| 1 |
use { |
| 2 |
anyhow::Result, |
| 3 |
regex_automata::{ |
| 4 |
nfa::thompson::{ |
| 5 |
self, |
| 6 |
pikevm::{self, PikeVM}, |
| 7 |
}, |
| 8 |
util::{prefilter::Prefilter, syntax}, |
| 9 |
PatternSet, |
| 10 |
}, |
| 11 |
regex_test::{ |
| 12 |
CompiledRegex, Match, RegexTest, SearchKind, Span, TestResult, |
| 13 |
TestRunner, |
| 14 |
}, |
| 15 |
}; |
| 16 |
|
| 17 |
use crate::{create_input, suite, testify_captures, untestify_kind}; |
| 18 |
|
| 19 |
/// Tests the default configuration of the hybrid NFA/DFA. |
| 20 |
#[test] |
| 21 |
fn default() -> Result<()> { |
| 22 |
let builder = PikeVM::builder(); |
| 23 |
let mut runner = TestRunner::new()?; |
| 24 |
runner.expand(&["is_match", "find", "captures"], |test| test.compiles()); |
| 25 |
runner.test_iter(suite()?.iter(), compiler(builder)).assert(); |
| 26 |
Ok(()) |
| 27 |
} |
| 28 |
|
| 29 |
/// Tests the PikeVM with prefilters enabled. |
| 30 |
#[test] |
| 31 |
fn prefilter() -> Result<()> { |
| 32 |
let my_compiler = |test: &RegexTest, regexes: &[String]| { |
| 33 |
// Parse regexes as HIRs so we can get literals to build a prefilter. |
| 34 |
let mut hirs = vec![]; |
| 35 |
for pattern in regexes.iter() { |
| 36 |
hirs.push(syntax::parse_with(pattern, &config_syntax(test))?); |
| 37 |
} |
| 38 |
let kind = match untestify_kind(test.match_kind()) { |
| 39 |
None => return Ok(CompiledRegex::skip()), |
| 40 |
Some(kind) => kind, |
| 41 |
}; |
| 42 |
let pre = Prefilter::from_hirs_prefix(kind, &hirs); |
| 43 |
let mut builder = PikeVM::builder(); |
| 44 |
builder.configure(PikeVM::config().prefilter(pre)); |
| 45 |
compiler(builder)(test, regexes) |
| 46 |
}; |
| 47 |
let mut runner = TestRunner::new()?; |
| 48 |
runner.expand(&["is_match", "find", "captures"], |test| test.compiles()); |
| 49 |
runner.test_iter(suite()?.iter(), my_compiler).assert(); |
| 50 |
Ok(()) |
| 51 |
} |
| 52 |
|
| 53 |
fn compiler( |
| 54 |
mut builder: pikevm::Builder, |
| 55 |
) -> impl FnMut(&RegexTest, &[String]) -> Result<CompiledRegex> { |
| 56 |
move |test, regexes| { |
| 57 |
if !configure_pikevm_builder(test, &mut builder) { |
| 58 |
return Ok(CompiledRegex::skip()); |
| 59 |
} |
| 60 |
let re = builder.build_many(®exes)?; |
| 61 |
let mut cache = re.create_cache(); |
| 62 |
Ok(CompiledRegex::compiled(move |test| -> TestResult { |
| 63 |
run_test(&re, &mut cache, test) |
| 64 |
})) |
| 65 |
} |
| 66 |
} |
| 67 |
|
| 68 |
fn run_test( |
| 69 |
re: &PikeVM, |
| 70 |
cache: &mut pikevm::Cache, |
| 71 |
test: &RegexTest, |
| 72 |
) -> TestResult { |
| 73 |
let input = create_input(test); |
| 74 |
match test.additional_name() { |
| 75 |
"is_match" => TestResult::matched(re.is_match(cache, input)), |
| 76 |
"find" => match test.search_kind() { |
| 77 |
SearchKind::Earliest => { |
| 78 |
let it = re |
| 79 |
.find_iter(cache, input.earliest(true)) |
| 80 |
.take(test.match_limit().unwrap_or(std::usize::MAX)) |
| 81 |
.map(|m| Match { |
| 82 |
id: m.pattern().as_usize(), |
| 83 |
span: Span { start: m.start(), end: m.end() }, |
| 84 |
}); |
| 85 |
TestResult::matches(it) |
| 86 |
} |
| 87 |
SearchKind::Leftmost => { |
| 88 |
let it = re |
| 89 |
.find_iter(cache, input) |
| 90 |
.take(test.match_limit().unwrap_or(std::usize::MAX)) |
| 91 |
.map(|m| Match { |
| 92 |
id: m.pattern().as_usize(), |
| 93 |
span: Span { start: m.start(), end: m.end() }, |
| 94 |
}); |
| 95 |
TestResult::matches(it) |
| 96 |
} |
| 97 |
SearchKind::Overlapping => { |
| 98 |
let mut patset = PatternSet::new(re.get_nfa().pattern_len()); |
| 99 |
re.which_overlapping_matches(cache, &input, &mut patset); |
| 100 |
TestResult::which(patset.iter().map(|p| p.as_usize())) |
| 101 |
} |
| 102 |
}, |
| 103 |
"captures" => match test.search_kind() { |
| 104 |
SearchKind::Earliest => { |
| 105 |
let it = re |
| 106 |
.captures_iter(cache, input.earliest(true)) |
| 107 |
.take(test.match_limit().unwrap_or(std::usize::MAX)) |
| 108 |
.map(|caps| testify_captures(&caps)); |
| 109 |
TestResult::captures(it) |
| 110 |
} |
| 111 |
SearchKind::Leftmost => { |
| 112 |
let it = re |
| 113 |
.captures_iter(cache, input) |
| 114 |
.take(test.match_limit().unwrap_or(std::usize::MAX)) |
| 115 |
.map(|caps| testify_captures(&caps)); |
| 116 |
TestResult::captures(it) |
| 117 |
} |
| 118 |
SearchKind::Overlapping => { |
| 119 |
// There is no overlapping PikeVM API that supports captures. |
| 120 |
TestResult::skip() |
| 121 |
} |
| 122 |
}, |
| 123 |
name => TestResult::fail(&format!("unrecognized test name: {name}")), |
| 124 |
} |
| 125 |
} |
| 126 |
|
| 127 |
/// Configures the given regex builder with all relevant settings on the given |
| 128 |
/// regex test. |
| 129 |
/// |
| 130 |
/// If the regex test has a setting that is unsupported, then this returns |
| 131 |
/// false (implying the test should be skipped). |
| 132 |
fn configure_pikevm_builder( |
| 133 |
test: &RegexTest, |
| 134 |
builder: &mut pikevm::Builder, |
| 135 |
) -> bool { |
| 136 |
let match_kind = match untestify_kind(test.match_kind()) { |
| 137 |
None => return false, |
| 138 |
Some(k) => k, |
| 139 |
}; |
| 140 |
let pikevm_config = PikeVM::config().match_kind(match_kind); |
| 141 |
builder |
| 142 |
.configure(pikevm_config) |
| 143 |
.syntax(config_syntax(test)) |
| 144 |
.thompson(config_thompson(test)); |
| 145 |
true |
| 146 |
} |
| 147 |
|
| 148 |
/// Configuration of a Thompson NFA compiler from a regex test. |
| 149 |
fn config_thompson(test: &RegexTest) -> thompson::Config { |
| 150 |
let mut lookm = regex_automata::util::look::LookMatcher::new(); |
| 151 |
lookm.set_line_terminator(test.line_terminator()); |
| 152 |
thompson::Config::new().utf8(test.utf8()).look_matcher(lookm) |
| 153 |
} |
| 154 |
|
| 155 |
/// Configuration of the regex parser from a regex test. |
| 156 |
fn config_syntax(test: &RegexTest) -> syntax::Config { |
| 157 |
syntax::Config::new() |
| 158 |
.case_insensitive(test.case_insensitive()) |
| 159 |
.unicode(test.unicode()) |
| 160 |
.utf8(test.utf8()) |
| 161 |
.line_terminator(test.line_terminator()) |
| 162 |
} |