case insensitive exact content search

[karottenreibe]

Searching in file names is case-insensitive by default. Searching content with c/ is case-sensitive by default. Intuitively, I'd have expected c/ to also be case insensitive. Adding /i at the end does not turn on case sensitivity as it does for cr/.

broot 1.11.1 on Ubuntu

Suggestions

I see several possible solutions (the one I like most ordered first):

• c/ could be "smart" by default, like vim with set smartcase: If all chars are lowercase in the search, search case-insensitively. If at least one is uppercase, search case-sensitively. This works very well for me in vim, matches my intuitive expectation and makes it trivial to turn on case sensitive search
• if this is not possible, c/ could be case insensitive by default
• If none of the above two can be implemented, c/ should at least accept the /i flag that cr/ accepts to turn case-sensitivity off

[Canop]

There are two kinds of content searches:

1. exact search, fast but absolutely not tunable, set to c/ by default
2. regular expression: a little less fast but very configurable, set to cr/ by default

Is there a reason for not using cr/ in your case ? You can even remap it to a one letter prefix if you want.

[karottenreibe]

@Canop thanks for your reply! Using cr/ means that I have to be aware of regex special characters while searching. While not a backbreaking problem, I find this uncomfortable. Also, my search is not iterative, since it only really does what I want once I have added /i at the end. So "iterative search term refinement" is a lot more clumsy:

• enter partial search, e.g. cr/asdf
• append /i
• now I see first actual results
• move cursor left by two characters
• keep refining search

This is much nicer with smartcasing. Smartcasing was also what I intuitively expected c/ to do (being used to it from vim).

Is there any use case where a "smartcased" c/ search would not be wanted?

As to performance, is case-insensitiveness a big performance problem for content search? I'd have though there should be ways to make case insensitive search fast, e.g. lowercasing the searched file contents (should be O(n) I guess, but haven't measured actual performance of lowercasing). I'd be happy to research this, if you think it's a problem.

[Canop]

OK, I understand why you don't find cr/ ideal.

I won't change the existing fast content search c/ but I'll probably add another search engine for easier case insensitive search.