Bloom filter for probabilistic membership testing
fix(lint): add gen_corpus.ml to all fuzz directories for E718
Add seed corpus generators with representative test data for:
ocaml-aos, ocaml-block, ocaml-bloom, ocaml-bpsec, ocaml-bundle,
ocaml-cfdp, ocaml-cgr, ocaml-clcw, ocaml-cookeio, ocaml-cpio
+32
+5
fuzz/dune
+5
fuzz/dune
+27
fuzz/gen_corpus.ml
+27
fuzz/gen_corpus.ml
···
1
+
(** Generate seed corpus for fuzz testing. *)
2
+
3
+
let () =
4
+
(try Unix.mkdir "corpus" 0o755
5
+
with Unix.Unix_error (Unix.EEXIST, _, _) -> ());
6
+
let write name data =
7
+
let oc = open_out_bin (Filename.concat "corpus" name) in
8
+
output_string oc data;
9
+
close_out oc
10
+
in
11
+
(* Simple strings that would be added to a bloom filter *)
12
+
write "seed_000" "hello";
13
+
write "seed_001" "world";
14
+
write "seed_002" "";
15
+
write "seed_003" (String.make 256 'x');
16
+
(* Binary patterns *)
17
+
write "seed_004" "\x00\x00\x00\x00";
18
+
write "seed_005" "\xFF\xFF\xFF\xFF";
19
+
(* Serialized bloom filter bytes - create a small one and serialize it *)
20
+
let bf = Bloom.v ~error_rate:0.01 100 in
21
+
Bloom.add bf "test";
22
+
Bloom.add bf "fuzz";
23
+
write "seed_006" (Bytes.to_string (Bloom.to_bytes bf));
24
+
(* Empty bloom filter serialized *)
25
+
let bf_empty = Bloom.v ~error_rate:0.1 10 in
26
+
write "seed_007" (Bytes.to_string (Bloom.to_bytes bf_empty));
27
+
print_endline "gen_corpus: wrote 8 bloom seed files"