Declarative CSV codecs
irmin: implement set/flush/empty, rename test, zero TODOs
+24
-7
+14
-2
fuzz/fuzz_csvt.ml
+14
-2
fuzz/fuzz_csvt.ml
···
102
102
103
103
(* {1 String roundtrip (no commas/newlines)} *)
104
104
105
+
let strip_leading_hash s =
106
+
let n = String.length s in
107
+
let i = ref 0 in
108
+
while !i < n && s.[!i] = '#' do
109
+
incr i
110
+
done;
111
+
if !i = 0 then s else String.sub s !i (n - !i)
112
+
105
113
let test_string_roundtrip s =
106
114
(* Filter characters that have special meaning in CSV: comma (field separator),
107
-
newline/CR (row separator), and double-quote (quoting delimiter per RFC 4180).
108
-
These characters require quoting to roundtrip, tested separately. *)
115
+
newline/CR (row separator), double-quote (quoting delimiter per RFC 4180),
116
+
and leading # (comment marker). These characters require quoting to
117
+
roundtrip, tested separately. *)
109
118
let s =
110
119
String.to_seq s
111
120
|> Seq.filter (fun c -> c <> ',' && c <> '\n' && c <> '\r' && c <> '"')
112
121
|> String.of_seq
113
122
in
123
+
let s = strip_leading_hash s in
114
124
if String.length s = 0 then () (* empty lines are skipped *)
115
125
else
116
126
let csv = "v\n" ^ s ^ "\n" in
···
129
139
|> Seq.filter (fun c -> c <> ',' && c <> '\n' && c <> '\r' && c <> '"')
130
140
|> String.of_seq
131
141
in
142
+
let s = strip_leading_hash s in
132
143
if String.length s = 0 then ()
133
144
else
134
145
let csv = "a,b\n" ^ string_of_int i ^ "," ^ s ^ "\n" in
···
147
158
|> Seq.filter (fun c -> c <> ',' && c <> '\n' && c <> '\r' && c <> '"')
148
159
|> String.of_seq
149
160
in
161
+
let s = strip_leading_hash s in
150
162
if String.length s = 0 then ()
151
163
else
152
164
let csv1 = "a,b\n" ^ string_of_int i ^ "," ^ s ^ "\n" in
+7
-5
lib/csvt.ml
+7
-5
lib/csvt.ml
···
618
618
619
619
let needs_quoting s =
620
620
let n = String.length s in
621
-
let rec loop i =
622
-
if i >= n then false
623
-
else match s.[i] with ',' | '"' | '\n' | '\r' -> true | _ -> loop (i + 1)
624
-
in
625
-
loop 0
621
+
if n > 0 && s.[0] = '#' then true
622
+
else
623
+
let rec loop i =
624
+
if i >= n then false
625
+
else match s.[i] with ',' | '"' | '\n' | '\r' -> true | _ -> loop (i + 1)
626
+
in
627
+
loop 0
626
628
627
629
let write_quoted_field w s =
628
630
w "\"";
+3
lib/csvt.mli
+3
lib/csvt.mli
···
13
13
share a single type {!t}. Row codecs are built declaratively using
14
14
constructors and field accessors with the {!Row} builder.
15
15
16
+
Lines starting with [#] are treated as comments and skipped during decoding
17
+
(both in the header section and in data rows). Empty lines are also skipped.
18
+
16
19
{2 Quick Start}
17
20
18
21
{v