Determining which field causes Doctrine to re-query the database - php

I'm using Doctrine with Symfony in a couple of web app projects.
I've optimised many of the queries in these projects to select just the fields needed from the database. But over time new features have been added and - in a couple of cases - additional fields are used in the code, causing the Doctrine lazy loader to re-query the database and driving the number of queries on some pages from 3 to 100+
So I need to update the original query to include all of the required fields. However, there doesn't seem an easy way for Doctrine to log which field causes the additional query to be issued - so it becomes a painstaking job to sift through the code looking for usage of fields which aren't in the original query.
Is there a way to have Doctrine log when a getter accesses a field that hasn't been hydrated?

I have not had this issue, but just looked at Doctrine_Record class. Have you tried adding some debug output to the _get() method? I think this part is where you should look for a solution:
if (array_key_exists($fieldName, $this->_data)) {
// check if the value is the Doctrine_Null object located in self::$_null)
if ($this->_data[$fieldName] === self::$_null && $load) {
$this->load();
}

Just turn on SQL logging and you can deduce the guilty one from alias names. For how to do it in Doctrine 1.2 see this post.
Basically: create a class which extends Doctrine_EventListener:
class QueryDebuggerListener extends Doctrine_EventListener
{
protected $queries;
public function preStmtExecute(Doctrine_Event $event)
{
$query = $event->getQuery();
$params = $event->getParams();
//the below makes some naive assumptions about the queries being logged
while (sizeof($params) > 0) {
$param = array_shift($params);
if (!is_numeric($param)) {
$param = sprintf("'%s'", $param);
}
$query = substr_replace($query, $param, strpos($query, '?'), 1);
}
$this->queries[] = $query;
}
public function getQueries()
{
return $this->queries;
}
}
And add the event listener:
$c = Doctrine_Manager::connection($conn);
$queryDbg = new QueryDebuggerListener();
$c->addListener($queryDbg);

Related

Laravel Eloquent - Model extends other model

I have a question about extending my own Models eloquent.
In the project I am currently working on is table called modules and it contains list of project modules, number of elements of that module, add date etc.
For example:
id = 1; name = 'users'; count = 120; date_add = '2007-05-05';
and this entity called users corresponds to model User (Table - users) so that "count" it's number of Users
and to update count we use script running every day (I know that it's not good way but... u know).
In that script is loop and inside that loop a lot of if statement (1 per module) and inside the if a single query with count. According to example it's similar to:
foreach($modules as $module) {
if($module['name'] == 'users') {
$count = old_and_bad_method_to_count('users', "state = 'on'");
}
}
function old_and_bad_method_to_count($table, $sql_cond) {}
So its look terrible.
I need to refactor that code a little bit, because it's use a dangerous function instead of Query/Builder or Eloquent/Model and looks bad.
I came up with an idea that I will use a Models and create Interface ElementsCountable and all models that do not have an interface will use the Model::all()->count(), and those with an interface will use the interface method:
foreach ($modules as $module) {
$className = $module->getModelName();
if($className) {
$modelInterfaces = class_implements($className);
if(isset($modelInterfaces[ElementsCountable::class])) {
/** #var ElementsCountable $className */
$count = $className::countModuleElements();
} else {
/** #var Model $className */
$count = $className::all()->count();
}
}
}
in method getModelName() i use a const map array (table -> model) which I created, because a lot of models have custom table name.
But then I realize that will be a good way, but there is a few records in Modules that use the same table, for example users_off which use the same table as users, but use other condition - state = 'off'
So it complicated things a little bit, and there is a right question: There is a good way to extends User and add scope with condition on boot?
class UserOff extends User
{
protected static function boot()
{
parent::boot();
static::addGlobalScope(function (Builder $builder) {
$builder->where('state', '=', 'off');
});
}
}
Because I have some concerns if this is a good solution. Because all method of that class NEED always that scope and how to prevent from method withoutGlobalScope() and what about other complications?
I think it's a good solution to create the UserOff model with the additional global scope for this purpose.
I also think the solution I would want to implement would allow me to do something like
$count = $modules->sum(function ($module) {
$className = $module->getModelName();
return $className::modulesCount();
}
I would create an interface ModulesCountable that mandates a modulesCount() method on each of the models. The modulesCount() method would return either the default count or whatever current implementation you have in countModuleElements().
If there are a lot of models I would probably use a trait DefaultModulesCount for the default count, and maybe the custom version too eg. ElementsModuleCount if that is consistent.

