:]}"

    def compare(self, otherEIN: 'EINData') -> dict:
        try:
            compareDict = {
                "buinessName" : True if self.buinessName == otherEIN.buinessName else False,
                "address" : True if self.address1 == otherEIN.address1 else False,
                "city": True if self.city == otherEIN.city else False,
                "state":  True if self.state == otherEIN.state else False,
                "zip" : True if self.zip == otherEIN.zip else False
            }
        except Exception as e:
            print(f"""Exception:\n{e}\nSelf:{self}\nOther: {otherEIN}\n""")
            return None
        score = 0
        for v in compareDict.values():
            score += 1 if v else 0
        compareDict["score"] = score
        return compareDict


class EINService(ABC):
    @classmethod
    @abstractmethod
    def search_ein(self, ein: str) -> EINData:
        """
        Takes a an ein and returns information from a search using the service
        """

    @classmethod
    @abstractmethod
    def search_eins(self, eins: list[EINData]):
        """
        
        """
    
    @classmethod
    def _isEIN(self, ein: str) -> bool:
        return re.search("\d{2}(-|)\d{7}", ein) != None


class EINTaxIDService(EINService):
    _url = "https://eintaxid.com"

    def search_ein(self, ein: str) -> EINData:
        try:
            if not self._isEIN(ein):
                print(f"{ein} is not a valid EIN!\nValid formats are: XX-XXXXXXX and XXXXXXXXX")
                raise Exception("Invalid EIN")
        except Exception as e:
            print(e)
            return None

        req = rq.request("POST",self._url + "/search-ajax.php", data={"query": ein}, \
            headers={'X-Requested-With': 'XMLHttpRequest'})
        soup = bsp(req.content, "html.parser")
        try:
            text = soup.text.splitlines()[4]
        except:
            print(f"Failed: {ein} | {soup}")
            return None
        data = self._parse_return(text)

        return EINData(
            ein,
            data["company"],
            data["address1"],
            data["city"],
            data["state"],
            data["zip"],
            data["phone"]
        )

    def search_eins(self, eins: str) -> list[EINData]:
        return [self.search_ein(ein) for ein in eins]


    def _parse_return(self, content: str) -> dict:
        m = re.search("EIN Number:", content)
        company = content[0:m.start()].strip()

        m = re.search("Doing Business As:.*Address:", content)
        dba = content[m.start()+18:m.end()-8].strip() if m != None else None
        
        m = re.search("Address:.*Phone:", content)
        address = content[m.start()+8:m.end()-6].strip() if m != None else None
        if address != None:
            addressPieces = address.split(',')
            address1 = addressPieces[0].strip()
            city = addressPieces[1].strip()
            state = addressPieces[2].strip().split(" ")[0].strip() 
            zip = addressPieces[2].strip().split(" ")[1].strip()
        else:
            address1 = None
            city = None
            state, zip = None

        phone = content[m.end():].strip() if m != None else None
        return {
            "company": company,
            "dba": dba,
            "address": address,
            "address1": address1,
            "city": city,
            "state": state,
            "zip": zip,
            "phone": phone
        }