most effective query? - mysql

I am creating a function to get all keywords from a database
The database has two tables
keywords [id | word | account] ( aliased as k )
keywordsTemplateLink [templateId | keywordId] ( aliased as ktl )
the functions signature is
getKeywords($id = null){}
so the way it works is,
if id != null a where clause is added which must limit the result set to keywords where ktl.templateId = $id
What would be the most effective way to achieve this query?
Im thinking SELECT id, keyword FROM keywords k, templatekeywordlink tkl WHERE tkl.templateId= $id AND tkl.keywordId = k.id AND k.account=$account
Is there a better way?

function getKeywords($id = null){
//query generated by function
$query .= ($id != null) ? ' where ktl.templateId = ' . $id : '';
}

It is generally bad practice to create sql by variable concatenation in this way..
If you don't want some script kiddie to pwn you via sql injection, use prepared queries.
$stm = $dbo->prepare("SELECT id, keyword FROM keywords k, templatekeywordlink tkl WHERE tkl.templateId= ? AND k.account=?);
$stm->execute(array($id,$account));

Related

Sql Alchemy filter / where clause joined by OR not AND

I want to select a bunch distinct records based off a composite key. In SQL I'd write something like this:
SELECT * FROM security WHERE (
exchange_code = 'exchange_code_1' AND code = 'code_1')
OR (exchange_code = 'exchange_code_2' AND code = 'code_2')
...
OR (exchange_code = 'exchange_code_N' AND code = 'code_N')
)
With SQLAlchemy I'd like to use the filter clause like:
query = sess.query(Security)
[query.filter(
and_(Security.exchange_code == security.exchange_code,
Security.code == security.code)
) for security in securities]
result = query.all()
The problem is filter and where join clauses with an AND not an OR... is there some way to use filter with OR?
Or is my only choice to generate a bunch of individual select's and UNION them? Something like:
first = exchanges.pop()
query = reduce(lambda query, exchange: query.union(exchange.pk_query),
first.pk_query())
query.all()
Use or_:
query = sess.query(Security).filter(
or_(*(and_(Security.exchange_code == security.exchange_code,
Security.code == security.code)
for security in securities)))
If your database supports it, you should use tuple_ instead.

SQL: Binding an undef (NULL) to the placeholder will not select rows which have a NULL value

I'm using perl 5.20 and MySQL 5.7, but I think the question is about SQL in general:
perldoc DBI says:
Binding an undef (NULL) to the placeholder will not select rows which have a NULL age! At least for database engines that conform to the SQL standard. Refer to the SQL manual for your database engine or any SQL book for the reasons for this. To explicitly select NULLs you have to say "WHERE age IS NULL".
I don't even know what to google for... My question is: What are the reasons behind = ? not matching a binding to NULL/undef? (Beyond "that is how it is defined and documented".)
I've discovered that MySQL has an operator <=> that allows comparisons with NULL and so:
my $sth = $dbh->prepare('select count(*) from table where field <=> ?');
$sth->execute(345);
$sth->execute(undef);
both work as expected. Unfortunately, the doc says:
The <=> operator is equivalent to the standard SQL IS NOT DISTINCT FROM operator.
And MySQL doesn't support the IS NOT DISTINCT FROM operator :-(. So there seems to be no portable way to do this. Except for the very hackish:
my $sth = $dbh->prepare('
select count(*) from table
where field = ? OR ( ? IS NULL AND field IS NULL )
');
$sth->execute(345, 345);
$sth->execute(undef, undef);
or the even more hackish snippet from perldoc DBI
$sql_clause = defined $age? "age = ?" : "age IS NULL";
$sth = $dbh->prepare(qq{
SELECT fullname FROM people WHERE $sql_clause
});
$sth->execute(defined $age ? $age : ());
Is there a portable way to do WHERE FIELD = ? and have it do what I mean also with NULLs/undefs? What did I miss?
EDIT: I also came up with this workaround, which especially works great if field is a numeric type such as an INT, so we're sure the string "NULL" is not a possible non-NULL value.
my $sth = $dbh->prepare('
select count(*) from table
where COALESCE(field, "NULL") = COALESCE(?,"NULL")
');
$sth->execute(345);
$sth->execute(undef);
But performance goes out the window, as I don't think any indexes can be used....
I understand that this is not exactly what you asked for but if you use DBIx::Class, the ORM will do that lifting for you.
my $res = $schema->resultset('table')->search({ field => [345, undef] });
print $res->count;
It will be translated to this SELECT COUNT( * ) FROM table me WHERE ( ( field = ? OR field IS NULL ) ): '345'

Is there a more efficient way to write this MySQL query?

I'm a newbie to mysql, I managed to scrape this together to get the result I wanted. Can it be coded better? Are there any security risks? Its being output in php.
$qwe = $product->virtuemart_product_id;
$id = mysql_real_escape_string($qwe);
$result = mysql_query('SELECT * FROM virtuemart_product_medias where virtuemart_product_id = ' . $id . ' LIMIT 1');
$row = mysql_fetch_assoc($result);
$matched = $row['virtuemart_media_id'];
$result2 = mysql_query('SELECT * FROM virtuemart_medias where virtuemart_media_id = ' . $matched . ' LIMIT 1');
$row2 = mysql_fetch_assoc($result2);
$matched2 = $row2['file_url_thumb'];
echo $matched2;
I don't know whether or not there is a security hole in the specific code you provided - that depends on what other validation exists elsewhere in your program, and what you consider to be a security hole. But the way you are coding means that there definitely could be security holes. Let's look at your first query:
$id = mysql_real_escape_string($qwe);
$result = mysql_query('SELECT *
FROM virtuemart_product_medias
WHERE virtuemart_product_id = ' . $id . ' LIMIT 1');
Imagine if $qwe is the string 0 OR 1=1 --. The mysql_real_escape_string only escapes certain characters such as quotes and backslashes.
mysql_real_escape_string() calls MySQL's library function mysql_real_escape_string, which prepends backslashes to the following characters: \x00, \n, \r, \\, ', " and \x1a.
The string 0 OR 1=1 -- that I mentioned above does not contain any of these characters so it will not be affected at all by mysql_real_escape_string. After you substitute in the value of $id, the resulting SQL query will look something like this:
SELECT *
FROM virtuemart_product_medias
WHERE virtuemart_product_id = 0 OR 1=1 -- LIMIT 1
As you can see, this will return all rows.
Long story short: Use PDO and parameterized queries.
Related
How can I prevent SQL injection in PHP?
Firstly, never use the mysql_* functions. They are deprecated and relying on them is highly discouraged. Use either MySQLi or PDO
The above query could be rewritten as
SELECT file_url_thumb FROM virtuemart_medias where virtuemart_media_id = (SELECT virtuemart_media_id FROM virtuemart_product_medias where virtuemart_product_id = ' . $id . ' LIMIT 1) LIMIT 1
Never do a SELECT *. Include only those fields in your query which you need in your code.
Use one query instead of two, and select only the fields you're using, like so:
SELECT `file_url_thumb` FROM virtuemart_medias where virtuemart_media_id = (SELECT `virtuemart_media_id` FROM virtuemart_product_medias where virtuemart_product_id = ' . $id . ' LIMIT 1) LIMIT 1
You can always use a join;
SELECT a.virtuemart_media_id, b.file_url_thumb
FROM virtuemart_product_medias a
LEFT JOIN virtuemart_medias b
ON a.virtuemart_media_id = b.virtuemart_media_id
WHERE virtuemart_product_id = $id
LIMIT 1
That'll always get you the virtuemart_media_id and, if it exists file_url_thumb.
Your query has a problem also, mysql_real_escape_string only escapes strings, since you're not quoting the $id in the query, it won't be handled as a string and the escaping will not help you. As other replies point out, you should really be using mysqli or PDO.
How about this:
SELECT a.file_url_thumb
FROM virtuemart_medias a
LEFT JOIN virtuemart_product_medias b on a.virtuemart_media_id=b.irtuemart_media_id
WHERE a.virtuemart_product_id=' . $id . ' LIMIT 1

mySQL: get hash value for each row?

