使用 LEFT JOIN 创建客户搜索。 如果在第二个表中找不到匹配项，则排除结果

Question

I'm trying to create a way to search through a database of customers, where some pieces of data like phone numbers and emails have a one-to-many relationship. I have a table that has a customer ID (unique, autonumber), a first name, and a last name. I have a second table containing a phone number, and the customer ID that that phone number is associated with. I have a third table containing an email address, and the customer ID that that email address is associated with.

Here is the query built in PHP:

$query = "SELECT *
            FROM customers
            LEFT OUTER JOIN phone_numbers ON customers.customer_id = phone_numbers.associated_customer
            LEFT OUTER JOIN email_addresses ON customers.customer_id = email_addresses.associated_customer
            WHERE first_name LIKE '%" . $_GET["fname"] . "%'
            AND last_name LIKE '%" . $_GET["lname"] . "%'
            AND phone_number LIKE '%" . $_GET["phone"] . "%'
            AND email_address LIKE '%" . $_GET["email"] . "%'";

With this, the customer I searched for only comes up if they have an phone number and email address. I also get duplicate results from the first table if the customer has multiple phone numbers or emails, so if the customer has 2 phone numbers and 2 emails, I would get 4 results back.

Am I structuring my tables correctly for to achieve this one-to-many relationship?

EDIT: Sample data for clarification.

table 1:

╔════════════════════════════════════════╗
║                customers               ║
╠═════════════╦════════════╦═════════════╣
║ customer_id ║ first_name ║ last_name   ║
╠═════════════╬════════════╬═════════════╣
║      1      ║ John       ║ Doe         ║
║      2      ║ John       ║ Wick        ║
║      3      ║ John       ║ Cena        ║
║      4      ║ John       ║ Krasinski   ║
║      5      ║ Jane       ║ Doe         ║
║      6      ║ Freddie    ║ Mercury     ║
╚═════════════╩════════════╩═════════════╝

table 2:

╔══════════════════════════════════════════════════╗
║                   phone numbers                  ║
╠════════════════╦═══════════════╦═════════════════╣
║  phone_number  ║ associated_id ║ primary_contact ║
╠════════════════╬═══════════════╬═════════════════╣
║   5555555555   ║       2       ║        0        ║
║   6692216251   ║       2       ║        1        ║
║   2025550174   ║       3       ║        1        ║
╚════════════════╩═══════════════╩═════════════════╝

table 3:

╔═══════════════════════════════════════════════════╗
║                   email_addresses                 ║
╠═════════════════╦═══════════════╦═════════════════╣
║  email_address  ║ associated_id ║ primary_contact ║
╠═════════════════╬═══════════════╬═════════════════╣
║ jdoe@aol.com    ║       1       ║        1        ║
║ jwick@email.com ║       2       ║        1        ║
║ jwick@aol.com   ║       2       ║        0        ║
╚═════════════════╩═══════════════╩═════════════════╝

search query:

first name: "John"
last name: ""
phone number: ""
email address: ""

The expected result is for it to return all matches for "John" in the first name field, and only include the primary methods of contact:

╔═════════════╦════════════╦═════════════╦══════════════╦═════════════════╗
║ customer_id ║ first_name ║  last_name  ║ phone_number ║  email_address  ║
╠═════════════╬════════════╬═════════════╬══════════════╬═════════════════╣
║      1      ║ John       ║ Doe         ║              ║ jdoe@aol.com    ║
║      2      ║ John       ║ Wick        ║  6692216251  ║ jwick@email.com ║
║      3      ║ John       ║ Cena        ║  2025550174  ║                 ║
║      4      ║ John       ║ Krasinski   ║              ║                 ║
╚═════════════╩════════════╩═════════════╩══════════════╩═════════════════╝

The actual result is that all results that do not have an associated phone number AND email address are excluded, and a duplicate is included for each phone number and each email address:

╔═════════════╦════════════╦═════════════╦══════════════╦═════════════════╗
║ customer_id ║ first_name ║  last_name  ║ phone_number ║  email_address  ║
╠═════════════╬════════════╬═════════════╬══════════════╬═════════════════╣
║      2      ║ John       ║ Wick        ║  5555555555  ║ jwick@email.com ║
║      2      ║ John       ║ Wick        ║  6692216251  ║ jwick@email.com ║
║      2      ║ John       ║ Wick        ║  5555555555  ║ jwick@aol.com   ║
║      2      ║ John       ║ Wick        ║  6692216251  ║ jwick@aol.com   ║
╚═════════════╩════════════╩═════════════╩══════════════╩═════════════════╝

