I am trying to generate and execute the following sql through CActiveRecord:
SELECT * FROM `bucket` `t`
WHERE bkt_user = unhex('A4FF2131E00C4696837689FCAAAC7DD2');
I came up with this:
$uuid = 'A4FF2131E00C4696837689FCAAAC7DD2';
$criteria = new CDbCriteria();
$expression = new CDbExpression(
'unhex(:value)',
array(':value'=>$uuid,));
$criteria->addCondition("bkt_user = :exp");
$criteria->params = array(':exp' => $expression);
$buckets = Bucket::model()->findAll($criteria);
The code executes without error or exception but does not return results as expected. A little debugging revealed that the snippet above generates the following sql:
SELECT * FROM `xpg_bucket` `t` WHERE bkt_user = :exp
and AFAIK only one CDbCommandBuilder::bindValue where :exp is bound to unhex(:value). There is no parameter binding done for :value. This is happening in CDbCommandBuilder::createFindCommand. CDbCommandBuilder::createInsertCommand, on the other hand seems to take care of a value itself being a CDbExpression and generated the appropriate bindValue's.
CDbExpression's documentation contains:
* CDbExpression is mainly used in {#link CActiveRecord} as attribute values.
* When inserting or updating a {#link CActiveRecord}, attribute values of
* type CDbExpression will be directly put into the corresponding SQL statement
* without escaping.
Does this mean CDbExpression is not meant to be used with CDbCriteria in findAll()? If yes, what is the alternative?
Actually, it turns out all that circus is not required. The following works with findAll:
$criteria->addCondition('bkt_user = unhex(:value)');
$criteria->params = array(':value'=>'665730BDEDA7489383E2519DB5DE6D60');
$buckets = Bucket::model()->findAll($criteria);
Be aware that the same will not work with add or update operations.
Related
I know this may sound silly but I've been stuck on this problem for too long!
I'm querying a PostgreSQL repository through JPA using native SQL queries. One of my queries looks like this:
#Query(value = "select * from gs where ?1 = ?2", nativeQuery = true)
public List<GsJsonStore> matchJson(String term, String value);
I'm testing the function using :
List<GsJsonStore> list = repo.matchJson("subject", "'Sub'");
The list is empty on running the query, however when I run the same query through PSQL command line using:
select * from gs where subject = 'Sub';
I get the correct output, records contatining the key-value pair are returned.
Where am I making the mistake?
You can't use parameter for column name. Your query resolves to
select * from gs where 'subject' = '''Sub'''
EDIT: just saw #pozs already posted the same in comment
my problem is about this kind of query :
select * from SOMETABLE where SOMEFIELD in ('STRING1','STRING2');
the previous code works fine within Sql Developer.
The same static query also works fine and returns me a few results;
Query nativeQuery = em.createNativeQuery(thePreviousQuery,new someResultSet());
return nativeQuery.getResultList();
But when I try to parameterize this, I encounter a problem.
final String parameterizedQuery = "select * from SOMETABLE where SOMEFIELD in (?selectedValues)";
Query nativeQuery = em.createNativeQuery(parameterizedQuery ,new someResultSet());
nativeQuery.setParameter("selectedValues","'STRING1','STRING2'");
return nativeQuery.getResultList();
I got no result (but no error in console).
And when I look at the log, I see such a thing :
select * from SOMETABLE where SOMEFIELD in (?)
bind => [STRING1,STRING2]
I also tried to use no quotes (with similar result), or non ordered parameter (:selectedValues), which leads to such an error :
SQL Error: Missing IN or OUT parameter at index:: 1
I enventually tried to had the parentheses set directly in the parameter, instead of the query, but this didn't work either...
I could build my query at runtime, to match the first (working) case, but I'd rather do it the proper way; thus, if anyone has an idea, I'll read them with great interest!
FYI :
JPA version 1.0
Oracle 11G
JPA support the use of a collection as a list literal parameter only in JPQL queries, not in native queries. Some JPA providers support it as a proprietary feature, but it's not part of the JPA specification (see https://stackoverflow.com/a/3145275/1285097).
Named parameters in native queries also aren't part of the JPA specification. Their behavior depends on the persistence provider and/or the JDBC driver.
Hibernate with the JDBC driver for Oracle support both of these features.
List<String> selectedValues = Arrays.asList("STRING1", "STRING2");
final String parameterizedQuery = "select * from SOMETABLE where SOMEFIELD in (:selectedValues)";
return em.createNativeQuery(parameterizedQuery)
.setParameter("selectedValues", selectedValues)
.getResultList();
Instead of:
nativeQuery.setParameter("selectedValues", params);
I had to use:
nativeQuery.setParameterList("selectedValues", params);
This worked for me in derby. parameter without "()".
List<String> selectedValues = Arrays.asList("STRING1", "STRING2");
final String parameterizedQuery = "select * from SOMETABLE where SOMEFIELD in
:selectedValues";
return em.createNativeQuery(parameterizedQuery)
.setParameter("selectedValues", selectedValues)
.getResultList();
Replace this:
nativeQuery.setParameter("selectedValues","'STRING1','STRING2'");
with
List<String> params;
nativeQuery.setParameter("selectedValues",params);
I also faced the same issue.
