The replace trick of Raul is nice.
After my initially failed (like Aaron's) attempt to use replace, I
solved the problem in a more array oriented way:
p1 =: [: +/ ".@({.,{:)@:(#~e.&Num_j_);._2
num =: ;:'zero one two three four five six seven eight nine'
fix =: [: ([: i:&1"1@|: 1,num (E.~>)~"0 1{.)}@|:,"0 1&Num_j_
p2 =: p1@fix
The juicy part is "(E.~>)~"0 1" which, when passed num on the left and a
string y on the right creates a boolean matrix of shape 10,#y , indicating
where which number is found. The rest massages it in the form required by
composite item (monadic u}) to replace the locations by digits: eighthree
=> 8igh3hree
On Reddit, I encountered a cute alternative for my "(E.~>)~"0 1": E.S:0 .
Jan-Pieter
On Sat, 2 Dec 2023, 03:53 Aaron Ash, <aaron....@gmail.com> wrote:
> I initially hit the same issue and then changed my approach to avoid using
> stringreplace.
> The logic I ended up with was: find the first and last match of any of the
> digits or number words,
> combine those two into a single string and now replace words with digits on
> those knowing that
> there's no chance of overlaps.
>
>
> SPOILERS BELOW:
>
>
> s =: 0 : 0
>
> two1nine
>
> eightwothree
>
> abcone2threexyz
>
> xtwone3four
>
> 4nineeightseven2
>
> zoneight234
>
> 7pqrstsixteen
>
> )
>
>
> NB. initial "replace" verb that had the same issue
>
>
> r=:rplc&(','cut'zero,0,one,1,two,2,three,3,four,4,five,5,six,6,seven,7,eight,8,nine,9')
>
>
>
> nums=:','cut'one,two,three,four,five,six,seven,eight,nine,1,2,3,4,5,6,7,8,9'
>
> NB. working "replace" verb
>
> r=:{{
>
> f=.(nums{~]i.<./@-.&0) ,(<./@-.&0)"1 I.E.&(' ',y) &> nums
>
> l=.(nums{~]i.>./@-.&0) ,>./"1 I.E.&(' ',y) &> nums
>
> ". -.&' ' (>f,l) rplc
>
> (','cut'zero,0,one,1,two,2,three,3,four,4,five,5,six,6,seven,7,eight,8,nine,9')
>
> }}
>
>
> +/ r &> LF cut s
>
> 281
>
>
>
>
> On Sat, Dec 2, 2023 at 12:29 PM David Pinchbeck <dpinc...@gmail.com>
> wrote:
>
> > I'm trying to improve my J using Advent of Code, but am already a bit
> > stuck on the second part of day 1. It seems late enough in the day now
> to
> > ask this question without spoiling anything.
> >
> > On the example code one needs to turn the string 'eightwothree' into
> '8wo3'
> > I tried the following, which fails because the priority of stringreplace
> > comes from the left argument ordering:
> >
> > digs=:'0123456789'
> >
> > s=:'zero';'one';'two';'three';'four';'five';'six';'seven';'eight';'nine'
> >
> > substr=:(s,. <"0 digs)
> >
> > g=:{{ substr stringreplace y}}
> >
> > g 'eightwothree'
> >
> > eigh23
> >
> > How can I get the earliest appearance on the right side to be replaced
> > first instead?
> >
> > Thanks,
> > David
> > --
> > David Pinchbeck
> > Associate Professor of Mathematics
> > St. Joseph's College
> > Standish, ME
> > ----------------------------------------------------------------------
> > For information about J forums see http://www.jsoftware.com/forums.htm
> >
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
