1 | use { |
2 | anyhow::Result, |
3 | regex::bytes::{RegexSet, RegexSetBuilder}, |
4 | regex_test::{CompiledRegex, RegexTest, TestResult, TestRunner}, |
5 | }; |
6 | |
7 | /// Tests the default configuration of the hybrid NFA/DFA. |
8 | #[test] |
9 | fn default() -> Result<()> { |
10 | let mut runner = TestRunner::new()?; |
11 | runner |
12 | .expand(&["is_match" , "which" ], |test| test.compiles()) |
13 | .blacklist_iter(super::BLACKLIST) |
14 | .test_iter(crate::suite()?.iter(), compiler) |
15 | .assert(); |
16 | Ok(()) |
17 | } |
18 | |
19 | fn run_test(re: &RegexSet, test: &RegexTest) -> TestResult { |
20 | match test.additional_name() { |
21 | "is_match" => TestResult::matched(re.is_match(test.haystack())), |
22 | "which" => TestResult::which(re.matches(test.haystack()).iter()), |
23 | name => TestResult::fail(&format!("unrecognized test name: {}" , name)), |
24 | } |
25 | } |
26 | |
27 | /// Converts the given regex test to a closure that searches with a |
28 | /// `bytes::Regex`. If the test configuration is unsupported, then a |
29 | /// `CompiledRegex` that skips the test is returned. |
30 | fn compiler( |
31 | test: &RegexTest, |
32 | _patterns: &[String], |
33 | ) -> anyhow::Result<CompiledRegex> { |
34 | let skip = Ok(CompiledRegex::skip()); |
35 | |
36 | // The top-level RegexSet API only supports "overlapping" semantics. |
37 | if !matches!(test.search_kind(), regex_test::SearchKind::Overlapping) { |
38 | return skip; |
39 | } |
40 | // The top-level RegexSet API only supports "all" semantics. |
41 | if !matches!(test.match_kind(), regex_test::MatchKind::All) { |
42 | return skip; |
43 | } |
44 | // The top-level RegexSet API always runs unanchored searches. |
45 | if test.anchored() { |
46 | return skip; |
47 | } |
48 | // We don't support tests with explicit search bounds. |
49 | let bounds = test.bounds(); |
50 | if !(bounds.start == 0 && bounds.end == test.haystack().len()) { |
51 | return skip; |
52 | } |
53 | // The bytes::Regex API specifically does not support enabling UTF-8 mode. |
54 | // It could I suppose, but currently it does not. That is, it permits |
55 | // matches to have offsets that split codepoints. |
56 | if test.utf8() { |
57 | return skip; |
58 | } |
59 | // If the test requires Unicode but the Unicode feature isn't enabled, |
60 | // skip it. This is a little aggressive, but the test suite doesn't |
61 | // have any easy way of communicating which Unicode features are needed. |
62 | if test.unicode() && !cfg!(feature = "unicode" ) { |
63 | return skip; |
64 | } |
65 | let re = RegexSetBuilder::new(test.regexes()) |
66 | .case_insensitive(test.case_insensitive()) |
67 | .unicode(test.unicode()) |
68 | .line_terminator(test.line_terminator()) |
69 | .build()?; |
70 | Ok(CompiledRegex::compiled(move |test| run_test(&re, test))) |
71 | } |
72 | |