Generate macOS model identifier / device name mapping from Apple support pages

generate_macos_model_identifiers_dict.py

import json
import re

import requests

DEVICE_NAME_MODEL_IDENTIFIER_REGEX = r"<strong>(.*)\s*(?:<br>\s*</strong>|</strong>\s*<br>)\s*.*\s*Model (?:Label|Identifier):\s*(.*)\s*<br>"

SUPPORT_URLS = [
    # iMac
    "https://support.apple.com/en-gb/HT201634",
    # MacBook Air
    "https://support.apple.com/en-gb/HT201862",
    # MacBook Pro
    "https://support.apple.com/en-gb/HT201300",
    # Mac mini
    "https://support.apple.com/en-gb/HT201894",
    # Mac Pro
    "https://support.apple.com/en-gb/HT202888",
]

identifiers = {}

for url in SUPPORT_URLS:
    html = requests.get(url).text.replace("&nbsp;", " ")
    for device_name, model_identifiers in reversed(
        re.findall(
            DEVICE_NAME_MODEL_IDENTIFIER_REGEX, html, re.IGNORECASE | re.MULTILINE
        )
    ):
        for model_identifier in model_identifiers.split(";"):
            identifiers[model_identifier.strip()] = device_name.strip()

print(json.dumps(identifiers, indent=4))