saphyr-serde/parser/tests/yaml-test-suite.rs

295 lines
9.2 KiB
Rust
Raw Normal View History

2022-01-17 23:44:29 +00:00
use std::fs::{self, DirEntry};
use libtest_mimic::{Arguments, Test, Outcome, run_tests};
use yaml_rust::{
parser::{Event, EventReceiver, Parser},
scanner::{TokenType, TScalarStyle},
ScanError,
Yaml,
YamlLoader,
yaml,
};
2022-01-17 23:44:29 +00:00
type Result<T, E=Box<dyn std::error::Error>> = std::result::Result<T, E>;
struct YamlTest {
yaml: String,
expected_events: String,
expected_error: bool,
is_xfail: bool,
}
fn main() -> Result<()> {
let mut arguments = Arguments::from_args();
if arguments.num_threads.is_none() {
arguments.num_threads = Some(1);
}
2022-01-17 23:44:29 +00:00
let tests: Vec<Vec<_>> = std::fs::read_dir("tests/yaml-test-suite/src")?
.map(|entry| -> Result<_> {
let entry = entry?;
let tests = load_tests_from_file(&entry)?;
Ok(tests)
})
.collect::<Result<_>>()?;
let mut tests: Vec<_> = tests.into_iter().flatten().collect();
tests.sort_by_key(|t| t.name.clone());
run_tests(
&arguments,
tests,
run_yaml_test,
).exit();
}
fn run_yaml_test(test: &Test<YamlTest>) -> Outcome {
let desc = &test.data;
let actual_events = parse_to_events(&desc.yaml);
let events_diff = actual_events.map(|events| events_differ(events, &desc.expected_events));
let error_text = match (events_diff, desc.expected_error) {
(Ok(_), true) => Some("no error when expected".into()),
(Err(_), true) => None,
(Err(e), false) => Some(format!("unexpected error {:?}", e)),
(Ok(Some(diff)), false) => Some(format!("events differ: {}", diff)),
(Ok(None), false) => None,
};
match (error_text, desc.is_xfail) {
(None, false) => Outcome::Passed,
(Some(text), false) => Outcome::Failed { msg: Some(text) },
(Some(_), true) => Outcome::Ignored,
(None, true) => Outcome::Failed { msg: Some("expected to fail but passes".into()) },
}
}
2022-01-17 23:44:29 +00:00
fn load_tests_from_file(entry: &DirEntry) -> Result<Vec<Test<YamlTest>>> {
let file_name = entry.file_name().to_string_lossy().to_string();
let test_name = file_name.strip_suffix(".yaml").ok_or("unexpected filename")?;
let tests = YamlLoader::load_from_str(&fs::read_to_string(&entry.path())?)?;
let tests = tests[0].as_vec().ok_or("no test list found in file")?;
2022-01-17 23:44:29 +00:00
let mut result = vec![];
let mut current_test = yaml::Hash::new();
for (idx, test_data) in tests.iter().enumerate() {
2022-01-17 23:44:29 +00:00
let name = if tests.len() > 1 {
format!("{}-{:02}", test_name, idx)
} else {
test_name.to_string()
};
let is_xfail = EXPECTED_FAILURES.contains(&name.as_str());
// Test fields except `fail` are "inherited"
let test_data = test_data.as_hash().unwrap();
2022-01-17 23:44:29 +00:00
current_test.remove(&Yaml::String("fail".into()));
for (key, value) in test_data.clone() {
2022-01-17 23:44:29 +00:00
current_test.insert(key, value);
}
2022-01-17 23:44:29 +00:00
let current_test = Yaml::Hash(current_test.clone()); // Much better indexing
2022-01-17 23:44:29 +00:00
if current_test["skip"] != Yaml::BadValue {
continue;
}
2022-01-17 23:44:29 +00:00
result.push(Test {
name,
kind: String::new(),
is_ignored: false,
is_bench: false,
data: YamlTest {
yaml: visual_to_raw(current_test["yaml"].as_str().unwrap()),
expected_events: visual_to_raw(current_test["tree"].as_str().unwrap()),
expected_error: current_test["fail"].as_bool() == Some(true),
is_xfail,
},
});
}
2022-01-17 23:44:29 +00:00
Ok(result)
}
fn parse_to_events(source: &str) -> Result<Vec<String>, ScanError> {
let mut reporter = EventReporter::new();
Parser::new(source.chars())
.load(&mut reporter, true)?;
Ok(reporter.events)
}
struct EventReporter {
events: Vec<String>,
}
impl EventReporter {
fn new() -> Self {
Self {
events: vec![],
}
}
}
impl EventReceiver for EventReporter {
fn on_event(&mut self, ev: Event) {
let line: String = match ev {
Event::StreamStart => "+STR".into(),
Event::StreamEnd => "-STR".into(),
Event::DocumentStart => "+DOC".into(),
Event::DocumentEnd => "-DOC".into(),
