Declarative CSV codecs
csvt: rename get_col → col, fix docs, shorten test name (merlint)
+17
-17
+2
-2
fuzz/fuzz_csvt.ml
+2
-2
fuzz/fuzz_csvt.ml
···
55
55
56
56
(* {1 Crash safety: arbitrary bytes never cause a crash} *)
57
57
58
-
let test_no_crash_on_arbitrary_input bytes =
58
+
let test_no_crash_arbitrary bytes =
59
59
(* Decoding arbitrary bytes should either succeed or return Error,
60
60
never crash *)
61
61
ignore (Csvt.decode_string int_codec bytes);
···
222
222
( "csvt",
223
223
[
224
224
test_case "no crash on arbitrary" [ bytes ]
225
-
test_no_crash_on_arbitrary_input;
225
+
test_no_crash_arbitrary;
226
226
test_case "int roundtrip" [ int ] test_int_roundtrip;
227
227
test_case "float roundtrip" [ float ] test_float_roundtrip;
228
228
test_case "bool roundtrip" [ bool ] test_bool_roundtrip;
+6
-6
lib/csvt.ml
+6
-6
lib/csvt.ml
···
272
272
273
273
The paper's [mem] combinator, adapted for CSV. *)
274
274
275
-
let get_col name t = Row.(obj Fun.id |> col name t ~enc:Fun.id |> finish)
275
+
let col name t = Row.(obj Fun.id |> col name t ~enc:Fun.id |> finish)
276
276
277
277
(* {1 Update support} *)
278
278
279
-
let find_col_by_name header name =
279
+
let col_by_name header name =
280
280
let name = String.trim name in
281
281
let n = Array.length header in
282
282
let rec go i =
···
287
287
go 0
288
288
289
289
let update_col name codec f header row =
290
-
let idx = find_col_by_name header name in
290
+
let idx = col_by_name header name in
291
291
if idx < 0 then Error (Missing_column name)
292
292
else if idx >= Array.length row then
293
293
Error
···
303
303
Ok result
304
304
305
305
let delete_col name header row =
306
-
let idx = find_col_by_name header name in
306
+
let idx = col_by_name header name in
307
307
if idx < 0 then (header, row)
308
308
else
309
309
let n = Array.length header in
···
330
330
r_dec : ('a, 'a) dec_fun;
331
331
}
332
332
333
-
let find_col header name =
333
+
let resolve_col header name =
334
334
let name = String.trim name in
335
335
let n = Array.length header in
336
336
let rec go i =
···
349
349
if i >= n then Ok { r_indices = indices; r_mems = mems; r_dec = m.dec }
350
350
else
351
351
let (Mem_dec mm) = mems.(i) in
352
-
let idx = find_col header mm.name in
352
+
let idx = resolve_col header mm.name in
353
353
indices.(i) <- idx;
354
354
if idx < 0 && Option.is_none mm.dec_absent then
355
355
Error (Missing_column mm.name)
+4
-4
lib/csvt.mli
+4
-4
lib/csvt.mli
···
77
77
[false]. *)
78
78
79
79
val nullable_float : float t
80
-
(** Like {!float} but treats ["NULL"] and empty strings as [nan]. *)
80
+
(** [nullable_float] is like {!float} but treats ["NULL"] and empty strings as [nan]. *)
81
81
82
82
val nullable_int : int t
83
-
(** Like {!int} but treats ["NULL"] and empty strings as [-1]. *)
83
+
(** [nullable_int] is like {!int} but treats ["NULL"] and empty strings as [-1]. *)
84
84
85
85
val option : 'a t -> 'a option t
86
86
(** [option t] treats ["NULL"] and empty strings as [None], otherwise decodes
···
104
104
105
105
(** {1:query Query} *)
106
106
107
-
val get_col : string -> 'a t -> 'a t
108
-
(** [get_col name t] creates a single-column row codec that extracts column
107
+
val col : string -> 'a t -> 'a t
108
+
(** [col name t] creates a single-column row codec that extracts column
109
109
[name] decoded with [t]. Useful for projecting a single column from a CSV
110
110
file without defining a full row type. *)
111
111
+5
-5
test/test_csvt.ml
+5
-5
test/test_csvt.ml
···
646
646
647
647
let test_get_col_int () =
648
648
let csv = "id,name,score,active\n1,alice,95.5,true\n2,bob,87.3,false\n" in
649
-
let id_codec = Csvt.get_col "id" Csvt.int in
649
+
let id_codec = Csvt.col "id" Csvt.int in
650
650
match Csvt.decode_string id_codec csv with
651
651
| Error e -> Alcotest.failf "get_col id: %s" (Csvt.error_to_string e)
652
652
| Ok ids -> Alcotest.(check (list int)) "ids" [ 1; 2 ] ids
653
653
654
654
let test_get_col_string () =
655
655
let csv = "id,name,score,active\n1,alice,95.5,true\n2,bob,87.3,false\n" in
656
-
let name_codec = Csvt.get_col "name" Csvt.string in
656
+
let name_codec = Csvt.col "name" Csvt.string in
657
657
match Csvt.decode_string name_codec csv with
658
658
| Error e -> Alcotest.failf "get_col name: %s" (Csvt.error_to_string e)
659
659
| Ok names -> Alcotest.(check (list string)) "names" [ "alice"; "bob" ] names
660
660
661
661
let test_get_col_missing () =
662
662
let csv = "id,name\n1,alice\n" in
663
-
let score_codec = Csvt.get_col "score" Csvt.float in
663
+
let score_codec = Csvt.col "score" Csvt.float in
664
664
match Csvt.decode_string score_codec csv with
665
665
| Ok _ -> Alcotest.fail "expected error for missing column"
666
666
| Error (Csvt.Missing_column "score") -> ()
667
667
| Error e -> Alcotest.failf "unexpected error: %s" (Csvt.error_to_string e)
668
668
669
669
let test_get_col_encode () =
670
-
let codec = Csvt.get_col "v" Csvt.int in
670
+
let codec = Csvt.col "v" Csvt.int in
671
671
let h = Csvt.Private.encode_header codec in
672
672
Alcotest.(check (array string)) "header" [| "v" |] h;
673
673
let row = Csvt.Private.encode_row codec 42 in
674
674
Alcotest.(check (array string)) "row" [| "42" |] row
675
675
676
676
let test_get_col_col_names () =
677
-
let codec = Csvt.get_col "score" Csvt.float in
677
+
let codec = Csvt.col "score" Csvt.float in
678
678
Alcotest.(check (list string)) "col_names" [ "score" ] (Csvt.col_names codec);
679
679
Alcotest.(check int) "col_count" 1 (Csvt.col_count codec)
680
680