Skip to content

Commit 1f9be1b

Browse files
sylvestresylvestre
committed
benchmark: refactor some code to avoid duplications
1 parent d453ea9 commit 1f9be1b

File tree

6 files changed

+408
-403
lines changed

6 files changed

+408
-403
lines changed

‎src/uu/numfmt/benches/numfmt_bench.rs‎

Lines changed: 27 additions & 32 deletions
Original file line numberDiff line numberDiff line change
@@ -4,56 +4,45 @@
44
// file that was distributed with this source code.
55

66
use divan::{Bencher, black_box};
7-
use tempfile::TempDir;
87
use uu_numfmt::uumain;
9-
use uucore::benchmark::{create_test_file, run_util_function};
10-
11-
/// Generate numeric data for benchmarking
12-
fn generate_numbers(count: usize) -> String {
13-
(1..=count)
14-
.map(|n| n.to_string())
15-
.collect::<Vec<_>>()
16-
.join("\n")
17-
}
18-
19-
/// Setup benchmark environment with test data
20-
fn setup_benchmark(data: String) -> (TempDir, String) {
21-
let temp_dir = tempfile::tempdir().unwrap();
22-
let file_path = create_test_file(data.as_bytes(), temp_dir.path());
23-
let file_path_str = file_path.to_str().unwrap().to_string();
24-
(temp_dir, file_path_str)
25-
}
8+
use uucore::benchmark::{run_util_function, setup_test_file, text_data};
269

2710
/// Benchmark SI formatting with different number counts
2811
#[divan::bench(args = [1_000_000])]
2912
fn numfmt_to_si(bencher: Bencher, count: usize) {
30-
let (_temp_dir, file_path_str) = setup_benchmark(generate_numbers(count));
13+
let data = text_data::generate_numbers(count);
14+
let file_path = setup_test_file(data.as_bytes());
15+
let file_path_str = file_path.to_str().unwrap();
3116

3217
bencher.bench(|| {
33-
black_box(run_util_function(uumain, &["--to=si", &file_path_str]));
18+
black_box(run_util_function(uumain, &["--to=si", file_path_str]));
3419
});
3520
}
3621

3722
/// Benchmark SI formatting with precision format
3823
#[divan::bench(args = [1_000_000])]
3924
fn numfmt_to_si_precision(bencher: Bencher, count: usize) {
40-
let (_temp_dir, file_path_str) = setup_benchmark(generate_numbers(count));
25+
let data = text_data::generate_numbers(count);
26+
let file_path = setup_test_file(data.as_bytes());
27+
let file_path_str = file_path.to_str().unwrap();
4128

4229
bencher.bench(|| {
4330
black_box(run_util_function(
4431
uumain,
45-
&["--to=si", "--format=%.6f", &file_path_str],
32+
&["--to=si", "--format=%.6f", file_path_str],
4633
));
4734
});
4835
}
4936

5037
/// Benchmark IEC (binary) formatting
5138
#[divan::bench(args = [1_000_000])]
5239
fn numfmt_to_iec(bencher: Bencher, count: usize) {
53-
let (_temp_dir, file_path_str) = setup_benchmark(generate_numbers(count));
40+
let data = text_data::generate_numbers(count);
41+
let file_path = setup_test_file(data.as_bytes());
42+
let file_path_str = file_path.to_str().unwrap();
5443

5544
bencher.bench(|| {
56-
black_box(run_util_function(uumain, &["--to=iec", &file_path_str]));
45+
black_box(run_util_function(uumain, &["--to=iec", file_path_str]));
5746
});
5847
}
5948

@@ -65,10 +54,11 @@ fn numfmt_from_si(bencher: Bencher, count: usize) {
6554
.map(|n| format!("{:.1}K", n as f64 / 1000.0))
6655
.collect::<Vec<_>>()
6756
.join("\n");
68-
let (_temp_dir, file_path_str) = setup_benchmark(data);
57+
let file_path = setup_test_file(data.as_bytes());
58+
let file_path_str = file_path.to_str().unwrap();
6959

7060
bencher.bench(|| {
71-
black_box(run_util_function(uumain, &["--from=si", &file_path_str]));
61+
black_box(run_util_function(uumain, &["--from=si", file_path_str]));
7262
});
7363
}
7464

@@ -80,37 +70,42 @@ fn numfmt_large_numbers_si(bencher: Bencher, count: usize) {
8070
.map(|n| (n * 1_000_000).to_string())
8171
.collect::<Vec<_>>()
8272
.join("\n");
83-
let (_temp_dir, file_path_str) = setup_benchmark(data);
73+
let file_path = setup_test_file(data.as_bytes());
74+
let file_path_str = file_path.to_str().unwrap();
8475

8576
bencher.bench(|| {
86-
black_box(run_util_function(uumain, &["--to=si", &file_path_str]));
77+
black_box(run_util_function(uumain, &["--to=si", file_path_str]));
8778
});
8879
}
8980

9081
/// Benchmark different padding widths
9182
#[divan::bench(args = [(1_000_000, 5), (1_000_000, 50)])]
9283
fn numfmt_padding(bencher: Bencher, (count, padding): (usize, usize)) {
93-
let (_temp_dir, file_path_str) = setup_benchmark(generate_numbers(count));
84+
let data = text_data::generate_numbers(count);
85+
let file_path = setup_test_file(data.as_bytes());
86+
let file_path_str = file_path.to_str().unwrap();
9487
let padding_arg = format!("--padding={padding}");
9588

9689
bencher.bench(|| {
9790
black_box(run_util_function(
9891
uumain,
99-
&["--to=si", &padding_arg, &file_path_str],
92+
&["--to=si", &padding_arg, file_path_str],
10093
));
10194
});
10295
}
10396

10497
/// Benchmark round modes with SI formatting
10598
#[divan::bench(args = [("up", 100_000), ("down", 1_000_000), ("towards-zero", 1_000_000)])]
10699
fn numfmt_round_modes(bencher: Bencher, (round_mode, count): (&str, usize)) {
107-
let (_temp_dir, file_path_str) = setup_benchmark(generate_numbers(count));
100+
let data = text_data::generate_numbers(count);
101+
let file_path = setup_test_file(data.as_bytes());
102+
let file_path_str = file_path.to_str().unwrap();
108103
let round_arg = format!("--round={round_mode}");
109104

110105
bencher.bench(|| {
111106
black_box(run_util_function(
112107
uumain,
113-
&["--to=si", &round_arg, &file_path_str],
108+
&["--to=si", &round_arg, file_path_str],
114109
));
115110
});
116111
}

‎src/uu/sort/benches/sort_bench.rs‎

Lines changed: 9 additions & 157 deletions
Original file line numberDiff line numberDiff line change
@@ -4,161 +4,13 @@
44
// file that was distributed with this source code.
55

