Notion to Hashnode and Dev.to using Pub/Sub Cloud Functions
Ads
A goal at CodingCat.dev is to reach the most students possible. In order to do that we use Dev.to and Hashnode to cross post our articles and leverage canonical urls to point those articles back the the original. For us it allows more eyeballs to get to the content and videos.
Searching for the Notion Data
In order to keep find what data that needs processing in notion the simplest way is to keep a record of the url that the new blog will be held in. For example when finding records that we will need to add to Dev.to we will add a column called devto
. We can then search the database for any records fitting a criteria with that field being empty using Notion’s API and filtered fields.
In the below example, the database from database_id
is searched for CodingCat.dev’s database. We also add the start_cursor because we want these going out in order we pass 1 to this field so that we can guarantee that our posts go out in order and don’t process out of sequence. We also look for podcasts that have a slug
, are Released
, have an Episode
assigned, include a start
date on or after today’s date.
export const queryPurrfectStreamDevTo = async (
page_size?: number,
start_cursor?: string | null
) => {
const raw = await notionClient.databases.query({
database_id: notionConfig.purrfectStreamsDb,
start_cursor: start_cursor ? start_cursor : undefined,
page_size,
filter: {
and: [
{
property: 'slug',
url: {
is_not_empty: true,
},
},
{
property: 'Status',
select: {
equals: 'Released',
},
},
{
property: 'Episode',
number: {
is_not_empty: true,
},
},
{
property: 'start',
date: {
on_or_before: new Date().toISOString(),
},
},
{
property: 'devto',
url: {
is_empty: true,
},
},
{
property: 'youtube',
url: {
is_not_empty: true,
},
},
{
property: 'spotify',
url: {
is_not_empty: true,
},
},
],
},
sorts: [
{
property: 'Season',
direction: 'ascending',
},
{
property: 'Episode',
direction: 'ascending',
},
],
});
return await formatPosts(raw, 'podcast');
};
Scheduled Cloud Functions
We can now create a Firebase function that creates a Google Cloud - Cloud Scheduler Job that can run at a given interval. This is a convenience method that uses pubsub functions. Notice below we will have this function run every 5 minutes
and check to see if there are any new articles that match our criteria. We could probably back this off to every 1 hour, but it is nice to see this trigger relatively quickly if you have the correct criteria.
const topicId = 'devtoCreateFromNotion';
export const scheduledNotionToDevto = functions.pubsub
.schedule('every 5 minutes')
.onRun(async () => {
// Check to see if ther are scheduled pods
console.log('Checking for scheduled pods');
const scheduledRes = await queryPurrfectStreamDevTo(1);
console.log('Scheduled Result:', JSON.stringify(scheduledRes));
if (scheduledRes?.results) {
const needCloudinaryPods = scheduledRes?.results;
console.log('Pods to add to pub/sub', JSON.stringify(needCloudinaryPods));
for (const pod of needCloudinaryPods) {
await sendTopic(topicId, pod);
}
}
console.log('Checking for devto missing');
const posts = await queryByDevto('post', 1);
console.log('Posts:', JSON.stringify(posts));
if (posts?.results) {
const needposts = posts?.results;
console.log('Posts to add to pub/sub', JSON.stringify(needposts));
for (const p of needposts) {
await sendTopic(topicId, p);
}
}
return true;
});
Now if there is something actually found in queryPurrfectStreamDevTo
function we can then take the data from that entry and pass this on to another pub/sub function. For CodingCat.dev we just look for 1 entry so that we can publish things in the order that they are created, as we have found with some of these API’s like dev.to there is not an easy way to set original publish date on the actual platform.
Pub/Sub Function to publish post
This pub/sub function is looking for anything that sends a topic of devtoCreateFromNotion
to it and then looks up the appropriate post information. This is often the easiest way for massive scaling to happen if you have several items you are trying to publish to the platform.
