How can I check if a MySQL table exists with PHP?

As simple in theory as it sounds I've done a fair amount of research and am having trouble figuring this out.

How can I check if a MySQL table exists and if it does do something. (I guess a simple php if/else statement could work for this)

Is there a way to do this?

This is what I have done with cwallenpoole's response:

my开发者_运维知识库sql_connect("SERVER","USERNAME","PASSWORD");
mysql_select_db('DATABASE');

$val = mysql_query('select 1 from `TABLE`');

if($val !== FALSE)
{
   print("Exists");
}else{
   print("Doesn't exist");
}

// Select 1 from table_name will return false if the table does not exist.
$val = mysql_query('select 1 from `table_name` LIMIT 1');

if($val !== FALSE)
{
   //DO SOMETHING! IT EXISTS!
}
else
{
    //I can't find it...
}

Admittedly, it is more Pythonic than of the PHP idiom, but on the other hand, you don't have to worry about dealing with a copious amount of extra data.

Edit

So, this answer has been marked down at least twice as of the time I am writing this message. Assuming that I had made some gargantuan error, I went and I ran some benchmarks, and this is what I found that my solution is over 10% faster than the nearest alternative when the table does not exist, and it over 25% faster when the table does exist:

:::::::::::::::::::::::::BEGINNING NON-EXISTING TABLE::::::::::::::::::::::::::::::
23.35501408577 for bad select
25.408507823944 for select from schema num rows -- calls mysql_num_rows on select... from information_schema.
25.336688995361 for select from schema fetch row -- calls mysql_fetch_row on select... from information_schema result
50.669058799744 for SHOW TABLES FROM test
:::::::::::::::::::::::::BEGINNING EXISTING TABLE::::::::::::::::::::::::::::::
15.293519973755 for good select
20.784908056259 for select from schema num rows
21.038464069366 for select from schema fetch row
50.400309085846 for SHOW TABLES FROM test

I tried running this against DESC, but I had a timeout after 276 seconds (24 seconds for my answer, 276 to fail to complete the description of a non existing table).

For good measure, I am benchmarking against a schema with only four tables in it and this is an almost fresh MySQL install (this is the only database so far). To see the export, look here.

AND FURTHERMORE

This particular solution is also more database independent as the same query will work in PgSQL and Oracle.

FINALLY

mysql_query() returns FALSE for errors that aren't "this table doesn't exist".

If you need to guarantee that the table doesn't exist, use mysql_errno() to get the error code and compare it to the relevant MySQL errors.

The cleanest way to achieve this in PHP is to simply use DESCRIBE statement.

if ( mysql_query( "DESCRIBE `my_table`" ) ) {
    // my_table exists
}

I'm not sure why others are posting complicated queries for a such a straight forward problem.

Update

Using PDO

// assuming you have already setup $pdo
$sh = $pdo->prepare( "DESCRIBE `my_table`");
if ( $sh->execute() ) {
    // my_table exists
} else {
    // my_table does not exist    
}

$res = mysql_query("SELECT table_name FROM information_schema.tables WHERE table_schema = '$databasename' AND table_name = '$tablename';");

If no records are returned then it doesn't exist.

SHOW TABLES LIKE 'TableName'

If you have ANY results, the table exists.

To use this approach in PDO:

$pdo         = new \PDO(/*...*/);
$result      = $pdo->query("SHOW TABLES LIKE 'tableName'");
$tableExists = $result !== false && $result->rowCount() > 0;

To use this approach with DEPRECATED mysql_query

$result      = mysql_query("SHOW TABLES LIKE 'tableName'");
$tableExists = mysql_num_rows($result) > 0;

125 microsecond table exists check

.000125 sec. (125µs)

mysql_unbuffered_query("SET profiling = 1");  // for profiling only

@mysql_unbuffered_query("SELECT 1 FROM `table` LIMIT 1 ");
if (mysql_errno() == 1146){

 // NO EXISTING TABLE CODE GOES HERE

}
elseif(mysql_errno() > 0){

  echo mysql_error();    

}

$results = mysql_query("SHOW PROFILE");  // for profiling only

Execution time, measured using mysql profiling, when a table exists, or not, is about .000125 sec. (125µs)

The LIMIT 1 is important for speed. This minimizes the Sorting Result and Sending Data query State times. And table size is not a factor.

