supercov-engine 4.9.0

//! Whether a run is still a record of a checkout whose test-side files have
//! moved on, judged file by file.
//!
//! A run is fingerprinted by domain -- source, tests, dependencies,
//! configuration -- and any change in a domain made it out of date. For the
//! source and the dependencies that is right. For the rest it was too blunt:
//! the end-to-end suite's Playwright configuration or helpers changing made
//! every unit run out of date, and a helper that only gained a function for a
//! new test made every run that loads it one. Each file that differs is
//! judged here by what it is and what the run's tests did with it.
use std::{
    collections::{BTreeMap, BTreeSet},
    path::Path,
};

use serde_json::Value;

use crate::{
    evidence_archive::{EvidenceArchiveEntry, read_archive_selected},
    run_store::StoredRun,
    source_manifest::{FileFingerprint, FileManifest, describe},
    source_units::only_added,
    test_loads::{Imports, configuration_runner, reads_imports},
};

/// The archive entry holding the Playwright configuration a run resolved.
pub const PLAYWRIGHT_CONFIGURATION: &str = "playwright-configuration.json";

/// What a run's archive says about how it ran, beyond its files.
#[derive(Debug, Clone, Default)]
pub struct Ran {
    /// The runners its tests ran under, where its archive says.
    pub runners: Option<BTreeSet<String>>,
    /// The Playwright configuration as it resolved, where it ran Playwright.
    pub playwright: Option<Value>,
    /// The files it measured, from its coverage manifest's scope. A file of
    /// the source outside them -- a simulator, a script -- is judged as a
    /// helper is.
    pub measured: Option<BTreeSet<String>>,
}

impl Ran {
    pub fn of(run: &StoredRun) -> Self {
        let merged = run.metadata.merged == Some(true);
        match read_archive_selected(&run.evidence_path, |path| {
            Self::wanted(path, merged).is_some()
        }) {
            Ok(entries) => Self::from_entries(&entries, merged),
            Err(_) => Self::default(),
        }
    }

    /// The same, from an archive already read.
    pub fn from_entries(entries: &[EvidenceArchiveEntry], merged: bool) -> Self {
        let mut ran = Self::default();
        for entry in entries {
            let Some(name) = Self::wanted(&entry.path, merged) else {
                continue;
            };
            let Ok(value) = serde_json::from_slice::<Value>(&entry.contents) else {
                continue;
            };
            if name == PLAYWRIGHT_CONFIGURATION {
                ran.playwright = Some(value);
            } else if name == "manifest.json" {
                ran.measured = value["scope"]["entries"].as_array().map(|entries| {
                    entries
                        .iter()
                        .filter(|entry| entry["status"] == "included")
                        .filter_map(|entry| entry["file"].as_str().map(str::to_owned))
                        .collect()
                });
            } else if let Some(runners) = value["runners"].as_array() {
                ran.runners.get_or_insert_with(BTreeSet::new).extend(
                    runners
                        .iter()
                        .filter_map(|runner| runner["runner"].as_str())
                        .filter(|runner| !matches!(*runner, "command" | "background"))
                        .map(str::to_owned),
                );
            }
        }
        ran
    }

    /// Which of the two entries a path is, in a run or a merge of runs.
    fn wanted(path: &str, merged: bool) -> Option<&'static str> {
        let name = if merged {
            path.rsplit('/').next()?
        } else {
            path
        };
        [PLAYWRIGHT_CONFIGURATION, "frontend.json", "manifest.json"]
            .into_iter()
            .find(|wanted| *wanted == name)
    }
}

/// The verdict on one run.
#[derive(Debug, Clone, Default, PartialEq, Eq)]
pub struct Judgment {
    /// Files whose change makes the run a record of something else, each with
    /// what moved in it.
    pub outdated: Vec<String>,
    /// Whether one of those is a file of the run's own tests: no option makes
    /// a changed test's record its own.
    pub own_tests: bool,
    /// Whether one of those is one no option accepts: a file of its own
    /// tests, or one it measured.
    pub settled: bool,
    /// Files that differ and do not reach the run, with why; and those the
    /// user accepted.
    pub across: Vec<String>,
}

