Improve our search results

[connyduck]

The Mastodon search can now do a lot more things than previously: mastodon/mastodon#10091

We should definitely take advantage of that, and, for example, present our search results in three (posts, users, hashtags) infinitely scrolling tabs. Or present the results as we do now, but with a "more" button for each section of the results.

Of course we need to wait until the majority of instances run this feature.

[pandasoft0]

I checked the search APIv2.

1. What is prefer solution:
   a) show first 5 items per category (statuses, accounts, hashtags) with "more" button below each category. On More click replace this summary fragment with category fragment and show infinite scroll list with items of selected category only;
   b) show 3 tabs at top, and show infinite scroll lists per tab.
   JFYI: in APIv2 we can't select what items to load. We can specify limit and offset, but anyway - it will load items for all 3 categories: accounts, statuses, hashtags.

2. Should i use the exists fragment and adapter to show statuses or create a new adapter in the same way i did for report's statuses? Should i implement all features of exists statuses adapter at the new?

3. Should i update exists activity+fragment code or create a new activity with fragments according to selected solution at p.1?

[connyduck]

I prefer solution B. Only showing results of one type after "more" clicked is a bad user experience.

I think the 3 tabs and the search bar should belong to the activity, and the tab contents are three different fragments. They should all share one ViewModel probably.
The ViewHolders we have for statuses and accounts should definitely be reused, other than that you can decide.

[pandasoft0]

@connyduck I am trying to use v2/search at mastodon.social but result looks strange.
So, first request. I want to get first 10 items per category:

https://mastodon.social/api/v2/search?q=test&resolve=true&limit=10&offset=0&following=false 
{
  "accounts": [
    {
      "id": "963"
    },
    {
      "id": "179860"
    },
    {
      "id": "1184"
    },
    {
      "id": "179862"
    },
    {
      "id": "820018"
    },
    {
      "id": "248304"
    },
    {
      "id": "725124"
    },
    {
      "id": "298091"
    },
    {
      "id": "217924"
    },
    {
      "id": "61339"
    }
  ],
  "statuses": [
    {
      "id": "102115228601373686"
    },
    {
      "id": "102146124881012866"
    },
    {
      "id": "102143335650922023"
    },
    {
      "id": "102246304503430384"
    },
    {
      "id": "102100944210578451"
    },
    {
      "id": "101836442594147121"
    },
    {
      "id": "102116886680668226"
    },
    {
      "id": "102117212948319447"
    },
    {
      "id": "102117313703633435"
    }
  ],
  "hashtags": [
    {
      "name": "test"
    },
    {
      "name": "test좋아"
    },
    {
      "name": "testてすと"
    },
    {
      "name": "testテスト"
    },
    {
      "name": "testちんぽっﾎﾞﾛﾝｯ"
    },
    {
      "name": "test1"
    },
    {
      "name": "test_1"
    },
    {
      "name": "test_111"
    },
    {
      "name": "test12"
    },
    {
      "name": "test123"
    }
  ]
}

I removed all fields except ids.
It returns 10 accounts, 10 hashtags and 9 statuses.
Ok, i want to load 5 items since 9 (PositionalDataSource behaviour)

https://mastodon.social/api/v2/search?q=test&resolve=true&limit=5&offset=9&following=false
{
  "accounts": [
    {
      "id": "963"
    },
    {
      "id": "179860"
    },
    {
      "id": "179862"
    },
    {
      "id": "1184"
    },
    {
      "id": "725124"
    }
  ],
  "statuses": [
    {
      "id": "102115228601373686"
    },
    {
      "id": "102146124881012866"
    },
    {
      "id": "102143335650922023"
    },
    {
      "id": "102246304503430384"
    },
    {
      "id": "102100944210578451"
    }
  ],
  "hashtags": [
    {
      "name": "test"
    },
    {
      "name": "test좋아"
    },
    {
      "name": "testてすと"
    },
    {
      "name": "testテスト"
    },
    {
      "name": "testちんぽっﾎﾞﾛﾝ"
    }
  ]
}

It returns some of items from the first request, but at the different order. looks like it ignores the offset and has no order... I don't know how to use it for infinite scroll. Have you ideas what i am doing wrong?

[pandasoft0]

Hm... Looks like search endpoint has an undocumented parameter "type"="statuses|accounts|hashtags" code and it works more accuracy if this parameter specified.

[connyduck]

Its probably fine to send three search requests, one for each tab? You could also look how Fedilab handles this, it has almost the same setup.

[pandasoft0]

@connyduck Yes, i started to work on the task. But i will need to change the adapter for statuses. It uses PagedList Adapter, but i want to change it to simple Adapter to support changes of statuses.
I will try to do that on this week.

[connyduck]

#1327