rust/library/core/tests/pattern.rs

504 lines
16 KiB
Rust
Raw Normal View History

2017-12-17 23:44:03 +01:00
use std::str::pattern::*;
// This macro makes it easier to write
// tests that do a series of iterations
macro_rules! search_asserts {
($haystack:expr, $needle:expr, $testname:expr, [$($func:ident),*], $result:expr) => {
let mut searcher = $needle.into_searcher($haystack);
2019-07-03 00:37:25 +02:00
let arr = [$( Step::from(searcher.$func()) ),*];
2017-12-17 23:44:03 +01:00
assert_eq!(&arr[..], &$result, $testname);
}
}
/// Combined enum for the results of next() and next_match()/next_reject()
#[derive(Debug, PartialEq, Eq)]
enum Step {
// variant names purposely chosen to
// be the same length for easy alignment
Matches(usize, usize),
Rejects(usize, usize),
InRange(usize, usize),
2019-12-22 23:42:04 +01:00
Done,
2017-12-17 23:44:03 +01:00
}
use self::Step::*;
2017-12-17 23:44:03 +01:00
impl From<SearchStep> for Step {
fn from(x: SearchStep) -> Self {
match x {
SearchStep::Match(a, b) => Matches(a, b),
SearchStep::Reject(a, b) => Rejects(a, b),
2019-12-22 23:42:04 +01:00
SearchStep::Done => Done,
2017-12-17 23:44:03 +01:00
}
}
}
impl From<Option<(usize, usize)>> for Step {
fn from(x: Option<(usize, usize)>) -> Self {
match x {
Some((a, b)) => InRange(a, b),
2019-12-22 23:42:04 +01:00
None => Done,
2017-12-17 23:44:03 +01:00
}
}
}
2017-12-22 06:49:50 +01:00
// FIXME(Manishearth) these tests focus on single-character searching (CharSearcher)
// and on next()/next_match(), not next_reject(). This is because
// the memchr changes make next_match() for single chars complex, but next_reject()
// continues to use next() under the hood. We should add more test cases for all
// of these, as well as tests for StrSearcher and higher level tests for str::find() (etc)
2017-12-17 23:44:03 +01:00
#[test]
fn test_simple_iteration() {
2019-12-22 23:42:04 +01:00
search_asserts!(
"abcdeabcd",
'a',
"forward iteration for ASCII string",
2017-12-22 06:49:50 +01:00
// a b c d e a b c d EOF
2019-12-22 23:42:04 +01:00
[next, next, next, next, next, next, next, next, next, next],
[
Matches(0, 1),
Rejects(1, 2),
Rejects(2, 3),
Rejects(3, 4),
Rejects(4, 5),
Matches(5, 6),
Rejects(6, 7),
Rejects(7, 8),
Rejects(8, 9),
Done
]
);
search_asserts!(
"abcdeabcd",
'a',
"reverse iteration for ASCII string",
2017-12-17 23:44:03 +01:00
// d c b a e d c b a EOF
2019-12-22 23:42:04 +01:00
[
next_back, next_back, next_back, next_back, next_back, next_back, next_back, next_back,
next_back, next_back
],
[
Rejects(8, 9),
Rejects(7, 8),
Rejects(6, 7),
Matches(5, 6),
Rejects(4, 5),
Rejects(3, 4),
Rejects(2, 3),
Rejects(1, 2),
Matches(0, 1),
Done
]
);
search_asserts!(
"我爱我的猫",
'我',
