Question

I have 27 tables in my database. One word table (a scrabble word list), and 26 association tables.

Table  Fields
================
word   [id,word]
a      [word_id]
b      [word_id]
...
z      [word_id]

I’m trying to figure out matching words given a string.

For example, if the given string is pant, I want to know: pant, apt, pat, tap, ant, tan, nap, pan, at, ta, pa, an, na.

My current strategy is to explode each letter in the string and find the associated words that match all the letters.

For example:

SELECT word.word
FROM word, p, a, n, t
WHERE
    word.id = p.word_id OR
    word.id = a.word_id OR
    word.id = n.word_id OR
    word.id = t.word_id

But this ends up printing all words that have a p,a,n or t in them.

And if I switch all the operators to AND, I’m stuck with only one match: pant.

Can you help me solve this riddle?

I’m also concerned with how to handle duplicate letters in the string. For example, PPANT should find a match for app, when plain PANT should not.

Am I on the right track with the association tables or is there a better way?

I’m trying to handle this fairly efficiently in php/mysql. I’m aware there are others who have solved this riddle before in C, perl, java and the like.

Answer 1

I’m not familiar with MySQL’s advanced capabilities, so I cannot say if there is a way to do this restriction procedurally, which may save you a good amount of storage space. Nonetheless, I’ll offer this possibility.

Say this was your word table:

+==========+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+
|   word   | a | b | c | d | e | f | g | h | i | j | k | l | m | n | o | p | q | r | s | t | u | v | w | x | y | z |
+==========+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+
| pant     | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 1 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 |
+==========+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+
| ppant    | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 2 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 |
+==========+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+
| app      | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 2 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
+==========+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+
| kick     | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 2 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
+==========+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+===+

Then your query might look like the following:

SELECT word.word FROM word
JOIN
(
  SELECT * FROM word WHERE word.word = "pant"
) AS root
ON
    word.a <= root.a
AND word.b <= root.b
AND word.c <= root.c
AND word.d <= root.d
AND word.e <= root.e
AND word.f <= root.f
AND word.g <= root.g
AND word.h <= root.h
AND word.i <= root.i
AND word.j <= root.j
AND word.k <= root.k
AND word.l <= root.l
AND word.m <= root.m
AND word.n <= root.n
AND word.o <= root.o
AND word.p <= root.p
AND word.q <= root.q
AND word.r <= root.r
AND word.s <= root.s
AND word.t <= root.t
AND word.u <= root.u
AND word.v <= root.v
AND word.w <= root.w
AND word.x <= root.x
AND word.y <= root.y
AND word.z <= root.z

Now, of course there are ways to normalize the table and multiple ways to create the query. You should experiment with what makes most sense for your situation.