Execute consultas do Data Explorer com a API do Discourse

Discourse · Junho 11, 2019, 7:43pm

This guide explains how to use the Discourse API to create, run, and manage queries with the Data Explorer plugin.

Required user level: Administrator

Virtually any action that can be performed through the Discourse user interface can also be triggered with the Discourse API.

This document provides a comprehensive overview for utilizing the API specifically in conjunction with the Data Explorer plugin.

For a general overview of how to find the correct API request for an action, see: Reverse engineer the Discourse API.

Running a Data Explorer query

To run a Data Explorer query via the API, make a POST request to /admin/plugins/explorer/queries/<query-id>/run. You can find the query ID by visiting it through your Discourse site and checking the id parameter in the address bar.

Below is an example query with an ID of 20 that returns topics by views on a specified date:

--[params]
-- date :viewed_at

SELECT
topic_id,
COUNT(1) AS views_for_date
FROM topic_views
WHERE viewed_at = :viewed_at
GROUP BY topic_id
ORDER BY views_for_date DESC

This query can be run from a terminal with:

curl -X POST "https://your-site-url/admin/plugins/explorer/queries/20/run" \
-H "Content-Type: multipart/form-data;" \
-H "Api-Key: <api-key>" \
-H "Api-Username: system" \
-F 'params={"viewed_at":"2019-06-10"}'

Note that you’ll need to replace the <api-key>, <your-site-url> with your API key and domain.

Creating a query via the API

To create a Data Explorer query via the API, you’ll need to make a POST request to /admin/plugins/explorer/queries.

You will also need to specify the query name and sql to use for the new query in the API call.

Below is an example of how to create a new query using the API:

curl -X POST "https://your-site-url/admin/plugins/explorer/queries" \
-H "Content-Type: multipart/form-data;" \
-H "Api-Key: <api-key>" \
-H "Api-Username: <username>" \
-F 'query[name]=Example Query' \
-F 'query[sql]=SELECT COUNT(*) FROM users'

This API call will return a response like:

{"query":{"id":49,"name":"Example Query","description":null,"username":"<username>","group_ids":[],"last_run_at":null,"user_id":1,"sql":"SELECT COUNT(*) FROM users","param_info":[],"created_at":"2025-03-13T18:41:44.226Z","hidden":false}}%

You can then run the query by using the query ID from the response (in this case, 49 ).

Returned results will be structured within the rows field.

Handling large datasets

The Data Explorer plugin limits results to 1000 rows by default. To paginate through larger datasets, you can use the example query below:

--[params]
-- integer :limit = 100
-- integer :page = 0
SELECT * 
FROM generate_series(1, 10000)
OFFSET :page * :limit 
LIMIT :limit

To fetch the results page-by-page, increment the page parameter in the request:

curl -X POST "https://your-site-url/admin/plugins/explorer/queries/27/run" \
-H "Content-Type: multipart/form-data;" \
-H "Api-Key: <api-key>" \
-H "Api-Username: system" \
-F 'params={"page":"0"}'

Stop when result_count is zero.

For additional information about handling large datasets, see: Result Limits and Exporting Queries

Removing `relations` data from the results

When Data Explorer queries are run through the user interface, a relations object is added to the results. This data is used for rendering the user in UI results, but you are unlikely to need it when running queries via the API.

To remove that data from the results, add a download=true parameter with your request:

curl -X POST "https://your-site-url/admin/plugins/explorer/queries/27/run" \
-H "Content-Type: multipart/form-data;" \
-H "Api-Key: <api-key>" \
-H "Api-Username: system" \
-F 'params={"page":"0"}' \
-F "download=true"

API authentication

Details about generating an API key for the requests can be found here: Create and configure an API key.

If the API key is only going to be used to run Data Explorer queries, you can select “Granular” from the Scope drop down menu, then select the “run queries” scope.

FAQs

Is there any api endpoint I can use to get the list of reports and the ID numbers? I want to build a dropdown with the list in it?

Yes, you can make an authenticated GET request to /admin/plugins/explorer/queries.json to get a list of all queries on the site.

Is it possible to send parameters with the post request?

Yes, include SQL parameters using the -F option, as shown in the examples.

Is CSV export for queries supported by the API?

While JSON output is standard, you can manually convert results to CSV. Native CSV export is no longer supported.