"forward iteration for Chinese string",
2017-12-17 23:44:03 +01:00
// 我 愛 我 的 貓 EOF
2019-12-22 23:42:04 +01:00
[next, next, next, next, next, next],
2017-12-17 23:44:03 +01:00
[Matches(0, 3), Rejects(3, 6), Matches(6, 9), Rejects(9, 12), Rejects(12, 15), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
"我的猫说meow",
'm',
"forward iteration for mixed string",
2017-12-17 23:44:03 +01:00
// 我 的 猫 说 m e o w EOF
2019-12-22 23:42:04 +01:00
[next, next, next, next, next, next, next, next, next],
[
Rejects(0, 3),
Rejects(3, 6),
Rejects(6, 9),
Rejects(9, 12),
Matches(12, 13),
Rejects(13, 14),
Rejects(14, 15),
Rejects(15, 16),
Done
]
);
search_asserts!(
"我的猫说meow",
'猫',
"reverse iteration for mixed string",
2017-12-17 23:44:03 +01:00
// w o e m 说 猫 的 我 EOF
2019-12-22 23:42:04 +01:00
[
next_back, next_back, next_back, next_back, next_back, next_back, next_back, next_back,
next_back
],
[
Rejects(15, 16),
Rejects(14, 15),
Rejects(13, 14),
Rejects(12, 13),
Rejects(9, 12),
Matches(6, 9),
Rejects(3, 6),
Rejects(0, 3),
Done
]
2017-12-17 23:44:03 +01:00
);
}
2017-12-18 00:05:29 +01:00
#[test]
fn test_simple_search() {
2019-12-22 23:42:04 +01:00
search_asserts!(
"abcdeabcdeabcde",
'a',
"next_match for ASCII string",
[next_match, next_match, next_match, next_match],
2017-12-18 00:05:29 +01:00
[InRange(0, 1), InRange(5, 6), InRange(10, 11), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
"abcdeabcdeabcde",
'a',
"next_match_back for ASCII string",
2017-12-18 00:05:29 +01:00
[next_match_back, next_match_back, next_match_back, next_match_back],
2019-12-22 23:42:04 +01:00
[InRange(10, 11), InRange(5, 6), InRange(0, 1), Done]
2017-12-18 00:05:29 +01:00
);
2019-12-22 23:42:04 +01:00
search_asserts!(
"abcdeab",
'a',
"next_reject for ASCII string",
[next_reject, next_reject, next_match, next_reject, next_reject],
2017-12-18 00:05:29 +01:00
[InRange(1, 2), InRange(2, 3), InRange(5, 6), InRange(6, 7), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
"abcdeabcdeabcde",
'a',
"next_reject_back for ASCII string",
[
next_reject_back,
next_reject_back,
next_match_back,
next_reject_back,
next_reject_back,
next_reject_back
],
[
InRange(14, 15),
InRange(13, 14),
InRange(10, 11),
InRange(9, 10),
InRange(8, 9),
InRange(7, 8)
]
2017-12-18 00:05:29 +01:00
);
}
// Á, 각, ก, 😀 all end in 0x81
// 🁀, ᘀ do not end in 0x81 but contain the byte
// ꁁ has 0x81 as its second and third bytes.
//
// The memchr-using implementation of next_match
// and next_match_back temporarily violate
// the property that the search is always on a unicode boundary,
// which is fine as long as this never reaches next() or next_back().
// So we test if next() is correct after each next_match() as well.
