A pretty horrible Python script to fix the SQL errors in the Rijksmonumenten dump mentioned here: https://github.com/clytras/AccessConverter/issues/5

fix-sql.py

#!/usr/bin/env python3
from tqdm import tqdm
from sys import argv, exit
from os.path import getsize

def fix_token(token):
    if token == ",":
        return "'',"
    elif token == "(,":
        return "('',"
    elif ",)," in token:
        return token.replace(",),", ",''),")
    elif token == "),":
        return "''),"
    elif token == ");":
        return "'');"
    elif ",);" in token:
        return token.replace(",);", ",'');")
    else:
        return token

def iter_tokens(path):
    with open(path) as f:
        data = ""
        statement = ""

        for line in f:
            for token in line.split(" "):
                yield token.strip()

            yield "\n"

def main():
    if len(argv) != 2:
        exit("Invalid arguments")

    path = argv[1]
    target = path.replace(".sql", "-fixed.sql")
    tokens = iter_tokens(path)
    tokens = tqdm(tokens, total = getsize(path) / 10)

    with open(target, "w") as f:
        statement = []
        values_open = False

        for token in tokens:
            if token == "INSERT":
                values_open = True

            if token == "CREATE" and values_open:
                # This should not happen, drop the whole current statement
                values_open = False
                statement = []

            if values_open:
                statement.append(token)
            else:
                f.write(token + " ")

            if ");" in token:
                statement = [fix_token(token) for token in statement]
                statement = " ".join(statement)
                f.write(f"\n{statement}\n")
                values_open = False
                statement = []

if __name__ == "__main__":
    main()