Question

In my MySQL table I have a simple postal codes table that I have converted it from an .xlsx file.

Country   |City    |MiniDistrict |Postal Code
---------------------------------------------
Bulgaria  |Sofia   |Oborishte    | 0844
Bulgaria  |Sofia   |Sredets      | 0988
Bulgaria  |Plovdiv |Montana      | 1299
Bulgaria  |Plovdiv |Dyavol       | 1288
Bulgaria  |Varna   |Andeevo      | 1574
France    |Paris   |Mantois      |38992
France    |Paris   |Valois       |38764
France    |Gard    |Lussan       |55980

As you can see Country and city names are repeating numerious times. also note that there are same city names in different countries. I want to write a PHP algorithm where I’ll generate a country table, a city table, and a minidistrict-postalcode table into MySQL. The country table is easy, SELECT DISTINCT. What algorythm should I write for the other tables? The algorytm idea will be enough for me, no need for the coding part.

Answer 1

The “algorithm” is this: you notice that every MiniDistrict has a PostalCode, and every PostalCode has its MiniDistrict. So you put these together in a single table, and associate each pair with an unique ID.

CREATE TABLE district_postalcode
(
    id integer not null primary key auto_increment,
    minidistrict varchar(???),
    postalcode char(5)
);

Then you notice that each city is a single object, and the same for the countries, so you create two tables for them, each with its own ID.

Now you must store a relationship between minidistricts and cities. But this is a many-to-one relationship: many minidistricts may belong to the same city, but the same minidistrict can’t belong to two cities.

So you add a reference to cities.id as foreign key city_id to minidistricts.

The relationship structure is the same between cities and countries. So again you add a foreign key, this time country_id to countries.id into cities.

At that point, you can also create a VIEW to see your data as before:

CREATE VIEW old_style AS SELECT
    countries.Country,
    cities.City,
    districts.MiniDistrict,
    districts.PostalCode
FROM districts
    JOIN cities ON (cities.id = districts.city_id)
    JOIN countries ON (country.id = cities.country_id);

You can also populate your new districts, cities and countries from your existing data, beginning with the table that has no foreign keys:

INSERT INTO countries ( Country ) SELECT DISTINCT Country FROM old_table;

INSERT INTO cities ( City, country_id )
    SELECT DISTINCT City, Country.id
    FROM old_table
    JOIN countries ON ( old_table.Country = countries.Country );

(This will insert twice a city if it exists with the same name in two countries – one with each country ID – but that’s as it should be: city names are normally not unique).