/// One checkout's files, read once for every run judged against it.
pub struct Checkout<'a> {
    pub root: &'a Path,
    /// Every file known to hold tests, from every run there is.
    pub tests: &'a BTreeSet<String>,
    imports: Imports,
    /// What all the known test files load: a helper none of them loads is one
    /// Supercov cannot place, so it counts against every run. Read when a
    /// helper first needs placing.
    loaded_by_tests: std::cell::OnceCell<crate::test_loads::Loads>,
}

impl<'a> Checkout<'a> {
    pub fn new(root: &'a Path, tests: &'a BTreeSet<String>) -> Self {
        Self {
            root,
            tests,
            imports: Imports::new(root),
            loaded_by_tests: std::cell::OnceCell::new(),
        }
    }

    /// Whether some known test file imports `file`: one that none does is
    /// loaded from where Supercov cannot see -- a pattern in a configuration,
    /// a command -- and may be loaded by any test. A run whose own imports
    /// are all followed and miss it does not load it.
    pub fn places(&self, file: &str) -> bool {
        self.loaded_by_tests
            .get_or_init(|| {
                self.imports.loads(
                    self.tests
                        .iter()
                        .filter(|file| reads_imports(file))
                        .map(String::as_str),
                )
            })
            .files
            .contains(file)
    }

    /// What a run's tests load: their files, and the configuration of each
    /// runner they ran under with what it names. `None` where a file of
    /// theirs is not one Supercov reads imports from.
    pub fn loads(
        &self,
        own: &BTreeSet<String>,
        runners: Option<&BTreeSet<String>>,
        configurations: &[String],
    ) -> Option<crate::test_loads::Loads> {
        if own.is_empty() || !own.iter().all(|file| reads_imports(file)) {
            return None;
        }
        let mut entries = own.iter().cloned().collect::<Vec<_>>();
        for configuration in configurations {
            let used = configuration_runner(configuration)
                .is_some_and(|runner| runners.is_some_and(|runners| runners.contains(runner)));
            if used {
                entries.push(configuration.clone());
                entries.extend(self.imports.named_by(configuration));
            }
        }
        Some(self.imports.loads(entries.iter().map(String::as_str)))
    }
}

/// A file that sets how tests or builds run, by the rule the fingerprint's
/// configuration domain uses.
pub fn configuration_file(file: &str) -> bool {
    crate::integrity::configuration_file(Path::new(file))
}

/// A file that sits with the tests by its path, as the fingerprint's test
/// domain finds them: in a `test`, `tests` or `__tests__` directory, or named
/// for tests.
pub fn beside_the_tests(file: &str) -> bool {
    let lower = file.to_ascii_lowercase();
    lower
        .split('/')
        .next()
        .is_some_and(|top| matches!(top, "test" | "tests" | "__tests__"))
        || lower
            .split(['/', '_', '.', '-'])
            .any(|part| matches!(part, "test" | "spec"))
}

pub fn dependency_file(file: &str) -> bool {
    let name = file.rsplit('/').next().unwrap_or(file);
    matches!(
        name,
        "package.json"
            | "package-lock.json"
            | "npm-shrinkwrap.json"
            | "pnpm-lock.yaml"
            | "yarn.lock"
            | "bun.lock"
            | "bun.lockb"
    )
}

/// The Playwright settings that decide what a test does, as opposed to when
/// and how often it runs: the rest -- timeouts, retries, workers, reporters,
/// which files and names to run -- schedules tests and changes nothing a test
/// that ran reaches.
fn playwright_shape(configuration: &Value) -> Value {
    const PROJECT: [&str; 5] = ["name", "use", "testDir", "dependencies", "teardown"];
    // At the top level `use` and `testDir` are what every project inherits.
    const CONFIG: [&str; 5] = [
        "globalSetup",
        "globalTeardown",
        "webServer",
        "use",
        "testDir",
    ];
    let pick = |value: &Value, keys: &[&str]| {
        Value::Object(
            keys.iter()
                .filter_map(|key| value.get(*key).map(|v| ((*key).to_owned(), v.clone())))
                .collect(),
        )
    };
    let mut shape = pick(configuration, &CONFIG);
    shape["projects"] = Value::Array(
        configuration["projects"]
            .as_array()
            .into_iter()
            .flatten()
            .map(|project| pick(project, &PROJECT))
            .collect(),
    );
    shape
}

