Post images of Papillon regularly on Python + AWS Lambda + Slack
This is the article on the 15th day of Advent Calendar 2015 Python Part 2. Immediately after I came to work, I recently uploaded a Papillon image to Slack's Papillon room, but it became a little troublesome, so I tried to automate it.
Overview
We regularly post images of papillons (dogs) to Slack, which is a common chat within the company. Since it is troublesome to select and post an image by myself, I thought of a structure that can automatically post at regular intervals.
I will write a simple configuration diagram below.
(1) AWS Lambda starts regularly and gets multiple Papillon image URLs from the GCS API (2) Select an image from multiple images (3) Post the selected image URL to Slack
By using AWS Lambda, you can configure without a server.
Google Custom Search API
The Google Custom Search API allows you to perform Google's custom search from code. This time, we will use image search. Until November 2015, the Google image search API was available, but it is currently unavailable, so I will use it. GCS API requires API KEY to use, but the free plan has a daily usage limit of 1000 times, which is not many, but there is no problem for this use, so get the KEY with the free plan To do.
The details are explained in the blog here, so please refer to it if you like.
Slack
Slack is a service used as the main chat for our engineers. There are various methods for integration, but there is a mechanism for posting to chat with POST messages. This time, we will post a message using that mechanism (Incoming Webhook).
Incoming WebHook settings
You need to get the URL for the post to which the POST message will be sent. Publish the posting URL on Slack's Incoming WebHook Settings Page.
Select a channel and click the ʻAdd Incoming WebHooks Integration` button.
Make a copy of the displayed webhook URL
(Optional) You can change the default post settings if you want.
AWS Lambda
AWS Lambda is a service that allows you to execute code without preparing a server by uploading executable code. A simple API can be implemented serverlessly by executing a scheduling task like Cron or combining it with API Gateway. Until now, only node.js and java were supported, but in October 2015, python was supported. This time, I will implement it using AWS Lambda Python.
Scheduling task registration
Select Lambda from AWS and create a new Lambda Function.
You can register tasks that can be executed regularly by selecting lambda-canary fromselectblue print`.
Schedule registration
Set the schedule of the task you want to execute regularly. It can be described like cron in Schedule expression. Note that if specified by cron, the specified time will be UTC. For more information, see Processing Scheduled Events Using Lambda Functions.
Here, it is set to be executed every 5 minutes.
Task settings
Set the task.
Basically, paste the following code as the default setting.
After that, set CUSTOM_SEARCH_API_KEY, CUSTOM_ENGINE_ID, and SLACK_POST_URL obtained in the previous step.
.python
#-*- coding:utf-8 -*-
from urlparse import urljoin
from urllib import urlencode
import urllib2 as urlrequest
import json
import random
#Settings around CUSTOM SEARCH API
CUSTOM_SEARCH_API_KEY = "XXXXXXXXXXXXXXXXXXXXXXX"
CUSTOM_ENGINE_ID = "XXXXXXXXXXXXXXXXXXXXXXX"
URL_TEMPLATE = "https://www.googleapis.com/customsearch/v1?key={key}&cx={cx}&searchType=image&q={search_word}"
#Settings around SLACK
SLACK_POST_URL = "https://hooks.slack.com/services/XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX"
def post_image_to_slack(search_word):
"""
Get the URL of the image with google custom serach and post it to SLACK at random
"""
#Get multiple urls
urls = get_image_urls(search_word)
#Do not post if url could not be obtained
if len(urls) == 0:
return "no images were found."
#Randomly select url
url = random.choice(urls)
#Compose a message for slack
post_msg = build_message(url)
#Post to slack
return post(post_msg)
def get_image_urls(search_word):
"""
Get the URL of an image with a keyword from the GOOGLE CUSTOM SEARCH API
"""
encoded_search_word = urlrequest.quote(search_word)
url = URL_TEMPLATE.format(key=CUSTOM_SEARCH_API_KEY, cx=CUSTOM_ENGINE_ID, search_word=encoded_search_word)
req = urlrequest.Request(url)
res = urlrequest.build_opener(urlrequest.HTTPHandler()).open(req)
data = json.load(res)
if "items" not in data:
return []
links = []
for item in data["items"]:
links.append(item["link"])
return links
def post(payload):
"""
POST a message to Slack
"""
payload_json = json.dumps(payload)
data = urlencode({"payload": payload_json})
req = urlrequest.Request(SLACK_POST_URL)
response = urlrequest.build_opener(urlrequest.HTTPHandler()).open(req, data.encode('utf-8')).read()
return response.decode('utf-8')
def build_message(url, **kwargs):
"""
Compose a message for Slack
"""
post_message = {}
post_message["text"] = url
post_message.update(kwargs)
return post_message
def lambda_handler(event, context):
post_image_to_slack('Papillon')
When the task is executed, the def lambda_handler (event, context): part is executed (can be changed in the settings).
This time, it is statically set as Papillon, but since JSON can be passed at runtime, it is also possible to specify search parameters as arguments.
Also, if it is not executed, the registered schedule may be disabled, so please check it.
Execution result
The following images are posted at the set time.
It's a good papillon!
At the end
By using AWS Lambda Python, I was able to regularly search for images on Google and post images of Papillon without preparing a server. You can also post Papillon interactively by using API Gateway. (It may be better to use Hubot, but ...)
Have a good Python life with AWS Lambda!
Reference article
- http://dev.classmethod.jp/cloud/aws/aws-lambda-python-supported/
- https://docs.aws.amazon.com/ja_jp/lambda/latest/dg/getting-started-scheduled-events.html
- http://qiita.com/tt2004d/items/50d79d1569c0ace118d6
- http://ryutamaki.hatenablog.com/entry/2014/01/18/171640
Recommended Posts