This is what I did:
List<String> sample = new ArrayList<String>();
sample.add("sample1");
sample.add("sample2");
And now you, can set the sample in params.
I'm not a JPA persistence criteria API guru and sometimes I get very terrible headaches using it.
Yesterday I noticed a new very weird behaviour. The code I will post is an adaptation of existing functioning code, so don't focus on trivial errors. I'm using glassfish 3.1.1 and the corresponding eclipse persistence plugin and Mysql DB.
I have written a criteriaQuery which filters data from different tables. If this criteriaquery is executed twice the second time it generates wrong SQL query. I cannot figure out why.
public CriteriaQuery createQuery4Count(EntityManager em) {
Calendar lastDate4Search = GregorianCalendar.getInstance();
javax.persistence.criteria.CriteriaBuilder cb = em.getCriteriaBuilder();
javax.persistence.criteria.CriteriaQuery cq = cb.createQuery();
javax.persistence.criteria.Root<Permessimercepath> checkPointRt = cq.from(Permessimercepath.class);
javax.persistence.criteria.Path<Permessimerce> permessimerceClass = checkPointRt.get(Permessimercepath_.permessimerce);
Predicate checkPointDatePredicate = cb.isNull(checkPointRt.get(Permessimercepath_.dataTransito));
Predicate checkPointAreaPredicate = cb.equal(checkPointRt.get(Permessimercepath_.iDArea), area);
Predicate datePredicate = cb.greaterThanOrEqualTo(permessimerceClass.get(Permessimerce_.datafine), lastDate4Search.getTime());
Predicate isValidPredicate = cb.lt(permessimerceClass.get(Permessimerce_.statopermesso), Permessimerce.COMPLETED);
cq.where(cb.and(checkPointAreaPredicate, checkPointDatePredicate, datePredicate, isValidPredicate));
cq.select(cb.countDistinct(checkPointRt));
return cq;
}
CriteriaQuery myCriteriaQuery = createQuery4Count(getEntityManager())
javax.persistence.Query q = getEntityManager().createQuery(myCriteriaQuery );
Long Result = ((Long) q.getSingleResult()).intValue();
// second query created with the same criteriaQuery
q = getEntityManager().createQuery(myCriteriaQuery );
Long Result2 = ((Long) q.getSingleResult()).intValue();
The generated sql is
// First and correct one
SELECT COUNT(t0.ID_permesso) FROM permessimercepath t0 WHERE EXISTS (SELECT t1.ID_permesso FROM permessimerce t2, permessimercepath t1 WHERE ((((t0.ID_permesso = t1.ID_permesso) AND (t0.CheckPointIndex = t1.CheckPointIndex)) AND ((((t1.ID_Area = ?) AND (t1.DataTransito IS NULL)) AND (t2.Data_fine >= ?)) AND (t2.Stato_permesso < ?))) AND (t2.ID_permesso = t1.ID_permesso)))
bind => [3 parameters bound]
// Second and wrong one
SELECT COUNT(t0.ID_permesso) FROM permessimercepath t0, permessimerce t2, permessimercepath t1 WHERE (((((t1.ID_Area = ?) AND (t1.DataTransito IS NULL)) AND (t2.Data_fine >= ?)) AND (t2.Stato_permesso < ?)) AND (t2.ID_permesso = t1.ID_permesso))
If nobody has an idea on why it happens I can try to reproduce it in a simpler way.
Thanks
Filippo
The problem seems to related with EntityManager.createQuery(CriteriaQuery) method. I checked the documentation, but nothing is mentioned about whether this method modifies CriteriaQuery or not. Normally one expects this method to not modify the passed parameter. However, what you are getting suggests that EntityManager.createQuery(CriteriaQuery) method modifies the passed parameter.
If this is the case you will need to call createQuery4Count(getEntityManager()) before calling EntityManager.createQuery(CriteriaQuery) in your code each time.
It looks likely that you spotted bug in implementation. Slightly similar (or so to say more complex but different) case works fine with Hibernate. Also, section "6.8 Query Modification" from JPA Specification encourages reuse of CriteriaQuery:
A CriteriaQuery object may be modified, either before or after
TypedQuery objects have been created and executed from it. For
example, such modification may entail replacement of the where
predicate or the select list. Modifications may thus result in the
same CriteriaQuery “base” being reused for several query instances.
I have some extremely complex queries that I need to use to generate a report in my application. I'm using symfony as my framework and doctrine as my ORM.
My question is this:
What is the best way to pass in highly-complex sql queries directly to Doctrine without converting them to the Doctrine Query Language? I've been reading about the Raw_SQL extension but it appears that you still need to pass the query in sections (like from()). Is there anything for just dumping in a bunch of raw sql commands?
$q = Doctrine_Manager::getInstance()->getCurrentConnection();
$result = $q->execute(" -- RAW SQL HERE -- ");
See the Doctrine API documentation for different execution methods.
