What’s the best way to filter a list based on some predicate?
I was able to use “extract value where … from somelist” for some basic arithmetic, but when I throw a function into the predicate I get unexpected results (“value < f[value]” always seems to be true even though it definitely isn’t).
I’m wondering if I’ve misunderstood the “where” clause and it is only supposed to be used in table columns or something.
In the context of query clauses, column names refer to the entire column as a list. The "where" clause expects an expression which yields a list of boolean values. Arithmetic operators like =, <, and + conform over list-list and list-scalar arguments.
Depending on how a predicate is written, it might naturally generalize to operating on lists for the same reason. For example:
on iseven x do 0=2%x end iseven[5] # 0 iseven[11,22,33] # (0,1,0) extract value where iseven[value] from 11,24,3,8 # (24,8)
If a predicate is only designed to operate on a single scalar value at a time, you can use the "@" operator to apply it to each element of a column, like so:
on seconde x do x[1]="e" end
extract value where seconde@value from "Lemon","Lime","Soda","Demon"
# ("Lemon","Demon")
This shorthand is semantically equivalent to
extract value where each v in value seconde[v] end from "Lemon","Lime","Soda","Demon"
(And of course in this particular case the "like" operator would be simpler:)
extract value where value like ".e*" from "Lemon","Lime","Soda","Demon"
Does that clear things up?
Thanks! That does explain why it seemed to work sometimes and not others. I will give it another try. For the record, It looks like my skimming skills failed me, because I now see where this is explicitly noted in the docs:
When computing columns, you're working with lists of elements, and taking advantage of the fact that primitives like<and+automatically "spread" to lists. When performing comparisons, be sure to use=rather than~! If you want to call your own functions- say, to average within a grouped column- write them to accept a list
It even calls out “where” in the next paragraph.