Ignore accented letters

[juanfal]

I love bfs and I using it since I met it, and telling others to do so. The only thing I really miss, and I sometimes resort to Spotlight terminal search, much much slower and more difficult to control, is when I want to ignore accented chars. That is, when mecánica.pdf is there but you don't know if it has the accent or not, so

bfs /Volumes/Libros/BOOKS -iregex ".*$*.*" 2>/dev/null

would fail. I miss -iuregex

[tavianator]

I can definitely see why that would be useful! It's also probably hard to implement.

Related: sharkdp/fd#638

[juanfal]

I totally agree, IMHO the problem is on the roof of the implementation of regular expressions, which are completely lacking in this very concept. Using

-E -iregex 'corazo.?n'

would match correctly corazon as well as corazón what is the closest I can get to it.

So

-E -iregex 'c.?o.?r.?a.?z.?o.?n.?'

could make a poor -iuregex

[tavianator]

That will only work if the ó is decomposed into o + a combining accent. This happens naturally on macOS due to NFD normalization, but on other platforms it's more likely a precomposed ó will be used, which won't be matched by o.?. -iregex 'coraz[[=o=]]n should work in both cases, if the regex engine supports it.

[tavianator]

Having just tested it, it appears Oniguruma does not support that syntax: kkos/oniguruma#288

[tavianator]

One potential idea would be to use https://github.com/laurikari/tre (agrep's regex implementation) to do approximate matching. E.g.

$ echo corazón | agrep -0 'corazon'  # Exact match
$ echo corazón | agrep -1 'corazon'  # Approximate match, within 1 edit
corazón

It could be something like:

$ bfs -aregex 1 '.*/corazon'
./corazon
./corazón

[juanfal]

It is a great approach! It also returns different letters as good but, it is a good compromise. $ echo corazén | agrep -1 'corazon' corazén The only (heuristic) solution I see based on what I have found (iconv, uregex <https://www.unicode.org/reports/tr18/#character_ranges>, equivalent non-accented chars <https://www.drillio.com/en/2011/java-remove-accent-diacritic/>, another <https://stackoverflow.com/questions/249087/how-do-i-remove-diacritics-accents-from-a-string-in-net>) is the heuristic list like the next into account when using this -uregex [ { "äæǽ", "ae" }, { "öœ", "oe" }, { "ü", "ue" }, ...