I’m working on a Python script to normalize street addresses that include corner street names. I’m using regex patterns to extract the relevant information, but I’m encountering difficulties with certain cases.

For example, I have following addresses like this:

addresses = [
    "corner hampstead road and east parkway northfield sa 5085",
    "burleigh west shopping centre, corner west burleigh road & reedy creek road, burleigh west qld 4219",
    "coburn central shopping centre, corner high & melton streets, melton vic 3337",
    "corner armadale road & alex wood drive forrestdale wa 6112",
    "corner beach road & hanna road, noarlunga centre, noarlunga sa 5168",
    "corner bray street & pacific highway, coffs harbour nsw 2450",
    "corner bruce hwy & geaney lane",
    "corner bruce hwy & highway plaza, mackay qld 4740",
    "shop 2  7 eleven service station. corner broardbeach nerang & chisholm roads",
    "61 sallys corner road",
    "corner princes fwy and hughes road",
]

I want to normalize it to the following format:

corner hampstead road and east parkway northfield sa 5085 -> {'street_number': 'cnr', 'street_name1': 'hampstead', 'street_type1': 'road', 'street_name2': 'east', 'street_type2': 'parkway'}
burleigh west shopping centre, corner west burleigh road & reedy creek road, burleigh west qld 4219 -> {'street_number': 'cnr', 'street_name1': 'west burleigh', 'street_type1': 'road', 'street_name2': 'reedy creek', 'street_type2': 'road'}
coburn central shopping centre, corner high & melton streets, melton vic 3337 -> {'street_number': 'cnr', 'street_name1': 'high', 'street_type1': 'street', 'street_name2': 'melton', 'street_type2': 'street'}
corner armadale road & alex wood drive forrestdale wa 6112 -> {'street_number': 'cnr', 'street_name1': 'armadale', 'street_type1': 'road', 'street_name2': 'alex wood', 'street_type2': 'drive'}
corner beach road & hanna road, noarlunga centre, noarlunga sa 5168 -> {'street_number': 'cnr', 'street_name1': 'beach', 'street_type1': 'road', 'street_name2': 'hanna', 'street_type2': 'road'}
corner bray street & pacific highway, coffs harbour nsw 2450 -> {'street_number': 'cnr', 'street_name1': 'bray', 'street_type1': 'street', 'street_name2': 'pacific', 'street_type2': 'highway'}
corner bruce hwy & geaney lane -> {'street_number': 'cnr', 'street_name1': 'bruce', 'street_type1': 'hwy', 'street_name2': 'geaney', 'street_type2': 'lane'}
corner bruce hwy & highway plaza, mackay qld 4740 -> {'street_number': 'cnr', 'street_name1': 'bruce', 'street_type1': 'hwy', 'street_name2': 'plaza', 'street_type2': 'highway'}
shop 2  7 eleven service station. corner broardbeach nerang & chisholm roads -> {'street_number': 'cnr', 'street_name1': 'broardbeach nerang', 'street_type1': 'roads', 'street_name2': 'chisholm', 'street_type2': 'roads'}
61 sallys corner road -> {'street_number': '61', 'street_name1': 'sallys', 'street_type1': 'roads', 'street_name2': 'N/A', 'street_type2': 'N/A'}
corner princes fwy and hughes road-> {'street_number': 'cnr', 'street_name1': 'princes', 'street_type1': 'fwy', 'street_name2': 'hughes', 'street_type2': 'roads'}

I’ve tried using regex patterns like b(?:cn?r|d+-d+|d+)b to match street numbers and r'b(?:corner|cn?r|shop|road|drive|highway|street|hwy|lane|plaza|roads|avenue|centre|center|place|square|boulevard|way|trailer park|terrace|crescent|roadhouse|crossing|wharf|court|fwy)b' to match street names, but I’m not getting the desired results.
Here is my current code:

import re


def parse_australian_address(address):
    # Normalize common terms for corner addresses
    address = re.sub(r"b[Cc]ornerb", "cnr", address)

    # Regex pattern to match the required address components
    pattern = re.compile(
        r"""
        (?:.*?bcnrbs)? 
        (?P<street_number>cnr|d+)s+
        (?P<street_name1>(?:bw+bs?)+?)s+
        (?P<street_type1>road|street|avenue|drive|lane|highway|hwy|fwy|parkway|roadways|streets|plaza)s+
        (?:and|&|near|at)s+
        (?P<street_name2>(?:bw+bs?)+?)s+
        (?P<street_type2>road|street|avenue|drive|lane|highway|hwy|fwy|parkway|roadways|streets|plaza)
        """,
        re.VERBOSE | re.IGNORECASE,
    )

    match = pattern.search(address)

    if not match:
        # If no match found, attempt to parse without the second street
        single_street_pattern = re.compile(
            r"""
            (?P<street_number>d+)s+
            (?P<street_name1>(?:bw+bs?)+?)s+
            (?P<street_type1>road|street|avenue|drive|lane|highway|hwy|fwy|parkway|roadways|streets|plaza)
            """,
            re.VERBOSE | re.IGNORECASE,
        )

        single_match = single_street_pattern.search(address)

        if not single_match:
            return None

        result = {
            "street_number": single_match.group("street_number"),
            "street_name1": single_match.group("street_name1").strip(),
            "street_type1": single_match.group("street_type1").strip(),
            "street_name2": "N/A",
            "street_type2": "N/A",
        }

        return result

    # Extract matched components
    result = {
        "street_number": match.group("street_number").lower(),
        "street_name1": match.group("street_name1").strip(),
        "street_type1": match.group("street_type1").strip(),
        "street_name2": match.group("street_name2").strip(),
        "street_type2": match.group("street_type2").strip(),
    }

    return result


addresses = [
    "corner hampstead road and east parkway northfield sa 5085",
    "burleigh west shopping centre, corner west burleigh road & reedy creek road, burleigh west qld 4219",
    "coburn central shopping centre, corner high & melton streets, melton vic 3337",
    "corner armadale road & alex wood drive forrestdale wa 6112",
    "corner beach road & hanna road, noarlunga centre, noarlunga sa 5168",
    "corner bray street & pacific highway, coffs harbour nsw 2450",
    "corner bruce hwy & geaney lane",
    "corner bruce hwy & highway plaza, mackay qld 4740",
    "shop 2  7 eleven service station. corner broardbeach nerang & chisholm roads",
    "61 sallys corner road",
    "corner princes fwy and hughes road",
]

for address in addresses:
    print(parse_australian_address(address))

Basically the current code covering most of the addresses except 2 addresses which is:

"coburn central shopping centre, corner high & melton streets, melton vic 3337","shop 2  7 eleven service station. corner broardbeach nerang & chisholm roads"

Can someone please help me with the correct regex pattern and logic to extract the street names and types in this format?

Thank you!