Implementing a S.O.L.I.D Domain Object Model in the following project

I have the following example in which I tend to use a couple of classes, to create a simple web app.
The file hierarchy seems like this.
> cupid
- libs
- request
- router
- database
- view
- bootstrap.php
- index.php
The index.php just calls the bootstrap.php which in turn contains something like this:
// bootstrap.php
namespace cupid
use request, router, database, view;
spl_autoload_register(function($class){ /* autoload */ });
$request = new view;
$response = new response;
$router = new router;
$database = new database;
$router->get('/blog/{id}', function($id) use ($database, $view) {
$article = $database->select("SELECT blog, content FROM foo WHERE id = ?",[$id]);
$view->layout('blogPage', ['article'=>$article]);
});
As you can probably tell, my problem is this line:
$article = $database->select("SELECT blog, content FROM foo WHERE id = ?", [$id]);
Which I don't want to use, and instead try a " Domain Object Model " approach.
Now, given that I will add another folder called domain, with blog.php
> cupid
- domain
- Blog.php
- libs
...
And fill blog.php with properties mapping table rows, and getter and setters ..
namespace App\Domain;
class Blog {
private $id, $title, $content, $author;
public function getTitle(){
return $this->title;
}
public function setTitle($title){
$this->title = $title;
}
...
}
My question is: Assuming my understanding of DOM is so far correct, and that I have a CRUD/ORM class, or a PDO wrapper to query the database;
"How can I tie together, i.e. the blog model with the PDO wrapper to fetch a blog inside my bootstrap file?"..
As far as a Domain Object you basically already have written one, your blog object. To qualify as a domain model all a class must to is to provide a representation along with any of the functionality of a concept within your problem space.
The more interesting problem here and the one you appear to be struggling with is how to persist a domain model. Keeping with the tenet of the single responsibility principle your Blog class should deal with being a blog post and doing the things that a blog post can do, not storing one. For that you would introduce the concept of a repository of blog posts that would deal with storing and retrieving objects of this type. Below is a simple implementation of how this can be done.
class BlogRepository {
public function __construct(\cupid\database $db){
$this->db = $db;
}
public function findById($id){
$blogData = $this->db->select("select * from blog where id = ?", [$id]);
if ($blogData){
return $this->createBlogFromArray($blogData);
}
return null;
}
public function findAllByTag($tag){...}
public function save(Blog $blog) {...}
private function createBlogFromArray(array $array){
$blog = new Blog();
$blog->setId($blogData["id"]);
$blog->setTitle($blogData["title"]);
$blog->setContent($blogData["content"]);
$blog->setAuthor($blogData["author"]);
return $blog;
}
}
Then your controller should look something like this.
$router->get('/blog/{id}', function($id) use ($blogRepository, $view) {
$article = $blogRepository->findById($id);
if ($article) {
$view->layout('blogPage', ['article'=>$article]);
} else {
$view->setError("404");
}
});
To truly be SOLID the above class should be a database specific implementation of a BlogRepository interface to adhere to IoC. A factory should also probably be supplied to BlogRepository to actually create the blog objects from data retrieved from the store.
In my opinion one of the great benefits of doing this is you have a single place where you can implement and maintain all of your blog related interactions with the database.
Other Advantages to this method
Implementing caching for your domain objects would be trivial
Switching to a different data source (from flat files, blogger api, Document Database Server,PostgresSQL etc.) could be done easily.
You can alternatively use a type aware ORM for a more general solution to this same problem. Basically this Repository class is nothing more than a ORM for a single class.
The important thing here is that you are not talking directly to the database and leaving sql scattered throughout your code. This creates a maintenance nightmare and couples your code to the schema of your database.
Personally I always tend to stick the database operations in a database class which does all the heavy lifting of initialising the class, opening the connection etc. It also has generic query-wrappers to which I pass the SQL-statements which contains the normal placeholders for the bound variables, plus an array of the variables to be bound (or the variable number of parameters approach if thats suits you better). If you want to bind each param individually and not use the $stmt->execute(array()); You just pass in the types with the value in a data structure of your choosing, multi dim array, dictionary, JSON, whatever suits your needs and you find easy to work with.
The model class it self (Blog in your case) then subclasses the Database. Then you have a few choices to make. Do you want to use the constructor to create only new objects? Do you want it to only load based on IDs? Or a mix of both? Something like:
function __construct(id = null, title = null, ingress = null, body = null) {
if(id){
$row = $this->getRow("SELECT * FROM blog WHERE id = :id",id); // Get a single row from the result
$this->title = $row->title;
$this->ingress = $row->ingress;
$this->body = $row->body;
... etc
} else if(!empty(title,ingress,body)){
$this->title = title;
... etc
}
}
Maybe neither? You can skip the constructor and use the new(title, ingress, body), save() and a load(id) methods if thats your preference.
Of course, the query part can be generalised even further if you just configure some class members and let the Database-superclass do the query building based on what you send in or set as member-variables. For example:
class Database {
$columns = []; // Array for storing the column names, could also be a dictionary that also stores the values
$idcolumn = "id"; // Generic id column name typically used, can be overridden in subclass
...
// Function for loading the object in a generic way based on configured data
function load($id){
if(!$this->db) $this->connect(); // Make sure we are connected
$query = "SELECT "; // Init the query to base string
foreach($this->columns as $column){
if($query !== "SELECT ") $query .= ", "; // See if we need comma before column name
$query .= $column; // Add column name to query
}
$query .= " FROM " . $this->tablename . " WHERE " . $this->idcolumn . " = :" . $this->idcolumn . ";";
$arg = ["col"=>$this->idcolumn,"value"=>$id,"type"=>PDO::PARAM_INT];
$row = $this->getRow($query,[$arg]); // Do the query and get the row pass in the type of the variable along with the variable, in this case an integer based ID
foreach($row as $column => $value){
$this->$column = $value; // Assign the values from $row to $this
}
}
...
function getRow($query,$args){
$statement = $this->query($query,$args); // Use the main generic query to return the result as a PDOStatement
$result = $statement->fetch(); // Get the first row
return $result;
}
...
function query($query,$args){
...
$stmt = $this->db->prepare($query);
foreach($args as $arg){
$stmt->bindParam(":".$arg["col"],$arg["value"],$arg["type"]);
}
$stmt->execute();
return $stmt;
}
...
}
Now as you see the load($id), getrow($query,$args) and query($query,$args) is completely generic. ´getrow()´is just a wrapper on query() that gets the first row, you may want to have several different wrappers that to or interpret your statement result in different ways. You may also even want to add object specific wrappers to your models if they cannot be made generic. Now the model, in your case Blog could look like:
class Blog extends Database {
$title;
$ingress;
$body;
...
function __construct($id = null){
$this->columns = ["title","ingress","body","id",...];
$this->idcolumn = "articleid"; // override parent id name
...
if($id) $this->load($id);
}
...
}
Use it as so: $blog = new Blog(123); to load a specific blog, or $blog = new Blog(); $blog->title = "title"; ... $blog->save(); if you want a new.
"How can I tie together, i.e. the blog model with the PDO wrapper to fetch a blog inside my bootstrap file?"..
To tie the two together, you could use an object-relational mapper (ORM). ORM libraries are built just for glueing your PHP classes to database rows. There are a couple of ORM libraries for PHP around. Also, most ORMs have a built in database abstraction layer, which means that you can simply switch the database vendor without any hassle.
Considerations when using an ORM:
While introducing a ORM also introduces some bloat (and some learning), it may not be worthwhile investing the time for simply a single Blog object. Although, if your blog entries also have an author, one or multiple categories and/or associated files, an ORM may soon help you reading/writing the database. Judging from your posted code, an ORM will pay off when extending the application in the future.
Update: Example using Doctrine 2
You may have a look at the querying section of the official Doctrine documentation to see the different options you have for read access. Reconsider the example you gave:
// current implementation
$article = $database->select("SELECT blog, content FROM foo WHERE id = ?",[$id]);
// possible implementation using Doctrine
$article = $em->getRepository(Blog::class)->find($id);
However, ideally you define your own repository to separate your business logic from Doctrines API like the following example illustrates:
use Doctrine\ORM\EntityRepository;
interface BlogRepositoryInterface {
public function findById($id);
public function findByAuthor($author);
}
class BlogRepsitory implements BlogRepositoryInterface {
/** #var EntityRepository */
private $repo;
public function __construct(EntityRepository $repo) {
$this->repo = $repo;
}
public function findById($id) {
return $this->repo->find($id);
}
public function findByAuthor($author) {
return $this->repo->findBy(['author' => $author]);
}
}
I hope the example illustrates how easily you can separate your business domain models and logic from the underlying library and how powerful ORMs can come into play.

Propel peer static classes: how to avoid to write duplicated code?

I have a few tables configured in Propel, with generated Peer static classes.
My problem is that I need to perform the same search operation on different but similar tables. Those tables have different Peer classes as it is how Propel works. This situation leads to duplicated code regarding to the queries that are performed on these tables.
I was wondering if there is some construct (avoiding the use of the function eval) that might help me in this case; I really would like to avoid writing duplicated code that performs the same exact calls on just different static Peer classes.
example code snippet from a (very long) method of a class I am writing:
$criteria = new Criteria();
$criteria->add(FoobarPeer::CONTRACTNR,$data['contractnr']);
$result = FoobarPeer::doSelect($criteria);
if(count($result) > 1){
throw new FoobarException("status: more than one row with the specified contractnr.");
}
if(count($result) == 0){
// no object with given contractnr. Create new one.
$obj = $this->factory->createORM("foobar");
$obj->setCreatedAt(time());
} else {
// use and update existing object.
$obj = $result[0];
}
As you can see I managed to write a factory method for the row object, but I could not find a way to do the same for static classes. In other words, I would like to have the access to the static classes dynamic and not in a way that is an ugly workaround.
Any ideas?
thanks :)
I'm not really sure I fully understand what you are asking, but here's a solution to what I think you are asking:
function orm_for($type) {
return strtolower($type);
}
function peer_for($type) {
return ucfirst($type)."Peer";
}
function exception_for($type) {
return ucfirst($type)."Exception";
}
function query($type, $data) {
$peer = $peer_for($type);
$exception = $exception_for($type);
$obj = null;
$criteria = new Criteria();
$criteria->add($peer::CONTRACTNR, $data["contractnr"]);
$result = $peer::doSelect($criteria);
if(count($result) > 1) {
throw new $exception("status: more than one row with the specified contractnr.");
} else if(count($result) == 0) {
$obj = $this->factory->createORM(orm_for($type));
$obj->setCreatedAt(time());
} else {
$obj = $result[0];
}
}
I think the code is self-explanatory. Let me know whether or not I interpreted your question correctly.
A live example (just a POC) can be found here
You should be able to use behaviors to achieve what you're trying to do. You can use behaviors to add custom code to the generated peer objects. See here.
Among other things, your behaviors can implement the following methods:
staticAttributes() // add static attributes to the peer class
staticMethods() // add static methods to the peer class
You should be able to use these to add the code you want to the peers. You only need to worry about writing the code once. Propel will duplicate the code during the code generation process, but this shouldn't be too much of a concern, as a lot of the generated is duplicated anyway. At least the duplication is only introduced by an automated process.

Advice extending PDO with basic CRUD functions

Just recently I started rewriting a previously procedurally written website by myself, I chose PDO as the wrapper since I'm also getting used to the OOP way of doing things. I would like some advice about the structure of the classes.
Mostly everything is database-driven, like adding categories and subcategories, brands of products, products, users, etc. I suppose each of them could be one class and since I need CRUD operations on all of them, I need a generic way of inserting, updating, deleting records in the MySql database. The problem is not the code, I'd like to (and already have) coded some of the CRUD operations by myself according to my needs, the real problem is the structure and how would I go to correctly distribute and extend those classes.
Right now I've coded 3 different approaches:
A class called 'Operations' which will be extended by all the other classes that need CRUD functions, this class contains pretty generic properties such as $id, $atributes, $fields and $table, and of course the generic methods to insert, update, delete. That way I can create, let's say my Product object with some parameters (name, category, price) and immediately Product->insert() it into the database, without passing any parameters to the insert function. The CRUD functions in this class don't accept parameters, they depend on the created object's properties.
Same as above but the CRUD functions accept parameters, making them (I suppose) more generic, in case I just need to insert something without creating an object with useless properties previously.
The 'Operations' class extends PDO, the way of working is similar to 2, but now they can be directly accessed when I create the database connection, not depending of other objects.
I'm leaning towards the first option because I think, for the most part, that it will satisfy everything I'll do with this website, again the website is already coded but procedurally, which has been a mess to maintain, so basically I need to re-do things but OO.
CMSs or already coded wrappers aside (the purpose of doing this is to learn PDO and getting used to OOP), which would be the best way to do that? not limited to the options I mentioned.
Here's the 'Operations' class I've managed to code so far, where I've been doing tests sandbox-like, don't mind the spanish variable names. Advices on the code are welcome too.
class Operaciones {
private $database;
protected $id;
protected $atributos;
protected $tabla;
protected $campos;
public function __construct($link) {
$this->database = $link;
}
public function insertar() {
if (!$this->verificarCamposNulos($this->atributos, $this->campos))
echo 'Campos nulos<br />';
else {
$this->prepararCampos();
$placeholders = $this->generarPlaceholders();
$stmt = $this->database->prepare("INSERT INTO {$this->tabla} ({$this->campos}) VALUES ({$placeholders})");
$valores = array_values($this->atributos);
$stmt->execute($valores);
$stmt = NULL;
echo 'Se ha insertado exitosamente';
}
}
public function modificar() {
if (!$this->verificarCamposNulos() || empty($this->id))
echo 'Campos nulos<br />';
else {
$this->prepararCampos('=?');
$stmt = $this->database->prepare("UPDATE {$this->tabla} SET {$this->campos} WHERE id = {$this->id}");
$valores = array_values($this->atributos);
$stmt->execute($valores);
$stmt = NULL;
echo 'Se ha modificado exitosamente';
}
}
private function generarPlaceholders() {
for($i=0;$i<count($this->atributos);$i++)
$qmarks[$i] = '?';
return implode(',', $qmarks);
}
// Check if the values to be inserted are NULL, depending on the field format given
private function verificarCamposNulos() {
$n_campos = explode(',', $this->campos);
$valores = array_values($this->atributos);
foreach($n_campos as $i => $result) {
if (strstr($result, '#'))
if (empty($valores[$i]))
return false;
}
return true;
}
// Removes the '#' from each field, used to check which fields are NOT NULL in mysql
private function prepararCampos($sufijo = NULL) {
$n_campos = explode(',', $this->campos);
foreach($n_campos as $i => $result)
$n_campos[$i] = str_replace('#', '', $result) . $sufijo;
$this->campos = implode(',', $n_campos);
}
}

Design pattern for repetitive switch in getters?

