csv.bqn doesn't handle empty fields at the end of data correctly

[shnarazk]

Hi,
I found Split in csv.bqn converts "ab,,bc," to ⟨ "ab" ⟨⟩ "bc" ⟩ instead of ⟨ "ab" ⟨⟩ "bc" ⟨⟩ ⟩.
And it emits the following error by "ab,,bc,,":

Error: ⊔: ≠𝕨 must be either ≠𝕩 or one bigger (5≡≠𝕨, 3≡≠𝕩)
test/../csv.bqn:22:
    (0∾s/+`e<n) ⊔ (d(¬⊸×-⊣)+`s) ⊔ 𝕩
                ^
test/csv.bqn:19:
  ! ⟨⟨"ab",⟨⟩,"bc",⟨⟩,⟨⟩⟩⟩ ≡ c.Split "ab,,bc,,"
                               ^^^^^

So I came up with the following.

$ diff -u csv.bqn csv-rev.bqn
--- csv.bqn     2023-07-20 07:32:01
+++ csv-rev.bqn      2023-07-20 14:19:51
@@ -24,7 +19,7 @@
   e ← ≠`q       # Escaped characters
   s ← e<c∨n     # Split points
   d ← s∨q>»e<q  # Characters to be dropped
-  (0∾s/+`e<n) ⊔ (d(¬⊸×-⊣)+`s) ⊔ 𝕩
+  (0∾s/+`e<n) ⊔ (1++´s)↑(d(¬⊸×-⊣)+`s) ⊔ 𝕩
 }
                                                                                        
 # Parse list of lines, e.g. from •file.Lines

This seems to work well. But I don't know if it's the best solution... 😄

[mlochbaum]

That fix is fine, although I might shorten 1++´s to 1+´s. I usually try to build the length into the left argument somehow, so that's what I did instead.

[shnarazk]

Wow. Anyway, thank you.