/// Whether the run's Playwright configuration and the reference's do the same
/// to the run's tests: every project it had is there and set alike, and the
/// servers and global setup are the same. A project added since runs none of
/// its tests. Where they differ, which settings did.
pub fn playwright_differences(run: &Value, reference: &Value) -> Vec<String> {
    let (run, reference) = (playwright_shape(run), playwright_shape(reference));
    let mut differences = Vec::new();
    for key in [
        "globalSetup",
        "globalTeardown",
        "webServer",
        "use",
        "testDir",
    ] {
        if run.get(key) != reference.get(key) {
            differences.push(key.to_owned());
        }
    }
    let projects = |shape: &Value| {
        shape["projects"]
            .as_array()
            .into_iter()
            .flatten()
            .map(|project| {
                (
                    project["name"].as_str().unwrap_or("").to_owned(),
                    project.clone(),
                )
            })
            .collect::<BTreeMap<_, _>>()
    };
    let now = projects(&reference);
    for (name, project) in projects(&run) {
        let label = if name.is_empty() {
            String::new()
        } else {
            format!("project {name}: ")
        };
        match now.get(&name) {
            None => differences.push(format!("{label}removed")),
            Some(current) => {
                for key in ["use", "testDir", "dependencies", "teardown"] {
                    if project.get(key) != current.get(key) {
                        differences.push(format!("{label}{key}"));
                    }
                }
            }
        }
    }
    differences
}

/// What the files a run was recorded with say against `now`, the files of the
/// checkout or of the later run it is merged into.
///
/// `text` gives a file's current text where it is the text `now` describes;
/// `reference` is how Playwright resolved for the later run, where one did;
/// `label` names the run in what is said about it.
#[allow(clippy::too_many_arguments)]
pub fn judge(
    checkout: &Checkout<'_>,
    label: &str,
    recorded: &FileManifest,
    now: &FileManifest,
    text: &dyn Fn(&str) -> Option<String>,
    own: Option<&BTreeSet<String>>,
    ran: &Ran,
    reference: Option<&Value>,
    allow_changed: bool,
) -> Judgment {
    let mut judgment = Judgment::default();
    let configurations = recorded
        .keys()
        .chain(now.keys())
        .filter(|file| configuration_file(file))
        .cloned()
        .collect::<BTreeSet<_>>()
        .into_iter()
        .collect::<Vec<_>>();
    let loads = own.and_then(|own| checkout.loads(own, ran.runners.as_ref(), &configurations));
    let mut changed = recorded
        .iter()
        .filter(|(file, was)| now.get(*file).is_none_or(|is| !is.same_bytes(was)))
        .map(|(file, _)| file.clone())
        .collect::<BTreeSet<_>>();
    // A configuration that did not exist then can change a runner too.
    changed.extend(
        now.keys()
            .filter(|file| !recorded.contains_key(*file) && configuration_file(file))
            .cloned(),
    );
    let refuse = |judgment: &mut Judgment, file: &str, what: String| {
        if allow_changed {
            judgment.across.push(format!("{file} (accepted)"));
        } else {
            judgment.outdated.push(what);
        }
    };
    for file in changed {
        let gone = !now.contains_key(&file);
        if own.is_some_and(|own| own.contains(&file)) {
            judgment.own_tests = true;
            judgment.settled = true;
            judgment.outdated.push(format!(
                "{file}, a file of its own tests, {}",
                if gone { "is gone" } else { "has changed" }
            ));
            continue;
        }
        // Another run's test: this run did not run it. Where the run does not
        // say which files are its own, every test file counts.
        if checkout.tests.contains(&file) && own.is_some() {
            continue;
        }
        if dependency_file(&file) {
            judgment.outdated.push(format!("{file} has changed"));
            continue;
        }
        if configuration_file(&file) {
            let runner = configuration_runner(&file);
            match (runner, ran.runners.as_ref()) {
                (Some(runner), Some(runners)) if !runners.contains(runner) => {
                    judgment.across.push(format!(
                        "{file} ({runner} settings; `{label}` ran under {})",
                        runners.iter().cloned().collect::<Vec<_>>().join(", ")
                    ));
                }
                (Some("playwright"), Some(_)) => match (&ran.playwright, reference) {
                    (Some(then), Some(reference)) => {
                        let differences = playwright_differences(then, reference);
                        if differences.is_empty() {
                            judgment
                                .across
                                .push(format!("{file} (only settings that schedule tests)"));
                        } else {
                            refuse(
                                &mut judgment,
                                &file,
                                format!("{file}: {} changed", differences.join(", ")),
                            );
                        }
                    }
                    _ => refuse(&mut judgment, &file, format!("{file} has changed")),
                },
                _ => refuse(&mut judgment, &file, format!("{file} has changed")),
            }
            continue;
        }
        // What the run measured has moved on: its numbers are of other code.
        // Where the run does not say what it measured, any file not beside
        // the tests may be.
        let measured = match &ran.measured {
            Some(measured) => measured.contains(&file),
            None => !beside_the_tests(&file),
        };
        if measured {
            judgment.settled = true;
            judgment.outdated.push(format!(
                "{file}, which it measured, {}",
                if gone { "is gone" } else { "has changed" }
            ));
            continue;
        }
        // A file tests may load: a helper, a fixture, a support module, or
        // source outside what is measured.
        let placed = checkout.places(&file);
        if let Some(loads) = &loads
            && loads.open.is_none()
            && placed
            && !loads.files.contains(&file)
        {
            judgment
                .across
                .push(format!("{file} (`{label}` does not load it)"));
            continue;
        }
        if gone {
            refuse(&mut judgment, &file, format!("{file} is gone"));
            continue;
        }
        let before = recorded.get(&file).and_then(|was| was.code.as_ref());
        let after = now.get(&file).and_then(|is| is.code.as_ref());
        if let (Some(before), Some(source)) = (before, text(&file))
            && let Some(added) = only_added(before, &file, &source)
        {
            judgment
                .across
                .push(format!("{file} (only added {})", added.join(", ")));
            continue;
        }
        let what = match (before, after) {
            (Some(before), Some(after)) => {
                format!(
                    "{file} changed: {}",
                    describe(before, after, &before.diff(after))
                )
            }
            _ => format!("{file} has changed"),
        };
        refuse(&mut judgment, &file, what);
    }
    judgment
}

