Search: Show Content Items based on model name that is searched

[zcolah]

Problem
I was in the Zesty instance and wanted to quickly see all of my articles. So I went to search and typed articles but only came across 3 content items that had the keyword "Articles" in them. This was despite the fact that there were over 300+ content items under the Articles model in my instance.

Solution
If a user searches for the term "Articles", and if we have a model called "Articles", then show all the content items for that model. See video for more context: https://www.loom.com/share/53463d6c0f134b1d9f5b469c19d1dbbc

[agalin920]

@zcolah. The search results all come from an API call so this should be an instance API ticket.

Additionally some more definition can be done for this. Does this need specific ordering? For example content items with the term 'Article' first and then content items with a model with the term 'Article'.

There might also be speed concerns that will be brought up by the BE team

[giseleblair]

@agalin920 in response to @zcolah 's feature request. This is currently feasible with the search endpoint by using the model zuid. So maybe there can be some sort of matching from the model name to the model zuid that would then return the items. Or maybe this can be a filter option like in Github where we can declare something like model:articles
I don't have an exact solution but here's a video of getting all items from a model:
https://www.loom.com/share/c40970dfed7a4cff8fa9881aa815108c

[finnar-bin]

@agalin920 so client side basically fetch all the models, find which models match the keyword "articles", then return all content that's under that model? Wouldn't we still need to call the search API after we get that model ZUID though to get all the content?

[agalin920]

@theofficialnar Yea. You are correct. I wanted to avoid the all models call but its not that large of a data set

[finnar-bin]

Alright, so I guess my question now is how do we go about ordering the responses @zcolah? Because as it would turn out, there's going to be

• Results from the initial search API call that returns whatever content matches the keyword "articles"
• Results from consequent search API calls that return the content from any model that has the keyword "articles". This could potentially be a lot as well since several models could have the keyword "articles" in their names (ex. News Articles, Blog Articles ...) and each model could have several content.

[giseleblair]

@theofficialnar @zcolah this needs to be scoped out. Not all searches will be for content models. How will the user select that they want to search for a content model? How does the user know what the results are based on their keywords (matching content vs matching content model name)

[finnar-bin]

@giseleblair Global Search Phase 4 adds support for schema in search results, which will let the user search for model names, and will add some buttons to specifically search for what kind of resource the user is searching for.

[zcolah]

@theofficialnar I think your response here is good. I've just updated the ordering slightly to accomodate models, code files, media items, etc:

1. Results from the initial search API call that returns whatever content matches the keyword "articles"
2. Results from the media search API
3. Results from the models API
4. Results from the code files API
5. And then Results from consequent search API calls that return the content from any model that has the keyword "articles". This could potentially be a lot as well since several models could have the keyword "articles" in their names (ex. News Articles, Blog Articles ...) and each model could have several content.

I do understand that it can be a large response, so we could potentially only load the first 5 content items for that model, and then only as the user scrolls further, make more API calls, that way we do not call all the data at once.

[giseleblair]

This flow may not match the intended query. The user wants to see content items that fall under the model name "articles" If the user filters by "Schema" what will the results be? Matching model names or matching content items with that model name? Intuitively, the user would click "I am looking for Content" since they want to pull up all article content items. Which would put us back at the original confusing flow that @zcolah described

[zcolah]

@giseleblair this enhancement is only for searches related to Content so as to present more relevant search results. It is not related to any filters or accelerators in the experience. Hence I think we can go forward with the experience described above.

[finnar-bin]

I do understand that it can be a large response, so we could potentially only load the first 5 content items for that model, and then only as the user scrolls further, make more API calls, that way we do not call all the data at once.

@zcolah the thing is that I don't think we can really limit it to just only load the first 5 content items first. Since if for example the user searches for "articles", I'm going to have to check client-side to see which models contain the word "articles". So, if I get like let's say 20 results, I'm then going to have to also make 20 api calls to the search endpoint to get all the content items inside those 20 models and only then be able to figure out which 5 content items from all collated content items (from the search api call to match the keyword AND the consecutive api calls for the model's content items) to show as the "suggestions".

So yeah, I guess the question here is, is it worth potentially slowing down the time to show the users the suggestions for this or not? Granted it would only be slow if there's a lot of api calls needed due to the keyword matching several models or if the user has bad internet.

[finnar-bin]

@zcolah can you provide a design on how the results would look like on the search page? Also, wouldn't it be confusing if the user searches for "articles" then get a content item with a title of "Beach" if there's a content item named beach on one of the models that contain the keyword articles? I'm just wondering why we'd like to show all the content items from models even if those content item names are not relevant to the keyword when the user can just go to the content page and then open up that model and see the content items instead?

[giseleblair]

+1 on @theofficialnar comment
The query returns all items that have matching meta content. If the user is searching for all items under a content model name then that's a different query and a separate endpoint