Event::SequenceStart(idx) => format!("+SEQ{}", format_index(idx)),
Event::SequenceEnd => "-SEQ".into(),
Event::MappingStart(idx) => format!("+MAP{}", format_index(idx)),
Event::MappingEnd => "-MAP".into(),
Event::Scalar(ref text, style, idx, ref tag) => {
let kind = match style {
TScalarStyle::Plain => ":",
TScalarStyle::SingleQuoted => "'",
TScalarStyle::DoubleQuoted => r#"""#,
TScalarStyle::Literal => "|",
TScalarStyle::Foled => ">",
TScalarStyle::Any => unreachable!(),
};
format!("=VAL{}{} {}{}",
format_index(idx), format_tag(tag), kind, escape_text(text))
}
Event::Alias(idx) => format!("=ALI *{}", idx),
Event::Nothing => return,
};
self.events.push(line);
}
}
fn format_index(idx: usize) -> String {
if idx > 0 {
format!(" &{}", idx)
} else {
"".into()
}
}
fn escape_text(text: &str) -> String {
let mut text = text.to_owned();
for (ch, replacement) in [
('\\', r#"\\"#),
('\n', "\\n"),
('\r', "\\r"),
('\x08', "\\b"),
('\t', "\\t"),
] {
text = text.replace(ch, replacement);
}
text
}
fn format_tag(tag: &Option<TokenType>) -> String {
if let Some(TokenType::Tag(ns, tag)) = tag {
let ns = match ns.as_str() {
"!!" => "tag:yaml.org,2002:", // Wrong if this ns is overridden
other => other,
};
format!(" <{}{}>", ns, tag)
} else {
"".into()
}
}
fn events_differ(actual: Vec<String>, expected: &str) -> Option<String> {
let actual = actual.iter().map(Some).chain(std::iter::repeat(None));
let expected = expected_events(expected);
let expected = expected.iter().map(Some).chain(std::iter::repeat(None));
for (idx, (act, exp)) in actual.zip(expected).enumerate() {
return match (act, exp) {
(Some(act), Some(exp)) => {
if act == exp {
continue;
} else {
Some(format!("line {} differs: expected `{}`, found `{}`", idx, exp, act))
}
}
(Some(a), None) => Some(format!("extra actual line: {:?}", a)),
(None, Some(e)) => Some(format!("extra expected line: {:?}", e)),
(None, None) => None,
}
}
unreachable!()
}
2022-01-17 23:44:29 +00:00
/// Convert the snippets from "visual" to "actual" representation
fn visual_to_raw(yaml: &str) -> String {
let mut yaml = yaml.to_owned();
for (pat, replacement) in [
("", " "),
("»", "\t"),
("", ""), // Tab line continuation ——»
("", "\r"),
("", "\u{FEFF}"),
("", ""), // Trailing newline marker
("\n", ""),
] {
yaml = yaml.replace(pat, replacement);
}
yaml
}
/// Adapt the expectations to the yaml-rust reasonable limitations
///
/// Drop information on node styles (flow/block) and anchor names.
/// Both are things that can be omitted according to spec.
fn expected_events(expected_tree: &str) -> Vec<String> {
let mut anchors = vec![];
expected_tree.split("\n")
.map(|s| s.trim_start().to_owned())
.filter(|s| !s.is_empty())
.map(|mut s| {
// Anchor name-to-number conversion
if let Some(start) = s.find("&") {
if s[..start].find(":").is_none() {
let len = s[start..].find(" ").unwrap_or(s[start..].len());
anchors.push(s[start+1..start + len].to_owned());
s = s.replace(&s[start..start + len], &format!("&{}", anchors.len()));
}
}
// Alias nodes name-to-number
if s.starts_with("=ALI") {
let start = s.find("*").unwrap();
let name = &s[start + 1 ..];
let idx = anchors.iter().enumerate().filter(|(_, v)| v == &name).last().unwrap().0;
s = s.replace(&s[start..], &format!("*{}", idx + 1));
}
// Dropping style information
match &*s {
"+DOC ---" => "+DOC".into(),
"-DOC ..." => "-DOC".into(),
s if s.starts_with("+SEQ []") => s.replacen("+SEQ []", "+SEQ", 1),
s if s.starts_with("+MAP {}") => s.replacen("+MAP {}", "+MAP", 1),
"=VAL :" => "=VAL :~".into(), // FIXME: known bug
s => s.into(),
}
})
.collect()
}
static EXPECTED_FAILURES: &[&str] = &[
// These seem to be API limited (not enough information on the event stream level)
// No tag available for SEQ and MAP
2022-01-17 23:44:29 +00:00
"2XXW",
"35KP",
"57H4",
"6JWB",
"735Y",
"9KAX",
"BU8L",
"C4HZ",
"EHF6",
"J7PZ",
"UGM3",
// Cannot resolve tag namespaces
2022-01-17 23:44:29 +00:00
"5TYM",
"6CK3",
"6WLZ",
"9WXW",
"CC74",
"U3C3",
"Z9M4",
"P76L", // overriding the `!!` namespace!
];