However, let's say my search query is the following:

first name: "John"
last name: ""
phone number: "5555555555"
email address: ""

The expected result would be:

╔═════════════╦════════════╦═════════════╦══════════════╦═════════════════╗
║ customer_id ║ first_name ║  last_name  ║ phone_number ║  email_address  ║
╠═════════════╬════════════╬═════════════╬══════════════╬═════════════════╣
║      2      ║ John       ║ Wick        ║  5555555555  ║ jwick@email.com ║
╚═════════════╩════════════╩═════════════╩══════════════╩═════════════════╝

I am aware of SQL injection attacks. I will have measures in place to prevent them, but this will be an internal system, so I'm not that worried about it.

Answer 1

Could you try the following query which for each table (phone and email) consider two cases 1. there's a non empty search criteria so you'll use it to filter the table 2. the search criteria is empty, you then only look for primary_contact. The situation where the client doesn't have a contact is handled with the outer join but then you need to allow for primary_contact to be null.

The situation where the customer only has only secondary contacts is not handled but it doesn't make sense if you have a contact phone (or email) one has to be the primary one.

The query is much longer but it's twice the same logic (one for email one for phone)

SELECT *
FROM customers c
LEFT OUTER JOIN phone_numbers p ON c.customer_id = p.associated_customer
LEFT OUTER JOIN email_addresses e ON c.customer_id = e.associated_customer
WHERE c.first_name LIKE '%" . $_GET["fname"] . "%'
AND c.last_name LIKE '%" . $_GET["lname"] . "%'
AND (( '%" . $_GET["phone"] . "%' <> '' and
        p.phone_number LIKE '%" . $_GET["phone"] . "%' )
     or
     ( '%" . $_GET["phone"] . "%' = '' and
        (p.primary_contact = 1 or p.primary_contact is null) ))
AND (( '%" . $_GET["email"] . "%' <> '' and
        e.email_address LIKE '%" . $_GET["email"] . "%' )
     or
     ( '%" . $_GET["email"] . "%' = '' and
        (e.primary_contact = 1 or e.primary_contact is null) ))

Answer 2

Your database design is fine.

Your conditions:

• If $_GET["phone"] <> '' then only matching phone rows, else only primary contacts.
• If $_GET["email"] <> '' then only matching email rows, else only primary contacts.

This leads to the following query:

SELECT *
FROM customers c
LEFT JOIN phone_numbers p ON p.associated_customer = c.customer_id
LEFT JOIN email_addresses e ON e.associated_customer = c.customer_id
WHERE c.first_name LIKE :fname
  AND c.last_name LIKE :lname
  AND
  (
    (:phone <> '' AND p.phone_number LIKE '%' || :phone || '%')
    OR
    (:phone = '' AND (p.primary_contact = 1 OR p.primary_contact IS NULL))
  )
  AND
  (
    (:email <> '' AND e.email_address LIKE '%' || :email || '%')
    OR
    (:email = '' AND (e.primary_contact = 1 OR e.primary_contact IS NULL))
  )
ORDER BY c.first_name, c.last_name, p.phone_numbers, e.email_addresses;

(I've replaced your GET commands with bind variables for readability.)

Anyway, with weak search patterns, say the phone number containing a certain digit and and the email address containing a certain letter, you could still get a cartesian product:

Eg: John Smith, phones: 123456, 234567, emails: one@company.com, two@company.com. Search: John Smith, phone contains 2, email contains o. Result:

fname | lname | phone  | email
------+-------+--------+----------------
John  | Smith | 123456 | one@company.com
John  | Smith | 123456 | two@company.com
John  | Smith | 234567 | one@company.com
John  | Smith | 234567 | two@company.com

You may want to consider this somehow, eg use string aggregation, so as to get the following result instead:

fname | lname | phones         | emails
------+-------+----------------+---------------------------------
John  | Smith | 123456, 234567 | one@company.com, two@company.com