I've an ORM model (PHP Active Record), say, for a blogging system. I've something that's a post model that stores the number of likes. The post could either be a picture or quote (say), and they are different tables (and hence models).
The schema is that a post holds data like number of shares, likes, description, etc. along with either a picture or a quote.
So when writing getters for the post model I'm having to write
public function getX() {
if ($this->isPicture()) {
return $this->picture->getX();
}
else if ($this->isQuote()) {
return $this->quote->getX()
}
else {
return self::DEFAULT_X
}
}
I'm currently having to write this structure for many getter. Is there something I can do to avoid that?
PS: Tagged as PHP because that's my code in.
EDIT
Changed comments to code.
This is a model (and a corresponding table in the DB) that has more data than just a picture and quote. Example, description that's part of the post and doesn't reside on either the picture or the quote.
There's tables for pictures and quotes.
Using PHP Active Record and each of the three classes extends the generic model class provided by PHP Active Record.
The picture model has it's own data. Same for quote.
To expand on the idea of the Strategy pattern mentioned in the comments:
class Post {
// get the correct 'strategy'
public function getModel() {
if ($this->isPicture()) {
return $this->picture;
}
if ($this->isQuote()) {
return $this->quote;
}
return null;
}
// using the strategy
public function getX() {
$model = $this->getModel();
if (null === $model) {
return self::DEFAULT_X;
}
return $model->getX();
}
}
Each strategy would presumably implement the same interface as the Post class for exposing those getters. Even better would be to provide a default strategy (rather than returning null) and have that return the default values. That way, the null check in each getter becomes redundant.
An alternative approach to this is a very basic form of metaprogramming. The idea is that you go a level higher than calling your methods by hand, and let the code do it for you.
(Assume that the method definitions are all part of Post)
public function getX($model = null) {
if ($model) return $model->getX();
else return self::DEFAULT_X;
}
// usage
$postModel->getX($pictureModel);
What's happening here is that, in this single instance of getX in your Post model, you're passing in the name of another class, and executing the `getX' method on that instance (if it exists and is callable).
You can extend this in other ways. For example, maybe you don't want to pass an instance in, when the method can do it anyway:
public function getX($model_name = null) {
if ($model_name && $class_exists($model_name) && is_callable(array($model_name, 'getX')) {
$model = new $model_name;
return $model->getX();
} else {
return self::DEFAULT_X;
}
}
// usage
$postModel->getX('Picture');
In this instance, you pass the model in as a string, and the method will do the rest. While this makes it quicker to get what you want, you might find that you don't want to work with fresh instances all the time (or you can't), so there's a bit of a trade-off with this 'convenient' way.
That still doesn't fully solve your problem, though, since you still have to repeat that for each getter, over and over again. Instead, you can try something like this:
public function __call($method, $args) {
$class = $args[0];
if (class_exists($class) && is_callable(array($class, $method))) {
$model = new $class;
return $model->$method();
}
}
// usage
$postModel->getX('Picture');
$postModel->getY('Quote');
$postModel->getZ('Picture');
If you call a function that doesn't exist on the Post model, that magic method will be called, and it'll fire up a new instance of the model name you supply as an argument, and call the getWhatever method on it, if it exists.
It's important to note that you must not define these getters in Post, unless you want to override the methods in the other classes.
There is still the problem of this creating new instances all the time, though, and to remedy this you can use a bit of dependency injection. This means that you let the Post class contains a list of other instances of classes that it wants to use in future, so you can add and remove them at will.
This is what I would consider the actual solution, with the other examples hopefully showing how I've got here (will edit to clarify things, of course).
public $models = array();
public function addModel($instance) {
$this->models[get_class($instance)] = $instance;
}
public function __call($method, $args) {
$class = $args[0];
if (array_key_exists($class, $this->models)) {
$model = $this->models[$class];
if (is_callable(array($model, $method)) {
return $model->$method();
}
}
}
// usage
$this->addModel($pictureModel);
$this->addModel($quoteModel);
$this->getX('Picture');
$this->getY('Quote');
Here, you're passing in your existing instances of models into the Post class, which then stores them in an array, keyed by the name of the class. Then, when you use the class as described in the last example, instead of creating a new instance, it will use the instance it has already stored. The benefit of this is that you might do things to your instances that you'd want reflected in the Post model.
This means that you can add as many new models as you like that need to plug into Post, and the only thing you need to do is inject them with addModel, and implement the getters on those models.
They all require you to tell the class what models to call at some point or another. Since you have an array of dependent models, why not add a way to get everything?
public function __call($method, $args) {
$class = $args[0];
if (array_key_exists($class, $this->models)) {
$model = $this->models[$class];
if (is_callable(array($model, $method)) {
return $model->$method();
}
} elseif ($class === 'all') {
// return an array containing the results of each method call on each model
return array_map(function($model) use ($method) {
if (is_callable(array($model, $method) return $model->$method();
}, $this->models);
}
}
// usage
$postModel->getX('all');
Using this, you'll get an array containing the return values of each getX method on each model you added with addModel. You can create pretty powerful functions and classes that do all this stuff without you having to repeat tedious logic.
I have to mention that these examples are untested, but at the very least I hope the concept of what you can do has been made clear.
Note:
The same thing can be applied to __GET and __SET methods, too, which are used for accessing properties. It's also worth saying that there may be the slight risk of a library already using these magic methods, in which case you'll need to make the code a little more intelligent.

Categories