I always use an unbuffered query when no results are required.

PROFILE RESULTS WHEN TABLE DOES NOT EXIST

QUERY STATE           SECONDS   
--------------------  -------
starting              0.000025  
checking permissions  0.000006  
Opening tables        0.000065  
query end             0.000005  
closing tables        0.000003  
freeing items         0.000013  
logging slow query    0.000003  
cleaning up           0.000003  
TOTAL                 0.000123  <<<<<<<<<<<< 123 microseconds

WHEN TABLE EXISTS

QUERY STATE           SECONDS   
--------------------  -------
starting              0.000024
checking permissions  0.000005
Opening tables        0.000013
System lock           0.000007
init                  0.000006
optimizing            0.000003
statistics            0.000009
preparing             0.000008
executing             0.000003
Sending data          0.000019
end                   0.000004
query end             0.000004
closing tables        0.000006
freeing items         0.00001
logging slow query    0.000003
cleaning up           0.000003
TOTAL                 0.000127 <<<<<<<<<<<< 127 microseconds

mysql_query("SHOW TABLES FROM yourDB");
//> loop thru results and see if it exists
//> in this way with only one query one can check easly more table 

or mysql_query("SHOW TABLES LIKE 'tblname'");

Don't use mysql_list_tables(); because it's deprecated

Even faster than a bad query:

SELECT count((1)) as `ct`  FROM INFORMATION_SCHEMA.TABLES where table_schema ='yourdatabasename' and table_name='yourtablename';

This way you can just retrieve one field and one value. .016 seconds for my slower system.

It was already posted but here it is with PDO (same query)...

$connection = new PDO ( "mysql:host=host_db; dbname=name_db", user_db, pass_db );

if ($connection->query ("DESCRIBE table_name"  )) {
    echo "exist";
} else {
    echo "doesn't exist";
}

Works like a charm for me....

And here's another approach (i think it is slower)...

if ($connection->query ( "SELECT * FROM INFORMATION_SCHEMA.TABLES WHERE table_schema = 'db_name' AND table_name ='tbl_name'" )->fetch ()) {
    echo "exist";
} else {
    echo "doesn't exist";
}

You can also play with this query:

SHOW TABLE STATUS FROM db_name LIKE 'tbl_name'

I think this was suggested to use in the mysql page.

DO NOT USE MYSQL ANY MORE. If you must use mysqli but PDO is best:

$pdo = new PDO($dsn, $username, $pdo); // proper PDO init string here
if ($pdo->query("SELECT table_name FROM INFORMATION_SCHEMA.TABLES WHERE table_schema = 'db_name'")->fetch()) // table exists.

<?php 
$connection = mysqli_connect("localhost","root","","php_sample_login_register"); 

if ($connection){
        echo "DB is Connected <br>";
        $sql = "CREATE TABLE user(
            id INT(6) UNSIGNED AUTO_INCREMENT PRIMARY KEY, 
            name VARCHAR(255)NOT NULL,
            email VARCHAR(255)NOT NULL,
            password VARCHAR(255) NOT NULL
            );";
        if(mysqli_query($connection,$sql)) {
            echo "Created user table";
        } else{
            echo "User table already exists";
        }
    } else {
        echo "error : DB isnot connected";
    } ?>

Or you could use

show tables where Tables_in_{insert_db_name}='tablename';

You can use many different queries to check if a table exists. Below is a comparison between several:

mysql_query('select 1 from `table_name` group by 1'); or  
mysql_query('select count(*) from `table_name`');

mysql_query("DESCRIBE `table_name`");  
70000   rows: 24ms  
1000000 rows: 24ms  
5000000 rows: 24ms

mysql_query('select 1 from `table_name`');  
70000   rows: 19ms  
1000000 rows: 23ms  
5000000 rows: 29ms

mysql_query('select 1 from `table_name` group by 1'); or  
mysql_query('select count(*) from `table_name`');  
70000   rows: 18ms  
1000000 rows: 18ms  
5000000 rows: 18ms  

These benchmarks are only averages: