> Documentation index: https://firebase.google.com/docs/firestore/llms.txt

## Description

Returns all documents from any collection with the specified collection ID,
regardless of its parent.

## Examples

### Web

```javascript
const results = await execute(db.pipeline()
  .collectionGroup("games")
  .sort(field("name").ascending())
  );
```

##### Swift

```swift
let results = try await db.pipeline()
  .collectionGroup("games")
  .sort([Field("name").ascending()])
  .execute()
```

### Kotlin

```kotlin
val results = db.pipeline()
    .collectionGroup("games")
    .sort(field("name").ascending())
    .execute()
```

### Java

```java
      Task<Pipeline.Snapshot> results = db.pipeline()
    .collectionGroup("games")
    .sort(field("name").ascending())
    .execute();
    
```

##### Python

```python
from google.cloud.firestore_v1.pipeline_expressions import Field

results = (
    client.pipeline()
    .collection_group("games")
    .sort(Field.of("name").ascending())
    .execute()
)
```

##### Java

```java
Pipeline.Snapshot results =
    firestore
        .pipeline()
        .collectionGroup("games")
        .sort(ascending(field("name")))
        .execute()
        .get();
```

##### Go

```go
snapshot := client.Pipeline().
	CollectionGroup("games").
	Sort(firestore.Orders(firestore.Ascending(firestore.FieldOf("name")))).
	Execute(ctx)
```

## Behavior

In order to use the `collection_group(...)` stage, it must appear as the first
stage in the pipeline.

The order of documents returned from the `collection_group(...)` stage is
unstable and cannot be relied upon. Cloud Firestore will attempt to
execute the query in the most efficient way possible, which can change the order
depending on the schema or index configuration. A subsequent
[`sort(...)`](https://firebase.google.com/docs/firestore/pipelines/stages/transformation/sort.md.txt) stage can be used to obtain a deterministic
ordering.

For example, for the following documents:

### Node.js

```javascript
await db.collection("cities/SF/departments").doc("building").set({name: "SF Building Deparment", employees: 750});
await db.collection("cities/NY/departments").doc("building").set({name: "NY Building Deparment", employees: 1000});
await db.collection("cities/CHI/departments").doc("building").set({name: "CHI Building Deparment", employees: 900});
await db.collection("cities/NY/departments").doc("finance").set({name: "NY Finance Deparment", employees: 1200});
```

The `collection_group(...)` stage can be used to return documents from every
departments collection across all parent collections in the database.

### Node.js

```javascript
const results = await db.pipeline()
  .collectionGroup("departments")
  .sort(field("employees").ascending())
  .execute();
```

This query produces the following documents:

```javascript
  { name: "SF Building Deparment", employees: 750 }
  { name: "CHI Building Deparment", employees: 900 }
  { name: "NY Building Deparment", employees: 1000 }
  { name: "NY Finance Deparment", employees: 1200 }
```