const STRESS: &str = "Áa🁀bÁꁁfg😁각กᘀ각aÁ각ꁁก😁a";
#[test]
fn test_stress_indices() {
// this isn't really a test, more of documentation on the indices of each character in the stresstest string
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'x',
"Indices of characters in stress test",
[
next, next, next, next, next, next, next, next, next, next, next, next, next, next,
next, next, next, next, next, next, next
],
[
Rejects(0, 2), // Á
Rejects(2, 3), // a
Rejects(3, 7), // 🁀
Rejects(7, 8), // b
Rejects(8, 10), // Á
Rejects(10, 13), // ꁁ
Rejects(13, 14), // f
Rejects(14, 15), // g
Rejects(15, 19), // 😀
Rejects(19, 22), // 각
Rejects(22, 25), // ก
Rejects(25, 28), // ᘀ
Rejects(28, 31), // 각
Rejects(31, 32), // a
Rejects(32, 34), // Á
Rejects(34, 37), // 각
Rejects(37, 40), // ꁁ
Rejects(40, 43), // ก
Rejects(43, 47), // 😀
Rejects(47, 48), // a
Done
]
);
}
#[test]
fn test_forward_search_shared_bytes() {
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'Á',
"Forward search for two-byte Latin character",
[next_match, next_match, next_match, next_match],
[InRange(0, 2), InRange(8, 10), InRange(32, 34), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'Á',
"Forward search for two-byte Latin character; check if next() still works",
[next_match, next, next_match, next, next_match, next, next_match],
[
InRange(0, 2),
Rejects(2, 3),
InRange(8, 10),
Rejects(10, 13),
InRange(32, 34),
Rejects(34, 37),
Done
]
);
search_asserts!(
STRESS,
'각',
"Forward search for three-byte Hangul character",
[next_match, next, next_match, next_match, next_match],
[InRange(19, 22), Rejects(22, 25), InRange(28, 31), InRange(34, 37), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'각',
"Forward search for three-byte Hangul character; check if next() still works",
[next_match, next, next_match, next, next_match, next, next_match],
[
InRange(19, 22),
Rejects(22, 25),
InRange(28, 31),
Rejects(31, 32),
InRange(34, 37),
Rejects(37, 40),
Done
]
);
search_asserts!(
STRESS,
'ก',
"Forward search for three-byte Thai character",
[next_match, next, next_match, next, next_match],
[InRange(22, 25), Rejects(25, 28), InRange(40, 43), Rejects(43, 47), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'ก',
"Forward search for three-byte Thai character; check if next() still works",
[next_match, next, next_match, next, next_match],
[InRange(22, 25), Rejects(25, 28), InRange(40, 43), Rejects(43, 47), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'😁',
"Forward search for four-byte emoji",
[next_match, next, next_match, next, next_match],
[InRange(15, 19), Rejects(19, 22), InRange(43, 47), Rejects(47, 48), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'😁',
"Forward search for four-byte emoji; check if next() still works",
[next_match, next, next_match, next, next_match],
[InRange(15, 19), Rejects(19, 22), InRange(43, 47), Rejects(47, 48), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'ꁁ',
"Forward search for three-byte Yi character with repeated bytes",
[next_match, next, next_match, next, next_match],
[InRange(10, 13), Rejects(13, 14), InRange(37, 40), Rejects(40, 43), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'ꁁ',
"Forward search for three-byte Yi character with repeated bytes; check if next() still works",
[next_match, next, next_match, next, next_match],
[InRange(10, 13), Rejects(13, 14), InRange(37, 40), Rejects(40, 43), Done]
);
}
#[test]
fn test_reverse_search_shared_bytes() {
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'Á',
"Reverse search for two-byte Latin character",
[next_match_back, next_match_back, next_match_back, next_match_back],
2019-12-22 23:42:04 +01:00
[InRange(32, 34), InRange(8, 10), InRange(0, 2), Done]
);
search_asserts!(
STRESS,
'Á',
"Reverse search for two-byte Latin character; check if next_back() still works",
[next_match_back, next_back, next_match_back, next_back, next_match_back, next_back],
[InRange(32, 34), Rejects(31, 32), InRange(8, 10), Rejects(7, 8), InRange(0, 2), Done]
);
search_asserts!(
STRESS,
'각',
"Reverse search for three-byte Hangul character",
[next_match_back, next_back, next_match_back, next_match_back, next_match_back],
[InRange(34, 37), Rejects(32, 34), InRange(28, 31), InRange(19, 22), Done]
);
search_asserts!(
