Parse Minted Contacts Print View

parse_print_contacts_html.py

import csv
import sys

from bs4 import BeautifulSoup as bs


"""
Minted allows you to import CSV/XLSX, etc, but doesn't provide an easy export. This helps with that.
The script parses contacts "exported" via Minted's print function into a CSV.
Not my best work, but it should save some time if anybody else needs to do this again.

Call like this, where "contacts-2021-12-12.html" is the html contents of the page presented for printing contacts (see: https://help2.minted.com/s/article/print-my-contact-list)

    python parse_print_contacts_html.py < contacts-2021-12-12.html > contacts-2021-12-12.csv

"""

soup = bs(sys.stdin, 'html.parser')

contacts = soup.body.find_all('li')

fieldnames = ['name', 'address_1', 'address_2', 'city', 'state', 'zip']
writer = csv.DictWriter(sys.stdout, fieldnames=fieldnames)

writer.writeheader()

for contact in contacts:

    name = contact.select(".contact-name")[0].string.strip()

    address_str = contact.select(".contact-address")[0].get_text().strip()
    address_parts = address_str.split("\n")

    if address_str == '':
        # Empty
        address_1 = ''
        address_2 = ''
        address_3 = ''
    elif len(address_parts) < 2:
        # 1 non-empty
        address_1 = address_parts[0].strip()
        address_2 = ''
        address_3 = ''
    elif len(address_parts) < 3:
        # 2 non-empty; 2nd part is city, state, zip
        address_1 = address_parts[0].strip()
        address_2 = address_parts[1].strip()
        address_3 = address_2
        address_2 = ''
    else:
        # 3 non-empty
        address_1 = address_parts[0].strip()
        address_2 = address_parts[1].strip()
        address_3 = address_parts[2].strip()

    if address_3:
        address_3_parts = address_3.split(",")
        city = address_3_parts[0].strip()
        state_zip = address_3_parts[1].strip()

        zip_code = state_zip.split(" ")[-1]
        state = " ".join(state_zip.split(" ")[0:-1])
    else:
        city = ''
        zip_code = ''
        state = ''

    writer.writerow({
        'name': name, 
        'address_1': address_1, 
        'address_2': address_2, 
        'city': city, 
        'state': state, 
        'zip': zip_code
    })