Yes. You can get a database handle from Doctrine using the following code:
$pdo = Doctrine_Manager::getInstance()->getCurrentConnection()->getDbh();
and then execute your SQL as follows:
$query = "SELECT * FROM table WHERE param1 = :param1 AND param2 = :param2";
$stmt = $pdo->prepare($query);
$params = array(
"param1" => "value1",
"param2" => "value2"
);
$stmt->execute($params);
$results = $stmt->fetchAll();
You can use bound variables as in the above example.
Note that Doctrine won't automatically hydrate your results nicely into record objects etc, so you'll need to deal with the results being returned as an array, consisting of one array per row returned (key-value as column-value).
I'm not sure what do you mean saying raw SQL, but you coud execute traditional SQL queries this way:
...
// $this->_displayPortabilityWarning();
$conn = Doctrine_Manager::connection();
$pdo = $conn->execute($sql);
$pdo->setFetchMode(Doctrine_Core::FETCH_ASSOC);
$result = $pdo->fetchAll();
...
The following method is not necsessary, but it shows a good practice.
protected function _displayPortabilityWarning($engine = 'pgsql')
{
$conn = Doctrine_Manager::connection();
$driver = $conn->getDriverName();
if (strtolower($engine) != strtolower($driver)) {
trigger_error('Here we have possible database portability issue. This code was tested on ' . $engine . ' but you are trying to run it on ' . $driver, E_USER_NOTICE);
}
}
You can also use Doctrine_RawSql(); to create raw SQL queries which will hydrate to doctrine objects.
It should be noted, that Doctrine2 uses PDO as a base, thus I would recommend using prepared statements over plain old execute.
Example:
$db = Doctrine_Manager::getInstance()->getCurrentConnection();
$query = $db->prepare("SELECT `someField` FROM `someTable` WHERE `field` = :value");
$query->execute(array('value' => 'someValue'));
Symfony insert raw sql using doctrine.
This in version Symfoney 1.3
$q = Doctrine_Manager::getInstance()->getCurrentConnection();
$result = $q->execute($query);
Setting the DBIC_TRACE environment variable to true:
BEGIN { $ENV{DBIC_TRACE} = 1 }
generates very helpful output, especially showing the SQL query that is being executed, but the SQL query is all on one line.
Is there a way to push it through some kinda "sql tidy" routine to format it better, perhaps breaking it up over multiple lines? Failing that, could anyone give me a nudge into where in the code I'd need to hack to add such a hook? And what the best tool is to accept a badly formatted SQL query and push out a nicely formatted one?
"nice formatting" in this context simply means better than "all on one line". I'm not particularly fussed about specific styles of formatting queries
Thanks!
As of DBIx::Class 0.08124 it's built in.
Just set $ENV{DBIC_TRACE_PROFILE} to console or console_monochrome.
From the documentation of DBIx::Class::Storage
If DBIC_TRACE is set then trace information is produced (as when the
debug method is set). ...
debug Causes trace information to be emitted on the debugobj
object. (or STDERR if debugobj has not specifically been set).
debugobj Sets or retrieves the object used for metric collection.
Defaults to an instance of DBIx::Class::Storage::Statistics that is
compatible with the original method of using a coderef as a callback.
See the aforementioned Statistics class for more information.
In other words, you should set debugobj in that class to an object that subclasses DBIx::Class::Storage::Statistics. In your subclass, you can reformat the query the way you want it to be.
First, thanks for the pointers! Partial answer follows ....
What I've got so far ... first some scaffolding:
# Connect to our db through DBIx::Class
my $schema = My::Schema->connect('dbi:SQLite:/home/me/accounts.db');
# See also BEGIN { $ENV{DBIC_TRACE} = 1 }
$schema->storage->debug(1);
# Create an instance of our subclassed (see below)
# DBIx::Class::Storage::Statistics class
my $stats = My::DBIx::Class::Storage::Statistics->new();
# Set the debugobj object on our schema's storage
$schema->storage->debugobj($stats);
And the definition of My::DBIx::Class::Storage::Statistics being:
package My::DBIx::Class::Storage::Statistics;
use base qw<DBIx::Class::Storage::Statistics>;
use Data::Dumper qw<Dumper>;
use SQL::Statement;
use SQL::Parser;
sub query_start {
my ($self, $sql_query, #params) = #_;
print "The original sql query is\n$sql_query\n\n";
my $parser = SQL::Parser->new();
my $stmt = SQL::Statement->new($sql_query, $parser);
#printf "%s\n", $stmt->command;
print "The parameters for this query are:";
print Dumper \#params;
}
Which solves the problem about how to hook in to get the SQL query for me to "pretty-ify".
Then I run a query:
my $rs = $schema->resultset('SomeTable')->search(
{
'email' => $email,
'others.some_col' => 1,
},
{ join => 'others' }
);
$rs->count;
However SQL::Parser barfs on the SQL generated by DBIx::Class:
The original sql query is
SELECT COUNT( * ) FROM some_table me LEFT JOIN others other_table ON ( others.some_col_id = me.id ) WHERE ( others.some_col_id = ? AND email = ? )
SQL ERROR: Bad table or column name '(others' has chars not alphanumeric or underscore!
SQL ERROR: No equijoin condition in WHERE or ON clause
So ... is there a better parser than SQL::Parser for the job?