66
use divan::{Bencher, black_box};
7-
use std::path::PathBuf;
87
use uu_sort::uumain;
9-
use uucore::benchmark::{create_test_file, run_util_function};
10-
11-
/// Helper function to generate test data from a list of words
12-
fn generate_data_from_words(words: &[&str], num_lines: usize) -> Vec<u8> {
13-
let mut data = Vec::new();
14-
for i in 0..num_lines {
15-
let word = words[i % words.len()];
16-
let number = i % 1000;
17-
data.extend_from_slice(format!("{word}_{number:03}\n").as_bytes());
18-
}
19-
data
20-
}
21-
22-
/// Helper function to generate test data from a list of words without number suffix
23-
fn generate_data_from_words_simple(words: &[&str], num_lines: usize) -> Vec<u8> {
24-
let mut data = Vec::new();
25-
for i in 0..num_lines {
26-
let word = words[i % words.len()];
27-
data.extend_from_slice(format!("{word}\n").as_bytes());
28-
}
29-
data
30-
}
31-
32-
/// Generate test data with ASCII-only text
33-
fn generate_ascii_data(num_lines: usize) -> Vec<u8> {
34-
let words = [
35-
"apple",
36-
"banana",
37-
"cherry",
38-
"date",
39-
"elderberry",
40-
"fig",
41-
"grape",
42-
"honeydew",
43-
"kiwi",
44-
"lemon",
45-
"mango",
46-
"nectarine",
47-
"orange",
48-
"papaya",
49-
"quince",
50-
"raspberry",
51-
"strawberry",
52-
"tangerine",
53-
"ugli",
54-
"vanilla",
55-
"watermelon",
56-
"xigua",
57-
"yellow",
58-
"zucchini",
59-
"avocado",
60-
];
61-
62-
generate_data_from_words(&words, num_lines)
63-
}
64-
65-
/// Generate test data with accented characters that require locale-aware sorting
66-
fn generate_accented_data(num_lines: usize) -> Vec<u8> {
67-
let words = [
68-
// French words with accents
69-
"café",
70-
"naïve",
71-
"résumé",
72-
"fiancé",
73-
"crème",
74-
"déjà",
75-
"façade",
76-
"château",
77-
"élève",
78-
"côte",
79-
// German words with umlauts
80-
"über",
81-
"Müller",
82-
"schön",
83-
"Köln",
84-
"Düsseldorf",
85-
"Österreich",
86-
"Zürich",
87-
"Mädchen",
88-
"Bär",
89-
"größer",
90-
// Spanish words with tildes and accents
91-
"niño",
92-
"señor",
93-
"año",
94-
"mañana",
95-
"español",
96-
"corazón",
97-
"María",
98-
"José",
99-
"más",
100-
"también",
101-
];
102-
103-
generate_data_from_words(&words, num_lines)
104-
}
105-
106-
/// Generate test data with mixed ASCII and non-ASCII characters
107-
fn generate_mixed_data(num_lines: usize) -> Vec<u8> {
108-
let words = [
109-
// Mix of ASCII and accented words
110-
"apple",
111-
"café",
112-
"banana",
113-
"naïve",
114-
"cherry",
115-
"résumé",
116-
"date",
117-
"fiancé",
118-
"elderberry",
119-
"crème",
120-
"über",
121-
"grape",
122-
"Müller",
123-
"honeydew",
124-
"schön",
125-
"niño",
126-
"kiwi",
127-
"señor",
128-
"lemon",
129-
"año",
130-
"mango",
131-
"María",
132-
"orange",
133-
"José",
134-
"papaya",
135-
];
136-
137-
generate_data_from_words(&words, num_lines)
138-
}
139-
140-
/// Generate test data with uppercase/lowercase variations
141-
fn generate_case_sensitive_data(num_lines: usize) -> Vec<u8> {
142-
let base_words = [
143-
"apple", "Apple", "APPLE", "banana", "Banana", "BANANA", "café", "Café", "CAFÉ", "über",
144-
"Über", "ÜBER",
145-
];
146-
147-
generate_data_from_words_simple(&base_words, num_lines)
148-
}
149-
150-
fn setup_test_file(data: &[u8]) -> PathBuf {
151-
let temp_dir = tempfile::tempdir().unwrap();
152-
let file_path = create_test_file(data, temp_dir.path());
153-
// Keep temp_dir alive by leaking it - the OS will clean it up
154-
std::mem::forget(temp_dir);
155-
file_path
156-
}
8+
use uucore::benchmark::{run_util_function, setup_test_file, text_data};
1579

15810
/// Benchmark sorting ASCII-only data
15911
#[divan::bench(args = [100_000, 500_000])]
16012
fn sort_ascii_only(bencher: Bencher, num_lines: usize) {
161-
let data = generate_ascii_data(num_lines);
13+
let data = text_data::generate_ascii_data(num_lines);
16214
let file_path = setup_test_file(&data);
16315

16416
bencher.bench(|| {
@@ -169,7 +21,7 @@ fn sort_ascii_only(bencher: Bencher, num_lines: usize) {
16921
/// Benchmark sorting accented/non-ASCII data
17022
#[divan::bench(args = [100_000, 500_000])]
17123
fn sort_accented_data(bencher: Bencher, num_lines: usize) {
172-
let data = generate_accented_data(num_lines);
24+
let data = text_data::generate_accented_data(num_lines);
17325
let file_path = setup_test_file(&data);
17426

17527
bencher.bench(|| {
@@ -180,7 +32,7 @@ fn sort_accented_data(bencher: Bencher, num_lines: usize) {
18032
/// Benchmark sorting mixed ASCII/non-ASCII data
18133
#[divan::bench(args = [100_000, 500_000])]
18234
fn sort_mixed_data(bencher: Bencher, num_lines: usize) {
183-
let data = generate_mixed_data(num_lines);
35+
let data = text_data::generate_mixed_data(num_lines);
18436
let file_path = setup_test_file(&data);
18537

18638
bencher.bench(|| {
@@ -191,7 +43,7 @@ fn sort_mixed_data(bencher: Bencher, num_lines: usize) {
19143
/// Benchmark case-sensitive sorting with mixed case data
19244
#[divan::bench(args = [100_000, 500_000])]
19345
fn sort_case_sensitive(bencher: Bencher, num_lines: usize) {
194-
let data = generate_case_sensitive_data(num_lines);
46+
let data = text_data::generate_case_sensitive_data(num_lines);
19547
let file_path = setup_test_file(&data);
19648

19749
bencher.bench(|| {
@@ -202,7 +54,7 @@ fn sort_case_sensitive(bencher: Bencher, num_lines: usize) {
20254
/// Benchmark case-insensitive sorting (fold case)
20355
#[divan::bench(args = [100_000, 500_000])]
20456
fn sort_case_insensitive(bencher: Bencher, num_lines: usize) {
205-
let data = generate_case_sensitive_data(num_lines);
57+
let data = text_data::generate_case_sensitive_data(num_lines);
20658
let file_path = setup_test_file(&data);
20759

20860
bencher.bench(|| {
@@ -216,7 +68,7 @@ fn sort_case_insensitive(bencher: Bencher, num_lines: usize) {
21668
/// Benchmark dictionary order sorting (only blanks and alphanumeric)
21769
#[divan::bench(args = [100_000, 500_000])]
21870
fn sort_dictionary_order(bencher: Bencher, num_lines: usize) {
219-
let data = generate_mixed_data(num_lines);
71+
let data = text_data::generate_mixed_data(num_lines);
22072
let file_path = setup_test_file(&data);
22173

22274
bencher.bench(|| {
@@ -251,7 +103,7 @@ fn sort_numeric(bencher: Bencher, num_lines: usize) {
251103
/// Benchmark reverse sorting with locale-aware data
252104
#[divan::bench(args = [100_000, 500_000])]
253105
fn sort_reverse_locale(bencher: Bencher, num_lines: usize) {
254-
let data = generate_accented_data(num_lines);
106+
let data = text_data::generate_accented_data(num_lines);
255107
let file_path = setup_test_file(&data);
256108

257109
bencher.bench(|| {
@@ -290,7 +142,7 @@ fn sort_key_field(bencher: Bencher, num_lines: usize) {
290142
/// Benchmark unique sorting with locale-aware data
291143
#[divan::bench(args = [100_000, 500_000])]
292144
fn sort_unique_locale(bencher: Bencher, num_lines: usize) {
293-
let data = generate_accented_data(num_lines);
145+
let data = text_data::generate_accented_data(num_lines);
294146
let file_path = setup_test_file(&data);
295147

296148
bencher.bench(|| {

0 commit comments

Comments
 (0)