Inital commit | Basic working service for eintaxid.com. Only single search. Returns dict

.gitignore

@@ -0,0 +1 @@
+/venv

EINService.py

@@ -0,0 +1,101 @@
+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
+
+
+SAMPLE_EIN = "59-1571026"
+
+@dataclass
+class EINResult:
+    found: bool
+    match: bool
+    businessName: str
+    address: str
+    city: str
+    state: zip
+
+
+class EINService(ABC):
+    @classmethod
+    @abstractmethod
+    def search_ein(self, ein: str) -> EINResult:
+        """
+        Takes a an ein and returns information from a search using the service
+        """
+
+    @classmethod
+    @abstractmethod
+    def search_eins(self, eins: list[EINResult]):
+        """
+        
+        """
+    
+    @classmethod
+    def _isEIN(self, ein: int) -> bool:
+        return re.search("\d{2}(-|)\d{7}", str(ein)) != None
+
+
+class EINTaxIDService(EINService):
+    _url = "https://eintaxid.com"
+    _htmlSearchInputID = "searchterm"
+
+    def search_ein(self, ein: str) -> EINResult:
+        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")
+        text = soup.text.splitlines()[4]
+        data = self._parse_return(text)
+
+        return data
+
+    def search_eins(eins: list[EINResult]):
+        pass
+
+    def _parse_return(self, content: str):
+        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
+        }        
+
+
+testService = EINTaxIDService()
+print(dt.now())
+prt(testService.search_ein(SAMPLE_EIN))