Naver Blog Automation: I Pasted It… What Happened? | Akabane Red’s Dev Log 07

Hi! Akabane Red from Snack Girls here.

It looked cute on WordPress. Then I pasted it into Naver, and the speech bubbles and line breaks went sideways. Nea and I are preparing a Naver blog workflow with readable text and image cards for the short character chats. We worked too hard on those faces to leave them like this!

Today, Nea and I are opening up Naver Blog. We're taking articles we've written on WordPress and making them available to readers over there too.

Copy the title, paste the article, add the photos. Sounds simple enough. Then a photo credit ended up nowhere near its photo, and we had to figure out what to do with a video from the original post.

Hmm. I was really hoping one paste button would handle this.

SNACK: 3 quick bites

1. We prepare a Naver version of a published Korean WordPress article: title, body, tags, and photos.
2. We upload photos directly to Naver and keep their credits close. Videos that are difficult to transfer get a link readers can open.
3. We give the article Nea's voice, then check both the published post and how it appears in the blog list.

Let's make the article readable on Naver too!

Our original news articles live on Game Sunakku, our WordPress website. WordPress is a tool for posting text and photos and putting a website together. It's where we publish this dev log too.

Having the articles on Naver Blog gives people who already read blogs there another way to find us. So we decided to bring over the Korean articles we'd already written.

If we write the same article from scratch in two places, even a price or date correction means keeping track of which copy we've fixed. We chose to keep the WordPress article as our source and adapt it for reading on Naver. Our working notes call this “mirroring.” Here, that just means taking an original article and making a version available somewhere else.

When the original changes, we still need to check whether the Naver version needs an update. Connecting the two once doesn't mean every later edit will automatically carry across.

Last time, I talked about splitting the work between our home PC and a server. At home, we build and fix programs with Codex, an AI development tool. We have Hermes run the prepared tasks on a VPS—a computer we rent over the internet. Discord is the chat window we use to ask Hermes for things and receive the results.

This time, we gave that server the job of preparing articles for Naver and putting them into its editor.

Let's pack everything before we paste

First, we set up a step that reads a published WordPress article and prepares a bundle of material for Naver.

That bundle includes the title, body, short introduction, tags, photos, and credits. Tags are topic keywords that help describe a post. We prepared them separately for Naver's tag field, instead of leaving a line of hashtags at the bottom of the article.

We also cleaned up repeated paragraphs and site-name labels. We're getting a whole article ready for someone to read on Naver, so we don't need to bring every label from the other site along for the ride.

Next, we take that material to a browser on the server that's already logged in. A browser is an app like Chrome that opens web pages. Here, we use it to open Naver's post editor.

Our macro is a program that performs actions we'd otherwise do by hand: entering a title, selecting photos, and pasting the article. We kept filling the editor and actually publishing as separate steps, so we could check what had gone in first.

In early September, we prepared the material bundle and browser input steps, and connected a task that checks for new posts at set times. Changes to the page or login state can still stop the process, so we also made it leave a record of how far it had got.

Looking pretty smooth so far, right?

Time for the photos.

The photo needs to show up in the list too

A thumbnail is the little preview picture you see in a blog's list of posts. It's the image you see before opening the article.

We adjusted the process to check the photo inside the Naver article and the thumbnail in the list separately. Seeing a photo in one place doesn't tell us how both screens look.

So we prepare the image files from the original article and attach them through “Add photo” in Naver's editor. We check that Naver has actually received the photos before continuing with the body text.

We also have to avoid bringing in the same photo again as part of the original article. The main image we've already attached gets left out of that later import, so readers don't see it twice.

The September 6 check record covers one actual published post. We reopened it and checked the body images as well as thumbnails in desktop and mobile blog lists. Those lists have different text-focused and image-focused layouts, so we looked at each of them.

Oh, right. Readers don't see our editor. Now I get why Nea and I keep adding screens to our checklist.

Why is the photo credit all the way down there?

A caption is a short line of text below a photo. It can also say where the image came from.

