Detect long vowels indicated with 〜

[Tomalak]

If it's feasible, 10ten could detect long vowels in colloquial texts, e.g. "行こ〜" could be identified as "行こう".

This would mean one additional lookup, depending on the character preceding the 〜. This search could probably be limited to Hiragana.

Preceding character	Substitute for ~
あぁかがさざただなはばぱまやゃらわ	あ
いぃきぎしじちぢにひびぴみ　　り	い
うぅくぐすずつづぬふぶぷむゆゅる	う
えぇけげせぜてでねへべぺめ　　れ	い
おぉこごそぞとどのほぼぽもよょろを	う

Some of these combinations are highly unlikely, such as "を～", so the list could be refined to cut back on obviously unnecessary attempts.

[birtles]

We can definitely do it. We already do it for ー, e.g. here:

10ten-ja-reader/src/background/word-search.ts

Lines 66 to 67 in 30d7917

	// Expand ー to its various possibilities
	variations.push(...expandChoon(input));

It's complicated a bit by the fact that お～ can represent おう or おお. (In theory it shouldn't be used for おお but in practice it is. e.g. search for オーサカ and you'll see it used.) So you end up with multiple possibilities to try out.

(I think the actual implementation and tests roughly match your chart.)

[birtles]

Do you have any particular content you're reading where this is common? I don't doubt it's used it's just helpful to have real-life content to refer to when prioritizing, creating test cases, verifying, and later announcing the feature.

[Tomalak]

Things like messaging services and social media tend to have this, on occasion.

[Tomalak]

I think the actual implementation and tests roughly match your chart.

I've taken a look and indeed it's pretty much exactly the same thing. Maybe using [〜ー]+ for the long-vowel mark, instead of just ー+, would already be enough to sufficiently implement this.