Currently, when you search a word to add to a lesson and there are multiple options, there is no way to pick the "best" one.
Using "Add all" will add all the options, and using "Add all without duplicates" will not add those with multiple options.
I suggest a third "Add all exact matches" which will take the option that:
Matches the spelling (kanji, katakana, okurigana) exactly. This may require that e.g どきどき/どきどき be allowed to specify the spelling is all hiragana.
Matches the reading exactly if there is more than one possible and one is specified e.g. 尊い/とうとい and not たっとい.
Is not conjugated.
And if nothing matches, then don't add it. Basically, add the word if it matches a dictionary entry exactly.
Example of current result for 居る/いる, where this new action will add the first entry, which is 居る with reading いる in JMDict (though tagged as usually kana).
One more thing that might be bit more niche but maybe easier to implement: match by JMDict sequence id. For example, コロコロ/コロコロ/1004840 matches the entry with sequence id 1004840 (the onomatopoeia) with spelling コロコロ and reading コロコロ. As far as I can think of, this might be the easiest way to disambiguate with another entry with the same spelling and reading, as コロコロ also matches an animal.
the icecream and さぼる ones are working now. The OUT is not going to work for the time being, because the system is expecting JAPANESE/JAPANESE for this specific tool, and those weird ones like ＯＵＴ just don't work well with the search system.
As to adding edict entities, it's probably not something I'll be adding at the moment.
煙草/タバコ finds たばこ but not タバコ, while タバコ/タバコ does not find anything at all. The JMDict entry contains 煙草 as one of the spellings, and both タバコ and たばこ as readings, for reference.