Currently I'm manually creating a string where I concatenate all the values in each row in my table. I'm hashing this string for each row to get a hash value for the current values (/status) of the row, which I'm later is using to determine if the row has changed.
Instead of doing this manually, is there an build-in way i mySQL to get a unique hash value for each row?
you could do something like
SELECT MD5(concat(field1, field2, field3, ...)) AS rowhash
but you can't get away from listing which fields you want, as concat(*) is not an option (syntax error).
It's better to use concat_ws(). e.g. two adjacent column: 12,3 => 1,23 .
Sorry, this still has some problems. Think about the null value, empty string, string can contain ',', etc...
A program is required to generate the hash statement, which should replace null to specific value (for null-able columns), and also use the seldom used char/byte as separator.
There are problems with CONCAT, e.g. CONCAT('ab', 'c') vs CONCAT('a', 'bc'). Two different rows, but result is the same. You could use CONCAT_WS(';', 'ab', 'c') to get ab;c but in case of CONCAT_WS(';', ';', '') vs CONCAT_WS(';', '', ';') you still get the same result.
Also CONCAT(NULL, 'c') returns NULL.
I think the best way is to use QUOTE:
SELECT MD5(CONCAT(QUOTE(c1), QUOTE(c2), QUOTE(c3))) AS row_hash FROM t1;
Result of: select (concat(quote('a'), quote('bc'), quote('NULL'), quote(NULL), quote('\''), quote('')));
is: 'a''bc''NULL'NULL'\''''
Also, don't use GROUP_CONCAT() to get hash of table, it has limit: https://dev.mysql.com/doc/refman/8.0/en/server-system-variables.html#sysvar_group_concat_max_len
Instead, CHECKSUM TABLE might be better, but you can't skip columns with CHECKSUM TABLE https://dev.mysql.com/doc/refman/5.7/en/checksum-table.html
Well I made a little script that could do excactly what you want, and maybe what others want... so here it goes...for PHP that is...
first you have to make a list of columns of the table, then you make a "case when" statement for each column based on their type and put that in the concat_ws statement and then you hash it with sha1...i've used this method on very large tables (600000+ records) and the speed is quite good when selecting all records. also I think that it is faster to concat the required data in a concat_ws and explode it in php or whatever you are using, but that is just a hunch...
<?
$query= mysql_query("SHOW COLUMNS FROM $table", $linklive);
while ($col = mysql_fetch_assoc($query)) {
$columns[] = mysql_real_escape_string($col['Field']);
if ($col['Key'] == 'PRI') {
$key = mysql_real_escape_string($col['Field']);
}
$columnsinfo[$col['Field']] = $col;
}
$dates = array("date","datetime","time");
$int = array("int","decimal");
$implcols = array();
foreach($columns as $col){
if(in_array($columnsinfo[$col]['Type'], $dates)){
$implcols[] = "(CASE WHEN (UNIX_TIMESTAMP(`$col`)=0 || `$col` IS NULL) THEN '[$col EMPTY]' ELSE `$col` END)";
}else{
list($type, $rest) = explode("(",$columnsinfo[$col]['Type']);
if(in_array($columnsinfo[$col]['Type'], $dates)){
$implcols[] = "(CASE WHEN ( `$col`=0 || `$col` IS NULL ) THEN '[$col EMPTY]' ELSE `$col` END)";
}else{
$implcols[] = "(CASE WHEN ( `$col`='' || `$col` IS NULL ) THEN '[$col EMPTY]' ELSE `$col` END)";
}
}
}
$keyslive = array();
//echo "SELECT $key SHA1(CONCAT_WS('',".implode(",", $columns).")) as compare FROM $table"; exit;
$q = "SELECT $key as `key`, SHA1(CONCAT_WS('',".implode(", ",$implcols).")) as compare FROM $table";
?>

Why doesn't this SELECT query return the results I expect?