STRESS,
'각',
"Reverse search for three-byte Hangul character; check if next_back() still works",
[
next_match_back,
next_back,
next_match_back,
next_back,
next_match_back,
next_back,
next_match_back
],
[
InRange(34, 37),
Rejects(32, 34),
InRange(28, 31),
Rejects(25, 28),
InRange(19, 22),
Rejects(15, 19),
Done
]
);
search_asserts!(
STRESS,
'ก',
"Reverse search for three-byte Thai character",
[next_match_back, next_back, next_match_back, next_back, next_match_back],
[InRange(40, 43), Rejects(37, 40), InRange(22, 25), Rejects(19, 22), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'ก',
"Reverse search for three-byte Thai character; check if next_back() still works",
[next_match_back, next_back, next_match_back, next_back, next_match_back],
[InRange(40, 43), Rejects(37, 40), InRange(22, 25), Rejects(19, 22), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'😁',
"Reverse search for four-byte emoji",
[next_match_back, next_back, next_match_back, next_back, next_match_back],
[InRange(43, 47), Rejects(40, 43), InRange(15, 19), Rejects(14, 15), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'😁',
"Reverse search for four-byte emoji; check if next_back() still works",
[next_match_back, next_back, next_match_back, next_back, next_match_back],
[InRange(43, 47), Rejects(40, 43), InRange(15, 19), Rejects(14, 15), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'ꁁ',
"Reverse search for three-byte Yi character with repeated bytes",
[next_match_back, next_back, next_match_back, next_back, next_match_back],
[InRange(37, 40), Rejects(34, 37), InRange(10, 13), Rejects(8, 10), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'ꁁ',
"Reverse search for three-byte Yi character with repeated bytes; check if next_back() still works",
[next_match_back, next_back, next_match_back, next_back, next_match_back],
[InRange(37, 40), Rejects(34, 37), InRange(10, 13), Rejects(8, 10), Done]
);
}
2018-01-05 06:41:39 +01:00
#[test]
fn double_ended_regression_test() {
// https://github.com/rust-lang/rust/issues/47175
// Ensures that double ended searching comes to a convergence
2019-12-22 23:42:04 +01:00
search_asserts!(
"abcdeabcdeabcde",
'a',
"alternating double ended search",
[next_match, next_match_back, next_match, next_match_back],
2018-01-05 06:41:39 +01:00
[InRange(0, 1), InRange(10, 11), InRange(5, 6), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
"abcdeabcdeabcde",
'a',
"triple double ended search for a",
[next_match, next_match_back, next_match_back, next_match_back],
2018-01-05 06:41:39 +01:00
[InRange(0, 1), InRange(10, 11), InRange(5, 6), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
"abcdeabcdeabcde",
'd',
"triple double ended search for d",
[next_match, next_match_back, next_match_back, next_match_back],
2018-01-05 06:41:39 +01:00
[InRange(3, 4), InRange(13, 14), InRange(8, 9), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'Á',
"Double ended search for two-byte Latin character",
[next_match, next_match_back, next_match, next_match_back],
2018-01-05 06:41:39 +01:00
[InRange(0, 2), InRange(32, 34), InRange(8, 10), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'각',
"Reverse double ended search for three-byte Hangul character",
[next_match_back, next_back, next_match, next, next_match_back, next_match],
2018-01-05 06:41:39 +01:00
[InRange(34, 37), Rejects(32, 34), InRange(19, 22), Rejects(22, 25), InRange(28, 31), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'ก',
"Double ended search for three-byte Thai character",
[next_match, next_back, next, next_match_back, next_match],
2018-01-05 06:41:39 +01:00
[InRange(22, 25), Rejects(47, 48), Rejects(25, 28), InRange(40, 43), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'😁',
"Double ended search for four-byte emoji",
[next_match_back, next, next_match, next_back, next_match],
2018-01-05 06:41:39 +01:00
[InRange(43, 47), Rejects(0, 2), InRange(15, 19), Rejects(40, 43), Done]
);
2019-12-22 23:42:04 +01:00
search_asserts!(
STRESS,
'ꁁ',
"Double ended search for three-byte Yi character with repeated bytes",
[next_match, next, next_match_back, next_back, next_match],
2018-01-05 06:41:39 +01:00
[InRange(10, 13), Rejects(13, 14), InRange(37, 40), Rejects(34, 37), Done]
);
}