/// What several judgments let through, one entry per file: `file (why;
/// why)`.
pub fn across(items: impl IntoIterator<Item = String>) -> Vec<String> {
    let mut files = BTreeMap::<String, Vec<String>>::new();
    for item in items {
        let (file, why) = match item.split_once(" (") {
            Some((file, why)) => (file.to_owned(), why.trim_end_matches(')').to_owned()),
            None => (item, String::new()),
        };
        let whys = files.entry(file).or_default();
        if !why.is_empty() && !whys.contains(&why) {
            whys.push(why);
        }
    }
    files
        .into_iter()
        .map(|(file, whys)| {
            if whys.is_empty() {
                file
            } else {
                format!("{file} ({})", whys.join("; "))
            }
        })
        .collect()
}

/// The checkout's own fingerprints of the files a run was recorded with, and
/// of every configuration file there is now.
pub fn checkout_files(
    root: &Path,
    recorded: &FileManifest,
    configurations: &[String],
) -> FileManifest {
    recorded
        .keys()
        .chain(configurations)
        .filter_map(|file| {
            let text = std::fs::read_to_string(root.join(file)).ok()?;
            Some((file.clone(), FileFingerprint::read(file, &text)))
        })
        .collect()
}

#[cfg(test)]
mod tests {
    use super::*;
    use serde_json::json;

    #[test]
    fn scheduling_settings_do_not_change_what_a_test_reaches() {
        let then = json!({
            "webServer": [{"command": "npm start", "url": "http://127.0.0.1:3000"}],
            "projects": [{"name": "chromium", "use": {"baseURL": "http://127.0.0.1:3000"}, "timeout": 30000, "retries": 0, "testMatch": "*.spec.ts"}],
            "workers": 1,
        });
        let mut later = then.clone();
        later["workers"] = json!(4);
        later["projects"][0]["timeout"] = json!(45000);
        later["projects"][0]["testMatch"] = json!(["*.spec.ts", "new/*.spec.ts"]);
        later["projects"]
            .as_array_mut()
            .unwrap()
            .push(json!({"name": "firefox", "use": {"browserName": "firefox"}}));
        assert_eq!(playwright_differences(&then, &later), Vec::<String>::new());

        let mut shaped = then.clone();
        shaped["projects"][0]["use"]["extraHTTPHeaders"] = json!({"x-suite": "e2e"});
        shaped["webServer"][0]["command"] = json!("npm run start:mock");
        assert_eq!(
            playwright_differences(&then, &shaped),
            ["webServer", "project chromium: use"]
        );
        // A configuration without projects sets `use` for all of them.
        let single = json!({"use": {"baseURL": "http://127.0.0.1:3000"}, "timeout": 30000});
        let mut longer = single.clone();
        longer["timeout"] = json!(45000);
        assert_eq!(
            playwright_differences(&single, &longer),
            Vec::<String>::new()
        );
        longer["use"]["locale"] = json!("de-DE");
        assert_eq!(playwright_differences(&single, &longer), ["use"]);
    }
}