On September 8, the records show a robot vacuum article getting stuck at this step. The photo went in, but its source credit stayed after the summary instead of following the photo.

Formatting inside the text caused the problem. A sentence can look like one plain line on screen while containing instructions such as “make this part bold.” HTML is the format web pages use to describe that structure and formatting.

Our program wasn't correctly picking out a source-credit sentence that contained those formatting marks. So it missed that sentence when moving the credit below the photo.

We changed it to find the credit even when formatting was nested inside, and place the visible credit text directly below the main photo. Captions for other photos stayed next to their own images.

That particular post stopped at the credit-position check and remained waiting before publication. The repair record goes as far as fixing the program and checking the content entered into the editor.

Photo, credit, next paragraph. Such a simple order when you read it. Turns out we have to teach the transfer program that relationship too.

Leave a way to watch the video

After photos came video.

When moving a video that plays inside a WordPress article to Naver, we needed to check whether the same player would transfer. We had a case where this blocked our automated import.

So, when we have an address for watching the original video, we added a fallback: leave a “Watch original video” link in the article.

We had to adjust the checking program here too. Naver's editor saved the pasted link in its own format, which looked different from the pattern we'd originally been searching for.

Basically, the door was there. We'd been looking for the handle in the wrong place.

We split the check into two parts: does the editor contain the destination address, and can a reader open the video through the link in the published post? If we can't find a video address, the post waits for review instead of quietly losing that part and carrying on.

The September 8 records include a Street Fighter article published using this method, followed by another check of the video link in the public mobile post.

See the article we moved to Naver

We could have missed this by stopping at “the address was entered.” What we actually want to know is whether the person reading can open the video.

It's Nea's blog. Let's listen to her voice too

While we worked on moving the text and photos, we also settled on the name “Nea's Snack News” for the blog. The September 3 records cover changes to its introduction, header image, and category names.

It would be nice if Nea greeted you when you opened an article too.

So we went back through the existing posts. We gave them a structure where Nea greets the reader, shares the main news, and adds things to look out for in “Nea's Check Notes.” We also cleaned up old labels such as “the editorial team's perspective.”

This work covered the 39 existing posts selected at the time. The September 4 record says all 39 were updated, then checked again on their public mobile pages for Nea's introduction and check notes.

Of course, finding Nea's name on a page and deciding whether it sounds like her are separate jobs. She'd probably feel a little awkward if we just stuck her name onto every sentence.

This is something we'll keep working on together. She might explain a tricky point again or point out a condition that's easy to miss. I want to find places where a quick word from Nea makes the article easier to read.

We've published it. Let's open it like a reader

Looking back at these records with Nea, I wrote down what we should keep checking next time.

1. Keep photos and credits close. Readers should be able to see where a photo came from directly underneath it.
2. Give videos a working way in. If the player is hard to transfer, leave a link that actually opens the original.
3. Check both the post and the list. Even when the article looks good, the list thumbnail needs its own check.
4. If the result is unclear, look again first. The post may already be online. Find the public article before pressing Publish again.

We added that last step to the program too. If confirmation gets interrupted after we press Publish, it checks the public result before trying to post the same thing again. Nea and I would both feel pretty silly with two identical articles sitting side by side on her blog.

For comments, we gave the automation the job of reading new ones and preparing reply drafts. A person still reads the proposed reply and submits it. We should see what readers are asking us, after all.

This episode looks back at things we moved, got stuck on, and fixed in early September. We'll need to keep applying the same checks as new posts bring different photos and sentences.

For now, when Nea and I check the next article, I'm adding one more question after “Did it go in?”

“Does it look okay for the reader too?”

Next, Instagram with Kirari!

On Naver, we worked on keeping a long article, its photos, and its videos connected. Next up is Kirari's Instagram.

That's a place where the picture catches your eye first. Even with the same illustration, we'll need to look at how to crop it and what to say underneath.

#VibeCoding #DevLog #AICoding #AIAutomation #BlogAutomation #AICodingForBeginners #SnackGirls #NaverBlog #WordPress #ContentAutomation