For our example we change the body_markdown
depending on our post type. An important note here is that we are sending canonical_url
so that search engine bots don’t think this is original content and know where to add 301 if necessary.
export const devtoToNotionPubSub = functions.pubsub
.topic(topicId)
.onPublish(async (message, context) => {
console.log('The function was triggered at ', context.timestamp);
console.log('The unique ID for the event is', context.eventId);
const page = JSON.parse(JSON.stringify(message.json));
console.log('page', page);
let data;
if (page._type === 'podcast') {
data = {
article: {
title: page.title,
published: true,
tags: ['podcast', 'webdev', 'javascript', 'beginners'],
series: `codingcatdev_podcast_${page.properties.Season.number}`,
main_image: `https://media.codingcat.dev/image/upload/b_rgb:5e1186,c_pad,w_1000,h_420/${page?.coverPhoto?.public_id}`,
canonical_url: `https://codingcat.dev/${page._type}/${page.slug}`,
description: page.excerpt,
organization_id: '1009',
body_markdown: `Original: https://codingcat.dev/${page._type}/${
page.slug
}
{% youtube ${page.properties.youtube.url} %}
{% spotify spotify:episode:${page.properties.spotify.url
.split('/')
.at(-1)
.split('?')
.at(0)} %}
`,
},
};
} else {
console.log(
`Getting ${page._type}: ${page.id} markdown, with slug ${page?.properties?.slug?.url}`
);
const post = await getNotionPageMarkdown({
_type: page._type,
slug: page?.properties?.slug?.url,
preview: false,
});
console.log('Block Result', post);
if (post && post?.content) {
data = {
article: {
title: page.title,
published: true,
tags: ['podcast', 'webdev', 'javascript', 'beginners'],
main_image: `https://media.codingcat.dev/image/upload/b_rgb:5e1186,c_pad,w_1000,h_420/${page?.coverPhoto?.public_id}`,
canonical_url: `https://codingcat.dev/${page._type}/${page.slug}`,
description: page.excerpt,
organization_id: '1009',
body_markdown: post.content,
},
};
}
}
if (data) {
try {
console.log('addArticle to devto');
const response = await addArticle(data);
console.log('addArticle result:', response);
const devto = response?.data?.url;
if (!devto) {
console.log('devto url missing');
return;
}
const update = {
page_id: page.id,
properties: {
devto: {
id: 'remote',
type: 'url',
url: devto,
},
},
};
console.log('Updating page with: ', JSON.stringify(update));
const purrfectPagePatchRes = await patchPurrfectPage(update);
console.log(
'Page update result:',
JSON.stringify(purrfectPagePatchRes)
);
return purrfectPagePatchRes;
} catch (error) {
console.error(error);
}
} else {
console.error('No Data matched for article');
}
return;
});
Bonus: To test locally
I would recommend testing these functions locally before having them run on their own. You can use the firebase emulation suite for this. Then update your code for both the schedule and http version to call the same function like below scheduleCheck()
.
const scheduleCheck = async () => {
// Check to see if ther are scheduled pods
console.log('Checking for scheduled pods');
const scheduledRes = await queryPurrfectStreamDevTo(1);
console.log('Scheduled Result:', JSON.stringify(scheduledRes));
if (scheduledRes?.results) {
const needCloudinaryPods = scheduledRes?.results;
console.log('Pods to add to pub/sub', JSON.stringify(needCloudinaryPods));
for (const pod of needCloudinaryPods) {
await sendTopic(topicId, pod);
}
}
for (const _type of ['post', 'tutorial']) {
console.log('Checking for devto missing');
const posts = await queryByDevto(_type, 1);
console.log('Posts:', JSON.stringify(posts));
if (posts?.results) {
const needposts = posts?.results;
console.log('Posts to add to pub/sub', JSON.stringify(needposts));
for (const p of needposts) {
await sendTopic(topicId, p);
}
}
}
};
export const httpNotionToDevto = functions.https.onRequest(async (req, res) => {
await scheduleCheck();
res.send({ msg: 'started' });
});
export const scheduledNotionToDevto = functions.pubsub
.schedule('every 5 minutes')
.onRun(async () => {
await scheduleCheck();
return true;
});
Full Source
Full Dev.to example can be found at
Full Hashnode example can be found at
Note that this is TypeScript and all functions need to be exposed in the index.ts
file to be deployed.
https://github.com/CodingCatDev/codingcat.dev/blob/dev/backend/firebase/functions/src/index.ts