How to Loop Through a Workflow in Bubble.io 

Hey thanks for the kind words, I'll try and answer but if you can point to the timestamp in the video that would help me be more precise. What it sounds like you're asking about is how Bubble parses JSON in their dynamic blue (or red if it's not fully complete!) expressions. Here's how it works: first, whenever you're using a dynamic expression in Bubble, it has an expected data format to receive. It could be things like a number, a text, a yes/no boolean, AND on top of that it could either a single item of number/text or a list of items of numbers/text. That's one part. Next, whenever we're at a level within the JSON hierarchy where is a list of things below it, it will include an "each item" option and for example, if we did each item's comment, that is how we'd create a list of all the comments. But if it were a single item it is expecting, then it's going to force a choice of :firstitem or item # 4 or whatever. That's a rough overview but without knowing the exact time stamp let's see if that oriented things for how they're working. Hope that helps.

Hey thanks for the kind words. In the structured JSON response that comes back from RU-vid or any API, Bubble is going to transform it to a list of things you can operate on, i.e. use one of the colon statements like :count. If you can know the dynamic count of however many items there are then you're good to go. So it doesn't matter, as you say, "and you do not know how many there are" because that's not true, you do know if you use the :count operator. I will do my best to explain where 13:36 at where I say "actually you know what" I am clicking that list and you can see the :count operator displayed there. If I had clicked on :count in that exact moment, it's what you'd put into the total_item_count field and it would know how many loops to run through regardless of what comes back from the API for a given time period. Hope that helps.

Hey you're welcome. Good question, you would create a loop within a loop and the inner loop's total_item_count variable is the 'count' of the varying number of items that come back. You'd pass over whatever data needs processing to the inner loop and store the data as it processes each iteration in whatever format makes sense (i.e. maybe you're building a list and using the 'add' for each pass it takes through the inner loop, but hard to say for sure without knowing your exact setup). Also, you could always tell chatgpt or claude your understanding of the single loop and ask it for the structure of the loop within a loop to get clarity on the structure of what's going on. Then pair that knowledge with this video's knowledge and you'd be pretty close to piecing it together. Hope that helps. If you're open to sharing, let us know how it worked out for you, I'm sure others are interested in solving the same problem you're facing.

I love NSFW content, mainly memes. Am I allowed to say that?

Do you know of an API that might process images you give it to return a NSFW-filtered image back? That would be the first step.

I believe Clarifai and Microsoft Azure allow you receive tags from images that are being uploaded. I wish to filter any tags that have NSFW content. I wish to keep my users safe and to ensure that kids don't use the application for nefarious sharing.

@@roshalangovender7444 I'm unsure about those personally so couldn't speak to them, but do see Google has an image detector in their Cloud Vision API here cloud.google.com/vision/docs/detecting-safe-search. My suggestion would be to view and study the implementation of the videos at ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-1XUu7-yIoUY.html and ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-Qma2aDLETqw.html. Then, instead of implementing for the RU-vid API, you'd turn on the Cloud Vision API and implement for that. Note you'll make the method be POST request to URL of 'vision.googleapis.com/v1/images:annotate'. In the Body, you will need to provide the details of what you want the Vision API to do. This will be in JSON format. You will need to encode your image into base64 format (use something like www.base64encode.org/) and include it in the request. Check the Google Cloud Vision API documentation to find out exactly what the body of your request should look like. Click 'Initialize call' to make sure everything is set up correctly. If it is, you should see a successful response from the Google Vision API. Good luck.

Hi can you include the timestamp in the video you're asking about? Or do you mean how did I get access to the RU-vid API? If so, search my channel for the RU-vid API video.

8:04 on this video, thanks@@NoCodeAcademy

Hey Joaquin, to understand you correctly, you're looping through a list of data that has both Name and Content values? So your list of name data looks like: Buyer_Persona_001 Buyer_Persona_002 Buyer_Persona_003 Buyer_Persona_004 and your list of Content values looks like: Young, milenials with interest in basket Old, milenials with interest in basket Young, gen z with interest in basket Old, gen z with interest in basket Or it looks like something else? I'll stop there bc I took large assumptions to write the above, so first to confirm that. Next, onto your point about 'user writes "create an ad for #buyer_persona_001"', if the user is specifying which buyer_persona, wouldn't you just have an option set for Name that has all the pre-defined values. Then, if the user is selecting one, you know which one it is, so would there still be a need for a loop to go through it? Or why not use an option set that has Content attached as an Attribute to a given Name's value? The option set would be Personas with Buyer_Persona_001, Buyer_Persona_002, Buyer_Persona_003, Buyer_Persona_004 as the options and then Attributes of Content attached to each option set item for Young, milenials with interest in basket.. Old, milenials with interest in basket.. Young, gen z with interest in basket.. Old, gen z with interest in basket.. etc. Hope that helps.

Hey@@NoCodeAcademy , thanks for taking your time to reply. Regarding the first part: correct. That's exactly what I meant. Regarding the idea of skipping the loop by using option sets, I think the issue here is that the user is the person creating these variables. It's a genertive ai app for digital marketers where they can create content using their own variables. For example they can easily create: NAME 1. Brand_Tone_Voice 2. Buyer_persona_001 3. Cold_Emails_Benchmark CONTENT 1. The brand tone of voice is friendly, warm and elegant. Lorem ipsum... 2. The buyer persona for this campaign is an audience of young milenials with interests in... 3. (here the content would be a list of 10 cold emails to use as benchmark) And then they can chat with gpt like this: "Create a cold email campaign for this audience Buyer_persona_001 usign this tone of voice Brand_Tone_Voice. Here some inspiration so you can imitate the style: Cold_Emails_Benchmark" The loop would be to send the complete message to chat gpt replacing "NAME" with "CONTENT". This is just a simple example but it can get very complex when you include all the channels (facebook, tik tok, google, email mkt, blog, etc). Thanks again for your help.

@@JoaquinTorroba any luck on a solution? Because sometimes the answers come from the act of explaining the problem clearly. Here's more to consider -- if you have a datatype of PromptVariables that contains Name as a text field and Content as a text field, when you are sending the prompt to ChatGPT, do the substitution directly before sending the prompt. Now, I can see that the question arises of how to determine the substitution of the Name for the Content. But if it were me, I would work with the find and replace operator doing a search for a PromptVariable where the Name value matched part of the statement "Create a cold email campaign for this audience Buyer_persona_001 usign this tone of voice Brand_Tone_Voice. Here some inspiration so you can imitate the style: Cold_Emails_Benchmark" where you want to make the substitution. You would also want a safety mechanism so people can't duplicate Name values.. but that's easy it's the same as making it so people can't have duplicate usernames on a platform. Just append something to the end of the Name value when a duplicate is found. Hope that helps.

Thanks a lot for the follow up@@NoCodeAcademy . I could solve this but in a non-traditional way. My idea was to do exactly what you say: a find&replace loop but it takes more time than what I was expecting (a couple of seconds but in kind of ruins the UX). Anyway, I think I have find a "temporary" solution. I hope I can learn a lot more in the coming months so this becomes more robust. Again: thanks a lot for the follow up!