Matching publication numbers / ids from 3rd parties to EPO Patstat (USPTO / EPO)

match.py

def clean_uspto_publn_nr_to_patstat_format(dirty:str):
    """
    removes 0s from uspto ids
    "2010015376"  --> "201015376"
    "20100015376" --> "2010015376"
    """
    if len(dirty) < 9:
        return dirty
    if dirty[0] == 3:
        raise ValueError("Can not process id")
    
    if dirty[4] == "0":
        return dirty[0:4] + dirty[5:]
    else:
        return dirty
      

def clean_epo_publn_nr_to_patstat_format(dirty:str):
    """
    adds 0s from to epo ids
    "1235"  --> "0001235"
    """
    return dirty.zfill(7)