Read records

A monitor delivers what it finds as records. Read them with the paged cursor, or hold a stream open and take them as they arrive. Both start from a monitor you already created — see the Quickstart.

from seltz import Seltz

client = Seltz()
monitor_id = client.monitor.list(name="my-example-monitor").monitors[0].monitor_id

import { Seltz } from "seltz";

const client = new Seltz();
const monitorId = (await client.monitor.list({ name: "my-example-monitor" })).monitors[0].monitorId;

Page through records

Records are returned oldest first. since is exclusive: store the record_id of the last record you handled and pass it back to resume without re-delivery. There is no cursor token — the record id is the cursor. A page can hold fewer records than limit, so a short page is normal. Page until has_more is false rather than until you get fewer records than you asked for.

cursor = {}  # resume from a stored record id with {"since": last_id}

while True:
    page = client.monitor.list_records(monitor_id, limit=100, **cursor)

    for record in page.records:
        document = record.search_result.document
        print(record.record_id, document.url)
        cursor["since"] = record.record_id

    if not page.has_more:
        break

let lastId: bigint | undefined;

for (;;) {
  const page = await client.monitor.listRecords(monitorId, {
    since: lastId,
    limit: 100,
  });

  for (const record of page.records) {
    lastId = record.recordId;
    if (record.payload.case !== "searchResult") continue;
    console.log(record.recordId, record.payload.value.document?.url);
  }

  if (!page.hasMore) break;
}

MONITOR_ID="your-monitor-id"
LAST_RECORD_ID=0

curl "https://api.seltz.ai/v1/monitors/$MONITOR_ID/records?since=$LAST_RECORD_ID&limit=100" \
  -H "x-api-key: $SELTZ_API_KEY"

record_id is a bigint in TypeScript and a string over REST — see Reference. To omit document content and get a cheaper list view, pass include_content: include_content=False in Python, includeContent: false in TypeScript, and ?include_content=false over REST. Records stay readable for 30 days, so persist anything you need to keep. See Reference.

Stream records

The SDKs can hold a long-lived stream open and receive records as the monitor produces them, instead of polling. since works the same way it does on the paged call: pass the last record id you handled to resume. A stream ends when the server closes it. Reopen it from the last id you saw. The stream yields a response message carrying record, not a bare record.

Streaming records is server-streaming gRPC. There is no REST equivalent — no cURL form of this call exists, so poll GET /v1/monitors/{monitor_id}/records if you are working over REST.

from seltz import Seltz

client = Seltz()
monitor_id = client.monitor.list(name="my-example-monitor").monitors[0].monitor_id

last_id = None  # or the last record id you handled

while True:
    for response in client.monitor.stream_records(monitor_id, since=last_id):
        record = response.record
        document = record.search_result.document
        print(record.record_id, document.url)
        last_id = record.record_id

import { Seltz } from "seltz";

const client = new Seltz();
const monitorId = (await client.monitor.list({ name: "my-example-monitor" })).monitors[0].monitorId;

let lastId: bigint | undefined; // or the last record id you handled

for (;;) {
  for await (const response of client.monitor.streamRecords(monitorId, {
    since: lastId,
  })) {
    const record = response.record;
    if (!record) continue;
    lastId = record.recordId;
    if (record.payload.case !== "searchResult") continue;
    console.log(record.recordId, record.payload.value.document?.url);
  }
}

Getting Started

Web Tools

Data

Tutorials

Chat

Integrations

SDKs

Read records

Page through records

Stream records

​Page through records

​Stream records

Page through records

Stream records