Declarative CSV codecs
Add update_col and delete_col to csvt
Typed column updates: decode → transform → re-encode a specific
column by name. delete_col removes a column from header and row.
4 new tests (50 total).
+107
+46
lib/csvt.ml
+46
lib/csvt.ml
···
242
242
243
243
let get_col name t = Row.(obj Fun.id |> col name t ~enc:Fun.id |> finish)
244
244
245
+
(* {1 Update support} *)
246
+
247
+
let find_col_by_name header name =
248
+
let name = String.trim name in
249
+
let n = Array.length header in
250
+
let rec go i =
251
+
if i >= n then -1
252
+
else if String.equal (String.trim header.(i)) name then i
253
+
else go (i + 1)
254
+
in
255
+
go 0
256
+
257
+
let update_col name codec f header row =
258
+
let idx = find_col_by_name header name in
259
+
if idx < 0 then Error (Missing_column name)
260
+
else if idx >= Array.length row then
261
+
Error
262
+
(Truncated_row { row = 0; expected = idx + 1; got = Array.length row })
263
+
else
264
+
let s = row.(idx) in
265
+
match decode_field codec s with
266
+
| Error msg -> Error (Bad_value { row = 0; column = name; value = s; msg })
267
+
| Ok v ->
268
+
let v' = f v in
269
+
let result = Array.copy row in
270
+
result.(idx) <- encode_field codec v';
271
+
Ok result
272
+
273
+
let delete_col name header row =
274
+
let idx = find_col_by_name header name in
275
+
if idx < 0 then (header, row)
276
+
else
277
+
let n = Array.length header in
278
+
let new_header =
279
+
Array.init (n - 1) (fun i ->
280
+
if i < idx then header.(i) else header.(i + 1))
281
+
in
282
+
let nr = Array.length row in
283
+
let new_row =
284
+
Array.init
285
+
(max 0 (nr - 1))
286
+
(fun i ->
287
+
if i < idx then row.(i) else if i + 1 < nr then row.(i + 1) else "")
288
+
in
289
+
(new_header, new_row)
290
+
245
291
(* {1 Header resolution} *)
246
292
247
293
type header = string array
+20
lib/csvt.mli
+20
lib/csvt.mli
···
157
157
val decode_row : 'a resolved -> int -> string array -> ('a, error) result
158
158
(** [decode_row resolved row_num fields] decodes a single CSV row. *)
159
159
160
+
(** {1:update Update} *)
161
+
162
+
val update_col :
163
+
string ->
164
+
'a t ->
165
+
('a -> 'a) ->
166
+
header ->
167
+
string array ->
168
+
(string array, error) result
169
+
(** [update_col name codec f header row] decodes column [name] from [row] using
170
+
[codec], applies [f] to the typed value, re-encodes, and returns the updated
171
+
row. Other columns are unchanged. *)
172
+
173
+
val delete_col : string -> header -> string array -> header * string array
174
+
(** [delete_col name header row] removes the column [name] from both the header
175
+
and the row. Returns the new header and row. If [name] is not found, returns
176
+
both unchanged. *)
177
+
178
+
(** {1:encode Encoding} *)
179
+
160
180
val encode_header : 'a t -> header
161
181
(** [encode_header t] returns the CSV header for encoding. *)
162
182
+41
test/test_csvt.ml
+41
test/test_csvt.ml
···
672
672
Alcotest.(check (list string)) "col_names" [ "score" ] (Csvt.col_names codec);
673
673
Alcotest.(check int) "col_count" 1 (Csvt.col_count codec)
674
674
675
+
let test_update_col_int () =
676
+
let header = [| "id"; "name"; "score" |] in
677
+
let row = [| "1"; "alice"; "95" |] in
678
+
match Csvt.update_col "score" Csvt.int (fun x -> x + 5) header row with
679
+
| Ok row' ->
680
+
Alcotest.(check string) "id unchanged" "1" row'.(0);
681
+
Alcotest.(check string) "name unchanged" "alice" row'.(1);
682
+
Alcotest.(check string) "score updated" "100" row'.(2)
683
+
| Error e -> Alcotest.failf "update_col: %s" (Csvt.error_to_string e)
684
+
685
+
let test_update_col_missing () =
686
+
let header = [| "id"; "name" |] in
687
+
let row = [| "1"; "alice" |] in
688
+
match Csvt.update_col "score" Csvt.int (fun x -> x + 5) header row with
689
+
| Ok _ -> Alcotest.fail "should fail on missing column"
690
+
| Error (Csvt.Missing_column "score") -> ()
691
+
| Error e -> Alcotest.failf "wrong error: %s" (Csvt.error_to_string e)
692
+
693
+
let test_delete_col () =
694
+
let header = [| "id"; "name"; "score" |] in
695
+
let row = [| "1"; "alice"; "95" |] in
696
+
let header', row' = Csvt.delete_col "name" header row in
697
+
Alcotest.(check int) "header len" 2 (Array.length header');
698
+
Alcotest.(check int) "row len" 2 (Array.length row');
699
+
Alcotest.(check string) "h0" "id" header'.(0);
700
+
Alcotest.(check string) "h1" "score" header'.(1);
701
+
Alcotest.(check string) "r0" "1" row'.(0);
702
+
Alcotest.(check string) "r1" "95" row'.(1)
703
+
704
+
let test_delete_col_missing () =
705
+
let header = [| "id"; "name" |] in
706
+
let row = [| "1"; "alice" |] in
707
+
let header', row' = Csvt.delete_col "score" header row in
708
+
Alcotest.(check int) "header unchanged" 2 (Array.length header');
709
+
Alcotest.(check int) "row unchanged" 2 (Array.length row')
710
+
675
711
let suite =
676
712
( "csvt",
677
713
[
···
735
771
Alcotest.test_case "get_col missing" `Quick test_get_col_missing;
736
772
Alcotest.test_case "get_col encode" `Quick test_get_col_encode;
737
773
Alcotest.test_case "get_col col_names" `Quick test_get_col_col_names;
774
+
(* Update *)
775
+
Alcotest.test_case "update_col int" `Quick test_update_col_int;
776
+
Alcotest.test_case "update_col missing" `Quick test_update_col_missing;
777
+
Alcotest.test_case "delete_col" `Quick test_delete_col;
778
+
Alcotest.test_case "delete_col missing" `Quick test_delete_col_missing;
738
779
] )