I need help with a select query, but before asking the question, I will give a short description of how my system works:
My database has a many-to-many relationship:
table product:
prd_cod(pk) //stores the product code ex: 0,1,2
cat_cod(fk)
prd_name //stores the product name, ex: tv, gps, notebook
table description_characteristc:
prd_cod(fk)
id_characteristic(fk)
description //stores the description of the characteristic, ex: sony, 1kg, hj10
table characteristic:
id_characteristic (pk)
name_characteristic //store the name of characteristic, ex: brand, weight, model
I have already made a suggest jQuery (in the index.php), where every word I type calls suggest.php, which makes a select and returns the result into the suggestion box in the index:
<?php
header('Content-type: text/html; charset=UTF-8');
$hostname = 'localhost';
$username = 'root';
$password = '';
$dbname = 'cpd';
mysql_connect($hostname, $username, $password)or die('Erro ao tentar conecta o banco
de dados.');
mysql_select_db( $dbname );
if( isset( $_REQUEST['query'] ) && $_REQUEST['query'] != "" )
{
$q = mysql_real_escape_string( $_REQUEST['query'] );
if( isset( $_REQUEST['identifier'] ) && $_REQUEST['identifier'] == "sugestao")
{
$sql = "SELECT p.prd_name, d.description
FROM product p
INNER JOIN description_characteristc d using (prd_cod)
WHERE '".$q."' like concat(p.prd_name, '%') AND
concat(p.prd_name, ' ', d.description) like concat('".$q."', '%')LIMIT 10";
$r = mysql_query( $sql );
if ( $r )
{
echo '<ul>'."\n";
$cont = 0;
while( $l = mysql_fetch_array( $r ) ){
$p = $l['nome'];
$p = preg_replace('/(' . $q . ')/i', '<span style="font-
weight:bold;">$1</span>',
$l['prd_nome'].' '.$l['descricao'].' '.$l['descricao']);
echo "\t".'<li id="autocomplete_'.$cont.'"
rel="'.$l['prd_nome'].'.'.$l['descricao'].'">'. utf8_encode( $p ) .'</li>'."\n";
$cont++;
}
echo '</ul>';
}
}
}
?>
Here are my questions:
Currently when the user types 't', the select brings nothing, only when the user type 'tv' is bringing the result:
tv led
tv plasm
tv samsumg
I would like that when the user type 't' the select bring me 'tv'.
When you type 'tv plasm' it's bringing the same name_characteristic twice:
ex: tv plasm plasm
Currently my select selects the prd_name and the descriptions of table description_characteristc:
tv led
I would like my select could make a inverse select too, ex: led tv.
I would like that when the results of the select were shown, there could be a cache feature that shows the order of the most sought for the less sought; remembering that prd_name stores only 'tv'.
The help I'm looking for can be in the form of select, as in the form of procedure. Also, I can edit the php file.
You should split and join your search query on PHP side like this:
<?php
$words = preg_split("/[^\\w]+/", $q);
$first = $words[0] + "%";
$all = implode(" ", $words) + "%";
?>
then use the variables $first and $all in this query:
SELECT p.prd_name, d.description
FROM product p
JOIN description d
ON d.prd_cod = p.prd_cod
WHERE p.prd_name LIKE '$first'
AND CONCAT(p.prd_name, ' ', d.description) LIKE '$all'
Create an index on product (prd_name) for this to work fast.
If you want the words matched in any order, you will have to create a FULLTEXT index on your tables (this is only possible in MyISAM):
CREATE FULLTEXT INDEX fx_product_name ON product (prd_name)
CREATE FULLTEXT INDEX fx_description_name ON description (description)
and write a query like this:
SELECT p.prd_name, d.description
FROM (
SELECT prd_cod
FROM product pi
WHERE MATCH(prd_name) AGAINST ('lcd tv' IN BOOLEAN MODE)
UNION
SELECT prd_cod
FROM description di
WHERE MATCH(description) AGAINST ('lcd tv' IN BOOLEAN MODE)
) q
JOIN product p
ON p.prd_cod = q.prd_cod
JOIN description d
ON d.prd_cod= p.prd_cod
WHERE MATCH(p.prd_name, d.description) AGAINST ('+lcd +tv' IN BOOLEAN MODE)
Note the search term syntax change: 'lcd tv' in the inner query and '+lcd +tv' in the outer one.
You may also want to set ##ft_min_word_len to 1 for the shorter words like tv or gps to match.
Since MySQL cannot build a fulltext index from two or more tables at once, it would be more simple if you denormalized you tables and put the prd_name into the description table. This way, you could get rid of the joins and just write:
SELECT prd_name, description
FROM description d
WHERE MATCH(prd_name, description) AGAINST ('+lcd +tv' IN BOOLEAN MODE)
You're using the LIKE clause badly and you don't seem to know what "AND" means. It's important to separate "and" as used in casual speech from "AND" as used in programming. AND in programming means "BOTH MUST BE TRUE". "and" in casual speech can mean "one of these conditions, you know what I mean?"
Also, you shouldn't be building SQL like this, it's an accident waiting to happen. You really should find a way to bind variables into SQL statements. I don't know PHP, so I can't help with that.
First, you should be using this in your WHERE clause p.prd_name LIKE '$q%'. Try this outside PHP -- outside the web -- just as a simple SQL query: SELECT * FROM PRODUCT P WHERE P.PRD_NAME LIKE 'T%'.
Second, you should fix "AND" to be "OR", since you want one condition OR the other condition to be true. If you want for BOTH conditions to be true, hardly anything will match.