query-builder.md 15.2 KB
Newer Older
Alexander Makarov committed
1 2 3
Query Builder and Query
=======================

4 5 6 7
Yii provides a basic database access layer as described in the [Database basics](database-basics.md) section.
The database access layer provides a low-level way to interact with the database. While useful in some situations,
it can be tedious and error-prone to write raw SQLs. An alternative approach is to use the Query Builder.
The Query Builder provides an object-oriented vehicle for generating queries to be executed.
Alexander Makarov committed
8

9
A typical usage of the query builder looks like the following:
Alexander Makarov committed
10 11

```php
12
$rows = (new \yii\db\Query())
13
    ->select('id, name')
14
    ->from('user')
15 16
    ->limit(10)
    ->all();
17 18

// which is equivalent to the following code:
Alexander Makarov committed
19

20
$query = (new \yii\db\Query())
21
    ->select('id, name')
22
    ->from('user')
23
    ->limit(10);
Alexander Makarov committed
24

25
// Create a command. You can get the actual SQL using $command->sql
Alexander Makarov committed
26
$command = $query->createCommand();
27 28

// Execute the command:
Alexander Makarov committed
29 30 31
$rows = $command->queryAll();
```

32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54
Query Methods
-------------

As you can see, [[yii\db\Query]] is the main player that you need to deal with. Behind the scene,
`Query` is actually only responsible for representing various query information. The actual query
building logic is done by [[yii\db\QueryBuilder]] when you call the `createCommand()` method,
and the query execution is done by [[yii\db\Command]].

For convenience, [[yii\db\Query]] provides a set of commonly used query methods that will build
the query, execute it, and return the result. For example,

- [[yii\db\Query::all()|all()]]: builds the query, executes it and returns all results as an array.
- [[yii\db\Query::one()|one()]]: returns the first row of the result.
- [[yii\db\Query::column()|column()]]: returns the first column of the result.
- [[yii\db\Query::scalar()|scalar()]]: returns the first column in the first row of the result.
- [[yii\db\Query::exists()|exists()]]: returns a value indicating whether the query results in anything.
- [[yii\db\Query::count()|count()]]: returns the result of a `COUNT` query. Other similar methods
  include `sum()`, `average()`, `max()`, `min()`, which support the so-called aggregational data query.


Building Query
--------------

55 56 57 58
In the following, we will explain how to build various clauses in a SQL statement. For simplicity,
we use `$query` to represent a [[yii\db\Query]] object.


59
### `SELECT`
Alexander Makarov committed
60

61
In order to form a basic `SELECT` query, you need to specify what columns to select and from what table:
Alexander Makarov committed
62 63 64

```php
$query->select('id, name')
65
    ->from('user');
Alexander Makarov committed
66 67
```

68 69
Select options can be specified as a comma-separated string, as in the above, or as an array.
The array syntax is especially useful when forming the selection dynamically:
Alexander Makarov committed
70 71

```php
72
$query->select(['id', 'name'])
73
    ->from('user');
Alexander Makarov committed
74 75
```

76 77 78 79 80
> Info: You should always use the array format if your `SELECT` clause contains SQL expressions.
> This is because a SQL expression like `CONCAT(first_name, last_name) AS full_name` may contain commas.
> If you list it together with other columns in a string, the expression may be split into several parts
> by commas, which is not what you want to see.

81
When specifying columns, you may include the table prefixes or column aliases, e.g., `user.id`, `user.id AS user_id`.
82
If you are using array to specify the columns, you may also use the array keys to specify the column aliases,
83
e.g., `['user_id' => 'user.id', 'user_name' => 'user.name']`.
84 85 86 87

To select distinct rows, you may call `distinct()`, like the following:

```php
88
$query->select('user_id')->distinct()->from('post');
89 90
```

91
### `FROM`
92 93 94 95

To specify which table(s) to select data from, call `from()`:

```php
96
$query->select('*')->from('user');
97 98 99
```

You may specify multiple tables using a comma-separated string or an array.
100
Table names can contain schema prefixes (e.g. `'public.user'`) and/or table aliases (e.g. `'user u'`).
101 102 103 104
The method will automatically quote the table names unless it contains some parenthesis
(which means the table is given as a sub-query or DB expression). For example,

```php
105
$query->select('u.*, p.*')->from(['user u', 'post p']);
106 107 108 109 110 111
```

When the tables are specified as an array, you may also use the array keys as the table aliases
(if a table does not need alias, do not use a string key). For example,

```php
112
$query->select('u.*, p.*')->from(['u' => 'user u', 'p' => 'post']);
113 114 115 116
```

You may specify a sub-query using a `Query` object. In this case, the corresponding array key will be used
as the alias for the sub-query.
Qiang Xue committed
117

118
```php
119
$subQuery = (new Query())->select('id')->from('user')->where('status=1');
120 121
$query->select('*')->from(['u' => $subQuery]);
```
Qiang Xue committed
122

123

124
### `WHERE`
Alexander Makarov committed
125

126
Usually data is selected based upon certain criteria. Query Builder has some useful methods to specify these, the most powerful of which being `where`. It can be used in multiple ways.
Alexander Makarov committed
127

128
The simplest way to apply a condition is to use a string:
Alexander Makarov committed
129 130

```php
Alexander Makarov committed
131
$query->where('status=:status', [':status' => $status]);
Alexander Makarov committed
132 133
```

134
When using strings, make sure you're binding the query parameters, not creating a query by string concatenation. The above approach is safe to use, the following is not:
Alexander Makarov committed
135

136 137 138 139 140
```php
$query->where("status=$status"); // Dangerous!
```

Instead of binding the status value immediately, you can do so using `params` or `addParams`:
Alexander Makarov committed
141 142 143

```php
$query->where('status=:status');
Alexander Makarov committed
144
$query->addParams([':status' => $status]);
Alexander Makarov committed
145 146
```

147
Multiple conditions can simultaneously be set in `where` using the *hash format*:
Alexander Makarov committed
148 149

```php
Alexander Makarov committed
150
$query->where([
151 152 153
    'status' => 10,
    'type' => 2,
    'id' => [4, 8, 15, 16, 23, 42],
Alexander Makarov committed
154
]);
Alexander Makarov committed
155 156
```

157
That code will generate the following SQL:
Alexander Makarov committed
158 159 160 161 162

```sql
WHERE (`status` = 10) AND (`type` = 2) AND (`id` IN (4, 8, 15, 16, 23, 42))
```

163
NULL is a special value in databases, and is handled smartly by the Query Builder. This code:
Alexander Makarov committed
164 165

```php
Alexander Makarov committed
166
$query->where(['status' => null]);
Alexander Makarov committed
167 168
```

169
results in this WHERE clause:
Alexander Makarov committed
170 171 172 173 174

```sql
WHERE (`status` IS NULL)
```

175 176 177 178 179 180 181 182 183 184 185 186 187 188
You can also create sub-queries with `Query` objects like the following,

```php
$userQuery = (new Query)->select('id')->from('user');
$query->where(['id' => $userQuery]);
```

which will generate the following SQL:

```sql
WHERE `id` IN (SELECT `id` FROM `user`)
```


Alexander Makarov committed
189
Another way to use the method is the operand format which is `[operator, operand1, operand2, ...]`.
Alexander Makarov committed
190 191 192 193

Operator can be one of the following:

- `and`: the operands should be concatenated together using `AND`. For example,
Alexander Makarov committed
194
  `['and', 'id=1', 'id=2']` will generate `id=1 AND id=2`. If an operand is an array,
Alexander Makarov committed
195
  it will be converted into a string using the rules described here. For example,
Alexander Makarov committed
196
  `['and', 'type=1', ['or', 'id=1', 'id=2']]` will generate `type=1 AND (id=1 OR id=2)`.
Alexander Makarov committed
197 198 199 200
  The method will NOT do any quoting or escaping.
- `or`: similar to the `and` operator except that the operands are concatenated using `OR`.
- `between`: operand 1 should be the column name, and operand 2 and 3 should be the
   starting and ending values of the range that the column is in.
Alexander Makarov committed
201
   For example, `['between', 'id', 1, 10]` will generate `id BETWEEN 1 AND 10`.
Alexander Makarov committed
202 203
- `not between`: similar to `between` except the `BETWEEN` is replaced with `NOT BETWEEN`
  in the generated condition.
204 205 206 207
- `in`: operand 1 should be a column or DB expression. Operand 2 can be either an array or a `Query` object.
  It will generate an `IN` condition. If Operand 2 is an array, it will represent the range of the values
  that the column or DB expression should be; If Operand 2 is a `Query` object, a sub-query will be generated
  and used as the range of the column or DB expression. For example,
Alexander Makarov committed
208
  `['in', 'id', [1, 2, 3]]` will generate `id IN (1, 2, 3)`.
Alexander Makarov committed
209
  The method will properly quote the column name and escape values in the range.
210 211
  The `in` operator also supports composite columns. In this case, operand 1 should be an array of the columns,
  while operand 2 should be an array of arrays or a `Query` object representing the range of the columns.
Alexander Makarov committed
212 213 214
- `not in`: similar to the `in` operator except that `IN` is replaced with `NOT IN` in the generated condition.
- `like`: operand 1 should be a column or DB expression, and operand 2 be a string or an array representing
  the values that the column or DB expression should be like.
215
  For example, `['like', 'name', 'tester']` will generate `name LIKE '%tester%'`.
Alexander Makarov committed
216
  When the value range is given as an array, multiple `LIKE` predicates will be generated and concatenated
217
  using `AND`. For example, `['like', 'name', ['test', 'sample']]` will generate
Alexander Makarov committed
218
  `name LIKE '%test%' AND name LIKE '%sample%'`.
219 220 221 222 223 224
  You may also provide an optional third operand to specify how to escape special characters in the values.
  The operand should be an array of mappings from the special characters to their
  escaped counterparts. If this operand is not provided, a default escape mapping will be used.
  You may use `false` or an empty array to indicate the values are already escaped and no escape
  should be applied. Note that when using an escape mapping (or the third operand is not provided),
  the values will be automatically enclosed within a pair of percentage characters.
Alexander Makarov committed
225 226 227 228 229 230
- `or like`: similar to the `like` operator except that `OR` is used to concatenate the `LIKE`
  predicates when operand 2 is an array.
- `not like`: similar to the `like` operator except that `LIKE` is replaced with `NOT LIKE`
  in the generated condition.
- `or not like`: similar to the `not like` operator except that `OR` is used to concatenate
  the `NOT LIKE` predicates.
231
- `exists`: requires one operand which must be an instance of [[yii\db\Query]] representing the sub-query.
232 233
  It will build a `EXISTS (sub-query)` expression.
- `not exists`: similar to the `exists` operator and builds a `NOT EXISTS (sub-query)` expression.
Alexander Makarov committed
234

Qiang Xue committed
235
If you are building parts of condition dynamically it's very convenient to use `andWhere()` and `orWhere()`:
Alexander Makarov committed
236 237 238 239 240

```php
$status = 10;
$search = 'yii';

Alexander Makarov committed
241
$query->where(['status' => $status]);
Alexander Makarov committed
242
if (!empty($search)) {
243
    $query->andWhere(['like', 'title', $search]);
Alexander Makarov committed
244 245 246 247 248 249 250 251 252
}
```

In case `$search` isn't empty the following SQL will be generated:

```sql
WHERE (`status` = 10) AND (`title` LIKE '%yii%')
```

Qiang Xue committed
253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277
#### Building Filter Conditions

When building filter conditions based on user inputs, you usually want to specially handle "empty inputs"
by ignoring them in the filters. For example, you have an HTML form that takes username and email inputs.
If the user only enters something in the username input, you may want to build a query that only tries to
match the entered username. You may use the `filterWhere()` method achieve this goal:

```php
// $username and $email are from user inputs
$query->filterWhere([
    'username' => $username,
    'email' => $email,
]);
```

The `filterWhere()` method is very similar to `where()`. The main difference is that `filterWhere()`
will remove empty values from the provided condition. So if `$email` is "empty", the resulting query
will be `...WHERE username=:username`; and if both `$username` and `$email` are "empty", the query
will have no `WHERE` part.

A value is *empty* if it is null, an empty string, a string consisting of whitespaces, or an empty array.

You may also use `andFilterWhere()` and `orFilterWhere()` to append more filter conditions.


278
### `ORDER BY`
Alexander Makarov committed
279

280
For ordering results `orderBy` and `addOrderBy` could be used:
Alexander Makarov committed
281 282

```php
Alexander Makarov committed
283
$query->orderBy([
284 285
    'id' => SORT_ASC,
    'name' => SORT_DESC,
Alexander Makarov committed
286
]);
Alexander Makarov committed
287 288 289 290
```

Here we are ordering by `id` ascending and then by `name` descending.

291 292
```

293
### `GROUP BY` and `HAVING`
Alexander Makarov committed
294 295 296 297 298 299 300 301 302 303

In order to add `GROUP BY` to generated SQL you can use the following:

```php
$query->groupBy('id, status');
```

If you want to add another field after using `groupBy`:

```php
Alexander Makarov committed
304
$query->addGroupBy(['created_at', 'updated_at']);
Alexander Makarov committed
305 306 307 308 309 310
```

To add a `HAVING` condition the corresponding `having` method and its `andHaving` and `orHaving` can be used. Parameters
for these are similar to the ones for `where` methods group:

```php
Alexander Makarov committed
311
$query->having(['status' => $status]);
Alexander Makarov committed
312 313
```

314
### `LIMIT` and `OFFSET`
Alexander Makarov committed
315 316 317 318 319 320 321 322 323 324 325 326 327

To limit result to 10 rows `limit` can be used:

```php
$query->limit(10);
```

To skip 100 fist rows use:

```php
$query->offset(100);
```

328 329 330 331 332 333 334 335 336 337 338
### `JOIN`

The `JOIN` clauses are generated in the Query Builder by using the applicable join method:

- `innerJoin()`
- `leftJoin()`
- `rightJoin()`

This left join selects data from two related tables in one query:

```php
339 340 341
$query->select(['user.name AS author', 'post.title as title'])
    ->from('user')
    ->leftJoin('post', 'post.user_id = user.id');
342 343 344 345 346 347 348 349
```

In the code, the `leftJoin()` method's first parameter
specifies the table to join to. The second parameter defines the join condition.

If your database application supports other join types, you can use those via the  generic `join` method:

```php
350
$query->join('FULL OUTER JOIN', 'post', 'post.user_id = user.id');
351 352 353 354 355 356 357 358 359 360 361 362 363 364
```

The first argument is the join type to perform. The second is the table to join to, and the third is the condition.

Like `FROM`, you may also join with sub-queries. To do so, specify the sub-query as an array
which must contain one element. The array value must be a `Query` object representing the sub-query,
while the array key is the alias for the sub-query. For example,

```php
$query->leftJoin(['u' => $subQuery], 'u.id=author_id');
```


### `UNION`
Alexander Makarov committed
365 366 367 368 369

`UNION` in SQL adds results of one query to results of another query. Columns returned by both queries should match.
In Yii in order to build it you can first form two query objects and then use `union` method:

```php
370
$query = new Query();
371
$query->select("id, 'post' as type, name")->from('post')->limit(10);
Alexander Makarov committed
372

373
$anotherQuery = new Query();
374
$anotherQuery->select('id, 'user' as type, name')->from('user')->limit(10);
Alexander Makarov committed
375 376 377 378

$query->union($anotherQuery);
```

379 380 381 382 383 384 385 386 387 388 389 390 391 392

Batch Query
-----------

When working with large amount of data, methods such as [[yii\db\Query::all()]] are not suitable
because they require loading all data into the memory. To keep the memory requirement low, Yii
provides the so-called batch query support. A batch query makes uses of data cursor and fetches
data in batches.

Batch query can be used like the following:

```php
use yii\db\Query;

393
$query = (new Query())
394
    ->from('user')
395
    ->orderBy('id');
396

Qiang Xue committed
397
foreach ($query->batch() as $users) {
398
    // $users is an array of 100 or fewer rows from the user table
399 400
}

Qiang Xue committed
401
// or if you want to iterate the row one by one
402
foreach ($query->each() as $user) {
403
    // $user represents one row of data from the user table
404 405
}
```
406

Qiang Xue committed
407 408 409 410 411 412 413 414 415
The method [[yii\db\Query::batch()]] and [[yii\db\Query::each()]] return an [[yii\db\BatchQueryResult]] object
which implements the `Iterator` interface and thus can be used in the `foreach` construct.
During the first iteration, a SQL query is made to the database. Data are since then fetched in batches
in the iterations. By default, the batch size is 100, meaning 100 rows of data are being fetched in each batch.
You can change the batch size by passing the first parameter to the `batch()` or `each()` method.

Compared to the [[yii\db\Query::all()]], the batch query only loads 100 rows of data at a time into the memory.
If you process the data and then discard it right away, the batch query can help keep the memory usage under a limit.

416 417 418 419 420 421
If you specify the query result to be indexed by some column via [[yii\db\Query::indexBy()]], the batch query
will still keep the proper index. For example,

```php
use yii\db\Query;

422
$query = (new Query())
423
    ->from('user')
424
    ->indexBy('username');
425

Qiang Xue committed
426
foreach ($query->batch() as $users) {
427
    // $users is indexed by the "username" column
428 429 430 431 432
}

foreach ($query->each() as $username => $user) {
}
```