Database Reference
In-Depth Information
Building an autocomplete function
So far, we've been focused on storing users and their status updates, but we can use our
knowledge of compound primary keys to make it a bit easier to write status updates too.
Let's introduce a hashtagging function into the status update composition interface, and
then autocomplete hashtags as users type them.
First, we'll set up a table to store hashtags using the following query:
CREATE TABLE "hash_tags" (
"prefix" text,
"remaining" text,
"tag" text,
PRIMARY KEY ("prefix", "remaining")
);
The structure of our table is a bit unusual but it will work very well for our purposes. The
partition key is
prefix
, which we'll use to store the first two letters of each hashtag. The
clustering column,
remaining
, will store the remaining letters of the hashtag, and
tag
will contain the entire hashtag start to finish.
By partitioning the table this way, we'll make things easy for Cassandra by immediately
narrowing down the list of possible autocomplete tags to those in the partition, identified
by the two-letter prefix. This does, of course, mean that we can't autocomplete entries of
fewer than two letters, but this is a pretty standard limitation.
To see how the autocomplete works, let's add some seed data into the table:
INSERT INTO "hash_tags" ("prefix", "remaining", "tag")
VALUES ('ca', 'ssandra', 'cassandra');
INSERT INTO "hash_tags" ("prefix", "remaining", "tag")
VALUES ('ca', 'ssette', 'cassette');
INSERT INTO "hash_tags" ("prefix", "remaining", "tag")
VALUES ('ca', 'sual', 'casual');
INSERT INTO "hash_tags" ("prefix", "remaining", "tag")
VALUES ('ca', 'ke', 'cake');
Now, let's say the user starts typing a hashtag, and has so far typed
ca
. That's easy: we can
just ask for the
ca
partition in our table:
