Redash Bot in Mattermost: charts automation

Automating Redash Dashboards: Sending Charts to Mattermost

A Python script solves the task of automatically delivering charts from Redash to Mattermost. It fetches data via API, recreates visualizations in matplotlib, combines them into a single dashboard, and sends it to a channel. Ideal for teams where data updates with a delay and manual dashboard review is inefficient.

The pipeline runs on a cron job: it checks data freshness, builds charts with annotations, and publishes the final report before the workday begins.

Fetching Data from Redash API

Redash provides an API to execute queries by query_id with parameters. The main function:

import requests

BASE_URL = 'https://your-redash-instance'
HEADERS = {'Authorization': 'Key YOUR_API_KEY'}

def fetch_query(query_id, params):
    response = requests.post(
        f'{BASE_URL}/api/queries/{query_id}/results',
        headers=HEADERS,
        json={'parameters': params, 'max_age': 0},
    )
    return response.json()

If the result isn't ready, the API returns a job — implement polling to wait for completion. For dashboards, parse parameters from the URL:

from urllib.parse import parse_qs, urlparse

def parse_dashboard_url(url):
    parsed = urlparse(url)
    params = parse_qs(parsed.query)
    return {
        key: value[0]
        for key, value in params.items()
    }

Convert dynamic dates like 'last 7 days' manually:

from datetime import datetime, timedelta

def last_7_days():
    today = datetime.utcnow()
    return {
        'start': (today - timedelta(days=7)).strftime('%Y-%m-%d'),
        'end': today.strftime('%Y-%m-%d')
    }

Building and Annotating Charts

Visualizations are built in matplotlib, mimicking Redash styles: colors, line types, sizes. A basic line chart example:

import matplotlib.pyplot as plt

def build_chart(df):
    fig, ax = plt.subplots(figsize=(6, 4))

    ax.plot(df['date'], df['value'], linewidth=2)
    ax.plot(df['date'], df['plan'], linewidth=2)

    ax.set_title('Metric dynamics')

    return fig

The key step is annotations for readability in static images. Add the latest value, week-over-week change, deviation from plan:

ax.annotate(
    f"value: {last_value}",
    xy=(last_date, last_value),
    xytext=(10, 10),
    textcoords='offset points'
)

This allows metrics to be understood without interactivity.

Combining Charts into a Dashboard

Multiple charts (by metrics, categories) are merged into one image using PIL for better UX:

from PIL import Image

def merge_images(images, output_path):
    base = Image.new('RGB', (1200, 800), 'white')

    for img, pos in images:
        base.paste(img, pos)

    base.save(output_path)

Add section headers by department or theme for navigation.

Checks and Duplicate Protection

The script checks for the target date in the data:

if df['date'].max() != target_date:
    raise Exception('Data not updated yet')

Protection against duplicate sends via a state file:

from pathlib import Path

STATE_FILE = Path('last_sent.txt')

def was_sent(date):
    return STATE_FILE.exists() and STATE_FILE.read_text() == str(date)

Sending to Mattermost

Use mattermostdriver to upload the file and post:

from mattermostdriver import Driver

mm = Driver({
    'url': 'your-mattermost',
    'token': 'BOT_TOKEN',
})

mm.login()

mm.posts.create_post({
    'channel_id': channel_id,
    'message': 'Report ready',
    'file_ids': [file_id],
})

Key Points

• Automation reduces manual effort: charts arrive in the messenger automatically upon data updates.
• Annotations on charts ensure full informativeness without hover interactions.
• Combining into one dashboard improves UX in chat.
• Freshness and duplicate checks guarantee pipeline reliability.
• Easy to scale: add new query_ids and parameters.

— Editorial Team