Bugs and Suggestions

On current version of Telegram, users can not find "pineapple" if the search keyword is "apple". Telegram should support PARTIAL MATCHING on the next version.

Fadi Hannoun Jan 11, 2021

Yes, and I think Telegram should revamp their search algorithms altogether, because sometimes even searching for a word in English yields totally unrelated results.

Edward Li Jan 13, 2021

Wouldn't mind if this will be the paid feature, full text search is essential

Deleted Account Jan 13, 2021

this problem exits with english too, for example you can just type a part of email address or part of word to search , it will get no results , i understand that this kind of search will consume lots of processing power from telegram servers , @telegram you can add a "slow search" that will compare char by char but when you have idle processing power , with giving priority to smaller chats, the search can be done on background and user get notified when the result is available , for general search you can add advance search option to selects only a few folders/chats not 300channel i have joint and never read some of them in months this saves loots of processing power to

artemetra 🇺🇦 Jan 20, 2021

yes!! this feature must be added asap, it would ease up searching in chats so, so much, especially when you know only a part of the word. as an option i suggest placing a switch inside the search menu that would toggle between classic/per-char search algorithms.

Admin Dog Jan 22, 2021

Deleted Account

I am not entirely sure if this problem is related to space usage. I've had issues while searching in Korean (which utilizes spaces), as well as Chinese and Japanese. The problem might be double-byte character set.

Could you describe the issues you had in more detail?

Deleted Account Jan 25, 2021

I know Telegram handle searches on their server, which has a number of benefits; but it's also undeniably true that the current search function is essentially useless for a major part of the userbase, which is a serious issue that has to be fixed. Just look at the number of thumb-ups this thread has, it speaks volumes about how people have been longing for this.
Therefore I have a proposal: just split search function into 2 parts: first is the server-side search that is exactly the same as the current search; then there is this client-side search, which enables user to use the computational power of their own devices to search for partial text in the data that has bee synced to their devices (maybe put up a remider that you will only be able to search for what is already been synced to your device, any chat history that is only on the cloud will not be searched). Then give us a toggle to switch between those 2. It is a compromise from both sides that I think will satisfy most people.
Whether the devs adopt my idea or not, please make this search function revamp a priority, all the affected users will be tremendously grateful!

Deleted Account Jan 26, 2021

Admin Dog

The issue is a little stranger than I had initially thought. It's always been my experience that when I try to search for anything on Telegram, I only get partial results, often missing the message that I was trying to find. I had chalked it up to Telegram not being able to run partial searches, but I just ran a few tests and even that doesn't seem entirely true.

When I am searching for text in Chinese or Japanese, I get almost no result back, as entire sentences are written with no spaces in between words. My experience with Korean is a lot better, as they DO use spaces to break up words, but they don't break up everything. Most notably, (I guess I can only write in English here,,, words in brackets are expressed as a single, unbroken chain of character in Korean)

BE verb is expressed through a suffix added to the noun itself
[Airplane is] flying

Some compund nouns are not separated by spaces
[Airplane ticket]

In this scenario, searching for ”airplane” will return neither of those examples.

....except, that it SOMETIMES does, and I cannot figure out why only SOME of those are returned when searched. All of this ends up giving the users a very inconsistent search experience

Mingye Wang Jan 28, 2021

In addition to allowing naive substring search, true word search can be quite easily implemented with SQLite's FTS tokenizers. The ICU tokenizer works well for a lot of languages. It might oversplit a bit, but that doesn't mess up searching.

Deleted Account Jan 30, 2021

Even though O have my conversations in English mostly, I'd like to see this coming. I can confirm this on Telegram Desktop for Windows.

Show next 100 comments

30614 Cards

Log In