Additional resources

37Rb · Maio 10, 2024, 2:47am

Este comentário parece implicar que você pode fazer exportação CSV da API. Isso é possível? Só estou curioso porque preciso dos dados como CSV. Sempre posso obtê-los como JSON e converter para CSV, mas se houver uma maneira integrada de obter CSV, seria um pouco mais fácil.

francis_exalate · Janeiro 4, 2025, 1:50pm

É possível fazer uma consulta ‘como criado ou atualizado nos últimos 50 segundos’?

Canapin · Janeiro 4, 2025, 2:21pm

A IA diz

-- [params]
-- int :seconds = 50

SELECT
    p.id AS post_id,
    p.created_at,
    p.updated_at,
    p.raw AS post_content,
    p.user_id,
    t.title AS topic_title,
    t.id AS topic_id
FROM posts p
INNER JOIN topics t ON t.id = p.topic_id
WHERE
    (EXTRACT(EPOCH FROM (NOW() - p.created_at)) <= :seconds
    OR EXTRACT(EPOCH FROM (NOW() - p.updated_at)) <= :seconds)
    AND p.deleted_at IS NULL
    AND t.deleted_at IS NULL
ORDER BY p.created_at DESC
LIMIT 50

Fiz um teste rápido, parece que funciona

Moin · Janeiro 4, 2025, 2:34pm

Não vejo o “like ” que adicionei a uma postagem usando sua consulta.

Acho que você precisa usar post_actions

--[params]
--string :timespan = 50 seconds

SELECT post_id,
       user_id,
       created_at,
       updated_at,
       deleted_at
FROM  post_actions
WHERE post_action_type_id=2 AND updated_at > NOW() - INTERVAL :timespan

Versão com resultados mais bonitos

--[params]
--string :timespan = 50 seconds
--boolean :include_in_timespan_deleted = false

SELECT
  post_id,
  user_id,
  CASE
    WHEN EXTRACT(EPOCH FROM (NOW() - created_at)) < 60 THEN CONCAT(ROUND(EXTRACT(EPOCH FROM (NOW() - created_at))), ' seconds ago')
    WHEN EXTRACT(EPOCH FROM (NOW() - created_at)) < 3600 THEN CONCAT(ROUND(EXTRACT(EPOCH FROM (NOW() - created_at)) / 60), ' minutes ago')
    WHEN EXTRACT(EPOCH FROM (NOW() - created_at)) < 86400 THEN CONCAT(ROUND(EXTRACT(EPOCH FROM (NOW() - created_at)) / 3600), ' hours ago')
    ELSE CONCAT(ROUND(EXTRACT(EPOCH FROM (NOW() - created_at)) / 86400), ' days ago')
  END AS relative_created_at,
  CASE
    WHEN EXTRACT(EPOCH FROM (NOW() - updated_at)) < 60 THEN CONCAT(ROUND(EXTRACT(EPOCH FROM (NOW() - updated_at))), ' seconds ago')
    WHEN EXTRACT(EPOCH FROM (NOW() - updated_at)) < 3600 THEN CONCAT(ROUND(EXTRACT(EPOCH FROM (NOW() - updated_at)) / 60), ' minutes ago')
    WHEN EXTRACT(EPOCH FROM (NOW() - updated_at)) < 86400 THEN CONCAT(ROUND(EXTRACT(EPOCH FROM (NOW() - updated_at)) / 3600), ' hours ago')
    ELSE CONCAT(ROUND(EXTRACT(EPOCH FROM (NOW() - updated_at)) / 86400), ' days ago')
  END AS relative_updated_at,
  CASE
    WHEN deleted_at IS NULL THEN 'no'
    WHEN EXTRACT(EPOCH FROM (NOW() - deleted_at)) < 60 THEN CONCAT(ROUND(EXTRACT(EPOCH FROM (NOW() - deleted_at))), ' seconds ago')
    WHEN EXTRACT(EPOCH FROM (NOW() - deleted_at)) < 3600 THEN CONCAT(ROUND(EXTRACT(EPOCH FROM (NOW() - deleted_at)) / 60), ' minutes ago')
    WHEN EXTRACT(EPOCH FROM (NOW() - deleted_at)) < 86400 THEN CONCAT(ROUND(EXTRACT(EPOCH FROM (NOW() - deleted_at)) / 3600), ' hours ago')
    ELSE CONCAT(ROUND(EXTRACT(EPOCH FROM (NOW() - deleted_at)) / 86400), ' days ago')
  END AS relative_deleted_at
FROM
  post_actions
WHERE
  post_action_type_id = 2
  AND updated_at > NOW() - INTERVAL :timespan
  AND (
    :include_in_timespan_deleted = false
    OR (deleted_at IS NOT NULL AND deleted_at > NOW() - INTERVAL :timespan)
  )

Canapin · Janeiro 4, 2025, 3:37pm

Oooh, eu li errado! Eu li a frase como "É possível fazer uma consulta como ‘criado ou atualizado nos últimos 50 segundos’?

(note a posição da aspa simples)"

francis_exalate · Janeiro 4, 2025, 6:58pm

Obrigado a todos,

Postei aqui porque foi para onde o ask.discourse me direcionou.
O que estou procurando é se a API está fornecendo essa capacidade.

Canapin · Janeiro 4, 2025, 7:09pm

Sim, claro.

Você cria a consulta no explorador de dados, em seguida, executa a consulta através da API, conforme descrito neste guia.

Acabei de executar a consulta de Moin através da API e ela retornou corretamente os resultados esperados.

tanya_byrne · Janeiro 14, 2025, 10:57pm

Eu também estava me perguntando sobre isso. JSON é a única forma de exportar dados via API ou a exportação CSV também é suportada para o Data Explorer?

francis_exalate · Janeiro 19, 2025, 11:18pm

Obrigado a todos,

Desculpe pela resposta tardia - estive offline por um tempo.

O que estou fazendo atualmente é pesquisar todos os tópicos/postagens criados hoje e filtrar os tópicos/postagens atualizados antes do carimbo de data/hora.

Canapin · Janeiro 19, 2025, 11:35pm

Se você não quiser “colocar a mão na massa”, pode perguntar ao bot em ask.discourse.com. Ele geralmente é bem preciso em relação a consultas SQL relacionadas ao Discourse (mas não assuma que está certo, verifique o código para ter certeza).

srinivas.chilukuri · Maio 6, 2025, 4:06pm

O cabeçalho Content-Type está correto?
Nas ferramentas do desenvolvedor, ao inspecionar uma consulta do Data Explorer com parâmetros, o cabeçalho Content-Type aparece como:

Content-Type: application/x-www-form-urlencoded; charset=UTF-8

No entanto, o comando cURL atual inclui:

-H "Content-Type: multipart/form-data;"

blake · Maio 6, 2025, 4:30pm

multipart/form-data
application/x-www-form-urlencoded
application/json

são todos content-types válidos que você pode usar ao fazer uma requisição de API.

srinivas.chilukuri · Maio 6, 2025, 4:37pm

@blake
linguagem python
biblioteca requests
Você poderia fornecer uma consulta de exemplo da API Data Explorer que inclua três parâmetros
consulte Tópico que diz que os parâmetros precisam estar estritamente entre aspas duplas

blake · Maio 6, 2025, 5:15pm

Claro, aqui está um exemplo usando python:

import json
import requests

API_KEY      = "SUA_API_KEY"
API_USERNAME = "system"
QUERY_ID     = 20
SITE_URL     = "https://seu-site-url"

# todos os valores devem ser strings
params = {
    "user_id":   "2",
    "viewed_at": "2019-06-10",
    "limit":     "5"
}

# O Data Explorer espera os parâmetros como uma string codificada em JSON
payload = {
    "params": json.dumps(params)
}

url = f"{SITE_URL}/admin/plugins/explorer/queries/{QUERY_ID}/run"
headers = {
    "Api-Key":       API_KEY,
    "Api-Username":  API_USERNAME,
    "Content-Type":  "application/json"
}

r = requests.post(url, headers=headers, json=payload)
r.raise_for_status()
print(r.json())

Tópico		Respostas	Visualizações
Discourse Data Explorer Plugin official , data-explorer , included-in-core	31	80354	22 de Outubro de 2025
Discourse AI + Data Explorer? Feature data-explorer , completed , ai , ai-bot	49	3587	13 de Janeiro de 2024
Data Explorer Tutorial - Part 1 - Writing Your First Query Data & reporting data-explorer , sql-tutorial	0	1417	5 de Setembro de 2023
How to create a data explorer query with the APIs? Dev data-explorer , rest-api	6	801	9 de Novembro de 2023
Download result of queries into Google Spreadsheet Data & reporting sql-query	10	3576	30 de Setembro de 2020