> ## Documentation Index
> Fetch the complete documentation index at: https://api.docs.flare.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Search in All of Flare's Events

Global searches in Flare is exposed through the
[global/\_search <Icon icon="code" size={16} />](/api-reference/v4/endpoints/global-search)
endpoint.

This guide details a typical search use case and provides an example for how to impement it.

## Use Case

* Export all chat messages collected by Flare in the last 6 hours that match the `fraud` keyword.
* Save our last page cursor so that we can resume fetching results in a future execution.

## Body Parameters

To achieve the desired results, the following body parameters will be used:

| Parameter                         | Value         | Justification                                                                                                            |
| --------------------------------- | ------------- | ------------------------------------------------------------------------------------------------------------------------ |
| query.query\_string               | fraud         | The keyword we are looking for.                                                                                          |
| filter.types                      | chat\_message | Retrieve only the chat\_message events.                                                                                  |
| order                             | asc           | Retrieve results in ascending order so that we can resume fetching in the future.                                        |
| filter.estimated\_created\_at.gte | `<from_date>` | Replace `<from_date>` with a timestamp that corresponds to 6 hours that we can fetch all results until the current time. |

### Query string examples

Global search supports the Lucene Query Syntax, which supports, among other things, boolean operators and regexes. Here are some common example queries.

| Description                                                | Example                                   |
| ---------------------------------------------------------- | ----------------------------------------- |
| Search for a telegram channel named "Best Telegram Source" | conversation\_name:"Best Telegram Source" |

## Paging

The search endpoint uses parameters that match the
[Flare standard paging pattern <Icon icon="book" size={16} />](/concepts/paging).

## Fetching new results in future executions

It is possible to save the `next` in a database and use it to resume fetching new results in the future.
However, it is important that future requests use the same parameters for everything else but `next`. Even the time filter.

## End-to-End Examples

These are end-to-end examples in various programming languages.

<AccordionGroup>
  <Accordion title="Python SDK Example">
    ```python theme={null}
    import datetime

    from flareio import FlareApiClient
    from flareio.ratelimit import Limiter


    api_client = FlareApiClient.from_env()

    limiter_search = Limiter.from_seconds(1)
    limiter_default = Limiter.from_seconds(0.25)

    from_timestamp: str = (
        datetime.datetime.now(tz=datetime.timezone.utc) - datetime.timedelta(hours=1)
    ).isoformat()

    last_from: str | None = None
    fetched_pages: int = 0

    for resp in api_client.scroll(
        method="POST",
        url="/firework/v4/events/global/_search",
        json={
            "size": 10,
            "order": "asc",
            "from": last_from,
            "filters": {
                "type": ["chat_message"],
                "estimated_created_at": {"gte": from_timestamp},
            },
            "query": {
                "type": "query_string",
                "query_string": "hello",
            },
        },
    ):
        limiter_search.tick()

        resp_data: dict = resp.json()
        items: list[dict] = resp_data["items"]

        fetched_pages += 1
        print(f"Fetched page {fetched_pages} with {len(items)} results...")

        # Save the last "next" value.
        last_from = resp_data.get("next") or last_from

        # (Optional): Get the full data
        for item in items:
            limiter_default.tick()

            event_response = api_client.get(
                url="/firework/v2/activities/",
                params={
                    "uid": item["metadata"]["uid"],
                },
            )

            full_data = event_response.json()
            print(f"Here is the full data of the event: {full_data}")

    print(f"The next execution could resume using {last_from=}.")
    ```
  </Accordion>
</AccordionGroup>